WOLBACHIA INFECTION AND DRAMATIC INTRASPECIFIC MITOCHONDRIAL DNA DIVERGENCE IN A FIG WASP

Authors

  • Jin-Hua Xiao,

    1. Key Laboratory of Zoological Systematics and Evolution, Institute of Zoology, Chinese Academy of Sciences, Beijing 100101, China
    2. These authors contribute equally
    Search for more papers by this author
  • Ning-Xin Wang,

    1. These authors contribute equally
    2. College of Plant Protection, Shandong Agricultural University, Tai’an 271018, China
    Search for more papers by this author
  • Robert W. Murphy,

    1. State Key Laboratory of Genetic Resources and Evolution, Kunming Institute of Zoology, Chinese Academy of Sciences, Kunming 650223, China
    2. Department of Natural History, Royal Ontario Museum, 100 Queen's Park, Toronto, Ontario, M5S 2C6, Canada
    Search for more papers by this author
  • James Cook,

    1. School of Biological Sciences, University of Reading, Reading, Berkshire, RG6 6BX, United Kingdom
    Search for more papers by this author
  • Ling-Yi Jia,

    1. Key Laboratory of Zoological Systematics and Evolution, Institute of Zoology, Chinese Academy of Sciences, Beijing 100101, China
    2. Graduate School of the Chinese Academy of Sciences, Beijing 100039, China
    Search for more papers by this author
  • Da-Wei Huang

    1. Key Laboratory of Zoological Systematics and Evolution, Institute of Zoology, Chinese Academy of Sciences, Beijing 100101, China
    2. College of Plant Protection, Shandong Agricultural University, Tai’an 271018, China
    3. E-mail: huangdw@ioz.ac.cn
    Search for more papers by this author

Abstract

Mitochondria and Wolbachia are maternally inherited genomes that exhibit strong linkage disequilibrium in many organisms. We surveyed Wolbachia infections in 187 specimens of the fig wasp species, Ceratosolen solmsi, and found an infection prevalence of 89.3%. DNA sequencing of 20 individuals each from Wolbachia-infected and -uninfected subpopulations revealed extreme mtDNA divergence (up to 9.2% and 15.3% in CO1 and cytochrome b, respectively) between infected and uninfected wasps. Further, mtDNA diversity was significantly reduced within the infected group. Our sequencing of a large part of the mitochondrial genome from both Wolbachia-infected and -uninfected individuals revealed that high sequence divergence is common throughout the mitochondrial genome. These patterns suggest a partial selective sweep of mitochondria subsequent to the introduction of Wolbachia into C. solsmi, by hybrid introgression from a related species.

Mitochondria play a vital role in cell life by providing cellular energy. They harbor their own small, maternally inherited genome that is generally nonrecombining with other mitochondrial lineages. The genome comprises 13 protein-coding genes, two rRNAs, and 22 tRNAs, the products of which cooperate with nuclear encoded mitochondrial proteins involved in the process of oxidative phosphorylation (OXPHOS) (Gershoni et al. 2009). Another intracellular, maternally inherited genetic element, the bacteria Wolbachia, is found in many arthropods (Hilgenboecker et al. 2008). It can alter the reproduction system of its hosts in many ways, such as inducing cytoplasmic incompatibility (CI), creating parthenogenesis, causing feminization of genetic males, and killing embryonic males (O’Neill et al. 1997; Werren 1997; Stevens et al. 2001). These reproductive alterations can increase the frequency of infected female hosts thereby favoring the spread and persistence of its infection throughout many host generations.

mtDNA evolves more rapidly than nuclear DNA and the greatest variation in mtDNA often occurs in the control region. Intraspecific mtDNA divergence is often less than 2%, which makes the molecule useful as a genetic marker in population, biogeographic, and phylogenetic studies, as well as for DNA barcoding (Xiao et al. 2010). When used as DNA barcode for species identification in fig wasp, co1 can easily help identify most species, including cryptic species, although sometimes it is puzzling with the existence of NUMTs (mitochondrial genes integrated into the nuclear genomes) or Wolbachia influence. Regardless, a large number of taxa have high levels of intraspecific mtDNA divergence, including both invertebrates (Avise et al. 1994; Thomaz et al. 1996) and vertebrates (Walker et al. 1997; Waters and Burridge 1999; Rawlings and Donnellan 2003). For example, divergence of up to 12.9% occurs in the 16s gene of the snail Cepaea nemoralis, and 14.6% in cytochrome b (cob), 6.0% in 16s in the freshwater fish Galaxias maculates, and more than 12% in both cob and atp8–atp6 in the black-tailed brush lizard, Urosaurus nigricaudus (Lindell et al. 2008). Hymenoptera appear to have very high mitochondrial mutation rates compared with other insects (Oliveira et al. 2008; Raychoudhury et al. 2009). For example, the mitochondrial genes in Nasonia wasps are evolving about 40 times faster than the nuclear genes; when making a comparison between Nasonia longicornis and N. giraulti, the average nucleotide diversity for the whole mitochondrial genome is 10%, whereas the mitochondrial protein-coding genes is 15% (Oliveira et al. 2008; Raychoudhury et al. 2009).

Because both mtDNA and Wolbachia are maternally inherited, linkage disequilibrium is expected to occur between them. Many studies report the dynamic coupling of Wolbachia and mtDNA in a variety of insect species. In Drosophila simulans, infections can cause a selective sweep of mtDNA by reducing sequence diversity. Infections may also induce further mtDNA introgression after the introgression of Wolbachia (Turelli and Hoffmann 1995; Ballard 2000b,a). In the butterflies Acraea encedon and A. encedana, Wolbachia not only induces a selective sweep of the mtDNA, it is also responsible for the introgression of mtDNA from A. encedana to A. encedon (Jiggins 2003). In butterflies of the genus Eurema, two sibling species groups (Y-type and B-type) have mtDNA introgression yet the association between mtDNA and Wolbachia-infection varies (Narita et al. 2006; Narita et al. 2007). Within a Y-type species (Eurema hecabe), the association is very strong; infected and uninfected individuals form two mtDNA groups that have much higher intergroup as opposed to intragroup sequence divergence (Narita et al. 2006). Wolbachia has also moved between two Nasonia wasp species by hybridization, and the associated mitochondria are introgressed (Raychoudhury et al. 2009).

No study to date associates three primary elements involved in our study system: extremely high intraspecific mtDNA divergence, variable patterns of infection by Wolbachia, and sympatric infected and uninfected forms. Herein, we survey infections of Wolbachia in a fig wasp, Ceratosolen solmsi, and compare intraspecific mtDNA divergence between and within groups of infected and uninfected individuals. Fig wasps live inside the compact syconia (inflorescences) of figs (Ficus: Moraceae) and these insects have many biological adaptations to their peculiar living environment (Weiblen 2002). Importantly, hundreds of individuals shelter inside the small syconium, which has a diameter of 10–20 mm. All individuals live sympatrically, in fact syntopically. Species of fig wasp have a very high incidence of Wolbachia infection (Shoemaker et al. 2002; Haine and Cook 2005; Chen et al. 2010) and this might influence wasp speciation because cryptic pollinator species can have different infections of Wolbachia (Haine et al. 2006).

We investigated the effects of Wolbachia infections on C. solmsi using roughly 200 individual wasps by comparing intraspecific mtDNA divergence in the genes co1 and cob. We first confirmed that all the individuals were conspecific via morphological evaluation and nucleotide sequences of the nuclear genes opsin and its2. Subsequently, we attempted to amplify the entire mitochondrial genome in one infected and uninfected individual each. Our analyses obtained the following results: (1) Wolbachia infected 89.3% of the wasps and the sequences of co1 and cob in the infected and uninfected groups were extremely divergent; (2) comparison of the large mitochondrial elements also showed extreme divergence between infected and uninfected individuals; (3) the divergence within each group was much less than that between groups and polymorphism was greatly reduced in the infected group compared to the uninfected group; and (4) the highly diverged co1 and co2 genes remained functional. We then discuss the relationship between infections of Wolbachia and dramatic mtDNA divergences.

Material and Methods

SAMPLES

All the specimens of C. solmsi from F. hispida were collected in 2008 from five trees on the campus of the Chinese Academy of Tropical Agricultural Sciences (Danzhou, Hainan province, China). For each crop (tree) of fruits, wasps were sampled from three to five individual syconia (“figs”), identified morphologically and stored in 95% ethanol at –20°C. A total of 200 individuals were then randomly chosen for genetic analysis from the combined sample of wasps.

DNA EXTRACTION, PRIMER DESIGN, PCR, AND SEQUENCING

DNA was extracted from wasps by methods applicable for long-fragment PCR, which included use of entire tiny wasps (Hu et al. 2007). Some individuals were not destructed and saved as vouchers. We used HiFi Taq (TransGen, Beijing, China) for PCR using the manufacturer's instructions. DNA quality was assessed with amplification of a short co1 fragment using primers COI1751/COI2191 (Simon et al. 1994). Only 189 positive templates were used for further analysis.

Infections of Wolbachia were identified by amplification with wsp (Wolbachia surface protein gene) primers: 81F (TGGTCCAATAAGTGATGAAGAAAC) and 691R (AAAAATTAAACGCTACTCCA) (Zhou et al. 1998). Genotyping using RsaI RFLP for 16s rDNA of Wolbachia was screened from five infected individuals. We also implemented multilocus sequence typing (MLST) for the Wolbachia strain using five housekeeping genes (gatB, coxA, hcpA, ftsZ, and fbpA), with the primers and protocols listed in the PubMLST website (http://pubmlst.org/wolbachia/) (Baldo et al. 2006). For amplification of partial regions of co1 and cob, we used two conserved primer pairs: LCO1490 (GGTCAACAAATCATAAAGATATTGG) with HCO2198 (TAAACTTCAGGGTGACCAAAAAATCA), which obtained a 652 bp fragment of co1, and CB10933 (TATGTACTACCATGAGGACAAATATC) with CB11367 (ATTACACCTCCTAATTTA TTAGGAAT), which amplified a 389 bp fragment of cob. We attempted to amplify the entire mitochondrial sequence for Wolbachia-infected and -uninfected wasps from one individual each. Initial PCR used conserved primers spanning the region from co1 to co3, and from Nd1 to 12s rRNA and Nd5 (Simon et al. 2006). Specific primers were then designed based on the obtained sequences to amplify the region between co1 and Nd1.

Amplicons were purified and then sent to BioSune (Beijing) for sequencing. When multiple peaks appeared, the products were cloned in pEasy-T1 (for fragments shorter than 1.5 kb) or pEasy-T3 (for longer fragments) (TransGen), and from three to five positive clones were picked for sequencing.

IDENTIFICATION AND ANNOTATION OF MITOCHONDRIAL GENES

The protein-coding and rRNA genes were identified by Blast and aligned to the orthologous mitochondrial genes of Nasonia, Apis mellifera, and an unpublished expressed sequence tag (EST) dataset. Positional confirmation and annotation of the tRNAs were accomplished using the online program tRNAscan-SE 1.21 (Lowe and Eddy 1997).

GENETIC DIVERGENCE ANALYSES

Nucleotide divergence for co1 and cob was summarized using neighbor-joining (NJ) trees constructed in MEGA 4.12 (Tamura et al. 2007). The software packages DnaSP 4.0 (Rozas et al. 2003) and MEGA 4.12 were used to measure the DNA polymorphism and divergence within and between infected and uninfected groups.

SEQUENCES USED IN THE ANALYSIS

All novel sequences were submitted to GenBank (accession numbers JF816299–JF816394 and JF816395–JF816396). The alignment is available on request.

Results

WOLBACHIA INFECTION SCREENING

One hundred sixty-seven of the 187 individuals (89.3%) were infected and among these the sequences from wsp indicated that infected individuals had only one strain of Wolbachia. Genotyping using RsaI RFLP for 16s rDNA also detected only a single strain in each individual (data not shown). MLST with the five standard housekeeping genes (gatB 7, coxA 6, hcpA 7, ftsZ 3, and fbpA 8) identified the Wolbachia strain in C. solmsi as ST-19. This supergroup A Wolbachia strain infects a diversity of hosts such as ants, moths, and butterflies and is widely distributed in Australia, South Africa, and South Asia (http://pubmlst.org/wolbachia/).

DIVERGENCE IN CO1 AND COB

After assigning individuals to either the infected or uninfected group, 20 individuals from each group were chosen at random for sequencing of co1 and cob. Ultimately, we obtained 39 co1 and cob sequences (Table 1).

Table 1.  DNA diversity comparison of the co1 and cob fragments between Ceratosolen solmsi infected and uninfected by Wolbachia.
 Segregating sites (S)Haplotype number (N)Haplotype diversity (Hd)Nucleotide diversity (Pi)Number of sequencesNumber of individuals
  1. In, Wolbachia-infected individuals; Un, Uninfected individuals.

In-co1440.2600.000522221
Un-co1780.8460.002481711
In-cob120.1730.000452221
Un-cob340.7060.002271716

The NJ trees for the two mtDNA genes were very similar in overall topology (Fig. 1). Infected and uninfected lineages were clearly resolved. Nucleotide divergence between the two lineages in co1 was 9.2%, and up to 15.3% in cob. The corresponding differences between lineages at the amino acid level were 3.7% for co1 and 13% for cob. The pattern that infected individuals and uninfected ones are located in different lineages suggests that the maternal transmission rate of Wolbachia is very high. Otherwise, on the tree, there may be one lineage with uninfected individuals only, and the other lineage with mostly infected but some uninfected individuals, due to uninfected wasps that have inherited mtDNA from an infected mother.

Figure 1.

Neighbor-joining (NJ) trees based on partial sequences of co1 (A) and cob (B) from Wolbachia-infected and -uninfected fig wasps, Ceratosolen solmsi. Sequences from Apocrypta bakeri (Chalcidoidea; Sycoryctinae) are used as the outgroup for displaying divergence. In, infected; Un, uninfected; Cs, C. solmsi; COI, co1; Cytb, cob; Wol+, Wolbachia infected individuals; Wol–, uninfected individuals.

We evaluated the segregating sites (S), haplotype number (N), and haplotype diversity (Hd) based on the partial sequences of co1 and cob from multiple individuals (Table 1). The intralineage divergence for both genes did not exceed 1%; co1 sequence divergence in the infected lineage was 0.052% versus 0.248% in the uninfected lineage. Similarly, for cob gene, divergence in the infected lineage was 0.045% but 0.227% in the uninfected lineage (Table 1). Both genes had much lower intragroup divergence than that between groups and polymorphism was about five times lower in the infected group compared to the uninfected group, consistent with patterns previously reported for other insects (Ballard 2000a; Jiggins 2003; Shoemaker et al. 2004).

Comparatively, the lepidopteran E. hecabe exhibited the same pattern as C. solmsi. Infected and uninfected specimens formed distinct groups that had much higher intergroup versus intragroup sequence divergence (Narita et al. 2006). Regardless, differences occurred between C. solmsi and E. hecabe. The former species had dramatically higher levels of interlineage divergence. We made a comparison and found that divergence in nad5 from C. solmsi was 0.1292, yet only 0.0166 in E. hecabe. Similarly, divergence in co1 was 0.1074 for C. solmsi but only 0.0265 in E. hecabe.

MITOCHONDRIAL GENOME DIVERGENCE

We evaluated whether the pattern observed for co1 and cob was typical of the entire mitochondrial genome by attempting to amplify the genome from one infected and one uninfected individual. Many factors made this difficult, including the small quantity of DNA obtained from a single specimen and AT richness of the genome. Observation from species of Nasonia indicated that chalcidoids likely experienced a series of dramatic rearrangements in gene order including not only the tRNAs, but also several protein-coding genes (Oliveira et al. 2008). The uninfected specimen proved more difficult to sequence than the infected wasp. Ultimately, we obtained a 12,546-bp fragment from the infected individual consisting of 11 protein-coding genes, 13 tRNAs, and two rRNA genes; this covered only partial sequences for co3 and 12s rRNA (Fig. 2; Table 2). We failed to amplify and sequence nad2 and nad3. As for the uninfected individual, two fragments spanning about 10-kb sequences were generated. The longer fragment, 6938 bp, contained seven protein-coding genes, and the shorter fragment, 3006 bp, contained nad1 and both rRNAs.

Figure 2.

Gene order in the mitochondrial genome of Ceratosolen solmsi. Arrows pointing to the right indicate genes on the heavy strand, whereas arrows pointing to the left denote the light strand (Black arrows indicate the protein-coding genes and rRNAs, whereas gray ones the tRNAs). Double lines indicate an unknown junction between the two fragments in the uninfected individual. The question mark indicates an unannotated space of 87 bp.

Table 2.  Annotation of the amplified mitochondrial genome sequences for Wolbachia-infected and -uninfected Ceratosolen solmsi. CeraIn, Wolbachia-infected C. solmsi; CeraUn, Uninfected C. solmsi. Note that there is a 100 Ns linking between the two fragments of mitochondrial genome for CeraUn.
GeneCeraInCeraUn
co3729 (ATG)-1749(ATG)-1
atp61397(ATG)-729(TAA)1417(ATG)-749(TAA)
atp81549(ATT)-1391(TAA)1569(ATT)-1411(TAA)
asp(D)1648–15821668–1602
lys(K)1717–16491737–1669
Co22390(ATT)-1706(T-)2410(ATT)-1736(T-)
leu(L2)2454–23912474–2411
co14058(ATA)-2525(T-)4062(ATG)-2536(TAG)
glu(E)4073–41394089–4154
phe(F)4206–41404220–4155
nad55885(ATT)-4206(TAA)5899(ATT)-4220(TAA)
his(H)5982–59205992–5929
nad47306(ATG)-5983(T-)6939–5993(T-)
nad4l7581(ATT)-7300(TAA) 
thr(T)7597–7664 
pro(P)7738–7603 
nad67740(ATG)-8300(TAA) 
cob8300(ATG)-9421(TAA) 
arg(R)9433–9498 
Ser(S)9498–9563 
nad110482(ATT)-9562(TAA)7831(ATT)-7040
Leu(L1)10569–105047920–7853
16s11981–105819304–7930
gln(Q)12074–12006 
ala(A)12138–120719459–9392
12s12546–1213910044–9460

Compared with other insects, C. solmsi exhibited a series of gene rearrangements (Fig. 2). As with Nasonia, it had a large inversion in the region spanning from co1 to co3 that might have contained nad3, which we did not successfully amplify in C. solmsi. The positional change of the protein-coding genes was shared only by Nasonia and Ceratosolen, suggesting that it might have occurred relatively recently in the common ancestor of parasitic wasps belonging to the superfamily Chalcidoidea. Compared to Nasonia (family Pteromalidae), C. solmsi (family Agaonidae) was more plesiomorphic in its gene order in two respects (see Fig. 2). First, the relative orientation of tRNA–Lys occurred in a “hot spot” for rearrangements in the Hymenoptera (Dowton and Austin 1999). The orientation of tRNA–Lys was not changed in Ceratosolen but reversed in Nasonia (Oliveira et al. 2008). Second, in Nasonia, the position of nad2 shifted to the position occurring between the protein-coding genes cob and nad1. However, in C. solmsi, we did not amplify nad2 yet it was not located in this region; we amplified the region spanning from cob to nad1 in one PCR reaction. Thus, the pattern of gene rearrangement in the region extending from cob to nad1 in C. solmsi was more plesiomorphic than in Nasonia. Notwithstanding, an insertion of tRNA–Arg occurred between cob and tRNA–Ser in Ceratosolen, unlike other insects. Another impressive rearrangement in C. solmsi was the presence of tRNA–Gln and tRNA–Ala in the middle of the two rRNAs. Nasonia had only tRNA–Ala in this position, where most arthropods had tRNA–Val (Boore 1999; Oliveira et al. 2008).

A high level of sequence divergence was observed in C. solmsi between infected and uninfected individuals, not only in the partial regions of co1 and cob, but also throughout the genome (Table 3; Fig. 3). The uncorrected average pairwise genomic sequence divergence (Pi) between the two groups was 0.1222. Although the average divergence of the seven protein-coding genes was 0.1228 (Fig. 3), that of the eight tRNAs was only 0.0509. The tRNAs were therefore much more conserved than the protein-coding genes in C. solmsi. In C. solmsi, divergence in co3 was the highest of the seven protein-coding genes, and dramatically higher than that of all interspecific divergences in Nasonia and Drosophila. Divergence in atp8 was also very high as evidenced from Pi and Ka/Ks values (Fig. 3). Among the seven genes, atp6 was the most conserved in C. solmsi. The extent of mtDNA divergence within C. solmsi substantially exceeded interspecific divergences in Drosophila and that between rapidly evolving N. giraulti and N. longicornis, which had an average Pi of 0.0995. Although these two sister species of Nasonia speciated around 0.4 million years ago (Mya), their hybrids were still viable (Campbell et al. 1994; Breeuwer and Werren 1995; Oliveira et al. 2008). The levels of synonymous substitution (Ks) were similar between infected/uninfected haplotypes of our fig wasp and the species pairs of Nasonia, and they were much higher than the levels for the sibling species D. melanogaster and D. simulans, and the more distantly related pair of D. melanogaster and D. yakuba (Fig. 3). The mtDNA evolution rates in our study species therefore appear very high, as in other chalcidoids (Oliveira et al. 2008; Raychoudhury et al. 2009).

Table 3.  Estimates of Ka and Ks between pairs of sequences in protein-coding regions in Ceratosolen solmsi, Drosophila, and Nasonia. Note synonymous and nonsynonymous mutation rates between pairs of sequences in protein-coding regions.
 co3atp6atp8co2co1nad5nad4Genome1 
  1. 1The estimates are based on the genome: for C. solmsi infected and uninfected individual, only based on the seven protein-coding genes; for Nasonia and Drosophila, the values are cited from Oliveira et al. (2008).

  2. Ks, the number of synonymous substitutions per synonymous site; Ka, the number of nonsynonymous substitutions per nonsynonymous site.

KsCeraIn-CeraUn0.61540.3910.30840.51010.51360.58640.60040.455
 Dromel-Drosim0.18320.18210.14680.22510.2120.18990.16760.189
 Dromel-Droyak0.34540.28610.31630.30540.35410.39150.2641
 Drosim-Droyak0.30480.27720.14160.36080.3420.31540.3135 
 Nv/Ng0.44950.65950.52810.49510.57610.47990.44970.519
 Nv/Nl0.47960.46740.56910.42420.56370.42690.49660.476
 Ng-Nl0.43340.43720.23830.44950.3930.32580.36590.389
KaCeraIn-CeraUn0.13020.03690.08910.04710.03010.06160.04390.0683
 Dromel-Drosim0.01080.02270.0160.00960.00260.01720.01640.014
 Dromel-Droyak0.01410.02770.01620.01160.0060.04210.0233 
 Drosim-Droyak0.01670.02080.03270.01160.00690.03960.0219
 Nv-Ng0.10150.08460.28160.06350.01560.09990.08630.100
 Nv-Nl0.08840.08410.33670.07270.02160.08850.07790.108
 Ng-Nl0.04040.03580.13930.03610.00840.05230.04570.051
Figure 3.

Pairwise comparison of sequence divergence for seven mitochondrial genes. Values of Ks and Ka are corrected by the JC method. Both uncorrected Pi and Ka values for all the genes show that the divergence between Wolbachia-infected and -uninfected Ceratosolen solmsi is similar to the interspecific divergence in Nasonia and higher than that in Drosophila. The fig wasp data lines are highlighted with heavy blue lines. CeraIn, Wolbachia-infected C. solmsi; CeraUn, uninfected C. solmsi; Dromel, Drosophila melanogaster; Drosim, D. simulans; Droyak, D. yakuba; Nv, Nasonia vitripennis; Nl, N. longicornis; Ng, N. giraulti.

ESTIMATE SUBSTITUTION RATES AND DIVERGENCE TIMES

We estimated the level of synonymous substitutions for seven protein-coding mitochondrial genes between Wolbachia-infected and -uninfected individuals. The Ks value was 45.5%, similar to the 45.24% for mitochondrial DNA of N. giraulti and N. longicornis. Assuming that mtDNA synonymous mutations in chalcidoids behaves in a clock-like manner, and that they had the same or similar mutation rates, the estimated divergence time between the mitochondrial genomes of infected and uninfected populations is virtually identical to that for N. giraulti and N. longicornis, that is, 0.40–0.51 Mya (Raychoudhury et al. 2009).

CONFIRMATION OF ONE SPECIES

The great extent of mtDNA sequence divergence between infected and uninfected C. solmsi suggested the possibility of two species, one of which may be immune to infection. Morphologically, the two groups were indistinguishable. Further, we confirmed conspecificity using two nuclear gene fragments: its2 and opsin. Opsin gene sequences were obtained from each of two infected and uninfected individuals. Seven infected and five uninfected individuals were sequenced for its2. Sequences alignments indicated that both fragments were virtually identical and did not fall into distinct clades. Sporadic differences may have been caused by PCR errors or individual variability. The same result was obtained even for opsin's introns, which vary between closely related species of fig wasps (Tables S2 and S3). Neither the morphological evaluation nor the nuclear genes suggested that the two groups were separate species. Compared to the nuclear gene sequences, divergence in the mitochondrial sequences between the Wolbachia-infected and -uninfected groups was very high.

NO DYSFUNCTION IN THE PROTEIN-CODING GENES

We questioned whether the dramatic genetic divergence caused mitochondrial gene dysfunction in either infected or uninfected individuals. All sequenced protein-coding genes translated into amino acids indicating that they continued to function. Two fully sequenced cytochrome c oxidase genes, co1 and co2, which encode two of the three subunits in complex IV of OXPHOS (Gershoni et al. 2009), were structurally compared to the well-characterized bovine orthologs to clarify the amino acid properties for each site (Betts and Russell 2003) (Table S1). The 26 bovine variable amino acids were compared to COI from infected and uninfected C. solmsi. The nine catalytic bovine sites in COI (240His, 242Glue, 244Tyr, 319Lys, 376His, 377Phe, 378His, 438Arg, and 439Arg) were all conserved in both infected and uninfected C. solmsi. The dramatically diverged co1 genes remained functional in both groups of wasps. Regarding COII, 16 variable sites were found among the three sequences, 13 of which were located at the N-terminus, the transmembrane domain. Only three mutated sites occurred on the C-terminal region and in positions 92 to 220 (cupredoxin); the copper center at aa159–207 (aa = amino acid) was completely conserved. Around 70% of the mutated amino acid sites did not exhibit changes in their physical properties (18 in COI; 11 in COII). Although four of the five changed sites in COII had the same type of amino acid in C. solmsi, this was not detected for COI.

Discussion

Extreme intraspecific divergence occurs between the mitochondrial genomic sequences in syntopic populations of fig wasps that shelter inside the same small syconia of figs. Two groups are identified and they correspond to whether the wasp is infected by Wolbachia or not.

EXTREME DIVERGENCE

The existence of extreme mitochondrial divergence is often reported in other organisms, including both invertebrates and vertebrates (Avise et al. 1994; Thomaz et al. 1996; Walker et al. 1997; Waters and Burridge 1999; Rawlings and Donnellan 2003; Lindell et al. 2008). Several lines of evidence suggest that the high level of divergence cannot be explained as an artifact of comparing mitochondrial and nuclear genes (Zhang and Hewitt 1996). First and foremost, the trees for cob and co1 are completely congruent (Fig. 1). This would not be expected if one or both of the genes were nuclear copies. Second, sequencing of most of the mitochondrial genome reveals great divergence between infected and uninfected individuals, and substantial divergence occurs in each gene region. If these are nuclear copies, we would expect identical sequences in some regions. Third, all amplified protein-coding genes can be translated. In contrast, nuclear integrated mitochondrial fragments typically have an advanced stop codon.

Several possible explanations exist for the unusually high levels of mtDNA diversity within C. solmsi. The coexistence of both mtDNA haplotypes may simply reflect ancient divergences coupled with morphological conservation. However, the absence of divergence in nuclear fragments, even in fast-evolving regions, precludes this explanation. Further, C. solmsi has a large brood size within one syconium, perhaps up to hundreds of individuals. Both infected and uninfected individuals may come from the same brood. If true, then it is exceptionally unlikely that ancient divergence will coexist in the same brood. These observations preclude antiquity as an explanation.

The second and most plausible explanation for the high levels of diversity involves the rapid rate of evolution of mtDNA. Parasitic hymenoptera commonly show elevated rates of mtDNA evolution. For example, species of Nasonia have the highest substitution rates found in animal mitochondria (Oliveira et al. 2008). The parasitic species C. solmsi also displays this attribute. Similar to Nasonia, it possesses many substantial gene rearrangements compared with other insects, and the dynamics of gene order rearrangements in the mtDNA often correlates with sequence substitution rates (Xu et al. 2006). A comparison of co1 sequences among pollinator fig wasps reveals very high levels of interspecific divergence, with an overall mean distance of 0.256, and for the genus Ceratosolen, the mean distance is 0.254 (unpubl. data in our lab.). The evolutionary rate of mtDNA may correlate with a correspondingly high rate in the nuclear genome, at least in those loci that encode proteins interacting in enzymatic complexes, such as ATP synthase. The high divergence rate in the mitochondrial genome might generate a corresponding divergence in the nuclear genome. Although we did not find any remarkable divergence in the nuclear markers opsin and its2, we might expect corresponding divergence in loci functionally associated with proteins encoded by the mitochondria.

WOLBACHIA AND HIGH mtDNA DIVERGENCE IN FIG WASPS

Because hymenopterans commonly have high rates mtDNA evolution, it is not surprising to discover high levels of intraspecific mtDNA divergence in species of fig wasp. However, the Wolbachia-infected population has little mtDNA variation, unlike the uninfected population that has much variation. Certainly, Wolbachia plays a vital role in determining levels of mtDNA divergence in this species.

MLST typing documents that the strain of Wolbachia that infects C. solmsi also occurs in a diversity of other host species. Consequently, either horizontal transfer from other host species or hybrid introgression from another infected species of fig wasp could explain the introduction of Wolbachia into C. solmsi (Raychoudhury et al. 2009). Considering the facts that the Wolbachia infection is accompanied by a distinct mitochondrial haplotype, which is different from the “genuine” mitochondrial haplotype of C.solmsi, but the divergence between both haplotypes is less than the average mtDNA divergence within the genus Ceratosolen, introgression of Wolbachia and associated mitochondrial haplotype between related Ceratosolen species is more possible. The observation of low mitochondrial polymorphism in the infected group and high mitochondrial polymorphism in the uninfected group also supports a relatively recent hybrid introgression of a Wolbachia infection and foreign mitochondria from a related species. Regardless, a subsequent partial selective sweep appears to have occurred both in Wolbachia and in the associated mitochondria of infected individuals. This scenario is consistent with the relatively high versus low levels of mtDNA divergence in uninfected and infected individuals, respectively.

Infections of Wolbachia in a diversity of host species cause various forms of reproductive parasitism, including CI and male killing. Such reproductive disturbances may be responsible for the observed patterns in C.solmsi. Both experimental and theoretical studies show that male-killing Wolbachia can reach high frequencies without going to fixation (Jaenike and Dyer 2008; Charlat et al. 2009; Dannowski et al. 2009). Theoretical studies also indicate that Wolbachia-infected populations can become genetic sinks, with constant transfer of the nuclear DNA from the uninfected to infected populations (Telschow et al. 2005; Jaenike et al. 2006; Kobayashi and Telschow 2010). Such events would cause cytoplasmic divergence without nuclear divergence, which is in accordance with our findings.

Reproductive isolation is pivotal to speciation. Although still debated, Wolbachia may be involved in this process. Recent theoretical modeling (Telschow et al. 2007) and laboratory studies on many organisms (Bordenstein et al. 2001; Jaenike et al. 2006; Miller et al. 2010), including both interspecific and inter-semispecific analyses, show that Wolbachia can promote speciation in their hosts by triggering reproductive isolation, including both post- and premating mechanisms. An intraspecific study of laboratory populations of D. melanogaster suggests that Wolbachia may contribute to premating isolation (Koukou et al. 2006).

The extreme intraspecific mitochondrial divergences associated with Wolbachia infections may be associated with historical events, that is Wolbachia may be involved in the speciation of fig wasps. In response to a recent invasion of Wolbachia, a new, highly diverged mitochondrial haplotype now hitchhikes its way through C. solmsi. This results in the formation of two lineages with extremely divergent mitochondrial genomes. High levels of divergence occur between mitochondrial-encoded OXPHOS subunit genes of C. solmsi infected and uninfected by Wolbachia. The nucleotide sequences are highly divergent as are the translated amino acid sequences. Although we cannot locate amino acid mutations in key functional sites, it is possible that mutated sites exert a subtle influence on protein functions, and if so are involved in incipient speciation.

A residual problem exists with the possible role Wolbachia plays in the process of speciation in fig wasps. The current models of speciation cannot explain the stable infection polymorphism of sympatric infected and uninfected individuals within a syconium. It is possible that we sampled fig wasps on five fig trees in a hybrid zone (Jaenike et al. 2006). Further studies on the putative effects of CI and male killing on the fig wasp's reproductive system, and fig wasp's fitness change subsequent to Wolbachia infection may help to determine whether Wolbachia plays a role in sympatric speciation in this promising system.

Associate Editor:L. Moyle

ACKNOWLEDGMENTS

We thank Dr. Li-Ming Niu for helping collecting the specimens and Dr. Wen Xin for providing the experimental reagents. This project was supported by National Natural Science Foundation of China (NSFC grant no. 31090253, 30900137), partially by Major Innovation Program of Chinese Academy of Sciences (KSCX2-YW-N-0807), by Program of Ministry of Science and Technology of the Republic of China (2006FY110500), by National Science Fund for Fostering Talents in Basic Research (Special subjects in animal taxonomy, NSFC- J0930004), and by a grant (No. O529YX5105) from the Key Laboratory of the Zoological Systematics and Evolution of the Chinese Academy of Sciences. Manuscript preparation was supported by a Visiting Professorship for Senior International Scientists from the Chinese Academy of Sciences to RWM. We thank the anonymous reviewers for their valuable comments and suggestions.

Ancillary