An early gall-inducing parasitic wasp adversely affects the fitness of its host Ficus tree but not the pollinator

The fig tree-fig pollinator mutualism is one of the most tightly knit symbiotic systems. The research on the ecology of non-pollinators which exploit the mutualism without providing services to the host is very limited and conclusions about the role they play in the maintenance of this mutualism are full of contradictions. The non-pollinating fig wasps species are highly diverse in their feeding habit and ecological function, which may result in different consequences on the mutualism. Sycophaga testacea is an early-ovipositing galler hosted by Ficus racemosa, which is a potencial competitor to the pollinators as they use the same female flowers in the fig as their ovipositing sites. In this study, we investigate the effect of S. testacea on the production of both pollinator and fig tree with a field control experiment. Seed production is decreased significantly when the figs were parasitized, while the offspring production of the pollinator is not significantly affected, which indicates that this galler species has a harmful effect on the fitness of its host fig tree but not the pollinator. The overall development ratio of the galls is decreased significantly when the figs were parasitized, and we found that the intrinsic low development ratio of S. testacea is responsible for the decrease in the overall development ratio.

www.nature.com/scientificreports/ Diverse species of non-pollinators co-exist within a fig inflorescence, and the knowledge on biology and ecology of the non-pollinating wasp community is very limited for us 12 . NPFW species may have different effects on the reproduction of fig tree and pollinator. Some studies found that NPFWs can reduce both seed production and the number of pollinating wasps 9,13,14 , other studies found that NPFWs only reduce the number of pollinating wasps 11,[15][16][17][18] , and some studies found that NPFWs have no negative effect on the pollinating wasps 19,20 . For example, by collecting and analyzing the natural maturing fruits that contain both gall inducer species Anidarnes bicolor and pollinators, Bronstein 15 found that the number of pollinator offspring decreased significantly with the increase of the number of A. bicolor, but no reduction in seed production occured. Conchou et al. 13 analyzed the effect of large-sized NPFWs on the production of pollinators and seeds through a controlled experiment. In their study, they clarified the community structure and feeding habits of NPFWs hosted by Ficus guianensis, and found a significant negative correlation between the number of NPFW and the production of pollinators and seeds 13 . The former studies have shown that the gall inducer can have a substantial impact on the reproductive success of their host fig tree and the pollinators. The community structure for F. racemosa in South China and India is known clearly, in which two (three in India) gall inducers oviposit sequentially, following by two parasitoids of the gall inducers and a parasitoid of the pollinator [21][22][23] . Although the non-pollinator community structure of F. racemosa is well studied, the effect of each species on the fitness of F. racemosa and its pollinator remains unknown.
The NPFWs of different feeding habits may have different effects on the mutualism. In order to get a better understanding of the effects of NPFWs, we need to do more controlled accurate experiments. In this research, we studied the effect of early gall-inducing non-pollinator on the fate of this mutualism using an early-ovipositing galler species S. testacea and pollinator species Ceratosolen fusciceps that hosted by Ficus racemosa.

Results
Size of mature figs. The experiments have three different treatements. To know and compare the growth of the figs in different treatments, the size of mature figs were measured. We added the length and the 2 diameters (mm), then divided the sum by 6 to represent the radius (r) of the fig. The volume of the fig was calculated as a sphere (Volume = 4/3*π*r 3 ). The size of mature figs varied significantly among treatment (F 2,46 = 80.213, P < 0.001; Fig. 1).
The size of mature figs in which only non-pollinating fig wasp S. testacea had oviposited was much smaller than that of figs into which pollinators had been introduced. The figs into which both S. testacea and Ceratosolen had oviposited were intermediate in volume and those into which only the pollinators had been introduced were largest. Development ratio. Development ratio reflects the percentage of the galls that produce wasps successfully. Numbers of galls and wasp offsprings. For figs in which only S. testacea had oviposited, the total number of galls was much lower than in figs into which pollinators were introduced. For the treatments with pollinators, figs with S. testacea contained a significantly higher number of galls than figs without S. Testacea (t = 2.824, df = 27, P < 0.05; Fig. 3).
The wasp production shows the same trends with the total number of galls. For figs in which only S. testacea had oviposited, the total number of waps produced was much lower than in figs into which pollinators were introduced. For the treatments with pollinators, figs with S. testacea contained slightly more total wasps than figs without S. testacea, but the difference is not significant (t = 1.72, df = 27, P = 0.097; Fig. 3).   www.nature.com/scientificreports/ of development of different types of NPFWs and their exact effect on the mutualism is limited by many interaction factors 13 and non-experimental empirical studies can underestimate or fail to detect the impact of NPFWs on this obligate mutualism when overall consideration was taken with different studies of pollinator-NPFW interactions in consideration 18 . For early-ovipositing gallers, a remarkable feature of those species is that their galls usually fill the space of B-phase syconium cavities and can greatly hinder pollinator movement and oviposition, thus result in their specific consequences on the mutualism 24 . For example, Conchou et al. 13 studied Ficicola spp. which is a genus of early-ovipositing large-sized gallers hosted by F. guianensis and found a substantial negative impact of this genus on the production of both pollinators and seeds. Here, we present another case that studies the effect of an early gall inducer.
Our study has explored the effect of early-ovipositing gallers on the growth of figs (fig size of matured figs) for the first time. In our study, it turns out that early-ovipositing gallers can greatly depress the growth of the figs.  28 argued that sanctions can be modular or individual. In a modular sanction, all the offspring produced in the fruit is punished due to involvement of non-cooperative individuals. Our experiment can be a good test of this hypothesis as we can investigate accurately the development of the pollinator and the cheater (S. testacea) respectively. By comparing the theoretical value and the actual number of galls for 2Cf + St treatment, we can explore the source that leads to the decrease of the overall development ratio in the 2Cf + St treatment compared with the 2Cf treatments, thus determine if the sanction works at fig level (modular) or not. We found there is no significant difference between theoretical value and the actual number for 2Cf + St treatment, which suggests that the development ratio of the two species is independent of each other, so the lower development ratio of 2Cf + St treatment compared with 2Cf treatment was caused mainly by the low development ratio of S. testacea. For 2Cf + St treatment, the development ratio of pollinator didn't decrease, and being pollinated didn't change the development ratio of S. testacea, which means pollination didn't increase the nutrition provided to galls of. S. testacea. Our results show that the sanction is not at a fig level, which is inconsistent with the sanction to the cheaters in pollinators (pollen-free pollinators) 28,30 . Perhaps the sanction mechanism is related to the identification of the cheaters. When the host can identify the cheaters, the sanction can work at individual level where they may only sanction the offspring of cheaters, such as S. testacea in this study.  www.nature.com/scientificreports/ The production of seeds is influenced greatly by the wasp S. testacea. A former study shows that galls produced by early-ovipositing gallers often fill B-phase syconium cavities and hinder pollinator movement and oviposition 24 , which can explain the decline of the seed production. In our study, the oviposition behavior of pollinators was much less affected than the pollen spreading behavior due to the block of galls produced by S. testacea.
For the treatments with the pollinators, figs with S. testacea (2Cf + St treatment) have more galls than figs without S. testacea (2Cf treatment). This indicts that the existence of galls produced by S. testacea has little effect on the egg-laying behavior of the pollinator. Since pollinators lay their eggs in the female flowers one by one and the number of eggs to be laid is much less than the number of pollen to be spread, it's easy to understand that oviposition is less affected than pollen spreading.
Our study shows that S. testacea has an obvious negative effect on the fig tree-fig pollinator mutualism. Oviposition by S. testacea leads to a drastic decrease in seed production, thus may harm the maintenance of stability of fig tree-fig pollinator mutualism if the fig is excessively parasitized by S. testacea. We think the mechanical injury from oviposition, the galling process, and the blocking of cavity by the galls of S. testacea may be responsible for the negative consequences. For figs in which only S. testacea oviposit, the total galls are much less than figs in which pollinators were introduced. There are two possible reasons. One possible reason is that the low density of the population of S. testacea results in the low oviposition rate. Another possible reason is that the fig tree may sanction the unpollinated figs, ie, the figs in which only S. testacea oviposit. Wang et al. 30,31 found that the sanction mechanism also works on F. racemosa. Further more, they found that the sanction strength became stronger with an increase in foundresses 30 (the wasps that enter the figs to oviposit). If too many S. testacea oviposit in figs, the figs are aborted more easily due to sanction by the tree. Only figs that contain a small number of galls survive.
Although many studies 9,11,[13][14][15][16][17][18]  Treatments. We made three treatments. In the 2Cf treatment, individual figs were protected in fine mesh bags early in development to preclude parasite oviposition. When the figs became receptive, 2 pollinators were introduced into each fig. When the figs are receptive, they attract pollinators that fly around them. In this experiment, we caught the pollinators in the air with a large fine mesh bag. We then used a fine paintbrush to pick the pollinator and deposit it on the surface of the fig. The wasps rapidly find their way to the ostiole. To ensure wasp vitality, all pollinator introductions were performed in the morning. After the pollinator introduction, the figs were bagged again. When the wasp offspring emerged from the figs, wasps and figs were collected in 70% alcohol. The figs were then opened and numbers of developed galls, seeds, and insects were counted.
In the St treatment, the figs were exposed to oviposition by S. testacea but not to pollinator oviposition. In this treatment, the figs were only bagged after the period of oviposition by S. testacea so that no other fig wasp species oviposited in them. Large numbers of figs aborted in this treatment, and we checked and cleaned the dropped figs every 2 or 3 days. Male S. testacea does not cut exit holes from the fig. So, in this treatment, when the wasps were ready to emerge from their galls, we cut a small hole near the ostiole and let the wasp emerge. www.nature.com/scientificreports/ Wasps and figs were then collected in 70% alcohol. Then the figs were opened and numbers of developed galls and insects were counted. In the St + 2Cf treatment, the figs were first exposed to oviposition by S. testacea and then they were protected in bags. At fig receptivity, two pollinators were introduced per fig as in the first treatment. When the wasps emerged from the figs, wasps and figs were collected in 70% alcohol. The figs were then opened and numbers of developed galls, seeds, and insects were counted.
The length (axis length, distance from the pedicel to the bracts of the fig) and width (the longest diameter and shortest diameter of the fig) of mature figs were measured for the three treatments to estimate fig size. The following development ratio will designate the ratio between the number of wasps and the number of galls. The assumption is that empty galls correspond to galls that contained a wasp larva that died during development.
Data analysis. We use SPSS 17.0 to do all the statistics. To compare the difference of fig volume and development ratio among treatments, one-way ANOVA was used in the analysis. The experiment was carried out on two trees, so we use a GLM model to test the tree effect. No significant effect of the tree was found, so we don't display them in the results. In order to test the effect of S. testacea on the wasp production of pollinators and seed production, an independent T-test was used to compare the difference of gall number, wasp offspring number and seed number between the two treatments that included pollinators.
We use the development ratio of S. testacea in St treatment and development ratio of C. fusciceps in Cf treatment to predict the theoretical total galls for the 2Cf + St treatment (predicted gall number = C. fusciceps offspring number in 2Cf + St treatment/development ratio of C. fusciceps in Cf treatment + S. testacea offspring number in 2Cf + St treatment/development ratio of S. testacea in St treatment), and use independent T-test to compare the difference between predicted and actual values of gall number, so as to determine the actual reason why the development ratio declines with the involvement of S. testacea in 2Cf + St treatment compared with 2Cf treatment.

Data availability
Data are available from the authors on request.