Host sex-specific parasites in a functionally dioecious fig: a preference way of adaptation to their hosts

Host–parasites interaction is a common phenomenon in nature. Diffusive coevolution might maintain stable cooperation in a fig–fig wasps system, in which the exploiter might diversify their genotype, phenotype, or behavior as a result of competition with pollinator, whereas the figs change flower syconia, fruits thickness, and syconia structure. In functionally dioecious Ficus auriculata, male figs and female figs contain two types of florets on separate plant, and share high similarities in outside morphology. Apocryptophagus (Sycophaginae, Chalcidoidea, Hymenoptera) is one of few groups of nonpollinating fig wasps that can reproduce within both male and female figs. On the basis of the morphology and DNA barcoding, evidence from partial sequences of mitochondrial cytochrome c oxidase I and nuclear internal transcribed spacer 2, we found that there are two nonsibling Apocryptophagus species living on male and female F. auriculata figs, respectively. We estimated that these two species diverged about 19.2 million years ago. Our study suggests that the host shift from Ficus variegate or Ficus prostrata fig species to male figs is a preference way for Apocryptophagus wasps to adapt to the separation of sexual function in diecious figs. Furthermore, to escape the disadvantage or sanction impact of the host, the exploiter Apocryptophagus wasps can preferably adapt to exploiting each sex of the figs, by changing their oviposition, niche shift, and habitat.


Introduction
The coevolution between hosts and their parasites or between mutualistic partners is common phenomenon in nature (Thrall et al. 2007;Elias et al. 2008). Compared with their free-living relatives, parasites or coevolved mutualistic partners often show dramatic changes in phenotype to adapt to their hosts or the other mutualistic partners (Mcleish et al. 2010). The changes in host/one mutual partners are often accompanied by the according changes in parasites or the other partners. The pollinating wasps and nonpollinating fig wasps (NPFW) on the same host figs constitute a classic example of the mutualistic and parasitic association in plant-insect coevolution (Weiblen 2003). Diffusive coevolution might maintain stable cooperation in a fig-fig wasps system, in which NPFW might diversify their genotype, phenotype, or behavior as a result of competition with pollinator, whereas the figs change flower syconia, fruits thickness, and syconia structure (Wang et al. 2010). It has been an ideal system for addressing an array of evolutionary ecology questions including sex allocation, precise adaptation (Weiblen 2002;Cook and Rasplus 2003;Molbo et al. 2003;Herre et al. 2008).
There are about 750 known fig species in the world (Berg 1989). About half of them are functionally dioecious and the other half are monoecious. In monoecious figs, seeds, pollinators and other NPFW are all produced in one fig (Kerdelhue et al. 2000). In functionally dioecious Ficus species, male figs (also known as gall figs) have short styles and produces pollinators that disperse fig pollen.
Female figs (also known as seed figs) have longer styles that are too long for pollinator's ovipositors to reach the ovules and only produce seeds (Ganeshaiah 1995). However, in most of monoecious Ficus species, pollinator could possess oviposition in flowers with long styles (Nefdt and Compton 1996), for example, in Ficus racemosa, the spatial constraints of female flowers cannot prevent pollinators from ovipositing more eggs, and showed sufficiently negative correlation between host and pollinator when the local resource is saturated (Wang et al. 2008), so that asymmetric interaction exist between cooperative players to maintain stable cooperation (Wang et al. 2011 (Weiblen 1999).
The separation of sexual function in dioecious figs seems to have enormous advantages to the fig-fig pollina-tor's mutualism . It facilitates exclusion of many NPFW in female figs. For most NPFW, they can only occur in male figs as they are dependent on the development of pollinator's larvae as food resource or the presence of pollinator to go through the fig development barrier. Only very few groups of NPFW such as Apocryptophagus can live inside male and female figs, independent of the absence of pollinators or other NPFW (Bou cek 1988). These Apocryptophagus wasps have been shown to only produce 10 times less offspring in female figs in the lack of male figs in few fig species that have been investigated (Peng et al. 2005). Molecular phylogeny reconstruction revealed multiple transitions from monoecy to dioecy in the evolution of Ficus (Weiblen 2000;Jousselin et al. 2003). Thus, each transition from monoecy to dioecy will make female figs free of parasitism of their original NPFW. As the arm race is well known as the main theme of host-parasite interaction, is there any according change in NPFW to adapt to the separation of sexual function in dioecious figs?
In this study, we collected hundreds of Apocryptophagus wasp specimens from male and female figs of Ficus auriculata. By using the combination of morphological examination and DNA barcoding analyses, we found that there are two nonsibling Apocryptophagus species living on male and female F. auriculata figs, respectively. Our study suggests that host shift from other fig species (i.e., Ficus variegate, Ficus prostrata) to male figs is a novel way to for Apocryptophagus wasps to adapt to the changes in hosts (i.e., separation of sexual function). China. It produces one of the largest figs in this area year around, with diameter averaging around 7 cm. It is pollinated by Ceratosolen emarginatus Mayr. It also harbors NPFW in genera Sycoscapter, Philotrypesis, and Apocryptophagus. Apocryptophagus sp. wasps oviposit just before the arrival of the pollinating wasps. It induces large gall than pollinator does. Their larvae appeared to feed on proliferating nucellus. Apocryptophagus sp. wasps compete with pollinators for floral resources . Yang et al. (2008) found three Apocryptophagus species, Apocryptophagus sp. 1 can reproduce in both female and male figs of F. auriculata in Xishuangbanna forests, but with strong preference to male figs (Yang et al. 2008).
The PCR products were confirmed using 2% agarose gel, stained with ethidium bromide, and purified using an Easy Pure PCR Purification kit (TransGen Biotech, Beijing, China). Then, the purified products were cloned into pEasy-T1Vector (TransGen Biotech, Beijing, China) and 3-5 positive clones were sequenced by Biosun Sequencing Center, Beijing.

Sequences and phylogeny analyses
Sequences were eye checked in BioEdit. We also downloaded 78 COI sequences from Genbank and Barcode of Life Data Systems from 31 fig species (Table S1). Fig pollinators (Agaonidae) and NPFW from several subfamilies were considered as outgroup. We also included two genera (Sycophaga sycomori, Idarnes) of Sycophaginae. All sequences were aligned using ClustalW 1.81. The alignment of COI was confirmed by translating into amino acids in MEGA5. Bayesian inference was employed to estimate phylogenetic relationships (MrBayes 3.12). The best-fitting model of nucleotide substitution was selected in the program of jModeltest based on the Aikake information criterion (Posada 2008). Four Markov Chain Monte Carlo (MCMC) chains were run for 20 million generations and sampled every 1000 generations with the first 20% trees discarded as burn-in. Adequate mixing of the MCMC chain was determined in TRACER version 1.5 (http://tree.bio.ed.ac.uk/software/tracer). Three independent runs were carried out. Heuristic searches under parsimony were conducted with PAUP (Swofford 2002) with 1000 random addition sequence replicates, and bootstrapping with 1000 replicates. Nonparametric bootstrap (BP) value greater than 70% and posterior probability (PP) value greater than 95% were considered as strong support. Divergence time was estimated in BEAST version 1.6.1 (Drummond et al. 2002;Drummond and Rambaut 2007). The GTR+I+G substitution model was employed. The MCMC chain was run for 20 million generations sampled every 1000 generations and the first 20% trees discarded as burn-in. The uncorrelated lognormal model was used to account for rate variation among lineages. Pegoscapus fossil (30 million years ago [MYA]) was used to calibrate the date estimation (Rønsted et al. 2005).

Result Morphological examination
We collected 196 specimens from 46 figs, including 31 specimens from three female figs. We examined the morphological diversity under a Nikon SMZ80 microscope and found that seven characters of female Apocryptophagus wasps were distinct between wasps from male figs and from female figs. These characters are located in antenna, head, thorax, and wings (details are shown in Table 1 and

DNA sequence analysis
We randomly selected 46 individuals from five geographical locations for DNA barcoding analyses. Of 46 individuals, we

Phylogenetic analyses and divergence time estimation
ACI tests indicate that TIM1+G model (Àln (L) = 9533.59, K = 203, and AIC = 19473.1798) was selected as the best-fitting model for COI gene. As we expected, the Bayesian tree and maximum parsimony tree based on COI fragments showed similar topologies to previous study (Cruaud et al. 2011) about the phylogenetic relationships of three genera included (Idarnes, Sycophaga, Apocryptophagus). All Apocryptophagus wasps were not formed into a monophylogenetic group. The sequences were uploaded to TREEBASE (http://purl.org/phylo/treebase/phylows/study/TB2:S13771). However, Apocryptophagus sp. 1 and Apocryptophagus sp. 2 were clustered into a well-supported clades (PP = 1; Fig. 1) with clade II having long branch. The mean divergence between two groups is 0.226, which is much large than 0.03, a criteria for delimiting cryptic species in most animal taxa (Haine et al. 2006). Phylogenetic analyses based on ITS2 sequences also showed that Apocryptophagus sp. 1 and Apocryptophagus sp. 2 formed two distinct clades (BP = 1 and PP = 1) with 1.25 mean genetic distance between two clades (Fig. 2). Thus, DNA barcoding support that the two Apocryptophagus morphospecies are two species and that they are not sibling species. Given the Pegoscapus fossil record (30 MYA) (Rønsted et al. 2005;Lopez-Vaamonde et al. 2009) and 2.3% mtDNA pairwise divergence/Myr (Brower 1994), we roughly calibrated that two species diverged about 19.2 MYA (Fig. 3).
Within clade II (Fig. 1), the mean divergence is low (0.01). Apocryptophagus sp. on figs of F. oligodon (HM770617/JN001530) from Yunnan province was clustered with all Apocryptophagus sp. 2 specimens from Hainan province and shared high similarity (100%). Therefore, we considered them as same species even they lived on different hosts thousands miles away. Clade I consists of all Apocryptophagus sp. 1 specimens from all five geographical locations, similar with Apocryptophagus sp. on figs of F. variegate from Indonesia and F. prostrata from China, shared similarity (64%). It seems that Apocryptophagus sp. 1 clade further diverged to two groups with mean divergences between two groups being mitochondrial heterogeneity.

Discussion
Apocryptophagus, also known as Platyneura in some references, is one-six known genera in subfamily Sycophaginae. It has been shown to be paraphyletic to Sycophaga. F. congesta, and F. hispidioides) host multiple Apocryptophagus wasps. There also have two cases that one Apocryptophagus wasp attack more than one fig species. For the cases of more than one Apocryptophagus wasps living on same host species, Apocryptophagus wasps usually differed in ovipositor length and oviposition timing (i.e., before, during or after pollination) (Kerdelhue and Rasplus 1996). Species with short ovipositors lay eggs prior to pollination when figs are small in diameter, whereas species with long ovipositors lay eggs after pollination when figs are larger (Weiblen and Bush 2002). There is tight correlation with the ovipositor length with the fig size when Apocryptophagus oviposit. Evidence that multiple parasite lineages colonized the same fig species independently (Weiblen and Bush 2002).
In host-parasites interaction, hosts usually dominate this interaction and there is strong natural selection for parasites to adapt to the changes in hosts. If there are major phenotype changes in hosts make it unsuitable place for parasites, parasites either go extinction or switch to other hosts that have similar habitats and less competition (Silvieus et al. 2007;Mcleish et al. 2010). To escape the disadvantage or sanction impact of the host, the exploiter Apocryptophagus wasps can preferably adapt to exploiting each sex of the figs, by changing their oviposition, niche shift, and habitat (Wang et al. 2010). For Apocryptophagus wasps, under frame of morphology difference between male and female figs, we found the galls only closed to the ostiole of female figs; however, the galls distributed covering male figs. There was no dissimilarity in the thickness or other structures between male and female figs. Our examination suggested that spatial niche partitioning may sufficiently favor exploiters in exploiting the female resource, and there was no competition with pollinators or other parasites. Unfortunately, in our fieldwork, we did not collect the Apocryptophagus wasps on male figs in Jintang location, the galls had been an empty house without wasps information. The thickness of the fig wall and the timing of oviposition with respect to fig development appear to be traits that could facilitate a host shift. Sister group comparisons showed that there is a tendency for Apocryptophagus to shift to figs with similar wall thickness (Weiblen and Bush 2002).
Reciprocal evolution between fig and fid wasp is a typical case of diversifying coevolution, in which the interaction cause at least one of the species to become subdivided into two or more reproductively isolated populations (Thompson 1989 (Peng et al. 2005). In addition, Apocryptophagus sp. 1 only exists in the male fig in other loca- tions of Hainan and Xishuang banna arboretum, Yunnan, there is no reproduction shift to female syconia, even less male figs on a tree. Our Apocryptophagus sp. 1 wasps are different from Peng's fig wasp species in the length of ovipositor, more similar to Apocryptophagus sp. on the figs of F. variegate and F. nodosa from Indonesia. Above all, this suggests that occurrence of Apocryptophagus sp. 1 in F. auriculata male syconia is likely to be a host shift event (Cook and Segar 2010), in ecologically associations similar to the yucca-yucca moth mutualisms (Kawakita and Kato 2006).
Stability of this mutualism depends on the relative allocation of floral resources to pollen, seeds, and pollinators. Given that pollinators also eat seeds, there is potential evolutionary conflict between seed production and seed consumption (Cook and Rasplus 2003). In functionally dioecious figs, this conflict is resolved by segregating the production of seeds and pollinators in two types of figs on separate plants . Molecular phylogeny suggested that dioecy arise independently multiple times in several lineages (Weiblen 2001;Jousselin et al. 2003). A key of maintaining mutualism in dioecious figs is that female figs can regularly deceive pollinators into visiting despite the absence of any reproductive reward (Grafen and Godfray 1991). Chemical volatiles are the primary cues that attract highly species-specific pollinators species to receptive figs (Hossaert-McKey et al. 1994;Grison-Pige et al. 2002). The same chemical volatiles are used for NPFW such as Apocryptophagus to search for host figs. Thus, male and female figs should also be visited comparable amount of times by Apocryptophagus. This is true for NPFW on few fig species that have been investigated including Apocryptophagus sp. 1 (Proffit et al. 2007). However, we do not know whether Apocryptophagus sp. 2 has developed ability to distinguish female figs from male figs and only visit female F. auriculata fig. If it were, pollinator might also have a chance to develop or have developed his ability to distinguish the sexual figs. In that case, the fig-fig wasp mutualism on F. auriculata is on the eve of collapse.