Detection and Characterization of Wolbachia Infections in Natural Populations of Aphids: Is the Hidden Diversity Fully Unraveled?

Aphids are a serious threat to agriculture, despite being a rather small group of insects. The about 4,000 species worldwide engage in highly interesting and complex relationships with their microbial fauna. One of the key symbionts in arthropods is Wolbachia, an α-Proteobacterium implicated in many important biological processes and believed to be a potential tool for biological control. Aphids were thought not to harbour Wolbachia; however, current data suggest that its presence in aphids has been missed, probably due to the low titre of the infection and/or to the high divergence of the Wolbachia strains of aphids. The goal of the present study is to map the Wolbachia infection status of natural aphids populations, along with the characterization of the detected Wolbachia strains. Out of 425 samples from Spain, Portugal, Greece, Israel and Iran, 37 were found to be infected. Our results, based mainly on 16S rRNA gene sequencing, indicate the presence of two new Wolbachia supergroups prevailing in aphids, along with some strains belonging either to supergroup B or to supergroup A.


Introduction
Wolbachia is a diverse group of obligatory intracellular and maternally transmitted a-Proteobacteria [1][2][3]. Several studies suggest that these bacteria are present in at least 65% of arthropod species as well as in filarial nematodes and in some plant parasitic nematodes [4][5][6][7][8]. Wolbachia strains infecting arthropod and nematode hosts are represented by a single species, Wolbachia pipientis [9]; however, there is extensive diversity which has resulted in the assignment of the bacterial strains into at least eleven Wolbachia supergroups, named A to F and H to L (supergroup G is considered a recombinant between A and B) [4,[10][11][12][13][14][15][16][17][18][19].Wolbachia diversity was initially characterized using the genes wsp, 16S rRNA, ftsZ, gltA and groEL as molecular markers, while strain genotyping is based on multi locus sequence typing systems (MLST), as well as on the amino acid sequences of the four hypervariable regions (HVRs) of the WSP protein [20,21].
Wolbachia have been reported in the somatic tissues of arthropod hosts; however, they mainly reside in the reproductive tissues and organs [2]. This tissue localization pattern has been associated with the induction of different reproductive alterations such as feminization, parthenogenesis, male killing and cytoplasmic incompatibility [2,22], which aid the spread of Wolbachia infections in host populations [23]. The widespread distribution of Wolbachia and their ability to manipulate the reproductive properties of arthropod hosts has attracted interest in its role in host biology, ecology and evolution, as well as in the development of novel, symbiont-based and environment friendly Wolbachia-based methods for pest and disease management [2,3,[24][25][26]. It has been suggested that Wolbachia-induced cytoplasmic incompatibility can be used either for the control of agricultural pests and disease vectors through the Incompatible Insect Technique (IIT), or by spreading a desirable genotype through populations, such as the inability of a vector species to transmit a pathogen [27][28][29][30][31][32][33]. The introduction of life-shortening Wolbachia strains could modify the population age structure of insect vector species, thus reducing pathogen transmission [34,35]. Furthermore, recent studies provide evidence that the presence of Wolbachia in some insect species may provide antiviral protection as well as inhibit the infection with and transmission of certain pathogens such as Dengue, Chikungunya and Plasmodium [35][36][37][38][39][40].
Aphids are a rather small group of insects but their threat to agricultural ecosystems is enormous. Currently, there are about 4,000 recognized species worldwide [41]. Aphids do great damage to their host plants in several ways [42]. They feed on plant sap and inject saliva (which can be phytotoxic) during feeding. Their honeydew is used by saprophytic ascomycetes that grow on plants.
More importantly, aphids have been shown to be vectors of numerous plant viruses. Due to their feeding behavior, they are by far the most important virus vectors, transmitting ,30% of all plant virus species [43].
Aphids exhibit many interesting biological traits. They have a complicated life cycle, being able to reproduce both sexually and asexually. They are specialized in probing and using phloem sap as sole food source, which leads to a tight association with their host plants. They are also important for the feeding of other insects; they modify phloem sap, which has a high ratio of non-essential to essential amino acids and elevated sugar content, and produce substances more suitable for other species [44].
Aphids have established sophisticated symbiotic relationships and many of their unique properties can be attributed to their symbiotic bacteria [45]. They have established an obligate mutualistic symbiosis with Buchnera aphidicola, whichprovides them with essential amino acids lacking from their phloem diet [46][47][48][49]. Occasionally, aphids harbour secondary or facultative symbionts that coexist with Buchnera, and can have positive effects on the aphid host [45]. It has been reported that 'CandidatusHamiltonella defensa' and 'CandidatusRegiella insecticola'can protect aphids against parasitoids [50,51], whereas CandidatusSerratia symbioticais implicated in heat tolerance [52]. Finally, studies showing lateral gene transfer from secondary symbionts to their aphid host and the fact that these genes are expressed in some cases [53,54], along with a reported case of metabolic complementation between B. aphidicola and ''Ca S. symbiotica'' in the aphid Cinara cedri [55,56] illustrate the very complex relationship between aphids and their symbionts. All the above suggest that aphids, together with their host plants and their microbial fauna, not only constitute an interesting biological model worth investigating, but that it is furthermore crucial to study and understand these relationships in order to devise appropriate control methods for these species and the plant diseases they transmit.
A small number of studies has investigated the presence of Wolbachia in aphids [6,[57][58][59][60]. Most of them failed to detect Wolbachia [57][58][59]. The first report of aphids (Toxoptera citricidusand Aphis craccivora) harboring Wolbachia was based on Long-PCR and the sequencing of the wsp gene [6]. Stronger evidence for the presence of Wolbachia in aphid species was based on 16S rDNA sequencing, electron microscopy and in situ localization of this endosymbiont in C. cedri [60]. It was recently reported that Chinese natural populations of the wheat aphid, Sitobion miscanthi, harbour single and/or double Wolbachia infections belonging to the A and B supergroup [61].
We undertook extensive screening and report here on the presence of Wolbachia infections in natural populations of aphid species. The characterization of these Wolbachia strains is based on the use of gene markers 16S rRNA, ftsZ, gltA, groEL, wsp and MLST. Our study suggests that neither the detection nor the unraveling of Wolbachia diversity in the aphid fauna is an easy task; they demand the development of novel tools.

Screening for Wolbachia infections in natural populations of aphids
A total of 425 natural samples of aphids were screened for the presence of Wolbachia with a 16S rRNA-based PCR approach using the wspecF/wspecR set of primers ( Figure S1). The samples were collected in five countries (Greece, Spain, Portugal, Israel and Iran) and on a variety of host plants (at least 165 different species). Collections were in some cases diachronic. This collection represents 144 different aphid species within 69 genera ofnine subfamilies of the family Aphididae (Aphidinae, Chaitophorinae, Pterocommatinae, Myzocallidinae, Drepanosiphinae, Thelaxinae, Lachninae, Mindarinae and Eriosomatinae) ( Table S1). The majority of samples screened belong to the subfamily Aphidinae (tribes Aphidini and Macrosiphini)followed by the subfamily Lachninae (mainly from the Eulachnini tribe), considered by recent studies as the most basal lineage among the aphid subfamilies [62],. and from all three tribes of Eriosomatinae subfamily (Pemphigini, Eriosomatini and Fordini).
The results of the screen, which are presented in Table 1, show that the prevalence of Wolbachia infections varied significantly between different aphid populations and can be summarized as follows: (a) Wolbachia infection was detected in only 37 out of 425 aphid populations tested; (b) Wolbachia was detected in aphid species of the subfamilies Lachninae, Aphidinae, Chaitophorinae Eriosomatinae, and Drepanosiphinae, while no infection was found in the rest of subfamilies; (c) at least eight species of the Lachninae subfamily were found infected including seven Cinara species (C. fresai, C. maritimae, C. juniperi, C. pinea,C. tujafilina, C. cedri and Cinara sp. from the Eulachnini tribe, Tuberolachnussalignus and Maculolachnus submacula from the Lachninitribe; (d) at least eleven species of the Aphidinae subfamily were found to be infected; nine of them belong to the Aphidini tribe, including three Aphis species (A. fabae, A. nerii and A. hederae ), three samples assigned as Aphis sp., and two Toxoptera species (T. auranti and T. citricidus). The remaining belong to the Macrosiphini tribe, two samples assigned as Cavariella sp., Macrosiphum euphorbiae, Metopolophium dirhodum and Aulacorthum solani; (e) a single infected species belongs to the Chaitophorinae subfamily (Sipha maydis); (f) a single sample of the Eriosomatinae subfamily, Baizongia pistaciae (tribe Fordini), was found to harbour Wolbachia; and (g) a single sample of the Drepanosiphinae subfamily, Neophyllaphis podocarpi, was found to harbour Wolbachia.
It should be noted that at least four different individuals of Cinara pinea (Madeira), Metopolophium dirhodum, Aphis fabae, Aphis hederae, Toxoptera citricidus (Madeira), Sipha maydis and Baizongia pistaciae populations were tested. All individuals were Wolbachia positive. For the rest of the populations, the screening was performed on a pool of four individuals.
Taken together, these results suggest that Wolbachia may be more abundant in aphids than previously thought, and that new universal primers coupled with new sequencing technologies will enable a better detection and investigation of the Wolbachia diversity.

Genotyping aphid Wolbachia strains
The current genotyping of Wolbachia strains is based on MLST approaches [20,21]. Efforts were made to amplify the MLST genes for the Wolbachia-infected aphid samples; however, the majority of PCRs failed. Only for a few of the samples, some of the genes were successfully amplified (Table 1). Due to these difficulties, attempts were undertaken to characterize the bacterial strains present in each of the thirty-seven Wolbachia-infected aphid Cinara sp. Cupressus sp.
Toxoptera aurantii Agapanthus sp. populations using a near-full length sequence of the 16S rRNA gene. Additional markers were also used, such as groEL,gltA, wsp and/or other individual MLST gene markers(gatB, coxA, ftsZ, hcpA and fbpA), which could be amplified from the Wolbachia-infected aphid samples. The results of these efforts can be summarized as follows: (a) a near-full length sequence of the 16S rRNA gene was amplified and analyzed for 35 out of the 37 Wolbachia-infected aphid samples, using PCR-sequencing approaches and the primers as presented in Table 1 and Figure S1. For two samples, GRA4 and GRA40, the amplification of 16S rRNA gene was not possible, and the characterization was based on other genes (see Table 1); (b) genes gatB, coxA, ftsZ, hcpA, fbpA, wsp, gltA and groEL were amplified only from ten, eight, two, two, four, seven, three and five Wolbachiainfected aphid samples, respectively (Table 1); (c) the sequence analysis of gatB, coxA, ftsZ, hcpA, fbpA and wsp revealed the presence of eight, three, two, two, four and two alleles respectively (Tables 2  and 3); (d) the sequence analysis also indicated the presence of novel alleles: seven for gatB, one for ftsZ, one for hcpA and three for fbpA (Tables 2 and 3); (e) gltA and groEL gene fragments were amplified only in three and five Wolbachia-infected aphid samples, respectively (Table 1).
These results indicate that there are differences in the Wolbachia infection status among different aphid species and populations and, more importantly, that the currently available genotyping tools of Wolbachia [10,20,21,63,64]cannot be universally applied for Wolbachia of aphids.

Phylogenetic analysis
Failure to amplify the majority of the MLST and/or other protein coding genes meant that the phylogenetic analysis had to be based mainly on partial 16S rRNA gene sequences (at least 1100 bp). Our Wolbachia sequences appear to cluster in four different groups: few belong either to supergroup A or to supergroup B, while the majority of the sequences form two new clades M and N, distinct from the so far reported supergroups, as depicted in Figure 1. Genetic distances of all the samples of the new groups are more than 2% from the so far reported A to L supergroups (Table S2). The 2% distance is a value necessary for the establishment of a new supergroup [65,66]. Supergroup M includes 30 new Wolbachia sequences and shows the smallest genetic distance to supergroup B (0.021) and the largest distance to supergroup J (0.059) (Table S3). Supergroup N includes 3 new Wolbachia sequences and shows the smallest genetic distance to supergroup K (0.022) and the largest to supergroup I (0.054) ( Table S3). The -within the group -genetic distances of these new groups are only 0.013 and 0.002 for supergroup M and N, respectively, supporting the classification of the strains in new clades. Given the tree topology, presence of recombination events was also examined between M, N and B supergroups, along with Wolbachia strains fromaphids placed into supergroups A and B. No indication of recombination events were detected using the RDP3 package.
A thorough phylogenetic analysis based on protein coding genes could not be completed, due to PCR amplification failure in most of the cases (Table 1); however the phylogenetic analysis that was based on the available aphid WolbachiagltA, gatB, fbpA and groEL gene sequences provided several important findings. gltA-based data indicate that the three amplified Wolbachia sequences belong to Supergroup B, while the corresponding 16S rRNA sequences group with the new supergroup M, which has, as stated above, the smallest genetic distance to supergroup B. Interestingly, all three sequences were amplified from C. cedri populations ( Figure S2). gatB sequence analysis indicates that almost all amplified sequences group together in a new phylogenetic lineage close to that of supergroup B, except one that groups with supergroup A ( Figure  S3). A similar picture can be seen with the fbpA-based data with three Wolbachia sequences forming a new phylogenetic cluster and one grouping with supergroup A sequences ( Figure S4). For the groEL-based data, one sequence makes a new phylogenetic cluster with supergroup L while the other four group with supergroup B sequences ( Figure S5).
Taken together, these results suggest that the aphid fauna may contain an unprecedented range of highly diverse Wolbachia

Extending our knowledge on Wolbachia infection of aphids
The presence of Wolbachia was investigated for 425 samples belonging to 153 different species and 70 genera, using a Wolbachia specific 16S rRNA-based PCR assay. The screen included aphid subfamilies with no previous reports of Wolbachia infection and included aphid species from different geographic locations and a variety of hosts (at least 165 different species) [6,[57][58][59][60]. Despite difficulties with PCR amplification (see below), Wolbachia infections were detected, adding important information to previous studies on aphids which had detected Wolbachia in only four species: three of these species belong to the Aphidinae subfamily and one to the Lachninae. The present analysis showed that the prevalence of Wolbachia infections varied significantly between different aphid populations (Table S1). Wolbachia were detectedin eighteen new aphid species, belonging to the subfamilies Chaitophorinae, Eriosomatinae and Drepanosiphidae, while they were not found in 146 species tested belonging to the seven aphid subfamilies: Aphidinae, Chaitophorinae, Pterocommatinae, Drepanosiphinae, Lachninae, Mindarinae and Eriosomatinae.
A direct comparison with previous screening efforts is difficult since: (a) aphid hosts, sample origin and even screening approaches differed and (b) not many aphid species were common in these studies. Our study confirmed previous results regarding the absence of Wolbachia in members of the subfamily Aphidinae: (i) Acyrthosiphon pisum [58,59]; (ii) different species of Uroleucon genus [59]; (iii) A. craccivora, Myzuspersicae, Rhopalosiphumpadi, Rhopalosiphummaidis and Schizaphis graminum [59] and (iv) Aphis jacobaeae, Capitophorus carduinis and Sitobium fragariae [57]. It should be noted that Wolbachia was not detected in any species tested of the generaUroleucon, Capitophorus, Myzus and Sitobion although Wolbachia infection was reported in a previous study [61].
Our study also confirmed previous results regarding the absence of Wolbachia in A. craccivora [6]and the presence [60] in all but one C. cedri samples tested (originating from different geographic locations: Spain, Portugal, Iran and Israel). Wolbachia were also detected in five more Cinara species (C. pinea, C. fresai, C. juniperi, C. tujafilina and C. maritimae), suggesting that the genus Cinara has a well-established symbiotic association with Wolbachia. However, it is difficult to speculate about a possible role of Wolbachia in this genus because in 20 out of the 37 samples screened, Wolbachia was not detected. In any case, most members of the Lachninae subfamily harbor S. symbiotica as a second symbiont [67,68] and, thus the possibility that these species are more prone to accept other infections cannot be ruled out. Finally, the possibility of a coevolution with the host can be discarded. First, samples from the same species and the same or different location are found in different supergroups (i.e. C. cedri from Israel and Valencia, Spain are found in M and B supergroups; samples from C. pinea are found in M and A supergroup). Second, due to the fact that several of the Cinara species were previously studied in a work analyzing the presence of Serratia in the subfamily Lachninae (94), we can compare the phylogenetic tree obtained in the present work, with those of Buchnera and Serratia previously obtained. The topology obtained regarding the samples from Cinara sp is non-congruent either with Buchnera or with Serratia. A very interesting result is the identification of multiple infections in C. cedri samples. PCRsequencing analysis of 16S rRNA clones from Israeli populations of C. cedri indicates the presence of two Wolbachia strains: one from supergroup B and a second from the new supergroup M (see below; Figure 1). The fact that DNA was extracted from a mix of four aphids leaves the possibility open that these two strains derive from different individuals.
There are two limitations in our study, regarding the detection of superinfections: the first is the low body mass of many aphid species, which did not allow isolation of high quality and quantity singleaphid DNA for multiple PCRs. The second is the small number of individuals analyzed per population, since we focused on the screening of as many populations as possible, which, in association with the differential abundance of strains and the non-optimized PCR protocols can lead to under-estimation of multiple infections. It should be noted that Wolbachia superinfections have repeatedly been reported in different insect taxa, including Chinese populations of the wheat aphidSitobion miscanthi [61,[69][70][71][72][73][74][75].

Extending our knowledge on Wolbachia diversity -Two new supergroups
The 16S rRNA gene sequence analysis strongly supports the existence of two new Wolbachia supergroups in aphids and raises questions about the robustness of supergroups E, F and H (Figure 1). Thirty-three of the aphid Wolbachia-specific 16S rRNA gene sequences cluster in two new clades, which are at least 2% genetically distant from all previously described supergroups and from each other (Table S3). However, the analysis also shows that the distance of supergroup A 16S rRNA gene sequences is less than 2% from the sequences present in supergroups E, F and H, suggesting that the overall classification of Wolbachia strains in supergroups (A to N) should be re-evaluated (see Table S3, figures in bold). Our analysis indicates that the within-supergroup diversity of M and N is 1.3% and 0.2%, respectively (Table S3). Similar phylogenetic analysis with the rest of the genes that are currently being used for the designation of supergroups, could not be completed due to failure of most PCR amplifications. However, the analysis performed with the limited available protein encoding gene sequences (groEL and some MLST genes) also support the presence of new supergroups ( Figures S3, S4 and S5).
Earlier efforts to characterize Wolbachia infections were based on the 16S rDNA and ftsZ genes, and later groEL and gltA were included [10,64,76]. In 2006, MLST-based systems were proposed for systematic genotyping and strain classification of Wolbachia infections [20,21]. However, the bacterial strains present in 37 Wolbachia-infected aphid populations, representing 25 aphid species, could not be genotyped using MLST analysis due to failure of PCR amplification despite great effort (Table 1). We managed to obtain sequences from ten samples for gatB, eight for coxA, two for ftsZ, two for hcpA, four forfbpA, seven for wsp, three for gltA and five for groEL (Table 1). Although the sequence analysis in the MLST and wsp databases indicated the presence of new alleles ( Tables 2 and 3), it also clearly shows that the currently available tools cannot be applied universally for the genotyping of the highly diverse aphid Wolbachia strains, and a new MLST system may need to be developed.

The challenge of detection and strain classification of Wolbachia infections in aphids
A major crossroad will be the choice of genes for a new MLST system, given that in the present study there were two cases [see Table 1: Aphis fabae (GRA4) and Metopolophium dirhodum (GRA40)] where Wolbachia-specific amplicons were obtained and confirmed by sequencing analysis, also for some MLST genes, but not for the 16S rRNA gene, which is considered one of the most conserved genes. Our data are in agreement with recent efforts on the assessment of PCR protocols for the detection of Wolbachia,which suggested that the current tools are far from optimal [77].
The development of robust and efficient Wolbachia detection and classification protocols is certainly hindered by the presence of low titre infections and multiple infections [78][79][80]. It has been reported that Wolbachia density may be affected and/ or regulated by co-infection with other Wolbachia strains or other vertically transmitted symbionts, as well as by host genotype [81][82][83].
Another important factor is horizontal transfer of Wolbachia genes to host genomes, which further complicates both Wolbachia detection and strain classification. Horizontal transfer events of Wolbachia genome fragments have been reported for several invertebrate species [84][85][86][87][88].It is evident that such phenomena can complicate phylogenetic analysis, since nuclear gene copies would evolve in a different way than cytoplasmic copies of Wolbachia genes. Also, Wolbachia detection is compromised in populations that carry nuclear copies of Wolbachia genes but lost the cytoplasmic Wolbachia [87]. The draft genome sequence of the pea aphid Acyrthosiphon pisum revealed the existence of 12 genes of bacterial origin [89], nine of which were intact and closely related to genes of a-Proteobacteria. There is, however, no evidence for horizontal transfer of Wolbachia genes in aphids, and Wolbachia was neither detected in the pea aphid in the present nor in previous studies [58,59]. The PCR detection of some MLST genes, but not of the 16S rRNA gene in two aphid samples of the present study (Aphis fabae (GRA4) and Metopolophium dirhodum (GRA40)) could be explained by the integration of genomic sequences of a former Wolbachia symbiont into the host genome although alternative causes can not be excluded.

Possible role of Wolbachia in aphids
Aphids feed on phloem sap, which has an unbalanced nitrogen/carbon content and is deficient in a number of nutrients, mainly amino acids, which insects cannot synthesize and are provided by Buchnera aphidicola, their primary endosymbiont. The relationship is mutualistic, since aphids need B. aphidicola for normal growth and reproduction, whereas the bacteriumcannot live outside the aphid [46,47,90]. In addition to B. aphidicola, some aphid populations harbor other heritable bacterial symbionts that are not required for host growth and reproduction, referred to as facultative or secondary symbionts [50,91]. The most common facultative symbionts found in aphids are 'Ca.Regiella insecticola', 'Ca.Hamiltonella defensa' and 'Ca.Seratia symbiotica' [45,91]. Several studies, mainly in A. pisum, a member of the Aphidinae subfamily, have shown that these symbionts can provide some benefits to the host; however, as mentioned above, no Wolbachia has so far been detected in A. pisum. The genome sequence of these endosymbionts shows that they have lost the ability to synthesize some amino acids and are thus dependent on Buchnera [92][93][94].
C. cedri, a member of the subfamily Lachninae that possess the B. aphidicola with the smallest genome reported so far, and has established a permanent association with the co-primary endosymbiont Serratia symbotica, deserves special attention. Both bacteria are needed for the survival of the whole consortium. When Wolbachia was found in C. cedri, it was postulated that its presence could increase the prevalence of asexual lineages, (C. cedri has a cyclic parthenogenetic life cycle) (see below). In the present study, Wolbachiahas been found in all analyzed C. cedri populations, corroborating their tight association with this species.
Facultative endosymbionts are a common feature of the Lachninae subfamily, to which C. cedri belongs [67,68]. These symbionts are somehow compensating the drastic metabolic losses that have occurred in B. aphidicola as it has been recently shown for C. tujafilina [95]. The presented data indicate that the members of the Lachninae subfamily tend to be infected with Wolbachia. The possibility that Wolbachia may have a nutritional function in these cases cannot be discarded, as it has been recently proven in thebedbug, Cimex lectularius [96].
Wolbachia is well known for its ability to induce reproductive alterations, such as parthenogenesis, feminization, male-killing and, most commonly, cytoplasmic incompatibility, in its hosts [2,3]. Aphids are known to have complicated life cycles, which include sexually and asexually reproducing species, as well as species with both sexual and asexual phases [97]. Whether Wolbachia is somehow involved in these phenomena remains to be investigated. Specifically, it would be interesting to check the life cycle of Wolbachia-infected versus non-infected aphids, as its presence could increase the prevalence of asexual lineages, as previously reported for the Hymenopteran group [22].

Conclusion
We report the largest screening effort so far for Wolbachia in aphids. Our results indicate the presence of two new supergroups prevailing in aphids, previously well hidden, probably due to low titer, genetic variability and lack of optimized identification and classification tools. Although Wolbachia was unambiguously identified only in a fraction of the samples analyzed, we believe that its presence is underestimated, and the development of more universal Wolbachia-screening tools is needed. Clarifying the Wolbachia status of aphids can help in the development of novel and environment-friendly methods for the efficient control of aphids, major pests and disease vectors.

Sample collection and DNA extraction
Aphid taxa examined in this study, information about their taxonomy, collection locations and the host plants they have been isolated from are listed in Table S1. Natural aphid populations were sampled in different years in Greece (2006Greece ( , 2007Greece ( , 2009 [98][99][100][101] and were stored in 100% ethanol at 220uC. Total DNA of the Greek aphid populations was extracted from single aphids (at least three individuals per sample) while for the Spanish, Portuguese, Israeli and Iranian samples, extractions were done from a pool of four adults. DNA extraction was performed as described previously [102]or by using a modified CTAB protocol [103].

PCR screen
A total of 425 specimens from five subfamilies of the 148 different aphid species were screened for the presence of Wolbachia strains. Detection was based on the amplification of a 16S rRNA gene fragment (438 base pairs) with the Wolbachia specific primers wspecF and wspecR ( Figure S1) [5]. For those samples that appeared negative for Wolbachia infection, the quality of DNA was further examined by amplifying part of the mitochondrial 12S rRNA gene (420 base pairs) using primers 12SCFR 59-GAGAGTGACGGGCGATATGT-39 and 12SCRR 59-AAAC-CAGGATTAGATACCCTATTAT-39 [104]. PCR amplifications were performed in 20 ml reactions containing 1 ml of DNA, 4 ml 56 reaction buffer (Promega), 1.6 ml MgCl 2 (25 mM), 0.1 ml deoxynucleotide triphosphate mixture (25 mM each), 0.5 ml of each primer (25 mM), 0.1 ml of Taq polymerase (Promega, 1 U/ml) and 12.2 ml water. Amplification was performed in a PTC-200 Thermal Cycler (MJ Research), using the following cycling conditions: 95uC for 5 min, followed by 34 cycles of 30 s at 94uC, 30 s at 54uC, 1 min at 72uC and a final extension of 10 min at 72uC. PCR reactions were electrophoresed on a 1.5% agarose gel. Positive samples were further analysed.
PCR, cloning and sequencing of 16S rRNA, groEl, gltA, wsp and MLST gene fragments Amplification of near full size 16S rRNA sequences proved to be a rather difficult task and required the deployment of a number of approaches (see Figure S1). These involved the use of (a) a new Wolbachia specific primer, W169F, designed for the purposes of this study and the universal eubacterial primer 1513R, followed by a nested PCR using the same forward primer (W169F) and wspecR and (b) the newly designed primer W169F and the new Wolbachia specific primer 16S_woR1 as reverse primer ( Figure S1). For some of the populations, a direct PCR with 16S_169F/16S_woR1 was used. PCR amplifications were performed in 20 ml reactions containing 1 ml of DNA, 4 ml 56 reaction buffer (Promega), 1.6 ml MgCl 2 (25 mM), 0.1 ml deoxynucleotide triphosphate mixture (25 mM each), 0.5 ml of each primer (25 mM), 0.1 ml of Taq polymerase (Promega 1 U/ml) and 12.2 ml water. Amplification was performed in a PTC-200 Thermal Cycler (MJ Research), using the following cycling conditions: 95uC for 5 min, followed by 34 cycles of 30 s at 94uC, 30 s at 51uC for W169F/1513R and 53uC for W169F/16SwolR1, 1 min at 72uC and a final extension of 10 min at 72uC. The annealing temperature for the nested PCR was 53uC.

Cloning and sequencing
To determine the sequence of 16S rRNA, wsp, groEL, gtlA and MLST gene fragments, PCR fragments were cloned in cases of poor sequencing quality or multiple chromatographic peaks in direct sequencing of PCR products. PCR products from 18 out of the 37 populations harboring Wolbachia were ligated into a Tvector (pGEM-T Easy) and then transformed into DH5a competent cells according to the manufacturer's instructions. Four to six clones were directly subjected to PCR using the primers T7 and SP6. The colony PCR products were purified using the PEG-NaCl method [105] or using NucleoFastH 96 PCR Plates (Macherey-Nagel) according to the manufacturer's instructions. Inserts were fully sequenced with the same primers and with the internal 16S rRNA primer 960R [106]. A dye terminator-labelled cycle sequencing reaction was conducted with the BigDye Terminator v3.1 Cycle Sequencing Kit (PE Applied Biosystems). Reaction products were analysed using an ABI PRISM 310 or an ABI 3730 Genetic Analyzer (PE Applied Biosystems).All Wolbachia gene sequences generated in this study were assembled and manually edited with SeqManII by DNAStar. For each sample, a majority-rule consensus sequence was created.

Nucleotide sequence accession numbers
All 16S rRNA, wsp, groEL, gtlA and MLST gene sequences generated in this study have been deposited in the GenBank database under accession numbers JN384025-JN384106.

Phylogenetic analysis
All Wolbachia 16S rRNA, gatB, fbpA, hcpA, ftsZ, coxA, groEL and gltA gene sequences generated in this study were aligned using MUSCLE [107] and ClustalW [108]. Sequences obtained from GenBank representing all currently known supergroups of Wolbachia were included in the analysis (Table S2). Phylogenetic analyses were performed using maximum-likelihood (ML) and Bayesian methods. PAUP version 4.0b10 was used to select the optimal evolution model by critically evaluating the selected parameters using the Akaike Information Criterion [109]. For the 16S rRNA and gltA gene sequence data the submodel GTR+I+G was selected. For the groEL, gatB and fbpA sequence data, the submodel GTR+G was selected. ML analysis was performed in PAUP using a heuristic search with a random addition of sequences with ten replicates and TBR swapping. The robustness was assessed with 1,000 bootstrap replicates. Bayesian analyses were performed as implemented in MrBayes 3.1 [110]. Analyses were initiated from random starting trees. Four separate runs, each composed of four chains were run for 6,000,000 generations. The cold chain was sampled every 100 generations, and the first 20,000 generations were discarded. Posterior probabilities were computed for the remaining trees.
Recombination events were examined with the default options of the RDP3 software package (Heath et al. 2006).To test for recombination events, we used the RDP3 software package, with all available softwares implemented in it [111]. We used the default options for all analyses. Figure S1 Position of the primers used in this study, relative to the 16S rRNA gene from wMel.