D. -W. Huang, Institute of Zoology, Chinese Academy of Sciences, 25 Beisihuanxilu, Haidian, Beijing 100080, P.R. China. Tel.: 86-10-62559639; fax: 86-10-62559639; e-mail: email@example.com
Figs (Ficus, Moraceae) and their associated fig waSPS (Hymenoptera, Chalcidoidea and Agaonidae) have attracted much attention and have been used as a model system for many studies. Fig waSPS belonging to the genus Philotrypesis are very common in most figs but their taxonomy, ecology and biology are currently poorly explored. A previous study on African Philotrypesis showed that their host association is phylogenetically conserved at subsection level. We reconstructed a molecular phylogeny with extended sampling from seven sections of figs. Our study suggested that the diversification of Philotrypesis is less constrained by host figs. Host switching is rampant between figs at species level and even at section level. We also investigated the evolution of the body colour forms in female Philotrypesis. Our study first suggested that female body colour is not evolutionarily stable and that there have been multiple transitions. Possible mechanisms for multiple colour transitions are expected to be determined in the near future.
Figs (Ficus, Moraceae) and their associated fig waSPS (Hymenoptera, Chalcidoidea, Agaonidae; Bouček, 1988) have attracted much attention and have been regarded as a model system for research on topics such as co-evolution, sex ratio evolution, virulence evolution and host-parasitoid interactions (Weiblen, 2002; Cook & Rasplus, 2003). According to whether they pollinate figs or not, fig waSPS are generally divided into fig-pollinating waSPS (fig pollinators) and nonpollinating fig waSPS (NPFW). Fig pollinators have long been considered as highly species-specific (Ramirez, 1970; Wiebes, 1979; Michaloud et al., 1996; Rasplus, 1996; but see Molbo et al., 2003) and have been suggested to have cospeciated with their hosts from low taxonomic level to high taxonomic level (Machado et al., 1996). Several morphological and behaviour traits are supposed to be evolutionary results of coadaptation between the two lineages. Figs and fig pollinators are a textbook example of obligate pollinating mutualism and an extreme case of coevolution (Ramirez, 1974; Wiebes, 1979; Ramirez, 1980; Weiblen, 2002). Compared with fig pollinators, most NPFWs have received less attention and their taxonomy, ecology, and biology have been poorly explored (Bouček, 1988). Nonpollinating fig waSPS had been placed in five subfamilies in Agaonidae but were suggested to be placed into Pteromalidae or left unplaced (Rasplus et al., 1998). Some species mimic pollinators by entering the fig cavity by the ostiole, but most NPFW species oviposit through the fig wall with the aid of their long ovipositors. Usually they lay one egg in a single ovule or in a previously gall-transformed fig flower. Their larvae develop in those ovules. They have potentially a great impact on the system by competing for oviposition sites or killing pollinator larvae, or by using fertilized ovaries and thus eating a certain proportion of seeds. Therefore, their influence on the mutualism cannot be ignored (Bronstein, 1991; West & Herre, 1994). Previous molecular phylogenetic studies suggested Sycoscapter (Sycoryctinae) have partially cospeciated with their hosts (Lopez-Vaamonde et al., 2001) but the level of cospeciation observed between Apocryptophagus (Sycophaginae) and their host figs seem to be low (Weiblen & Bush, 2002).
Among the NPFWs, species belonging to the genus Philotrypesis (Sycoryctinae) are very common in most figs and widely distributed from southern Europe (one sp. introduced with edible fig) throughout Africa and southern Asia to Australia (Bouček, 1988). The Philotrypesis species arrive mostly after the fig diameter stops growing. They do not enter the syconium and lay eggs in a number of different syconia on one or more fig trees. Like most NPFWs, the biology of most Philotrypesis species is still poorly known. Philotrypesis caricae, the one species whose biology is known, has been shown to be a parasitoid of the pollinator Blastophaga psenes but also to feed on the galled plant tissue (Joseph, 1958). Generally, their larvae are presumed to be parasite or inquiline of the pollinators. The taxonomy of Philotrypesis is less advanced than that of fig pollinators, with most species still being undescribed (Bouček, 1988). Molecular work and male morphology suggest that Philotrypesis are species specific in African figs (section Galoglychia) (Jousselin et al., 2004). The male intraspecific polymorphism is common in Philotrypesis and Jousselin et al. (2004) suggested that male morphology is evolutionarily labile and that it responds quickly to selection imposed by the mating environment.
Jousselin et al. (2004) indicated their host association is phylogenetically conservative at subsection level. Philotrypesis species associated with figs of each subsection in Galoglychia clustered together in a molecular phylogeny. However, owing to the limited number of samples included, the evolution of host association was not tested on the higher level. In addition, some fig species simultaneously harbour several morphospecies (hereafter referred to as species) that exhibit some divergence in oviposition time. For example, Ficus benjamina host two yellow and one black Philotrypesis species in China. The speciation mode of these coexisting species is still unknown. We can wonder whether these coexisting species are sister species or not.
In China, two types of female body colour forms are commonly observed in Philotrypesis. One is black, and the other is orange with black stripes on the dorsal part of the abdomen (Fig. 1). Female body colour form of a Philotrypesis species is stable across different environments or seasons. According to our present knowledge, no Philotrypesis species are polymorphic for the two colour forms. Colour form is an important trait that influences fitness in insects. For example, body colour forms may influence vulnerability to visually searching predators (Bock, 1980; Forsman & Appelqvist, 1998). It also has a profound influence on the capability of insects to live under extreme thermal conditions (Gilbert et al., 1998; Wilmer & Unwin, 1981), as well as in resistance to ultraviolet radiation (Hollocher et al., 2000), desiccation and (indirectly) to parasite infection (Wittkopp et al., 2003). Until now, the evolution of colour forms in female Philotrypesis has not been investigated. If colour forms are evolutionarily conservative, it could be a useful character for the taxonomy of Philotrypesis. On the other hand, if there are multiple transitions between colour forms, then ecological studies combined with results from phylogenetic analyses may possibly help to identify biological or ecological causes leading to colour changes.
Therefore, the main purposes of this paper are: (1) to construct the molecular phylogeny of Philotrypesis with extended sampling from figs of several sections, (2) to study the evolution of their host associations and to determine the speciation mode when more than one species simultaneously occupies the same host fig and (3) to investigate the evolution of female body colour forms in a phylogenetic context.
Materials and methods
All fig waSPS were mainly collected from fig fruits (syconia) of wild fig species in or nearby Xishuangbanna Tropical Garden, Yunnan province, China. Twenty-two Philotrypesis species associated with 17 fig species of seven sections were collected. Most species are currently undescribed. Dr. Huang identified these specimens mainly by the combined use of the following female morphological characters: length ratio of extended T7 and T8, length ratio of T7 plus T8 and gaster, length ratio of T7 plus T8 and ovipositor sheath, shape of T7 and T8, body colour pattern, basal cell of forewing hairy or hairless, ratio of pronotum breadth and length, shape of hind femur, funicular shape, body sculpture, tip shape of ovipositor sheath. Philotrypesis is highly host species-specific (Jousselin et al., 2004) but some fig species may host several Philotrypesis species. Similar species on one host fig species are distinguished on the basis of female body colour and ovipositor length. Detailed information is listed in Table 1. Species from the closely related genera Sycoscapter and Watshamiella were chosen as outgroups. All Cytb and ITS2 sequences of 24 species (22 Philotrypesis species; 2 species from closely related genera as outgroup) involved in Jousselin et al. (2004) were downloaded from GenBank for combined analysis. In addition, two ITS2 sequences (F. carica and F. rubiginosa) were kindly provided by Jousselin. The voucher specimens used in the prior study were unavailable for us. However, considering combined phylogeny and genetic divergence, we believe that all specimens used in Jousselin et al. (2004) are different from ours.
Adult female waSPS were killed and immediately preserved in 95% ethanol. Voucher specimens are kept in the Institute of Zoology, Chinese Academy of Sciences (IZCAS), China. Specimens were washed in TE buffer before DNA extraction. Total genomic DNA was extracted from a single individual using standard phenol–chloroform extraction method (Sambrook et al., 1989). In order to combine all sequences used in the previous study, the same two genes were initially amplified: the internal transcribed spacer (ITS2) of ribosomal DNA and the mitochondrial cytochrome B (Cytb). PCR reactions and primers were done following Jousselin et al. (2004) and references hereinto. The D2 domain of 28S ribosomal DNA was also amplified with primers D2F (5′-ACCCGCTGAATTTAAGCATAT3′) and D2R (5′-TTGGTCCGTGTTTCAAGACGGG-3′) (Campbell et al., 1993). The following cycling conditions were used for 28S rDNA: 5 min at 94°C; 30 s at 55°C, and1 min at 72°C (repeated 30 cycles); 5 min at 72°C. PCR amplifications were performed in a 50 μL reaction volume containing 2 mM MgCl2, 250 μM of each dNTP, 1 μM of each primer and one unit of Takara polymerase. The amplified DNA products were purified and automated DNA sequencing was carried with an ABI PRISM BigDye terminator cycle sequencing ready reaction kit (Perkin-Elmer, Applied Biosystems, Foster City, CA, USA). Sequence primers were trimmed of.
Sequence alignment and phylogenetic analyses
Sequences were initially aligned by using Clustal W (Thompson et al., 1994) with default multiple alignment parameters (gap opening penalty = 10, gap extension penalty = 5, delay divergent sequences = 40%). Unambiguous alignments were easily acquired for Cytb and 28S rDNA. The alignments for ITS2 were manually checked and the ambiguous insertion regions were discarded prior to phylogenetic analysis. The complete alignments used in the final analyses were submitted to Treebase.
Two phylogenetic reconstruction methods were used. Maximum parsimony analyses were conducted using paup4.0b10 (Swofford, 2002). All characters were unordered and equally weighted. Gaps were treated as a fifth character state. The most parsimonious trees were obtained by heuristic search with 1000 random stepwise taxon addition replicates and 10 trees held at each step. Maxtrees were set to 10, 000 trees. Nonparametric bootstrap resampling (1000 replicates) was used to estimate clade robustness.
Bayesian estimate of phylogeny was conducted using mrbayes 3.04b (Huelsenbeck & Ronquist, 2001). Appropriate models were selected by using hierarchical likelihood ratio test (hLRT; Goldman, 1993), which was implemented in the program Modeltest (Posada & Crandall, 1998). We ran the program for 1 million generations, sampling a tree every 100 generations and then calculated a 50% majority rule consensus tree in paup4.0b10 (Swofford, 2002). To ensure that our analyses were not trapped on local optima, we monitored the fluctuating value of the likelihood graphically and compared apparent stationary levels of at least two independent analyses (Huelsenbeck & Bollback, 2001). We excluded the first 5000 trees, which is enough to represent the ‘burn-in’ period. Posterior probabilities of 95% or greater are considered to be significantly supported.
Initially we performed phylogenetic analyses on each of the three datasets and then combined all Cytb and ITS2 sequences into a single analysis (the larger dataset). As 28S rDNA sequences were only available for our specimens, we also combined three data sets of fewer taxa (only our 24 species) into a single matrix for analysis (the smaller dataset). We tested the congruence of datasets using the incongruence length difference test (Farris et al., 1994) implemented in paup4.0b10. In spite of its limitations (Barker & Lutzoni, 2002; Darlu & Lecointre, 2002; Dowton & Austin, 2002), it provides a useful heuristic measure of incongruence when data partitions are of similar size and patterns of nucleotide substitution are similar across partitions. The test was run with 1000 replicates and 100 random stepwise additions of taxa. All parsimony uninformative sites were excluded before the test (Cunningham, 1997).
Wasp-host association evolution
We first mapped Ficus section/subsection onto the fig wasp phylogeny using the program macclade v. 4.0 (Maddison & Maddison, 1992) and then investigated whether wasp-host association is phylogenetically conserved using permutation tail probability tests (PTP test; Faith & Cranston, 1991) following other researches (Kelley & Farrell, 1998; Lopez-Vaamonde et al., 2003,2005). In our utilization of PTP test, we randomized the taxa 1000 times while holding the tree topology constant (as a constraint tree). These compare the number of host change steps in the actual tree with the number of steps observed in 1000 randomized trees. By mapping host-figs onto the fig wasp phylogeny, we found that Philotrypesis species associated with figs of some section did not form a monophyletic group. Alternative topologies that all Philotrypesis species from figs of each section are monophyletic were compared using Shimodaira-Hasegawa test (Shimodaira & Hasegawa, 1999) as implemented in paup4.0b10.
Colour form evolution
Dead female adults were mounted onto cards and body colour forms were directly observed under a stereomicroscope. The black form included seven species with totally black bodies. The black stripes on the dorsal abdomen vary a little within the yellow form. Here, we binarily coded body colour states and mapped them onto the phylogenetic trees using the program macclade v. 4.0 (Maddison & Maddison, 1992). Colour forms of all species involved in Jousselin et al. (2004) were kindly provided by Jousselin. Shimodaira-Hasegawa tests were used to compare alternative topologies that fig waSPS in each colour form are monophyletic.
Nucleotide characteristics and phylogenetic analyses
Sequences of 22 Cytb gene, 23 ITS2 and 23 28S rDNA D2 were newly added. Unfortunately, Cytb gene for two Philotrypesis species (P. sp. 64 and P. sp. 66), ITS2 sequences for one species (Waterstoniella sp.) and 28S rDNA for one species (P. sp. 49) were unsuccessfully amplified. For P. sp. 47, P. sp. 64 and P. sp. 66, comparisons of two sequences from the same specimen and between three individuals of the same species were conducted to check possible paralogues that may seriously confound phylogenies (Buckler et al., 1997). These sequences were nearly identical, suggesting paralogues were not highly divergent or not amplified. All sequences were submitted to GenBank under the accession number DQ270033-DQ270100.
The unambiguous alignments were easily acquired for Cytb and 28S rDNA. The final alignments were 435 bp for Cytb and 575 bp for 28S D2. The alignments of ITS2 were manually checked because sequences were too highly diverged to be easily aligned, especially between out groups and in groups. We used the alignments of previous study as reference. The ambiguous insertion regions were discarded prior to phylogenetic analysis. The alignment used in the phylogenetic reconstruction was 342 bp.
The frequency of adenine (A) and thymine (T) was higher in Cytb (A = 0.32, C = 0.14, G = 0.10, T = 0.44), especially at third codon positions, which is a well-known characteristic in insects (Crozier et al., 1989; Simon et al., 1994). In contrast, the nucleotide composition is more balanced for ITS2 (A = 0.25, C = 0.25, G = 0.25, T = 0.25) and 28S rDNA (A = 0.23, C = 0.31, G = 0.26, T = 0.19). Sequences divergence of Cytb between all Philotrypesis varied from 0 to 16.5%. ITS2 sequences showed higher levels of divergence (0–33.8%). The 28S rDNA gene is much more conservative and only shows little sequence divergence (0–0.049%).
For Cytb, when nucleotide sequences were analysed, we obtained low branch supports and low resolution of consensus trees. Better phylogenetic relationships were acquired but still lacked high branch supports (data not shown) when we translated the sequences into amino acids for analysis following other studies (Normark et al., 1991; Harry et al., 1998). This suggests Cytb is fast evolved and has little ability to solve high-level phylogenetic relationships. For ITS2, parsimony searches result in 64 trees (CI = 0.5048, 1137 steps).
For the larger dataset, there was no significant incongruence between Cytb and ITS2 data (P = 0.63) indicated by the ILD test. We thus combined Cytb and ITS2 data sets into a single matrix for phylogenetic analysis. Parsimony analysis resulted in four most-parsimonious trees (CI = 0.5529, 2758 steps). The phylogenetic relationships on the strict consensus tree were better resolved than trees based on either data set alone. Unfortunately, most deep nodes did not receive high confidence support. GTR + I + G models were selected as best-fit model by hLRT. Bayesian analyses were very similar to parsimony analysis in topology and node supports (Fig. 2).
The 28S is much conserved and parsimony analysis based on 38 parsimony informative characters resulted in 12 trees (CI = 0.7632, 159 steps). Bayesian analyses and parsimony analyses share identical topology. However, both were not well resolved in deep nodes (Fig. 3). The resolved topology is congruent with that based on a larger data set of Cytb and ITS2. ILD test also suggested that 28S are congruent with Cytb (P = 0.39) and ITS2 (P > 0.01) when only our specimens included. However, phylogenetic relationships were not better resolved than based on 28S when based on combined data set of three genes (28S, Cytb and ITS2) or only combined 28S and Cytb (data not shown).
The evolution of wasp-host association
Philotrypesis species sampled in Jousselin et al. (2004) were mainly associated with figs belonging to the three subsections in Galoglychia. Similar to the previous study, Philotrypesis species associated with figs of each subsection were clustered together (Fig. 2). Philotrypesis diversification in section Galoglychia was suggested to be constrained by their host association (Jousselin et al., 2004). In agreement with this, our PTP analyses also showed this host use conservatism at Ficus section/subsection level (P < 0.05), although not at species level (P = 0.80). However, when only our specimens were analysed, host use conservatism was not supported by PTP test at either Ficus section/subsection level (P = 0.08) or species level (P = 1.00). In fact, Philotrypesis species associated with figs of each section did not appear to be monophyletic in our analysis (Figs 4 and 5). The Shimodaira-Hasegawa test strongly rejects the alternative hypothesis that all Philotrypesis species from one section were clustered together (P < 0.001). In addition, F. concinna and F. microcarpa both host two Philotrypesis species. F. benjamina host three Philotrypesis species. Our analyses suggest that they are not sister species and some even appear to be distantly related (Figs 2 and 3).
The evolution of colour forms
By mapping the colour forms onto the phylogeny, we found that each female body colour form did not form a monophyletic clade and that there were multiple transitions in body colour (Figs 6 and 7). The alternative hypothesis that each colour form is monophyletic is strongly rejected by the Shimodaira–Hasegawa test (P < 0.001). This suggests that body colour form is not a reliable indicator for the high-level classification of Philotrypesis.
One mitochondrial gene (Cytb) and two nuclear genes (28S D2 and ITS) were used in our study. Cytb and ITS2 have long been used for revealing phylogenies at lower level (Jermiin & Crozier, 1994; Rokas et al., 2001; Hung et al., 2004). The 28S D2 is much more conservative and appropriate for higher taxonomic level (Campbell et al., 1993; Dowton & Austin, 2001). The combination of Cytb and ITS2 revealed a well-resolved phylogeny of Philotrypesis associated with figs of section Galoglychia in Jousselin et al. (2004). Unfortunately, phylogenetic analyses of the combined dataset of those two genes still left most deep nodes poorly supported in our study. This situation was not greatly changed even after adding 28S gene. This could have several reasons. First, our sampling was limited to seven sections of the Ficus genus and we did not encompass the diversity of the genus Philotrypesis. Extending the analyses to specimens associated to all sections of the fig classification would possibly improve the resolution of the phylogeny. Second, considering the less sequence divergence (of 28S D2 and Cytb) than Sycoscapter (Lopez-Vaamonde et al., 2001), one of its close relatives, a recent radiation on figs appears to be a more reasonable explanation. Denser taxon sampling and more genes would be welcome to construct the well-resolved phylogeny of Philotrypesis.
Philotrypesis diversification in section Galoglychia was suggested to be constrained by their host association (Jousselin et al., 2004). However, our study suggests that Philotrypesis diversification at species level and even at section level is not so strictly constrained by their host association. In other words, host switching may be common in the diversification of Philotrypesis fig waSPS. In our study, F. concinna and F. microcarpa both host two Philotrypesis species. F. benjamina host three Philotrypesis species. Intriguingly, they were suggested not to be sister species and some even to be distantly related. This suggests that there has already been a lot of host switching at the species level. Host switching was rampant in the evolution of Philotrypesis fig waSPS. One possible explanation for this discrepancy between our study and previous studies is that some unrecognized mechanisms effectively constrained the host-parasite association in section Galoglychia. If so, the possible factors are expected to be determined in the future. In fact, rampant host switching in Philotrypesis seems to be plausible considering the lack of physical barriers and chemical signalling by host figs, which were suggested to be possible reasons for host switching in Sycoscapter (Lopez-Vaamonde et al., 2001). Similar to Sycoscapter, Philotrypesis females use their long ovipositors to lay eggs through the wall of the inflorescence without entering it. The main physical barriers to host switching are the relationship between ovipositor style length and the thickness of their host fig wall. If their ovipositor styles are too short, they cannot reach the ovules owing to the thick fig wall. If their ovipositor styles are too long, they are believed to be less efficient at ovipositing. Possibly, it is much easier to switch between host figs whose wall is similar in thickness. Philotrypesis waSPS lay eggs into relatively mature figs whose chemical signals are likely to be different from those of receptive figs. This is suggested to increase the probability of ‘mistakes’ by nonpollinating fig waSPS (Lopez-Vaamonde et al., 2001).
Body colour form is of great importance for most animals to improve their fitness. The adaptive significance of body colour form has intrigued biologists (Poulton, 1890; Beddard, 1892). Camouflage, communication and physiological processes have been suggested as important evolutionary causes for the evolution of coloration forms in animals. But until now, little was known about the behaviour, ecology and biology of most Philotrypesis species. Certainly, the differences between species with these two colour forms are unknown. The possible causes leading to the multiple colour form evolution in female Philotrypesis are still not determined. One explanation is that colour form reflects their differences in activity time. The yellow form was supposed to be diurnal and the black form to be nocturnal (Simon Van Noort, personal communication). But our preliminary observation suggested that both forms are diurnal. Another explanation is that the colouration possibly plays an important role in escaping predator (Da-rong Yang, personal communication). Usually scores of Philotrypesis land on the surface of fig fruits and oviposit simultaneously. Ants are common predators of Philotrypesis. But the preference of those ants for the different colour forms has not been examined. We do not know whether the colouration has been involved in escaping ant predation. In addition, the colouration could also be involved in the niche partitioning and speciation. A lot of studies were encouraged to determine possible mechanisms of multiple colour form transitions in the near future.
We thank Emmanuelle Jousselin for kindly providing the final alignment of ITS2, the female body colour forms of species used in previous studies, two unpublished ITS2 sequences, and valuable comments on the manuscript. We thank Simon Van Noort and Da-rong Yang for the discussion about the possible mechanisms leading to the multiple colour form transitions. We also thank two anonymous reviewers for their valuable comments. This project was supported by the National Natural Science Foundation of China (NSFC Grant No. 30330090), partially by the National Science Fund for Fostering Talents in Basic Research (NSFC-J0030092).