Annals of Botany 111: 173–181, 2013 doi:10.1093/aob/mcs250, available online at www.aob.oxfordjournals.org

The mechanism of pollinator specificity between two sympatric fig varieties: a combination of olfactory signals and contact cues

Gang Wang1,2, Stephen G. Compton3 and Jin Chen1,* 1Key Laboratory of Tropical Forest Ecology, Xishuangbanna Tropical Botanical Garden, Chinese Academy of Sciences, Mengla, Yunnan 666303, China, 2University of Chinese Academy of Science, Beijing 100039, China and 3School of Biology, Faculty of Biological Sciences, University of Leeds, Leeds, West Yorkshire LS2 9JT, UK

* For correspondence. E-mail [email protected] Downloaded from https://academic.oup.com/aob/article/111/2/173/254661 by guest on 30 September 2021

Received: 4 April 2012 Returned for revision: 9 August 2012 Accepted: 11 October 2012 Published electronically: 23 November 2012

† Background and Aims Pollinator specificity facilitates reproductive isolation among plants, and mechanisms that generate specificity influence species boundaries. Long-range volatile attractants, in combination with mor- phological co-adaptations, are generally regarded as being responsible for maintaining extreme host specificity among the fig that pollinate fig trees, but increasing evidence for breakdowns in specificity is accumulat- ing. The basis of host specificity was examined among two host-specific Ceratosolen fig wasps that pollinate two sympatric varieties of semicordata, together with the consequences for the plants when pollinators entered the alternative host variety. † Methods The compositions of floral scents from receptive figs of the two varieties and responses of their polli- nators to these volatiles were compared. The behaviour of the wasps once on the surface of the figs was also recorded, together with the reproductive success of figs entered by the two Ceratosolen species. † Key Results The receptive-phase floral scents of the two varieties had different chemical compositions, but only one Ceratosolen species displayed a preference between them in Y-tube trials. Specificity was reinforced at a later stage, once pollinators were walking on the figs, because both species preferred to enter figs of their normal hosts. Both pollinators could enter figs of both varieties and pollinate them, but figs with extra-varietal were more likely to abort and contained fewer seeds. Hybrid seeds germinated at normal rates. † Conclusions Contact cues on the surface of figs have been largely ignored in previous studies of fig host preferences, but together with floral scents they maintain host specificity among the pollinators of sympatric F. semicordata varieties. When pollinators enter atypical hosts, post-zygotic factors reduce but do not prevent the production of hybrid offspring, suggesting there may be gene flow between these varieties.

Key words: Contact cues, Ficus semicordata, floral scents, host recognition, hybridization, , pollinator specificity, reproductive isolation.

INTRODUCTION number of species of pollinators, and thus contribute to the plants’ reproductive isolation. In the fig–fig wasp , Adaptations that favour pollination by a limited range of the volatile blends released from the figs of .40 Ficus species vectors have been a key factor in the diversification have been investigated (Ware et al., 1993; Grison et al., 1999; of flowering plants, with some of the most species-rich plant Song et al., 2001; Grison-Pige et al., 2002; Chen et al., 2009; groups characterized by extreme pollinator specificity Proffit et al., 2009; Soler et al., 2011), and their role in attracting (Darwin, 1876; Johnson, 2006; Raguso, 2008). Pollinator spe- their specific pollinators has been demonstrated in field trapping cificity is a pre-zygotic isolating mechanism that often deter- (Bronstein, 1987; van Noort et al., 1989; Ware and Compton, mines the extent of gene flow among taxa and consequently 1994b) and laboratory tests (Grison-Pige et al., 2002; Chen determines their species boundaries (Grant, 1994; Sargent, and Song, 2008; Chen et al., 2009; Proffit et al., 2009). It is 2004; Waser and Ollerton, 2006). Plants with plant-specific believed that species specificity in the fig–fig wasp system is pollinators often also display weak post-zygotic isolating generally mediated by species-specific volatile signals (van mechanisms, as in some Orchidaceae species (Schiestl and Noort et al., 1989; Chen et al., 2009; Hossaert-McKey et al., Schluter, 2009) and those genera associated with obligate 2010). Pollinator specificity that is driven primarily by plant nursery pollination mutualisms, such as Ficus (Ramirez, volatiles is also found in some other systems. Euglossine bee- 1970; Hossaert-McKey et al., 2010), Yucca (Pellmyr, 2003) pollinated neotropical orchids, and sexually deceptive orchids and Glochidion (Kato et al., 2003). Consequently, traits re- in Australia and Europe are examples (Ayasse et al., 2003; sponsible for maintaining pollinator specificity in these plant Schiestl et al., 2003; Eltz et al., 2005). Similarly, the groups with highly specific pollinators influence both repro- host-specific Epicephala moths that pollinate Breynia ductive isolation and the generation of novel diversity. vitis-idaea (Euphorbiaceae) have been shown to be attracted Specific floral scents are known to be a key element in pollin- by floral scent using both behavioural and electrophysiological ator attraction for many plants that have just one or a small methods (Svensson et al., 2010).

# The Author 2012. Published by Oxford University Press on behalf of the Annals of Botany Company. All rights reserved. For Permissions, please email: [email protected] 174 Wang et al. — Mechanism of pollinator specificity in Ficus

The one pollinator–one host Ficus relationship is not uni- Pollinator responses to volatiles have generally been moni- versal, and there are several reports of Ficus species with mul- tored using walking in Y-tube olfactometers, though tiple pollinator species (Rasplus, 1996; Kerdelhue et al., 1999; the responses of flying insects in the field have also been Molbo et al., 2003; Compton et al., 2009), or one pollinator tested (Bronstein, 1987; van Noort et al., 1989; Ware and species may be associated with more than one host taxon Compton, 1994b; Chen and Song, 2008; Chen et al., 2009; (Cornille et al., 2012). Those reports suggest that a relatively Proffit et al., 2009). Those studies have consistently detected loose form of has been occurring, at least in pollinator attraction to species-specific and developmental some lineages. Entry of pollinators into the figs of atypical stage-specific floral scents emanating from figs. However, at- Ficus hosts often results in fertile seeds, even among sympatric traction of fig wasps to their normal host does not necessarily Ficus species that are clearly distinct in terms of their ecology equate with the maintenance of absolute pollinator specificity, and morphology. This shows that post-zygotic barriers to hy- because infrequent arrivals at non-host figs have been recorded

bridization are often weak in this genus (Ramirez, 1970; under field conditions, as has the entry of pollinators into figs Downloaded from https://academic.oup.com/aob/article/111/2/173/254661 by guest on 30 September 2021 Janzen, 1979; Parrish et al., 2003; Kusumi et al., 2012). of atypical host species (Bronstein, 1987; Ware and Compton, Therefore, the strength of pollinator specificity of each fig 1994b; Moe et al., 2011). In addition to these ‘mistakes’ where wasp largely determines species boundaries of figs (Machado long-range cues have been by-passed, some fig wasps may also et al., 2005). Factors among Ficus species that are likely to in- be attracted to figs of species other than their typical hosts, as crease the likelihood of figs being visited by the ‘wrong’ fig with the pollinator of F. microcarpa, which is attracted to figs wasps include living in sympatry, co-flowering phenologies, of F. fulva as well as to those of its normal pollinator similarities in fig size and shared compounds in the floral (Grison-Pige et al., 2002). Furthermore, for those closely scents released from their receptive figs (Whitehead and related fig species that are living in sympatry, behavioural Peakall, 2009; Hossaert-McKey et al., 2010). The extent to tests of fig wasp responses to floral scent cues could supply which phylogeny determines the components of floral scents more powerful evidence to understand the mechanism of pol- among Ficus species is unclear (Hossaert-McKey et al., linator specificity. 2010), but if closely related taxa produce more similar volatile Here, we use two sympatric varieties of an Asian fig tree, blends, then pollinator ‘mistakes’ that result in genetic ex- Ficus semicordata, that are host to two different pollinators, change may be more likely to occur among sister taxa. to investigate mechanisms of pollinator specificity among Sympatric or sister taxa can provide powerful insights into sympatric closely related fig taxa. Both volatile and short- floral scent-mediated reproductive isolation (Whitehead and range contact cues were investigated and the following specific Peakall, 2009). Ideal systems for studying the consequences questions were addressed. (1) Are the floral scents emitted by of floral fragrance variation should control for other confound- receptive-phase figs of the two varieties distinct? (2) Do floral ing abiotic and biotic variables, such as geographical and tem- scents alone maintain pollinator specificity or are contact cues poral isolation, mechanical and ethological isolation and on the surface of the figs also utilized for host recognition by phylogenetic constraints (Knudsen, 1999; Schiestl and the fig wasps? (3) Do morphological barriers prevent entry into Ayasse, 2002; Okamoto et al., 2007). Whitehead and Peakall figs by the fig wasps associated with the other variety? (4) (2009) have summarized examples of studies that more or What are the reproductive consequence for the plants of less meet the requirements for sympatric, co-flowering and being pollinated by fig wasps carrying alien pollen? morphologically similar taxa, and emphasize their rarity. In the fig–fig wasp system, although co-flowering sympatric taxa are common (Berg and Corner, 2005), studies of the MATERIALS AND METHODS maintenance of pollinator specificity among sympatric Study species and sites closely related fig species rarely combine both behavioural and chemical investigations. Ficus semicordata is a functionally dioecious small free- Contact cues from the surfaces of plants or insects have a standing tree distributed in the Sino-Himalayan region, south- role in host attraction and selection among both parasitoids wards to the Malay Peninsula (Berg and Corner, 2005). It has and phytophagous insects (Vinson, 1976; Visser, 1988; large leaves with asymmetrical heart-shaped leaf bases and fla- Espelie et al., 1991; Giudice et al., 2010). Chemical and phys- gelliflorous or geocarpic figs. Molecular data suggest that its ical features on the surface of figs are thought to contribute to two varieties [Ficus semicordata Buch.-Ham. ex Sm. var. host recognition in fig wasps (Ware and Compton, 1994a; semicordata (FSS) and Ficus semicordata var. montana Gibernau et al., 1998), but their role in the maintenance of Amatya (FSM)] are sister taxa within subgenus Sycomorus, host specificity is largely unknown. Furthermore, mechanical Section Hemicardia [G. Wang, Xishuangbanna Tropical isolation for specificity in the fig–fig wasp system is also Botanical Garden (XTBG), China, unpubl. res.; Fig. 1]. The common. females need to have an appropriate two have often been recorded under the same name, so any dif- shape and size to allow passage through the ostiole into figs, ferences in their distributions are poorly understood, but the as well as an ovipositor of appropriate length for laying eggs two taxa are known to be sympatric in Nepal (Amatya, into the ovaries via the styles (Nefdt and Compton, 1996; 1996), central and northern Laos (G. Wang, XTBG, China, van Noort and Compton, 1996). Combinations of long-range pers. observ.) and south-western China [herbarium records of floral scents, short-range contact cues and physical matching the Chinese Virtual Herbarium (www.cvh.org.cn); G. Wang, may act together to help ensure host specificity among fig XTBG, China, pers. obs.]. The two varieties have similar wasps, especially among closely related fig taxa. growth forms, heights and leaf shape, but their figs and Wang et al. — Mechanism of pollinator specificity in Ficus 175

AB Pre-receptive figs of similar size on one or more leafless branches were covered with mesh bags to exclude pollinators. The figs were subsequently identified as being receptive on the basis of their loose ostiolar bracts and the presence of large numbers of pollinating wasps flying around or walking on the bagged figs. For each sample, the average number of recep- tive figs used was 57 for FSS and 355 for FSM – this reflected the difference in fig size and quantity of volatiles released by the two varieties. The volatiles they were releasing were col- lected in situ using the adsorption–desorption (dynamic head- space) technique described in Chen et al. (2009). To detect any

environmental contamination during the volatile collections, Downloaded from https://academic.oup.com/aob/article/111/2/173/254661 by guest on 30 September 2021 controls were conducted with ambient air collected at the

F IG. 1. Receptive-phase female figs of (A) F. semicordata var. semicordata same place using the same dynamic headspace technique. Any (FSS) and (B) F. semicordata var. montana (FSM). Arrows indicate the volatile compounds shared with this control treatment were ostioles, where pollinators enter the figs. Note the difference in colour removed before analyses. Two internal standards, octane and between figs of the two varieties at this developmental stage. Bars ¼ 10 mm. decyl acetate, were added into each sample, using the same method as Chen et al. (2009). The extracts were analysed in a coupled gas chromatography– habitat preferences are different (Amatya, 1996). Receptive mass spectrometry (GC-MS) system (Agilent HP6890GC/ figs of FSS are green or light brown in colour and about 5973MS) equipped with a HP-5MS column (length 30 m; 16 mm in diameter with 1611 + 18 (mean + s.e.; n ¼ 20) inner diameter 0.25 mm; film thickness 0.25 mm) with helium female flowers per female fig, whereas receptive figs of FSM as the carrier gas. Ionization was by electron impact (70 eV; are bright red, about half the diameter of receptive FSS figs source temperature 230 8C). For each sample, 2 mL was injected and the number of female flower per female fig is 730 + 26 (split with a 10:1 ratio), with the injector temperature at 250 8C. (n ¼ 20). The figs of FSM at XTBG are also typically The column temperature began at 60 8C, was increased at 3 8C almost free of surface hairs, whereas FSS figs are more pubes- min21 up to 80 8C, then up to 260 8Cat58C min21, and this tem- cent. FSS favours well-lit open areas, such as secondary perature was then maintained for 15 min. forests, whereas FSM prefers shadier habitats in or adjacent Compound identification was based on automated matching to rain forests, but it is also common to find the two varieties of the mass spectra with Wiley7n.1 libraries. Some compo- growing ,100 m apart (Amatya, 1996; G. Wang, XTBG, nents were confirmed by comparison of the retention data China, unpubl. res.). The flowering phenologies of the two with published data. The absolute amounts of all compounds fig varieties overlap, as both of them produce receptive figs were estimated using the average peak area of the two internal throughout the year. A few putative hybrid individuals with standards as a reference scale. The relative proportions of each intermediate morphological characteristics have been found compound were also calculated. in field surveys, but they are not included here. The typical form of FSS is pollinated by Ceratosolen grave- lyi Grandi (Wang et al., 2003). It is attracted to receptive-phase Behavioural bioassays figs of FSS by a simple volatile blend where one compound Adult female fig wasps were obtained by placing mature comprises .90 % of the total quantity of volatiles. This male figs into mesh bags during the afternoon before the ‘private channel’ is highly unusual and has not been recorded experiments. These emerged from the figs the following in any other Ficus species (Chen et al., 2009). The pollinator morning. Two potential components of chemosensory recogni- of FSM in China is an undescribed Ceratosolen species, which tion were examined: (1) an olfactory phase, where wind-borne has a longer ovipositor than C. gravelyi, despite having a no- volatiles could be used for long-range host localization; and ticeably smaller body. A phylogenetic tree based on COI (2) a contact stimuli phase operating once a female wasp had sequences suggests that, unlike their host figs, the sister landed on the surface of a suitable fig (Ware and Compton, species of C. gravelyi is C. emarginaus, pollinator of 1994b; Smadja and Butlin, 2009). F. auriculata complex, but not Ceratosolen sp. (host FSM) Male and female dioecious figs have broadly similar volatile (G. Wang, XTBG, China, unpubl. res.). profiles, reflecting intersexual chemical mimicry (Chen et al., Our experimental studies were carried out using trees (and 2009; Hossaert-McKey et al., 2010), but receptive female figs their associated wasps) that grow naturally in or around the ′ ′ were used in most of the olfactory tests and all the contact XTBG (21841 N, 101825 E)], Chinese Academy of Science stimuli tests. (CAS), Yunnan Province, south-western China. Collection of volatiles, behavioural tests and other experiments were con- ducted from April to September 2011. Olfactory tests These were carried out using Y-tube olfactometry. The equipment and procedures were as described in Chen et al. Collection and tentative identification of volatiles (2009). The responses of C. gravelyi and Ceratosolen sp. to Volatiles were collected from six FSS (two male and four three scent combinations were examined: receptive host figs female) and seven FSM (three male and four female) trees. vs. air, atypical host figs vs. air, and host figs vs. atypical 176 Wang et al. — Mechanism of pollinator specificity in Ficus host figs. The first two combinations simulated conditions Data analysis when receptive figs of only one variety were available Analyses were mainly performed using R version 2.12.0(R locally, which is likely to be the more common situation in Development Core Team, 2011). Non-metric multidimension- nature, and the third where volatiles from receptive figs of al scaling (NMDS) ordination using the Vegan package both varieties were being perceived simultaneously, as might (Oksanen et al., 2011) was used to compare scent composition occur at ecotones when the two varieties are flowering in between the two fig varieties. A multiple response permutation close proximity. Because of their difference in size and the procedure (MRPP; Vegan package) was used in a matrix of numbers of flowers they contain, volatiles from one receptive mean dissimilarities with 999 permutations to test the null hy- FSS fig and six FSM figs were used in the Y-tube olfactom- pothesis that there was no difference in the scent profiles. The etry. Experiments were performed between 0900 and 1100 h relative proportions of all the compounds recorded from the within 3 h of the figs being removed from the trees, and the seven FSM and six FSS samples were included in the multi-

figs were replaced with new ones every hour. Each experiment Downloaded from https://academic.oup.com/aob/article/111/2/173/254661 by guest on 30 September 2021 variate analyses. (comprising one scent combination for each wasp species) was x2 and binominal tests were used to determine whether fig replicated 23–41 times, with no more than five females from wasps were attracted to one of two odour sources and to deter- any one male fig utilized. mine fig wasp entry preferences after being placed on the surface of figs. Wilcoxon rank-sum tests were used to Contact stimuli on the surface of figs compare the total absolute quantities of scent emitted by the two fig varieties and the time spent by fig wasps on the fig sur- The fig surface experiments were conducted on receptive faces. Seed production was compared using t-tests. figs in situ between 0900 and 1200 h. Individual newly emerged female C. gravelyi and Ceratosolen sp. were intro- duced separately onto the surface of receptive figs of their RESULTS own or the atypical host. Each wasp was used only once. Variation in floral scents Their behaviour was observed; the number of wasps which entered figs and wasps which left after walking around on The absolute quantity of scent emitted over the course of 1 h the fig surface for .10 s or after trying to enter but failing by a single FSS fig {mean + s.d. ¼ 37.72 + 22.87 ng (n ¼ 6) to do so were recorded as either ‘Entered’ and ‘Failed to [56.04 + 35.23 ng (n ¼ 2), 28.56 + 11.06 ng (n ¼ 4) for enter’, respectively. The time from the wasp arriving in the male and female figs, respectively]} was .100 times higher ostiole for the first time to the point when its entire body than that from the smaller FSM figs {0.34 + 0.13 ng (n ¼ 7) had disappeared inside the ostiolar bracts was also recorded. [0.38 + 0.21 ng (n ¼ 3) and 0.32 + 0.03 ng (n ¼ 4) for male Wasps that spent ,10 s on the fig surface before leaving or and female, respectively]}. A total of 35 different compounds, flying away were considered to have displayed ‘no choice’ including fatty acid derivatives, terpenoids and shikimic com- and were excluded from the analyses. pounds, were found in the scents emitted by the receptive figs of the two varieties (see Supplementary Data Table S1). As has been recorded previously, the shikimic compound, 4- Artificial pollination and seed germination methylanisole, contributed almost 98 % (94.48–99.55 %) of Cross-pollination and subsequent germination experiments the total quantity of volatiles released from receptive FSS were performed to assess the consequences for the trees of figs. A further ten compounds were present in small quantities being pollinated by atypical wasp species carrying pollen (male and female figs combined). This dominance of a single from a different variety of F. semicordata. Pre-receptive figs component contrasts with the more diverse floral scent of on female trees were enclosed in mesh bags to prevent FSM, which included 31 compounds (male and female figs natural pollinator entry. Once the figs were receptive, two re- combined), of which nine, representing 8.29 % of the total cently emerged adult females of either C. gravelyi or scent emitted, remain unidentified. Of the 30 compounds, 18 Ceratosolen sp. were introduced into figs of FSS, and single were recorded from both sexes (76.47 % of the total quantity individuals of each species into the smaller figs of FSM. The of volatiles), with five compounds only recorded from male figs were located on two trees of each variety. The pollinators figs and eight only from female figs. Five compounds were placed separately on the surface of the figs and allowed to (b-caryophyllene, a-caryophyllene, 4-methylanisole, 1,8- enter unaided. The figs were then bagged again to prevent cineole and d-limonene), each representing .5 % of the entry by further pollinators. The figs matured 50–65 d later, total quantity, together constituted about 67.4 % of the total and their contents were then recorded. Seeds belonging to scent emitted from FSM figs. 4-Methylanisole was present in each of four types from two female trees were mixed together all three samples from male figs, but with very large variation separately; groups of 300 normal-looking seeds were chosen at in its relative percentage (5.47, 17.46, and 66.48 %), and it was random for germination trials. Germination rates of the seeds not detected at all in the four scent profiles obtained from were compared in Petri dishes lined with moistened filter female figs. Seven of the 11 compounds from FSS figs were paper in a continuously illuminated incubator maintained at shared with FSM figs, but six of these were in trace amounts 30 8C. The trial was repeated three times for each of the four and together represented ,1 % of the total quantity of seed sources (two varieties, each with their own or heterovar- volatiles. ietal pollen), with each replicate consisting of 100 seeds in a An ordination (NMDS with Bray–Curtis distances) of the single Petri dish. Germination was recorded daily until no 13 volatile samples illustrated the contrast between the volatile further seeds had germinated for 5 d. blends of the two varieties (Fig. 2). The overall scent profiles Wang et al. — Mechanism of pollinator specificity in Ficus 177 of the receptive figs differed significantly between varieties alternative, but in contrast to Ceratosolen sp. they showed a (MRPP, A ¼ 0.6314, P ¼ 0.002). strong preference for their normal host figs (FSS) when given a choice between these and receptive figs of FSM (Fig. 3). Y-tube experiments Female Ceratosolen sp. were attracted to receptive female Contact stimuli tests figs of both FSM (its regular host) and FSS when given a choice between them and clean air, but displayed no prefer- The searching behaviour of female Ceratosolen sp. on re- ence between the scents of receptive figs of the two varieties ceptive figs mainly involved walking across the surface (Fig. 3). Ceratosolen gravelyi females were also attracted to re- while keeping the head lower than the body. The antennae ceptive figs of both varieties when clean air was the were swept across the fig surface, while keeping their tips in

contact with the fig surface. Ceratosolen gravelyi females Downloaded from https://academic.oup.com/aob/article/111/2/173/254661 by guest on 30 September 2021 tended to walk more quickly and often kept their heads further from the substrate, and there was noticeably less anten- 1·0 Mf1 nal contact with the fig surface. These differences in behaviour on the general surface of the figs were consistent, irrespective of the fig variety. Once the ostiole was encountered, both species pressed about half the length of their antennae 0·5 against the outer ostiolar bracts, before deciding whether to Sf1 enter or not. Ceratosolen sp. females were more likely to resume walking and then return to the ostiole later than were NMDS2 Mm1 Sf4 Mf3 C. gravelyi females, which tended to enter the ostiole on the 0·0 Mf2 Sm2 first occasion they came into contact. Sm1 Sf2 Mm3 When C. gravelyi and Ceratosolen sp. females were placed Mf4 Sf3 on the surface of receptive figs, they were significantly more likely to enter their normal host figs than those of the alterna- –0·5 Mm2 tive variety (Fig. 4A). Some individuals of each species none- theless entered atypical host figs (34 % of C. gravelyi on FSM –1·0 –0·5 0·0 0·5 1·0 figs and 17 % of Ceratosolen sp. on FSS figs, compared with NMDS1 92 and 100 % entry when the wasps were placed on their normal hosts). There was a non-significant tendency for a F IG. 2. Non-metric multidimensional scaling (NMDS) ordination of scent profiles of receptive phase figs of F. semicordata var. montana (M, n ¼ 7), higher proportion of C. gravelyi than Ceratosolen sp. and F. semicordata var. semicordata (S, n ¼ 6). Female and male figs are indi- females to be willing to enter an atypical host once placed cated by f and m. Stress ¼ 0.057. on the fig surface (x2 with Yates’ correction ¼ 3.2, d.f. ¼ 1,

Choice for: Wasp tested Odour 1 Odour 2 P NC FSM 16 18 FSS ns 8 Ceratosolen sp. Air (host FSM) FSM 26 2 *** 7

Air 3 20 FSS *** 6

FSM 1 40 FSS *** 9

FSM 2 22 FSS *** 7 C.gravelyi (host FSS) Air 0 21 FSS *** 1

FSM 32 4 Air *** 29

100 50 0 50 100 Proportion of wasps choosing either odour

F IG. 3. Behavioural responses of female Ceratosolen wasps to olfactory stimuli in Y-tube tests. Numbers in or beside bars indicate the absolute numbers of wasps that chose each odour. Wasps that did not respond within 5 min (NC) were excluded from the statistical analysis. (x2 tests: ns, P . 0.05, ***, P , 0.001). FSS ¼ receptive figs of Ficus semicordata var. semicordata; FSM ¼ receptive figs of F. semicordata var. montana; Air ¼ purified air. Male and female figs are indicated as ‘F’ and ‘C’. 178 Wang et al. — Mechanism of pollinator specificity in Ficus

A Table 1). Those figs that completed their development also contained more seeds if they had been entered by their 100 usual pollinator (t-tests, FSS female figs, t ¼ –2.6, d.f. ¼ 44, P ¼ 0.01; FSM female figs, t ¼ –3.2, d.f. ¼ 46, P ¼ 0.003). Seed germination rates were uniformly high, irrespective of 50 the male parents (Table 1).

22 56 929 0 DISCUSSION 43 5 43 0 Ficus semicordata var. semicordata and F. semicordata var. montana are two very closely related and frequently sympatric 50 taxa that are pollinated by two different species of Ceratosolen Downloaded from https://academic.oup.com/aob/article/111/2/173/254661 by guest on 30 September 2021 * *** *** *** fig wasps. Their receptive figs release floral scents that contain Failed to enterFailed Entered different combinations of volatiles, but both scents are attract- 100 ive to both pollinators, and one species showed no preference B for volatiles released from figs of its normal host over those 1200 released by the other variety. These results contrast with the c generally accepted basis for pollinator specificity in the fig– 1000 fig wasp mutualism, that volatiles released from receptive figs are long-range olfactory cues that attract only the pollina- 800 tors associated with that particular species of fig tree. Short-range cues have largely been ignored in previous 600 studies of pollinator specificity, but the two pollinators of the d F. semicordata varieties displayed clear preferences for their 200 usual hosts when they were placed on the surfaces of receptive figs, and these contact cues, rather than long-range volatile a attractants, may be more critical in determining which pollina- Time before entry Time before into figs (s) b tors enter the figs. Some pollinators were nonetheless willing 2256 9 29 to enter the ‘wrong’ figs and, once inside, their pollination 0 resulted in the production of viable hybrid seeds, suggesting FSM FSM FSS FSS that gene flow between the two varieties of F. semicordata C. gravelyi Ceratosolen sp. could take place where plants are growing together. Hybrid

F IG. 4. Behavioural responses of female Ceratosolen placed on the surface of viability was not assessed beyond the seedling stage, but receptive figs. (A) The willingness of female fig wasps to enter receptive figs of there is clearly the potential for introgression between the normal or atypical hosts after they had been placed on the surface of the figs. two varieties of F. semicordata. Shaded bars indicate the wasps that entered the figs; open bars the wasps that Molecular evidence suggests that the two varieties of did not. Sample sizes (n wasps) are also indicated. (B) The time spent by female fig wasps on the surface of receptive figs of the two varieties of F. semicordata are sister taxa, yet their volatile profiles are F. semicordata, measured from when the wasps first arrived at the ostiole to very different. The diverse scent of FSM is typical of figs in when their entire body had entered the figs (mean + s.e.). Note that after general (Hossaert-McKey et al., 2010), whereas the very coming into contact with the ostiole for the first time, some females wandered simple composition of the FSS bouquet is rare and possibly away before then returning and attempting entry. Shaded bars indicate atypical host figs, open bars the normal host figs. Sample sizes (n wasps) are also indi- unique among Ficus species (Chen et al., 2009). However, cated. (A) x2 tests (*P , 0.05; ***P , 0.001) and (B) Wilcoxon rank-sum only a small proportion of the volatile compounds that make tests, with different letters indicating significant differences. FSS ¼ Ficus up the scent of FSM may be attractive to its Ceratosolen sp. semicordata var. semicordata, FSM ¼ F. semicordata var. montana. pollinator. 4-Methylanisole, which contributes the vast major- ity of the volatiles released from FSS figs, is clearly not essen- tial for attraction of Ceratolsolen sp. to FSM figs, because not P ¼ 0.07). Ceratosolen gravelyi females spent less time on the all FSM figs release it. Furthermore, the scent of FSM figs was surface of the figs than Ceratosolen sp. irrespective of whether attractive to C. gravelyi females, despite the rarity or absence they were on their normal or atypical host figs (Fig. 4B). of 4-methylanisole, so this pollinator will clearly respond to other volatiles. Similarly, the simple FSS blend also attracted Ceratosolen sp. females. Some compounds are components of both floral scents (b-caryophyllene, a-caryophyllene, Pollination and germination 4-methylanisole and a-copaene) and these may be responsible Both species of pollinators were able to enter figs of both for the figs of the two varieties being attractive to both species varieties of F. semicordata, though some individuals of of pollinators. C. gravelyi became trapped in the ostioles of FSM figs. Y-tube olfactometry is widely used to investigate pollinator Female figs were more likely to abort if they had been attraction to floral scents (Okamoto et al., 2007; Chen and entered by wasps carrying pollen from the other variety of Song, 2008; Chen et al., 2009; Proffit et al., 2009; Svensson F. semicordata (FSS female figs, x2 ¼ 12.9, d.f. ¼ 1, P , et al., 2010, 2011), but it is clearly a poor model for host 0.001; FSM female figs, x2 ¼ 11.5, d.f. ¼ 1, P , 0.001; choice under natural situations – the insects are usually Wang et al. — Mechanism of pollinator specificity in Ficus 179

TABLE 1. Intra- and inter-varietal crosses between F. semicordata var. semicordata (FSS) and F. semicordata var. montana (FSM)

Fig retention Seed production Seed germination

Female parent Male parent No. of figs Aborted figs (%) No. of figs Seeds (mean + s.e.) No. of seeds % (mean and range)

FSS FSS 40 20.0 32 1217 + 135.1 300 99.7 (99–100 ) FSM 28 50.0 14 658 + 104.3 300 98.7 (95–99) FSM FSM 30 20.0 24 279 + 26.6 300 99.0 (98–100 ) FSS 56 48.0 24 191 + 7.4 300 97.0 (98–100 )

One pollinator individual was introduced into each FSM fig and two individuals into each FSS fig. Downloaded from https://academic.oup.com/aob/article/111/2/173/254661 by guest on 30 September 2021 walking rather than flying, and concentrations of volatiles may rate once pollinators are on the fig surface and weak post- be unnaturally high. Trapping of fig wasps under more natural zygotic isolation (at least up to the seedling stage) will facili- field conditions has nonetheless given similar results to Y-tube tate gene flow between the two varieties of F. semicordata. experiments, and supported the conclusion that pollinator- Unidirectional hybrids resulting from Ceratosolen sp. (with specific responses to host-specific volatiles are largely respon- FSM as its normal host) visiting FSS figs are particularly sible for maintaining fig wasp host specificity (Bronstein, likely. Ongoing genetic studies will test this assumption. An 1987; van Noort et al., 1989; Ware and Compton, 1994b). understanding of how the two varieties maintain their iden- However, small numbers of unexpected fig wasp species tities will also help assess the likely role of hybridization reported in the above studies can nonetheless be trapped and introgression in Ficus speciation and diversification around receptive trees, and this may not necessarily reflect (Machado et al., 2005; Whitehead and Peakall, 2009). only chance encounters. In F. semicordata, the marked differ- In conclusion, our study shows that both long-range volatile ences in floral scents between the two varieties do not appear attractants and surface cues contribute to host specificity in to be sufficient to prevent both Ceratosolen species being two closely related and frequently sympatric species of fig attracted to both varieties. This result conflicts with previous trees. Species-specific floral scents, the major mechanism for assumptions that specific floral scent alone can maintain pollinator isolation in Ficus (Whitehead and Peakall, 2009; Ficus pollinator specificity (van Noort et al., 1989; Hossaert-McKey et al., 2010), were unable on their own to Grison-Pige et al., 2002; Chen et al., 2009; Hossaert-McKey maintain the species-specific relationship between the fig et al., 2010), a difference that may reflect the paucity of com- wasps and their hosts. Contact stimuli from the surface of parisons between very closely related fig trees. the figs played a complementary role in host recognition, but In the absence of clear-cut differences in responses to host failed to ensure that pollinators only entered their typical fig volatiles, contact cues from receptive figs appear to help host figs. Putative weak post-zygotic isolation between these the two pollinator species distinguish between the figs of figs species suggests that hybrid offspring production is FSS and FSM. Both physical (the fig surfaces differ in hairi- likely. The extent to which fig surface characters are important ness) and tactile chemical cues may be involved (Vinson, for host recognition among other species of fig wasps remains 1976). As with other pollinating fig wasps (Ware and to be determined. Compton, 1994a; Gibernau et al., 1998), females of both Ceratosolen species repeatedly tapped the surface of the figs with their antennae as they walked around, but their responses SUPPLEMENTARY DATA to cues emanating from the ostioles appeared to be more de- Supplementary data are available online at www.aob.oxford- cisive than those from the fig surface, with both species journals.org and consist of Table S1: relative abundances much more likely to avoid entry if the ostiole belonged to an and occurrence of the volatile compounds found in floral atypical host variety. Although the relative significance of scents emitted by figs of two varieties of F. semicordata at visual, chemical and physical cues was not examined, their receptive phase. contact chemical cues may be particularly important, based on studies of both parasitoids and phytophagous insects in general (Vinson, 1976; Visser, 1988; Espelie et al., 1991; ACKNOWLEDGEMENTS Giudice et al., 2010). Additionally, Gibernau et al. (1998) We thank Zhen Yu of the Kunming Institute of Botany, CAS showed that extracts of whole receptive figs of F. carica are at- for help with GC-MS analyses, Meng Li of the XTBG, CAS tractive to its pollinator ( psenes), but these for help with field experiments, and Zi Li of the Institute of extracts may have contained compounds from the fig surface Zoology, CAS for help with fig wasp identification. We also that elicit behavioural responses, as well as those responsible thank two anonymous referees for their constructive comments for long-distance attraction to the figs. and interest in our study. There is increasing evidence of hybridization and introgres- sion among fig trees (Parrish et al., 2003; Machado et al., 2005; Renoult et al., 2009; Kusumi et al., 2012). Sympatry, LITERATURE CITED similar flowering phenologies, a lack of pollinator specificity Amatya SM. 1996. A new variety of Ficus semicordata (Moraceae) from in response to long-range olfactory cues, a high misdirection Nepal. Novon 6: 323–324. 180 Wang et al. — Mechanism of pollinator specificity in Ficus

Ayasse M, Schiestl FP, Paulus HF, Ibarra F, Francke W. 2003. Pollinator Moe AM, Rossi DR, Weiblen GD. 2011. Pollinator sharing in dioecious figs attraction in a sexually deceptive orchid by means of unconventional che- (Ficus: Moraceae). Biological Journal of the Linnean Society 103: micals. Proceedings of the Royal Society B: Biological Sciences 270: 546–558. 517–522. Molbo D, Machado CA, Sevenster JG, Keller L, Herre EA. 2003. Cryptic Berg CC, Corner EJH. 2005. Moraceae (Ficus). In: Nooteboom HP. ed. species of fig-pollinating wasps: implications for the evolution of the fig– Flora Malesiana Series I – seed plants. Leiden: Nationaal Herbarium wasp mutualism, sex allocation, and precision of adaptation. Proceedings Nederland. of the National Academy of Sciences, USA 100: 5867–5872. Bronstein JL. 1987. Maintenance of species-specificity in a neotropical fig– Nefdt RJC, Compton SG. 1996. Regulation of seed and pollinator produc- pollinator wasp mutualism. Oikos 48: 39–46. tion in the fig–fig wasp mutualism. Journal of Animal Ecology 65: Chen C, Song QS. 2008. Responses of the pollinating wasp Ceratosolen 170–182. solmsi marchali to odor variation between two floral stages of Ficus van Noort S, Compton SG. 1996. Convergent evolution of agaonine and hispida. Journal of Chemical Ecology 34: 1536–1544. sycoecine (, Chalcidoidea) head shape in response to the Chen C, Song QS, Proffit M, Bessiere JM, Li ZB, Hossaert-McKey M. constraints of host fig morphology. Journal of Biogeography 23: 2009. Private channel: a single unusual compound assures specific pollin- 415–424. Downloaded from https://academic.oup.com/aob/article/111/2/173/254661 by guest on 30 September 2021 ator attraction in Ficus semicordata. Functional Ecology 23: 941–950. van Noort S, Ware AB, Compton SG. 1989. Pollinator-specific volatile Compton SG, Grehan K, van Noort S. 2009. A fig crop pollinated by three or attractants released from the figs of Ficus burtt-davyi. South African more species of agaonid fig wasps. African Entomology 17: 215–222. Journal of Science 85: 323–324. Cornille A, Underhill JG, Cruaud A, et al. 2012. Floral volatiles, pollinator Okamoto T, Kawakita A, Kato M. 2007. Interspecific variation of floral scent sharing and diversification in the fig–wasp mutualism: insights from composition in Glochidion and its association with host-specific pollinat- Ficus natalensis, and its two wasp pollinators (South Africa). ing seed parasite (Epicephala). Journal of Chemical Ecology 33: Proceedings of the Royal Society B: Biological Sciences 279: 1065–1081. 1731–1739. Oksanen J, Blanchet FG, Kindt R, et al. 2011. vegan: Community Ecology . Darwin C. 1876. The effects of cross-and self-fertilization in the vegetable Package. R package version 2 0–2. http://CRAN.R-project.org/ kingdom. London: John Murray. package=vegan. Eltz T, Roubik D, Lunau K. 2005. Experience-dependent choices ensure Parrish TL, Koelewijn HP, van Dijk PJ, Kruijt M. 2003. Genetic evidence species-specific fragrance accumulation in male orchid bees. Behavioral for natural hybridization between species of dioecious Ficus on island Ecology and Sociobiology 59: 149–156. populations. Biotropica 35: 333–343. Espelie KE, Bernays EA, Brown JJ. 1991. Plant and cuticular lipids Pellmyr O. 2003. Yuccas, yucca moths, and coevolution: a review. Annals of serve as behavioral cues for insects. Archives of Insect Biochemistry the Missouri Botanical Garden 90: 35–55. and Physiology 17: 223–233. Proffit M, Chen C, Soler C, Bessiere JM, Schatz B, Hossaert-McKey M. 2009. Can chemical signals, responsible for mutualistic partner encounter, Gibernau M, Hossaert-McKey M, Frey J, Kjellberg F. 1998. Are olfactory promote the specific exploitation of nursery pollination mutualisms? The signals sufficient to attract fig pollinators? Ecoscience, 5: 306–311. case of figs and fig wasps. Entomologia Experimentalis et Applicata, 131: Giudice DL, Peri E, Bue ML, Colazza S. 2010. Plant surfaces of vegetable 46–57. crops mediate interactions between chemical footprints of true bugs and R Development Core Team. 2011. R: a language and environment for stat- their egg parasitoids. Communicative and Integrative Biology 3: 70–74. istical computing. R Foundation for Statistical Computing, Vienna, Grant V. 1994. Modes and origins of mechanical and ethological isolation in Austria. http://www.R-project.org/. angiosperms. Proceedings of the National Academy of Sciences, USA 91: Raguso RA. 2008. Wake up and smell the roses: the ecology and evolution of 3–10. floral scent. Annual Review of Ecology, Evolution and Systematics 39: Grison L, Edwards AA, Hossaert-McKey M. 1999. Interspecies variation in 549–569. floral fragrances emitted by tropical Ficus species. Phytochemistry 52: Ramirez W. 1970. Host specificity of fig wasps (Agaonidae). Evolution 24: 1293–1299. 680–691. Grison-Pige L, Bessiere JM, Hossaert-McKey M. 2002. Specific attraction Rasplus JY. 1996. The one-to-one species specificity of the Ficus–Agaoninae of fig-pollinating wasps: role of volatile compounds released by tropical mutualism: how casual? In: van der Maesen LJ, van der Burgt XM, van figs. Journal of Chemical Ecology 28: 283–295. Medenbach de Rooy JM. eds. Biodiversity of African plants. Dordrecht: Hossaert-McKey M, Soler C, Schatz B, Proffit M. 2010. Floral scents: their Kluwer Academic Publishers, 639–649. roles in nursery pollination mutualisms. Chemoecology 20: 75–88. Renoult JP, Kjellberg F, Grout C, Santoni S, Khadari B. 2009. Janzen DH. 1979. How to be a fig. Annual Review of Ecology and Systematics Cyto-nuclear discordance in the phylogeny of Ficus section 10: 13–51. Galoglychia and host shifts in plant–pollinator associations. BMC Johnson SD. 2006. Pollinator-driven speciation in plants. In: Harder LD, Evolutionary Biology 9: 248. http://dx.doi.org/10.1186/1471-2148-9-248. Barrett SCH. eds. The ecology and evolution of flowers. New York: Sargent RD. 2004. Floral symmetry affects speciation rates in angiosperms. Oxford University Press, 295–310. Proceedings of the Royal Society B: Biological Sciences 271: 603–608. Kato M, Takimura A, Kawakita A. 2003. An obligate pollination mutualism Schiestl FP, Ayasse M. 2002. Do changes in floral odor cause speciation in and reciprocal diversification in the tree genus Glochidion sexually deceptive orchids? Plant Systematics and Evolution 234: (Euphorbiaceae). Proceedings of the National Academy of Sciences, 111–119. USA 100: 5264–5267. Schiestl FP, Schluter PM. 2009. Floral isolation, specialized pollination, and Kerdelhue C, Le Clainche I, Rasplus JY. 1999. Molecular phylogeny of the pollinator behavior in orchids. Annual Review of Entomology 54: Ceratosolen species pollinating Ficus of the subgenus Sycomorus sensu 425–446. stricto: biogeographical history and origins of the species-specificity Schiestl FP, Peakall R, Mant JG, et al. 2003. The chemistry of sexual decep- breakdown cases. Molecular Phylogenetics and Evolution 11: 401–414. tion in an orchid–wasp pollination system. Science 302: 437–438. Knudsen JT. 1999. Floral scent differentiation among coflowering, sympatric Smadja C, Butlin RK. 2009. On the scent of speciation: the chemosensory species of Geonoma (Arecaceae). Plant Species Biology 14: 137–142. system and its role in premating isolation. Heredity 102: 77–97. Kusumi J, Azuma H, Tzeng H-Y, et al. 2012. Phylogenetic analyses suggest Soler C, Hossaert-McKey M, Buatois B, Bessiere JM, Schatz B, Proffit M. a hybrid origin of the figs (Moraceae: Ficus) that are endemic to the 2011. Geographic variation of floral scent in a highly specialized pollin- Ogasawara (Bonin) Islands, Japan. Molecular Phylogenetics and ation mutualism. Phytochemistry 72: 74–81. Evolution 63: 168–79. Song QS, Yang DR, Zhang GM, Yang CR. 2001. Volatiles from Ficus Machado CA, Robbins N, Gilbert MTP, Herre EA. 2005. Critical review of hispida and their attractiveness to fig wasps. Journal of Chemical host specificity and its coevolutionary implications in the fig/fig-wasp mu- Ecology 27: 1929–1942. tualism. Proceedings of the National Academy of Sciences, USA 102: Svensson GP, Okamoto T, Kawakita A, Goto R, Kato M. 2010. Chemical 6558–6565. ecology of obligate pollination mutualisms: testing the ‘private channel’ Wang et al. — Mechanism of pollinator specificity in Ficus 181

hypothesis in the Breynia–Epicephala association. New Phytologist 186: Ware AB, Compton SG. 1994a. Dispersal of adult female fig wasps. 1. 995–1004. Arrivals and departures. Entomologia Experimentalis et Applicata 73: Svensson GP, Pellmyr O, Raguso RA. 2011. Pollinator attraction to volatiles 221–229. from virgin and pollinated host flowers in a yucca/moth obligate mutual- Ware AB, Compton SG. 1994b. Responses of fig wasp to host-plant volatile ism. Oikos 120: 1577–1583. cues. Journal of Chemical Ecology 20: 785–802. Vinson SB. 1976. Host selection by insect parasitoids. Annual Review of Ware A, Kaye P, Compton S, Noort S. 1993. Fig volatiles: their role in Entomology 21: 109–133. attracting pollinators and maintaining pollinator specificity. Plant Visser JH. 1988. Host-plant finding by insects – orientation, sensory input Systematics and Evolution 186: 147–156. and search patterns. Journal of Insect Physiology 34: 259–268. Waser NM, Ollerton J. 2006. Plant–pollinator interactions: from specializa- Wang QY, Yang DR, Peng YQ. 2003. Pollination behaviour and propagation tion to generalization. Chicago: University of Chicago Press. of pollinator wasps on Ficus semicordata in Xishuangbanna, China. Acta Whitehead MR, Peakall R. 2009. Integrating floral scent, pollination ecology Entomologica Sinica 46: 27–34. and population genetics. Functional Ecology 23: 863–874. Downloaded from https://academic.oup.com/aob/article/111/2/173/254661 by guest on 30 September 2021