| Edited by Yoshihiko Tsumura. Zhi-Hui Su: Corresponding author. E-mail: su.zhihui@brh.co.jp. Hiroshi Azuma: Present address: Department of Botany, Graduate School of Science, Kyoto University, Kyoto 606-8502, Japan. Rhett D. Harrison: Present address: Xishuangbanna Tropical Botanical Garden, Menglun, Mengla, Yunnan, China. |
The genus Ficus (Moraceae) is composed of about 750 species mainly distributed in tropical and subtropical regions of the world, exhibiting various life forms of tree, shrub, climber, and (hemi-)epiphyte (Corner, 1965; Berg, 1989, 1990). Ficus is one of the most diversified genera of woody plants in the tropics (Harrison, 2005), and because of the year-round production of the fruits, fig species play an important role as keystone plant resources for many terrestrial organisms (Lambert and Marshall, 1991; Shanahan and Compton, 2001; Shanahan et al., 2001; Kissling et al., 2007). Symbiotic interaction between plants and animals is considered as one of important factors contributing the biodiversity of the terrestrial ecosystem especially in tropical and subtropical regions, therefore, it is interesting to know how the interaction among Ficus and their associated animals has arisen and is being maintained (Janzen, 1979; Frank, 1989; Kjellberg and Maurice, 1989; McKey, 1989; Windsor et al., 1989; Berg, 1990; Herre, 1996; Nason et al., 1998; Harrison, 2003; Harrison et al., 2003; Herre et al., 2008).
In contrast to the diversity of organisms feeding on fig fruit, the pollination of figs is entirely dependent on the members of a single clade of Chalcid wasps (Agaonidae) (Ramírez, 1970, 1974; Janzen, 1979; Wiebes, 1979; Corner, 1985). The genus Ficus is taxonomically defined by their unique, closed inflorescence (syconium), which bears many small unisexual flowers on the inside. The pollen-carrying female wasps enter via a small entrance (ostiole), protected by bracts, at the top of the syconium. Inside the syconium, the female wasps pollinate the flowers and oviposit in some of the ovules. Wasp larval development is entirely completed within the galled ovaries. It was previously thought that the interaction was predominantly species-specific, and therefore that lineages of figs and fig-wasps should have co-evolved and co-speciated in isolation. Consequently, it was suggested that both groups should have basically identical phylogenetic relationships (Herre, 1996; Weiblen, 2001).
Recent advances in molecular phylogenetics afford us the opportunity to trace the co-evolutionary histories of figs and fig-wasps and to elucidate evolutionary trends of specific traits that contribute to the mutualism (e.g., Yokoyama, 1995, 2003; Herre et al., 1996; Machado et al., 1996, 2001, 2005; Kerdelhue et al., 1999; Weiblen, 2000, 2001, 2004; Weiblen and Bush, 2002; Cook and Rasplus, 2003; Jousselin et al., 2003; Datwyler and Weiblen, 2004; Jackson, 2004; Rønsted et al., 2005; Su et al., 2008). These studies reveal a broad scale pattern of co-speciation, but at a finer scale within clades of closely related species there is little support for strict sense co-speciation (e.g., Weiblen and Bush, 2002; Machado et al., 2005; Su et al., 2008).
The degree of host specificity in fig wasps is fundamental to understanding the co-evolutionary biology of the interaction (Cook and Rasplus, 2003; Machado et al., 2005). Examples of two or more fig wasp utilizing the same host and of multiple host use by a single fig wasp have been known (Wiebes, 1966; Rasplus, 1994). However, they were thought to be exceptions to “the one-to-one rule”. Recent studies, predominately from Central America, involving intense intraspecific sampling indicate that the breakdown in specificity is more prevalent, occurring in over a third of the species investigated. Population level molecular studies in both figs and fig-wasps have revealed high levels of intraspecific genetic heterogeneity, which may be a prerequisite for speciation, and suggest the presence of cryptic wasp species and hybridization among figs (Molbo et al., 2003; Machado et al., 2005; Haine et al., 2006; Su et al., 2008). It is thus important to conduct broad sampling throughout the distribution of each fig and fig-wasp species to understand the degree of intraspecific genetic variation.
Janzen (1979) proposed that the specificity between figs and fig-pollinating wasps might break down on islands or in very harsh environments towards the edge of a species’ distribution, where pollinator populations are likely to suffer periodic extirpations. Indeed there is some evidence of this from the volcanic Krakatau Islands (Parrish et al., 2003). In this study, therefore, we focus on fig and fig-wasp interactions found in the Ryukyu Islands to understand the specificity of fig and fig-wasp communities on islands in the northern edge of the species’ distribution. In addition, we included the Bonin (Ogasawara) Islands because they are oceanic islands (ca. 1000 km from the Japan mainland and ca. 1200–1600 km from the Ryukyu Islands) and only endemic fig species (plus one introduced species) are found there.
The purposes of this study were to investigate (1) the inter and intraspecific sequence variation of Japanese figs and fig-wasps and (2) whether the molecular phylogenies of figs and fig-wasps collected from various localities in the Ryukyu and Bonin Islands are congruent or not, and finally to know whether the phenomenon of breakdown of the strict relationship between figs and fig-pollinating wasps in the Ryukyu Islands has occurred. We conducted molecular phylogenetic analyses of Japanese figs and their associated fig-pollinating wasps using sequence data of six non-coding plastid regions and nuclear ribosomal internal transcribed spacer (ITS) region for the fig phylogeny, and those of the nuclear 28S rRNA gene and mitochondrial COI gene for the wasp phylogeny.
Sixteen species of Ficus are recognized as native figs in Japan (Fig. 1; Corner, 1965; Walker, 1976; Hatusima and Amano, 1994; Shimabuku, 1997; Yokoyama, 2006), nine of which (F. superba var. japonica, F. caulocarpa, F. microcarpa, F. ampelas, F. irisana, F. virgata, F. benguetensis, F. septica, and F. variegata) are widely distributed in the Southeast and East Asia (Corner, 1965). Four species (F. pumila, F. thunbergii, F. nipponica, and F. erecta) are mainly found in Japan, Korea, and China (Corner, 1965), and three species (F. boninsimae, F. nishimurae, and F. iidaiana) are endemic to the Bonin (Ogasawara) Islands.
 View Details | Fig. 1 Locality maps of the samples used in this study. The locality numbers (sample ID numbers) correspond to those in Table 1 and all figures. Approximate northern limits of the distribution of all Japanese fig species are also presented by lines. Ficus boninsimae, F. nishimurae, and F. iidaiana are endemic to the Bonin Islands. Taxonomy of Japanese Ficus was followed by Yokoyama (2006). |
The list of samples used in this study and their locality maps are given in Table 1 and Fig. 1, respectively. In the field, fresh leaves were collected and dried by silica gel for DNA analyses. The mature but unopened syconia (2 to 5) were also collected simultaneously from the same fig tree and kept in a plastic tube (50–100 mL). After the wasps had emerged from the syconia, we preserved them in 99.5% ethanol solution. We tried to collect all native figs and their fig-wasps. However, we failed to collect F. iidaiana in the Bonin Islands and the fig-pollinating wasps for F. nipponica and F. nishimurae, and were only able to collect one sample of fig-pollinating wasp for F. irisana, F. pumila, and F. thunbergii. In addition, we collected leaf sample and their associated fig-wasps from introduced F. microcarpa in the Bonin Islands. In total, we collected 136 leaf samples (individuals) from 15 species of Ficus for plastid DNA phylogeny (of which 49 samples were used for ITS phylogenetic analysis), and 95 samples (one wasp per fig tree) of the associated fig-pollinating wasps from 13 fig species (Table 1).
 View Details | Table 1 List of materials used in this study and their accession numbers |
Total DNA was isolated from the silica gel-dried leaves (1–2 g) using modified CTAB (Cetyltrimethylammonium Bromide) method of Doyle and Doyle (1987). The following six non-coding plastid regions were amplified; rps16 intron (primers; rps16/1F and rps16/2R), trnG intron (UCC2/F and UCC/1R), petB intron (petB/1F and petB/2R) (Nishizawa and Watano, 2000), trnL intron (c and d), trnL-trnF spacer region (e and f) (Taberlet et al., 1991), and atpB-rbcL spacer region (AT1 and RB) (Azuma et al., 1999).
The PCR mixture (20 μL) contained 1 μL (100–150 ng) of template DNA, 200 μmol/L of each dNTP, 1 μmol/L each primer, 2.5 mmol/L MgCl2, 2 μL of 10X Taq buffer, 1U of Taq polymerase (TaKaRa Ex Taq, TaKaRa BIO INC., Japan). The PCR was performed by a GeneAmp PCR System 9700 (Applied Biosystems) starting at 94°C (5 min), followed by 35 cycles of denaturation at 94°C (60 sec), annealing at 50°C (90 sec), and extension at 72°C (2 min), and final extension at 72°C (7 min). After checking a single band by electrophoresis on 1% agarose gel, the PCR products were purified by the QIAquick PCR Purification Kit (QIAGEN K. K., Tokyo, Japan). Direct sequencing of both strands were conducted on an ABI 3100 Genetic Analyzer (Applied Biosystems) using BigDye Terminator version 3.1 Cyclic Sequencing Ready Reaction Kit (Applied Biosystems) under conditions in manufacture’s protocol.
For sequencing of the nuclear ITS region, primers ITS-Y5 and ITS-Y4 were used for the amplification under the condition described in Kita and Ito (2000). The thermal conditions for the PCR were as follows: denaturation at 94°C (5 min); followed by 35 cycles of denaturation at 94°C (30 sec), annealing at 50°C (30 sec), and extension at 72°C (1 min); and final extension at 72°C (7 min). When direct sequencing suggested polymorphism of ITS sequences, molecular cloning was conducted in such cases, and more than five colonies were applied for sequencing.
Total DNA was extracted from a single wasp using a QIAamp DNA Mini Kit (QIAGEN K. K., Tokyo, Japan). PCR amplification and sequencing of the cytochrome oxidase subunit I (COI) and nuclear 28S rRNA genes were conducted according to Su et al. (2001, 2008). A fragment of the mitochondrial DNA containing an approximately 1000 bp 5’-region of COI gene was amplified by a primer set “COI1-1-Ple” and “COI-2M” (Su et al., 2008), and a fragment comprising the variable domains D1 and D2 of 28S rRNA gene were amplified by a primer set “28S-01” and “28SR-01” (Kim et al., 2000; Su et al., 2001).
Alignments of the DNA sequences were carried out by MAFFT ver.6 with FFT-NS-2 methods (Katoh et al., 2002; http://mafft.cbrc.jp/alignment/server/), which were then manually checked. Neighbor-joining (NJ), maximum parsimony (MP) and maximum likelihood (ML) criteria were used in the phylogenetic analyses. All analyses were performed by PAUP* 4.0b10 (Swofford, 2003), with heuristic search, 10 replications of random addition sequence, and TBR (tree bisection reconnection) branch swapping option (for MP and ML). One thousand (MP and NJ) and/or one hundred (ML) bootstrap replications were performed. Because of the huge amount of time required for the bootstrap analysis under the ML criterion, the bootstrap values were not estimated for ITS and COI trees. In MP analysis of ITS sequence data, the Maxtree was set to 10,000. For ML analyses, the nucleotide substitution model was selected by Akaike Information Criterion in Modeltest3.7 (Posada and Crandall, 1998). Samples which showed identical sequence were represented by one sample, ignoring differences in number of single-nucleotide repeats.
The sequence data of six non-coding plastid regions from each fig sample were combined into one sequence. Two species from different genera (Broussonetia papyrifera and Castilla elastica) in Moraceae were used as outgroups. Two species in Ficus subgenus Pharmacosycea (F. insipida and F. tonduzii) were also included in the plastid phylogenetic analysis to confirm the Pharmacosycea species is sister to all other Ficus species, because they were designated as outgroups in ITS phylogenetic analysis. For ITS phylogenetic analysis, the sequences of two Pharmacosycea species (F. insipida and F. maxima) were used as outgroups. For the phylogenetic analyses of fig-wasps, three species of genus Tetrapus (T. costaricanus, T. americanus, and T. ecuadoranus) which pollinate Pharmacosycea species were used as outgroups (Rønsted et al., 2005). Prior to doing this, we confirmed that these Tetrapus species were sister to the collected fig-wasps by using non-pollinating or parasitic wasps as outgroups.
A combined data matrix of six non-coding regions of plastid DNAs of Ficus and two genera (Broussonetia and Castilla) was composed of 3922 characters after alignment (total 24 OTUs) (accession numbers of new sequences are AB445493–AB445616, Table 2). There were 271 (6.9%) variable sites, of which 111 (2.8%) characters were parsimony-informative, among all OTUs, and 93 (2.3%) variable sites, of which 60 (1.5%) characters were parsimony-informative, within the ingroup (Ficus). The MP analysis generated two equally most parsimonious trees with tree length 294, CI (consistency index) excluding uninformative characters = 0.886, RI (retention index) = 0.929. The best-fit model for the ML analysis of plastid sequences was K81uf + G with gamma shape 0.5078. Topology of the strict consensus MP tree and NJ tree were identical to that of the ML tree (Fig. 2).
 View Details | Table 2 List of accession numbers and haplotypes of plastid DNA of Ficus |
 View Details | Fig. 2 Molecular phylogenetic tree of Japanese Ficus constructed by ML criterion using six plastid non-coding regions. The strict consensus tree by MP analysis also showed the same topology. Numbers on branches indicate bootstrap values (ML/NJ/MP > 50%). Number in parentheses indicates sample ID number (see Table 1), but for example, “F. virgata (all;14)” means all samples (14 individuals) of F. virgata showed an identical sequence. Subgeneric classification is also presented. |
The MP, NJ and ML trees showed that the subgenus Pharmacosycea was sister to the rest of Ficus species, which comprised subgenera Ficus, Synoecia, Sycidium, Sycomorus, and Urostigma, although bootstrap values supporting the clade were low [61(ML) / < 50(NJ) / 64(MP)]. Judging from this and previous studies (Weiblen, 2000; Datwyler and Weiblen, 2004; Zerega et al., 2005), subgenus Pharmacosycea appears to have sister relationship with the other Ficus examined. Intraspecific sequence variations were observed in F. erecta (6 haplotypes), F. pumila (2), F. benguetensis (4), and F. variegata (2). While, following four pairs of species did not show interspecific sequence difference: one haplotype of F. erecta and the Bonin endemic species (F. boninsimae and F. nishimurae); F. nipponica and F. thunbergii; F. ampelas and F. irisana; F. superba var. japonica and F. caulocarpa (Fig. 2).
There are five well [97/99/99 (subgenus Ficus), 100/99/100 (subg. Synoecia), 89/84/85 (subg. Urostigma) and 80/88/83 (subg. Sycomorus)] or moderately [61/55/59 (subg. Sycidium)] supported clades containing more than one species, which are basically concordant with circumscriptions of the subgenera (Berg and Corner, 2005) (Fig. 2).
An aligned sequence matrix of nuclear ITS sequences of Ficus was composed of 703 characters [total 84 OTUs from 49 leaf samples plus outgroups], in which F. insipida and F. maxima (subgenus Pharmacosycea) were set as outgroups (accession numbers of new sequences are AB485811–AB485920, Table 1). There are 198 (28.2%) variable sites, of which 130 (18.5%) characters were parsimony-informative, within the ingroup (Japanese Ficus), and 220 (31.3%) variable sites, of which 153 (21.8%) characters were parsimony-informative, among all OTUs. The MP analysis generated 10,000 (maxtree) equally parsimonious trees with tree length 388, CI excluding uninformative characters = 0.583, RI = 0.929. The best-fit model for the ML analyses was TVM + I + G with gamma shape 0.8720. Clades that appeared in the strict consensus tree of 10,000 MP trees and NJ tree are denoted by thick lines in the ML tree (Fig. 3). The bootstrap values noted on the branches were estimated by bootstrap analyses using NJ and MP criterions (Fig. 3). Major discordance between the MP, NJ and ML trees was found in phylogenetic relationship among subgeneric clades.
 View Details | Fig. 3 Molecular phylogenetic tree of Japanese Ficus constructed by ML criterion using nuclear ITS sequences. Branches which also appeared in the NJ tree and the strict consensus tree of 10000 MP trees were denoted by thick lines. Numbers on branches indicate bootstrap values (NJ/MP > 50%). Bootstrap analysis under ML criterion was not performed. Each OTU indicates sample ID (see Table 1) and clone ID (sample-clone), but “-a” at clone ID indicates that the sequence was determined by direct sequencing. Names of species and subgenera are presented at right side of OTUs. The sequences of two Pharmacosycea species were retrieved from the DNA Data Bank of Japan (DDBJ). |
Molecular cloning of amplified ITS sequences gave multiple copies in some samples but all copies derived from one species grouped together as a single clade with relatively high bootstrap values in MP, NJ and ML analyses, except for the following: copies of F. boninsimae and F. nishimurae (Bonin endemics) were nested with each other in a single clade within the clade of F. erecta, as was found in the plastid DNA phylogeny (clade A in Fig. 3 and see Fig. 2). Copies of F. caulocarpa formed a clade but the bootstrap value was low (clade E2 in Fig. 3). As in the plastid DNA sequences, ITS sequences of F. thunbergii and F. nipponica showed no difference (clade B in Fig. 3).
In MP, NJ and ML analyses, close affinity among species belonging to the same subgenus was supported again. Phylogenetic resolution at interspecific level was better in the ITS phylogeny than in the plastid DNA phylogeny.
An aligned sequence matrix of the nuclear 28S gene of fig-wasps was composed of 959 characters (total 19 OTUs; 89 samples plus outgroups) (accession numbers of new sequences are AB489204–AB489219 and AB576183-AB576185, Table 1). There were 330 (34.4%) variable sites, of which 246 (25.7%) characters were parsimony-informative, within the ingroup, and 366 (38.2%) variable sites, of which 296 (30.9%) characters were parsimony-informative, among all OTUs. The MP analysis generated one most parsimonious tree with tree length 876, CI excluding uninformative characters = 0.602, RI = 0.709. Clades which appeared in the MP and NJ trees are denoted by thick lines in the ML tree (Fig. 4). The best-fit models were GTR + I + G with gamma shape 0.4973 for 28S. The topologies of MP, NJ and ML trees were basically congruent, but the clade of Kradibia + Liporrhopalum branched first in the consensus MP and NJ trees. Species belonging to the same genera were grouped together in a single clade with high bootstrap values (Fig. 4).
 View Details | Fig. 4 Molecular phylogenetic tree of fig-pollinating wasps constructed by ML criterion using nuclear 28S gene sequences. Branches which also appeared in the NJ tree and the MP tree were denoted by thick lines. Numbers on branches indicate bootstrap values (ML/NJ/MP > 50%). Number in parentheses indicates sample ID (see Table 1), but for example, “Ceratosolen bisulcatus (all;14)” means all samples (14 individuals) of C. bisulcatus showed an identical sequence. Host figs from which the fig-wasps were collected and the subgeneric classification are also presented. |
Intraspecific variation was observed in Ceratosolen cornutus (from F. benguetensis) and Platyscapa sp.-2 (from F. caulocarpa). While, the other twelve species of fig-wasps did not show intraspecific variation. Fig-wasps emerged from fig species belonging to the same subgenera formed a single clade without exception. However, phylogenetic relationships among these clades were still unclear. Although F. boninsimae (Bonin endemic) showed no difference with one haplotype of F. erecta in the plastid DNA tree (Fig. 2) and nested within a clade of F. erecta in the ITS tree (Fig. 3), the fig-wasp (Blastophaga sp.) emerging from F. boninsimae is clearly diverged from B. nipponica (F. erecta) (Fig. 4). The wasp species pollinating F. superba and F. caulocarpa appear to be morphologically different from the known pollinating wasp species of these two figs (Wiebes, 1966). Hence, we tentatively treated them as Platyscapa sp.-1 and P. sp.-2 in this study (Fig. 4, Fig. 5 and Fig. 6). To confirm this result, further molecular and morphological analyses are in progress.
 View Details | Fig. 5 Molecular phylogenetic tree of fig-pollinating wasps constructed by ML criterion using mitochondrial COI gene sequences. Branches which also appeared in the NJ tree and the strict consensus tree of MP trees were denoted by thick lines. Numbers on branches indicate bootstrap values (NJ/MP > 50%). Bootstrap analysis under ML criterion was not performed. OTUs indicate sample ID (see Table 1). Names of fig-wasps, host figs and their subgeneric classification are also presented. The sequences of outgroup species (Tetrapus americanus, AB308327; T. costaricanus, AB308328; T. ecuadoranus, AB308322) were analyzed in previous study (Su et al., 2008). |
 View Details | Fig. 6 Summarized interspecific relationship of Japanese Ficus (left) and their associated fig-wasps (right). Branches which appeared in both trees (plastid and ITS or 28S and COI) are presented with bootstrap values (plastid/ITS or 28S/COI). Dashed lines indicate the branches collapsed in the plastid phylogenetic tree. Arrowed lines indicate interaction between the fig and their associated fig-wasps. COI sequences of Wiebesia sp. and Blastophaga sp. were not determined, therefore, the bootstrap values for clades of W. sp. and W. pumilae, and B. nipponica and B. sp. were not indicated. Ficus boninsimae and F. nishimurae have nested within a clade of F. erecta. Because fig-wasps from F. nipponica and F. nishimurae were not collected, they were placed near the closest relatives. |
An aligned sequence matrix of mitochondrial COI genes of fig-wasps was composed of 868 characters (total 52 OTUs; 75 samples plus outgroups) (accession numbers of new sequences are AB488704–AB488752). No indels (insertion or deletion) were required for the sequence alignment. There are 407 (46.9%) variable sites, of which 376 (43.3%) characters were parsimony-informative, within the ingroup, and 419 (48.3%) variable sites, of which 384 (44.2%) characters were parsimony-informative, among all OTUs. The MP analysis generated 550 equally most parsimonious trees with tree length 1111, CI excluding uninformative characters = 0.522, RI = 0.896. Clades which appeared in the strict consensus MP tree and NJ tree are denoted by thick lines in the ML tree (Fig. 5). The best-fit models were GTR + I + G with gamma shape 0.6710. The bootstrap values noted on the branches were estimated by bootstrap analyses using NJ and MP criterions. The interspecific topologies of MP, NJ and ML trees were basically concordant, but Wiebesia pumilae was sister to the Ceratosolen clade in the consensus MP tree, and to the clade with Ceratosolen, Kradibia and Liporrhopalum in the NJ tree.
Although all fig-wasp species showed intraspecific sequence variation (except Kradibia commuta and W. pumilae for which only one sample was examined), all species clearly formed a single clade with 100% bootstrap values. In addition to the 28S tree (Fig. 4), the COI tree also showed that fig-wasps emerging from figs belonging to the same subgenera formed a single clade. In the 28S phylogeny (Fig. 4), Ceratosolen cornutus and Platyscapa sp.-2 showed intraspecific sequence variation, however, the two species did not show the same pattern of variation in the COI phylogeny (Fig. 4 and Fig. 5).
Molecular phylogenetic analysis of the non-coding plastid sequences indicated the sister relationship between two lineages (sects. Urostigma and Conosycea) in subgenus Urostigma (Fig. 2). In contrast, the ITS phylogeny suggested that the two lineages were distantly related (clades E1 and E2 in Fig. 3), which were concordant with the results in previous ITS phylogenetic studies (Weiblen, 2000; Jousselin et al., 2003; Rønsted et al., 2005). Although the two lineages belong to different sections, the subgenus Urostigma is well characterized as being monoecious (the other subgenera studied here are all dioecious), producing aerial roots, and the presence of a single waxy gland (Berg and Corner, 2005). Moreover, it should be noted that it was shown that the fig-wasps collected from these host fig species were closely related to each other in both 28S and COI phylogenetic trees (P. sp.-1, P. sp.-2, Eupristina verticillata, and E. sp. in Fig. 4 and Fig. 5), consistent with previous studies (Machado et al., 2001; Weiblen, 2001; Jiang et al., 2006; Lopez-Vaamonde et al., 2009). These results suggest that monophyletic relationship of members in subgenus Urostigma detected in the plastid phylogeny may be a more reliable phylogenetic hypothesis than that found in the ITS tree, although more extensive analysis should be conducted.
On the whole, the interspecific relationships of figs were better resolved in the ITS tree than the plastid tree, indicating a high performance of the ITS sequence in phylogenetic inference of the lower-level phylogeny of Ficus. For example, relationships among members of subgenus Sycomorus (F. benguetensis, F. variegata, and F. septica) were not resolved (polytomous) in the plastid tree (Fig. 2), but they were clear in the ITS tree (Fig. 3). Likewise, F. ampelas and F. irisana in subgenus Sycidium showed an identical sequence of plastid DNA, but they were distinct from each other in the ITS sequences.
Although the inter-subgeneric relationships of figs were generally unclear, species taxonomically belonging to the same subgenus formed a single clade in both plastid and ITS trees (except for subgenus Urostigma in the ITS tree) with relatively high bootstrap supports (Fig. 2 and Fig. 3). As was expected, the same pattern was observed in the phylogenies of the associated fig-wasps, that is, the fig-wasps collected from the host figs belonging to the same subgenus formed a well-supported clade without exception, but the relationships among these clades are still uncertain in most cases (Fig. 4 and Fig. 5).
These results indicate that the fig and fig wasp lineages found in the Ryukyu Islands have been reproductively isolated throughout their evolutionary history and suggests that the host specificity of these fig-pollinators remains very high today, despite being at the northern edge of the species’ range and scattered across an island archipelago. There was no evidence of host switching among fig-wasps, as has been reported from elsewhere (e.g., Machado et al., 2005; Su et al., 2008). The nuclear ITS sequence data is sometimes used as a molecular marker to detect a historical or current hybridization event in plants, as it provides multiple copies (or a chimeric copy) from both maternal and paternal progenitor lineages (e.g., Sang et al., 1995; Barkman and Simpson, 2002; Siripun and Schilling, 2006). However, we found that all ITS copies derived from one species of fig were gathering in a single clade and supported by high bootstrap values (except for F. caulocarpa). Thus, it seems that hybridization among closely related Ficus species has not occurred in the Ryukyu Islands (Fig. 3).
Interestingly, F. nipponica and F. thunbergii (in subgenus Synoecia) are the only species which showed an identical ITS sequence, even if they are morphologically distinct. However, because both species are relatively rare in the Ryukyu Islands (one and two samples collected, respectively) and we did not perform molecular cloning procedures, more critical analysis should be conducted. It should also be noted that both species reproduce vegetatively.
In this study, we did find one example of two wasp species (E. verticillata and E. sp.) pollinating one fig species (F. microcarpa) (Fig. 4 and Fig. 5). These two species can be easily distinguished from each other by their morphological features. One appears to be an undescribed species. As seen in the phylogenetic trees (Fig. 4 and Fig. 5), E. sp. is closely related to the known pollinating wasp species (E. verticillata) of F. microcarpa, suggesting it may be a case of breakdown of the host specificity or recent speciation at one fig species. In this study we were able to analyze the pollinating wasps from a total of 14 fig trees of F. microcarpa. Eupristina sp. was only collected from two trees inhabiting the Iriomote island, where E. verticillata was also distributed.
To understand the evolutionary process of co-speciation between the fig and fig-wasp mutualism, it may be important to reveal intraspecific genetic variation (or structure) in both fig and fig-wasp, which sometimes leds to finding cryptic species (Molbo et al., 2003; Haine et al., 2006).
Molecular cloning produces multiple copies of ITS sequences from a single leaf in most cases and these copies scattered and nested with other copies originated from different individuals of the same species (Fig. 3). It is therefore difficult to infer intraspecific variation of each fig species based on the ITS sequence data. On the other hand, plastid DNA has often been used to detect intraspecific variation and indeed gave a clear geographic structure in some plant species native in Japan (e.g., Fujii et al., 1997; Ohi et al., 2003a, 2003b).
Phylogenetic analysis of six non-coding plastid DNA sequences of Ficus revealed that there was relatively large intraspecific variation in F. erecta (six haplotypes) (Fig. 2). Three haplotypes formed a subclade with low bootstrap values (63/62%), and the other three haplotypes showed an unclear relationship. All samples constituting the subclade were collected from northern area of the Ryukyu Islands and the mainland (i.e., OKI, KDI, MDI, AMA, YAK, AWA, and OSA; Fig. 1), and the rest samples were collected from the southern islands in the Ryukyu Islands (MIY, YON, ISH, and IRI; Fig. 1), suggesting that genetic differentiation between the northern and southern areas has occurred possibly by a geographical isolation, although the associated fig-wasp (B. nipponica) did not show clear intraspecific variation and geographic structure.
Ficus benguetensis and its pollinator, C. cornutus, showed intraspecific variations (Fig. 2, Fig. 4, and Fig. 5). However, a clear geographical structure was not evident from the phylogenetic trees, and the phylogenetic relationships among the haplotypes were not concordant between the fig and the fig-wasp, and between the 28S and COI trees of the wasp. Likewise, although there was no intraspecific variation in the plastid DNA of the host figs, the COI phylogenies indicated that K. sumatrana (F. ampelas) and L. philippinensis (F. virgata) each divided into two well-supported clades. However, the clades did not correspond to the island geography (Fig. 1 and Fig. 5).
Becuase the number of individuals analyzed and molecular approaches used in this study may be insufficient to detect or infer the intraspecific genetic structure of each species, we need to use more proper DNA markers for population genetic analysis to reveal intraspecific variation in figs and fig-wasps.
Because the Bonin (Ogasawara) Islands are typical oceanic islands, located ca. 1000 km south of the mainland of Japan and ca. 1200 km east of the Ryukyu Islands (Fig. 1), plants and animals living on the islands (or their ancestors) must be migrated from adjacent continents or continental islands. There are three endemic fig species (F. boninsimae, F. nishimurae, and F. iidaiana) in the Bonin Islands, all of which belong to section Ficus (Corner, 1965; Berg and Corner, 2005; Yokoyama, 2006). Although F. iidaiana could not be included in this study, a phylogenetic analysis using plastid non-coding sequences clearly indicated that all samples of F. boninsimae and F. nishimurae were identical to each other and to one haplotype of F. erecta (Fig. 2). All samples of F. erecta showing the identical haplotype with F. boninsimae and F. nishimurae were collected from the southern Ryukyu Islands (i.e., MIY, ISH, IRI, Fig. 1). Likewise, the ITS phylogenetic tree also showed that F. boninsimae and F. nishimurae were not distinct, but formed a single clade with lineages of F. erecta (Fig. 3).
Morphologically, F. boninsimae and F. nishimurae are considered to be closely related and are sometimes treated as a single species (Corner, 1965; Yokoyama, 2006), but there are a few morphological and ecological differences (Yokoyama, 2006). However, when the two species are compared with F. erecta, there are few good diagnostic characters (Yokoyama, 2006). The molecular phylogenetic analyses conducted here evidently suggest that a common ancestor of the two species originated from F. erecta in the southern Ryukyu Islands (a possible migration from Taiwan or southern China cannot be rejected). Having established in the Bonin Islands the ancestor diversified morphologically into the two species in response to the isolated island environments without genetic differentiation at plastid sequence level, and with little differentiation in ITS sequence or in the RAPD analysis by Yokoyama (2003). This situation is considered to be a consequence of rapid radiation, a typical phenomenon in oceanic islands (Carlquist, 1974; Crawford and Stuessy, 1997; Baldwin et al., 1998; Ito, 1998).
We were only able to collect fig-wasps from F. boninsimae in the Bonin Islands, and we were only able to determine the nuclear 28S sequence for phylogenetic analysis. The analysis of the 28S sequences of all fig-wasps collected clearly indicated that Blastophaga sp. from F. boninsimae was sister to B. nipponica (from F. erecta) (Fig. 4). There was no intraspecific sequence variation in either species and 2.92% base difference was observed between them. This value is almost equivalent to that between K. sumatrana (F. ampelas) and K. commuta (F. irisana) (3.99%) and between W. pumilae (F. pumila) and W. sp. (F. thunbergii) (2.19%), suggesting that B. sp. (F. boninsimae) and B. nipponica (F. erecta) have differentiated enough to establish reproductive isolation each other, even if the host-figs (F. boninsimae and F. erecta) have not clearly diverged into the two lineages.
A few individuals of F. microcarpa without its pollinating wasps were introduced into the Bonin Islands in the early Meiji era (more than one hundred years ago), but have produced a number of mature syconia in recent years. We were interested in which wasps pollinated the introduced F. microcarpa on the Bonin Islands.
All of the fig-wasps collected from F. microcarpa in the Bonin Islands are morphologically considered to be the pollinating fig-wasp, E. verticillata, and indeed phylogenetic analyses of the wasp clearly indicated that they were identical to a lineage found in the Ryukyu Islands in both 28S and COI phylogenies (OGS01, 04, 19 in Fig. 4 and Fig. 5), suggesting long distance dispersal, more than 1200 km, of the fig-wasp from the Ryukyu Island to the Bonin Islands. It may be considered that the recent migration was effected by human activities. However, there is no direct transportation between the Ryukyu and Bonin Islands and no regular airport in the Bonin Islands. Hence, human assisted migration is considered unlikely.
Summary of results of the phylogenetic analyses and interaction between figs and fig-wasps found in the Ryukyu and Bonin Islands is illustrated in Fig. 6. The host specificity between the figs and fig-pollinators is tightly maintained in the Ryukyu and Bonin Islands. There is no or little intraspecific variation in both figs and fig-wasps among islands, suggesting the dispersal within the Ryukyu Islands occurs easily, although in F. erecta there is some divergence between northern and southern areas. The two endemic species (F. boninsimae and F. nishimurae) or their common ancestor in the Bonin (Ogasawara) Islands and the associated fig-wasps (Blastophaga sp.) are apparently derived from F. erecta and its associated fig-wasp (B. nipponica), respectively, which probably migrated from the Ryukyu Islands (ca. 1200–1600 km) or a more distant area.
We thank Akihiro Seo for helping us to collect some samples of figs and fig-wasps, Jean-Yves Rasplus for identification of fig wasps, and Tsukuba Botanical Garden, National Museum of Nature and Science for providing a leaf material. We are also grateful to Takashi Miyata for useful comments and suggestions in advancing this study. Thanks are also due to Hideko Kanda and Masami Yamazaki for their skillful technical assistance in PCR and sequencing. This study was supported in part by Grant-in-Aid for Scientific Research (C) (No. 19570225) from Japan Society for the Promotion of Science.
|
Index