SEARCH

SEARCH BY CITATION

Keywords:

  • bioassay;
  • Breynia vitis-idaea;
  • coevolution;
  • Epicephala;
  • floral scent;
  • gas chromatography with electroantennographic detection (GC-EAD);
  • mass spectrometry;
  • pollination

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • Obligate mutualisms involving actively pollinating seed predators are among the most remarkable insect–plant relationships known, yet almost nothing is known about the chemistry of pollinator attraction in these systems. The extreme species specificity observed in these mutualisms may be maintained by specific chemical compounds through ‘private channels’. Here, we tested this hypothesis using the monoecious Breynia vitis-idaea and its host-specific Epicephala pollinator as a model.
  • Headspace samples were collected from both male and female flowers of the host. Gas chromatography with electroantennographic detection (GC-EAD), coupled gas chromatography–mass spectrometry, and olfactometer bioassays were used to identify the floral compounds acting as the pollinator attractant.
  • Male and female flowers of B. vitis-idaea produced similar sets of general floral compounds, but in different ratios, and male flowers emitted significantly more scent than female flowers. A mixture of 2-phenylethyl alcohol and 2-phenylacetonitrile, the two most abundant compounds in male flowers, was as attractive to female moths as the male flower sample, although the individual compounds were slightly less attractive when tested separately.
  • Data on the floral scent signals mediating obligate mutualisms involving active pollination are still very limited. We show that system-specific chemistry is not necessary for efficient host location by exclusive pollinators in these tightly coevolved mutualisms.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Obligate pollination mutualisms, where seed predators and their host plants trade pollination services for the rearing of larvae, are among the most specialized insect–plant interactions known. Classical examples of such ‘nursery’ pollination systems (sensuDufaÿ & Ansett, 2003) are those involving yucca moths (Pellmyr, 2003) and fig wasps (Weiblen, 2002), and several analogous associations have been discovered more recently, including Chiastocheta flies on globeflower Trollius europaeus (Pellmyr, 1989), the moth Upiga virescens on the senita cactus Lophocereus schottii (Fleming & Holland, 1998) and Epicephala moths on Phyllanthaceae plants (Kato et al., 2003; Kawakita & Kato, 2004a,b). Although these interactions have served as important focal systems for studies on the origin, stability, and reversal of mutualisms (Pellmyr & Huth, 1994; Pellmyr et al., 1996; Pellmyr & Leebens-Mack, 1999; Holland et al., 2002; Rønsted et al., 2005; Kawakita & Kato, 2009), as well as analyses of reciprocal diversification among interacting lineages within mutualisms (Kawakita et al., 2004; Machado et al., 2005), data on the specific sensory cues guiding obligate pollinators to their hosts are still very scarce. The floral scent has been chemically characterized for several taxa of Ficus (Grison et al., 1999; Grison-Pigéet al., 2002b), Yucca (Svensson et al., 2005, 2006), and Glochidion (Okamoto et al., 2007), and olfactory-based host attraction has been confirmed for both fig wasps (Hossaert-McKey et al., 1994; Ware & Compton, 1994; Gibernau et al., 1998; Song et al., 2001; Grison-Pigéet al., 2002a) and Epicephala moths (Okamoto et al., 2007), but the pollinator attractant has only been chemically identified for one such association (Chen et al., 2009).

Highly species-specific mutualistic or antagonistic interactions between plants and pollinators, or between fungi and spore dispersers, have been suggested to be mediated by a few, system-specific compounds through ‘private channels’ (e.g. Raguso, 2003; Steinebrunner et al., 2008), but to date few data are available with which to test this prediction. Unique compounds have indeed been documented as pollinator attractants in sexually deceptive orchids, for example one Ophrys species (Ayasse et al., 2003) and several Chiloglottis species (Schiestl et al., 2003; Franke et al., 2009), as well as a spore disperser attractant in Epichloë fungi (Steinebrunner et al., 2008). However, most other Ophrys and insect-attracting fungi use blends of ‘conventional’ compounds frequently occurring in flowering plants to attract respective pollinators and spore dispersers (Raguso & Roy, 1998; Schiestl et al., 1999). Thus, more studies are needed on the chemical ecology of plant–pollinator interactions to test whether a higher degree of specialization in such interactions is correlated with a more specific floral scent chemistry of the pollinator-attracting signal.

We tested the hypothesis that a ‘private channel’ mediates an obligate insect–plant mutualism, using the recently described mutualism between Breynia vitis-idaea and its species-specific Epicephala moth pollinator as a model (Kawakita & Kato, 2004b). To date, it is estimated that 500 Phyllanthaceae species are each pollinated by a host-specific Epicephala pollinator, which in turn lays eggs in female flowers, from which develop larvae that consume a fraction of the seeds (Kawakita & Kato, 2009). The ovipositing female moths actively pollinate host flowers to ensure that larval food (i.e. seeds) is produced for their offspring. At night, a female collects pollen from host male flowers using the proboscis, and then visits a female flower, on which she deposits pollen and subsequently lays an egg. Whether moths visit male and female flowers on the same plant, or fly to a new plant after collecting pollen, is as yet unknown. The pollen collection and deposition behaviours are highly distinct and stereotypic, and are accompanied by a morphological specialization: the female proboscis is equipped with numerous hairs that are absent in the males, which probably facilitate effective handling of pollen. Because pollination takes place only at night, the female moth is predicted to use olfactory cues for host location. Also, as male Epicephala moths are commonly found on host leaves, male moths may rely on floral cues for mate location, although they have not been observed on flowers and apparently do not pollinate the host. Furthermore, insects other than the pollinator Epicephala females have rarely been observed on flowers, which led us to hypothesize that a private signal mediating these intimate associations is present.

In the present study: we chemically characterized the odour bouquets of male and female flowers of B. vitis-idaea; we tested whether floral scent alone is sufficient for attraction of the host-specific Epicephala pollinator; and we identified the blend of floral volatiles functioning as the pollinator attractant in this mutualism.

Materials and Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

The model system

Breynia vitis-idaea Burm.f. (Phyllanthaceae) (Fig. 1) grows in tropical and subtropical forests from Pakistan to the southern parts of Japan (Govaerts et al., 2000). This shrub is monoecious with yellow male flowers and green female flowers arranged at the base and at the apex, respectively, of a branch (Fig. 1b). The small (2.5–3.0 mm i.d.), inconspicuous flowers lack petals. Male flowers (Fig. 1c) have fused calyx lobes with a small opening at the tip, and female flowers (Fig. 1d) have three styles, which are more or less fused at the upper part of the ovary, so the floral architecture in B. vitis-idaea makes pollen removal and deposition by floral visitors other than Epicephala females very unlikely. Red, globe-shaped fruits are produced within 3–4 wk after pollination. Flowering and fruit production occur throughout the year, but peak in March–May and August–October, respectively, in our study areas in southern Japan and Taiwan (Fig. 1). An Epicephala species (Fig. 1e), yet to be formally described, is the only documented pollinator of B. vitis-idaea (Kawakita & Kato, 2004b). Because of the small size of host flowers, visual cues may be of less importance for the nocturnally active moths in search for mating and oviposition sites. By contrast, flowers of B. vitis-idaea are fragrant to the human nose only at night, suggesting that floral scent may be crucial for pollinator attraction.

image

Figure 1. Breynia vitis-idaea and sites for odour and fruit collection from the species in this study. (a) A B. vitis-idaea plant with fruits, (b) male and female flowers on a branch, (c) a male flower, (d) a female flower, (e) a female Epicephala moth ovipositing into a female host flower, and (f) study sites.

Download figure to PowerPoint

Rearing of moths

Epicephala moths were reared from B. vitis-idaea fruits in order to obtain naïve insects that had no previous experience with host flowers for electrophysiological and behavioural experiments. Branches with fruits were collected at our field sites (Fig. 1f, Supporting Information Table S1) and transported to the laboratory in sealed plastic bags at ambient temperature. The bags containing fruits were further incubated at 25–28°C in the laboratory for an additional 4–5 d to induce larvae to exit fruits. Larvae and pupae were kept in small transparent plastic containers at 25–28°C, 70% relative humidity and a 16 h : 8 h light : dark cycle. The short-lived adults emerged c. 10 d after pupation under these environmental conditions, and were then separated by sex. One- to three-day-old insects were used in all experiments.

Odour collection

We performed headspace collections of B. vitis-idaea on four islands in the Ryukyu archipelago, southwest Japan, as well as in southern Taiwan, in September–October 2006 and in April 2008 (Fig. 1f, Table S1). Collections started at 20:00 h and continued for 2–3 h. We sampled odour from whole branches with intact flowers as well as from cut flowers. As B. vitis-idaea produces minute male and female flowers that occur together on branches, odour collection from individual flowers attached to the plant is extremely difficult. Thus, to obtain sufficient amounts of floral scent for reliable chemical identification of individual compounds, and to enable comparisons of odour bouquets of male and female flowers, we performed odour collections from flowers that had been removed from the plant, assuming that possible wounding compounds should be very similar from flowers of different sex. For each sex, we cut off 35–200 flowers per plant and put them in polyvinylacetate oven bags (Toppits; Melitta Scandinavia AB, Klippan, Sweden), which had a volume of < 20 ml. As filters, we used small Teflon tubes (3 mm i.d.), filled with 20 or 40 mg of Tenax GR 60/80 adsorbent (Alltech, State College, PA, USA; catalogue #4937). Small battery-operated pumps were used (custom-built or purchased from GroTech, Gothenburg, Sweden) and the flow rate was set to 200 ml min−1. Filters were eluted with 400 μl of hexane, and 500 ng of methyl stearate was added to each sample as an internal standard. Ambient air was also collected as a control to identify background contaminants. Before analysis, samples were concentrated down to 40 μl under N2.

Floral scent analysis

We used gas chromatography–mass spectrometry (GC-MS) to analyse headspace samples of B. vitis-idaea, using an HP 6890 gas chromatograph (Hewlett-Packard Co., Palo Alto, CA, USA), equipped with a medium-polar Innowax column (30 m × 0.25 mm × 0.25 μm film thickness; J&W Scientific, Folsom, CA, USA), and linked to an HP 5973 mass spectrometer (Hewlett-Packard Co.). Samples of 1 μl were injected in a pulsed-splitless mode. Helium was used as the carrier gas at a velocity of 30 cm s−1, and the injector temperature was 225°C. The column temperature was maintained at 30°C for 3 min after injection and then increased by 10°C min−1 to 225°C. Additional analyses were performed using an HP 5890 gas chromatograph (Hewlett-Packard Co.), equipped with a nonpolar HP-1 column (30 m × 0.25 mm × 0.25 μm film thickness; J&W Scientific), linked to an HP 5972 mass spectrometer (Hewlett-Packard Co.). Helium was used as the carrier gas at a velocity of 40 cm s−1, and the injector temperature was 220°C. The oven temperature was programmed to be maintained at 50°C for 2 min after injection and then increased at 10°C min−1 to 250°C. To identify floral compounds in samples, peak retention times and mass spectra were compared with those of authentic standards.

Statistical analysis of scent data

We performed principal component analyses (PCAs) to check for differences in the composition of the odours produced by male and female flowers of B. vitis-idaea. Comparisons were made using all identified scent compounds, as well as only those compounds that produced consistent antennal activity in Epicephala moths (see ‘Electrophysiology’ section). Proportions of compounds were arcsine square-root transformed to better fit a normal distribution of data, and variables were scaled to unit variance before the analysis. We also used unpaired t-tests to compare the total amounts of scent emitted from male and female flowers (all compounds or only electroantennographic detection (EAD)-active compounds).

Electrophysiology

We used coupled gas chromatography and electroantennographic detection (GC-EAD) to identify physiologically active compounds in headspace samples of male and female flowers of B. vitis-idaea. The head of an Epicephala moth was cut off and both antennae were mounted on a PRG-2 EAG (10× gain) probe (Syntech, Kirchzarten, Germany) using conductive gel (Blågel; Cefar, Malmö, Sweden). Charcoal-filtered and humidified air passed over the antennal preparation from a glass tube outlet at 5 mm distance from the preparation. The GC effluent to the antennae passed through a heated transfer line set at 230°C.

Odour samples were injected into an HP 5890 Series II plus gas chromatograph (Hewlett-Packard Co.), equipped with a nonpolar HP-1 column (30 m × 0.25 mm × 0.25 μm film thickness; J&W Scientific). Hydrogen was used as the carrier gas at a velocity of 40 cm s−1, and the injector temperature was 220°C. At the end of the column, a four-way splitter with nitrogen as the make-up gas allowed a 1 : 1 division of the GC effluent to the flame ionization detector (FID) and to the antennal preparation. The oven temperature programme was the same as for the additional GC-MS analyses with the same column type (see `Floral scent analysis' section). An antennal preparation was only used for a single GC-EAD trial. To check the quality of the antennal preparation before a GC run, a Pasteur pipette, loaded with 2 μl of the test stimulus to be analysed on a strip of filter paper, was inserted through a hole in the glass tube 10 cm from the outlet. The pipette was linked to an air control system (Syntech, Kirchzarten, Germany), which generated 0.5-s air puffs through the pipette into the air stream of the glass tube. GC-EAD data were analysed with Autospike Version 3.3 software (Syntech, Kirchzarten, Germany).

Behavioural analysis

Two-choice glass Y-tube olfactometers (Scientific Glass, Löberöd, Sweden) were used to study the attraction of the Epicephala pollinator to volatiles released from B. vitis-idaea. The Y-tube was connected to a battery-driven air pump (GroTech, Gothenburg, Sweden) and the total flow rate was set to 200 ml min−1. Experiments were performed 1–4 h into the scotophase in 20-Lux red light, at 25–28°C and > 65% relative humidity. At lower temperatures, moths became inactive and did not respond to olfactory stimulation. A moth was transferred to the Y-tube, and allowed 5 min to respond, that is, to walk into either of the two arms of the olfactometer. Moths not responding within the given time period were excluded from the protocol. Both male and female moths were used for the bioassay, and each moth was only used once. The stimulus and control arms were alternated on a regular basis to avoid position effects. Data on olfactory response were analysed, separately for each sex, with a binomial test (Siegel & Castellan, 1998) against the null hypothesis that moths would be equally attracted to both arms of the olfactometer.

In the first experiment, branches bearing a total of > 50 male and > 100 female flowers (also with leaves) were used as the olfactory stimulus. Branches were cut from a host tree 1–2 h into the scotophase and enclosed in a 25 × 38 cm oven bag, which was connected to one arm of the Y-tube, and an empty bag of the same size connected to the second arm served as a control. In subsequent experiments, headspace samples or synthetic standards of EAD-active floral compounds were used as odour stimuli. The test stimulus was applied to a small strip of filter paper inserted into a 10-ml plastic syringe which was connected to the stimulus arm. The test stimulus was applied at a volume of 10 μl. A second syringe loaded with 10 μl of hexane served as a control. The solvent was allowed to evaporate for 3 min before the test started. When tested against the hexane control, male and female headspace samples (concentrated to 40 μl) were used as the stimulus. When synthetic compounds were tested against the hexane control, a total amount of 1 μg was used. When the male flower sample was tested against synthetic compounds, it was diluted to match the amount of compounds used (200 ng or c. 7000 male flower h−1 equivalents of EAD-active compounds; see Fig. 2). During experiments, filters were replaced every 20 min, and new stimulus added.

image

Figure 2.  Total (mean ± SE) emission of scent compounds from male flower samples (n = 12) and female flower samples (n = 18) of Breynia vitis-idaea. Male flowers emitted significantly more scent compounds compared with female flowers, both when the analysis included all 19 identified compounds and when only the four electroantennographic detection (EAD)-active compounds were included (unpaired t-test: ***, < 0.001 for both analyses).

Download figure to PowerPoint

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Floral scent analysis

The sweet floral fragrance of B. vitis-idaea includes at least 19 compounds as identified by GC-MS (Table 1). Male and female flowers produced similar sets of floral compounds, except for phenylacetaldehyde oxime, isoeugenol and vanillin, which were found exclusively in male flowers, and Z-β-ocimene, which was only found in female flowers. 2-phenylacetonitrile (2-PAN) and 2-phenylethyl alcohol (2-PEA) dominated the male flower scent and constituted on average 80% of the blend, whereas no such dominance was observed in the female floral scent (Table 1). A male flower emitted on average significantly more scent than a female flower, both when all scent compounds were included in the analysis (35.2 ± 10.2 pg h−1 vs 2.1 ± 0.4 pg h−1; t = 4.01, < 0.001; Fig. 2) and when only the antennal-active compounds were included (29.1 ± 8.6 pg h−1 vs 1.0 ± 0.5 pg h−1; t = 4.03, < 0.001). No compound was found exclusively in branch samples, but the amounts of E-4,8-dimethyl-1,3,7-nonatriene and E-β-ocimene were higher in branch samples than in samples from cut flowers (data not shown). In addition, no compounds were exclusively detected in samples from cut flowers when compared with branch samples with intact flowers, indicating no production of wounding compounds. Three compounds in samples, d-limonene, nonanal and decanal, were probably contaminants, as they were often found in higher amounts in control samples than in floral samples, and these compounds were excluded from the analyses.

Table 1.   Mean percentages (± SD) of scent compounds detected in headspace samplesa from male and female flowers of Breynia vitis-idaea
CompoundsMale flowers (n = 12)Female flowers (n = 18)
  1. aVolatiles were collected using Tenax GR filters.

  2. bCompounds eliciting responses in moth antennae in gas chromatography with electroantennographic detection (GC-EAD) trials.

Terpenoids
 β-Myrcene0.04 ± 0.030.13 ± 0.20
 Z-β-Ocimene0.00 ± 0.000.17 ± 0.17
 E-β-Ocimene1.63 ± 5.851.24 ± 1.84
 E-4,8-Dimethyl-1,3,7-nonatriene5.68 ± 4.624.22 ± 3.97
 6-Methyl-5-hepten-2-one1.36 ± 1.2014.72 ± 6.34
 Linalool0.69 ± 1.1610.36 ± 7.88
 Geranyl acetoneb1.44 ± 1.4620.20 ± 15.41
Aromatics
 Benzaldehyde0.99 ± 0.571.91 ± 1.99
 Benzyl alcohol1.02 ± 0.611.82 ± 3.18
 2-Phenylethyl alcoholb29.66 ± 16.769.16 ± 9.59
 2-Phenylacetonitrileb50.70 ± 17.7013.48 ± 15.92
 2-Phenylethyl acetateb0.72 ± 0.870.89 ± 0.92
 Phenylacetaldehyde oxime3.14 ± 4.170.00 ± 0.00
 Indoleb0.15 ± 0.254.31 ± 5.38
 Isoeugenol1.69 ± 1.370.00 ± 0.00
 Vanillin0.36 ± 0.440.00 ± 0.00
Fatty acid derivatives
 Z-3-Hexen-1-ol0.33 ± 0.272.64 ± 1.92
 Hexyl acetate0.01 ± 0.023.28 ± 3.69
 Z-3-Hexen-1-ol acetate0.38 ± 0.5211.46 ± 9.00

In the first PCA, including all 19 scent compounds identified, the fragrance blends of male and female flowers were clearly separated in odour space (Fig. 3a). Five principal components with eigenvalues > 1 explained 74.2% of the variation observed in floral fragrance data (for loadings of compounds, see Table S2). In the second PCA, including only the four EAD-active compounds (see `Electrophysiology' section), the fragrance blends of male and female flowers were still well separated (Fig. 3b), and a single principal component with an eigenvalue > 1 explained 51.7% of the variation observed among the samples (for loadings of compounds, see Table S3).

image

Figure 3.  Score plots from principal component analyses (PCAs) showing the divergent odour profiles in male and female flowers of Breynia vitis-idaea. The first PCA (a) included all 19 identified compounds, whereas the second PCA (b) only included the four electroantennographic detection (EAD)-active compounds (male flowers, n = 12; female flowers, n = 18). Closed circles, male flowers, open circles, female flowers.

Download figure to PowerPoint

Electrophysiology

Five compounds were consistently found to elicit responses in olfactory receptors of both male and female Epicephala moths: 2-PEA, 2-PAN, 2-phenylethyl acetate, decanal (contaminant), and indole (Fig. 4). The physiological activity for these compounds was confirmed by stimulating antennae with synthetic standards. Compounds eliciting antennal responses in only a few GC-EAD runs were d-limonene (contaminant), nonanal (contaminant) and geranyl acetone. No apparent sexual dimorphism in sensory sensitivity to these scent compounds was observed in the moths. All EAD-active compounds were found in both male and female flowers of B. vitis-idaea (Table 1).

image

Figure 4.  Simultaneous responses of the flame ionization detector (FID) and electroantennographic detection (EAD) using antennae of female Epicephala moths to headspace samples of male and female flowers of Breynia vitis-idaea. Decanal is a contaminant. The first undefined response in the female flower sample is to an unknown compound, whereas the second undefined response is to geranyl acetone.

Download figure to PowerPoint

Behavioural analysis

In the first experiment, a high proportion (85%; < 0.01) of female Epicephala moths were attracted to the odour from cut branches with flowers of B. vitis-idaea, confirming that the pollinator uses olfaction to detect its host (Fig. 5). In subsequent experiments, 85% and 73% of female moths showed attraction to the male flower headspace sample (< 0.01) and the female flower headspace sample (< 0.05), respectively, when tested against the hexane control. The two most abundant EAG-active compounds (2-PEA and 2-PAN) were then tested individually at a dose of 1 μg against the hexane control. 2-PEA elicited significant attraction in female moths (75% were attracted to the stimulus arm; < 0.05), whereas 2-PAN was not attractive. However, when 2-PEA was tested against the male flower headspace sample at the same dose of ∼200 ng per stimulus, 89% of females preferred the flower sample (< 0.001). A 1 : 2 blend of 2-PEA and 2-PAN (total dose of 1 μg) was then tested against the hexane control and found to be attractive (74% were attracted to the stimulus arm; < 0.05). When this binary blend was tested against the male flower headspace sample at a dose of ∼200 ng per stimulus, female moths did not discriminate between the two stimuli (Fig. 5).

image

Figure 5.  Behavioural responses of female Epicephala moths to different olfactory stimuli in Y-tube olfactometer tests (binomial test: *, < 0.05; **, < 0.01; ***, < 0.001).

Download figure to PowerPoint

Although GC-EAD recordings revealed no difference in antennal responses between sexes of Epicephala moths, male moths showed significant attraction only to the odour from branches with flowers (data not shown). When tested against headspace samples from male and female flowers, or against the three most abundant EAD-active compounds individually, males were generally inactive and few individuals (< 50%) initiated upwind walking in the olfactometer.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Contrary to our expectation of a ‘private signal’ guiding the exclusive pollinator to host flowers in an obligate pollination seed-consuming mutualism, the volatiles functioning as the attractant in the EpicephalaBreynia interaction (2-PEA and 2-PAN) should be regarded as general floral compounds, as they occur in the floral scent of a large number of angiosperms. Knudsen et al. (2006) reported 2-PEA and 2-PAN as constituents of the floral fragrance in 234 species from 49 families, and in 64 species from 18 families, respectively, showing the widespread occurrence of these compounds as floral volatiles among flowering plants. These two compounds were also found to occur together in the floral scent of a number of species (Knudsen et al., 2006). Although there are still limited data on whether and to what extent 2-PEA and 2-PAN function as attractants, or possibly repellents, in insect–plant interactions, the frequent occurrence of the two compounds as floral volatiles may indicate that they are used as attractants by many insect species during, for instance, nectar foraging. For example, 2-PEA alone has been shown to be an attractant for the generalist moth Trichoplusia ni (Haynes et al., 1991).

The first identification of a floral scent signal mediating an obligate pollination seed-consuming mutualism was reported very recently for Ficus semicorda (Chen et al., 2009). A single compound, 4-methylanisole, constitutes the attractive signal for the obligate pollinator in this association, and thus may represent a typical ‘private channel’. This is in contrast to scent data from other associations between figs and their host-specific pollinators, where general floral volatiles have mainly been identified (Grison-Pigéet al., 2002a,b). Interestingly, floral scent data from five species of Glochidion (Okamoto et al., 2007), a close relative of Breynia, which is also exclusively pollinated by host-specific Epicephala moths, suggest that these interactions are mediated by conventional floral scent compounds, similar to findings for B. vitis-idaea. Chemical and electrophysiological analyses of the floral scent in the yucca–yucca-moth mutualism strongly indicate that pollinator attraction in this association is mediated by specific chemistry (Svensson et al., 2005, 2006; G. P. Svensson, O. Pellmyr & R. A. Raguso, unpublished data). No data are yet available on the chemical ecology of the other two documented cases of obligate pollination mutualism, involving the globeflower and senita cactus and their respective obligate pollinators.

Until recently, the only nursery pollination system for which the pollinator attractant had been chemically characterized was the association between Silene latifolia and its pollinating seed predator Hadena bicruris (Jürgens et al., 2002; Dötterl et al., 2006). Electrophysiological analyses revealed almost 20 compounds in the floral headspace of S. latifolia eliciting an antennal response in H. bicruris. When the behavioural response to such compounds was tested in a flight tunnel, moths were found to be highly attracted to lilac aldehydes alone, and no difference in attraction was observed between these compounds and the scent from single S. latifoila flowers, showing that few key compounds constitute the active signal in this pollinator–plant interaction (Dötterl et al., 2006). Interestingly, nursery pollination systems involving Silene species and moths of the genera Hadena (Pettersson, 1992) and Perizoma (Westerbergh, 2004) do not constitute obligate associations, do not involve active pollination, and range from antagonistic to potentially mutualistic interactions (Kephart et al., 2006). Thus, additional data from these less specialized interactions, as well as other obligate associations, should greatly improve our understanding of whether floral chemistry is in fact more specialized in obligate plant–pollinator interactions than in less specialized mutualisms (Raguso, 2003).

We found that a blend of volatiles rather than a single volatile constitutes the attractive signal in our model system. This is in contrast to the study by Chen et al. (2009), who reported that a single compound was responsible for the attraction of a fig wasp pollinator to its host. Similar variation has been observed for sexually deceptive orchids, where Ophrys species mimic the multi-component blend of the female-produced sex pheromone of the pollinator (Schiestl et al., 1999; Ayasse et al., 2003), whereas Chiloglottis species rely on individual compounds for pollinator attraction (Schiestl et al., 2003; Schiestl & Peakall, 2005). Available data from more generalized pollinator–plant interactions suggest that blends of compounds rather than unique volatiles function as pollinator attractants. Plepys et al. (2002) used GC-EAD analyses and flight tunnel assays to identify which floral compounds from Platanthera bifolia, an orchid predominantly pollinated by sphingid and noctuid moths, constitute the attractive signal for the generalist moth Autographa gamma. Although 13 floral compounds elicited consistent EAD responses in the moths, only a small subset of the physiologically active compounds (the lilac aldehydes) were found to be critical for attraction, similar to findings in the SileneHadena association (Dötterl et al., 2006).

Our PCAs revealed a clear sexual dimorphism in floral scent production in the monoecious B. vitis-idaea, irrespective of whether all identified compounds or only EAD-active compounds were included in the analysis (Fig. 3). The two compounds making up the active pollinator signal (2-PEA and 2-PAN) showed a clear dominance in male flower samples but not in female flower samples (Table 1). In addition, male flowers emitted much higher amounts of the EAD-active compounds compared with female flowers (Fig. 2). Despite this clear difference in odour blends between flower types, the same EAD-active compounds were detected in both male and female flowers, and female moths were attracted to both odour stimuli in Y-tube bioassays (Fig. 5). Although elucidation of the role of sexual dimorphism in floral scent in this mutualism requires further study, the present results have some interesting implications as to how the evolutionary outcomes of floral scent may differ between floral sexes.

One potential cause of the quantitative difference in scent production between the sexes is the difference in their floral sizes (Ashman, 2009). However, although Breynia male flowers are larger than the female flowers (Fig. 1), the difference is probably no greater than twofold, which is far smaller than the observed difference in the amount of scent produced (Fig. 2). Alternatively, and more probably, the observed quantitative difference may be an outcome of sexual selection. As a consequence of asymmetry in reproductive interest between floral sexes, wherein fitness through male function linearly increases with increased access to female flowers, while female fitness is mostly limited by available resources, monoecious plants are expected to allocate resources differently through male and female function to maximize overall fitness. Waelti et al. (2009) showed that, in the S. latifoliaH. bicruris association, male flowers emit larger amounts of pollinator-attracting compounds compared with female flowers, and the pollinator prefers male flowers over female flowers in flight tunnel assays, indicating that the divergence in the floral signal in S. latifolia is caused by sexual selection. A similar pattern has been observed in the gynodioecious strawberry Fragaria virginiana, with hermaphroditic flowers emitting more scent, resulting in more pollinator visits compared with female flowers (Ashman et al., 2005).

However, there was also a qualitative difference between floral scents produced by male and female Breynia flowers. A nonadaptive explanation for the observed difference is that different floral structures, such as stamens and pistils, produce nonidentical compounds, resulting in floral bouquets that differ in chemical composition between the sexes (Ashman, 2009). However, it is also possible that the divergent scents in male and female flowers are the direct result of the highly specialized Epicephala pollination. Because Epicephala females exhibit distinctly different behaviours on male and female flowers, that is, pollen collection on male flowers and pollen deposition and oviposition on female flowers, host plants may be selected to produce different fragrance blends to induce compatible behaviours on flowers of each sex. Further analyses of sexual floral scent dimorphism in related plants without Epicephala pollination, as well as more detailed moth behavioural assays, are necessary to test this adaptive hypothesis.

Although we have identified 2-PEA and 2-PAN as critical components of the floral attractant in B. vitis-idaea, the actual concentrations of the compounds individual Epicephala moths are exposed to under field conditions are difficult to quantify. The doses of compounds used in our bioassays were 200 ng or c. 7000 male flower h−1 equivalents, which was arbitrarily chosen because of the difficulties in predicting such concentrations. The number of scent-producing flowers per plant in Epicephala-pollinated taxa of Breynia, Glochidion and Phyllanthus could be up to hundreds of thousands; thus, the moths are probably guided to their host by the scent signal produced by a large fraction of the flowers on a plant rather than by odour plumes derived from individual flowers. Moreover, because male and female flowers differ in their scent profiles (Fig. 3), the whole bouquet of a single plant may change depending on the relative numbers of male and female flowers produced at a given time. Therefore, the questions of whether the strength of the odour attractant or the relative proportions of male and female flowers affect patterns of moth attraction should be addressed in more detail in future studies.

Another important question is how Epicephala moths discriminate among hosts in areas where several Phyllanthaceae plants coexist. Floral scent has been shown to be an important isolation mechanism between closely related sympatric species in more generalized pollination systems (Waelti et al., 2008), and this trait may also play a key role in the tightly coevolved mutualisms involving pollinating seed predators. If heterospecific pollinations will not lead to seed maturation, an Epicephala female making such a mistake will have zero fitness, as her progeny will starve to death. Thus, to facilitate host specificity in these interactions, strong divergent selection should act on the floral scent profiles among sympatrically occurring species. Scent data are not yet available for additional Breynia species, but analyses on Glochidion plants show that sympatric species differ in the composition of the floral odour, and that female moths discriminate host from nonhost on the basis of floral scent (Okamoto et al., 2007), although the specific compounds responsible for this host specificity have not been identified. Similar odour-based host discrimination has been observed in fig wasps (Grison-Pigéet al., 2002a; Chen et al., 2009).

In summary, our study shows that B. vitis-idaea uses a blend of conventional floral scent compounds as an attractant for its exclusive Epicephala pollinator, demonstrating that system-specific chemistry is not a necessity for efficient host location by species-specific seed predators in a tightly coevolved pollination mutualism. The Phyllantheae–Epicephala association has recently emerged as a promising model system for studies on various aspects of mutualism, and this study and the previous study by Okamoto et al. (2007) show that this association is very suitable for research on how floral scent influences host location and host discrimination in herbivorous insects. Studies of floral scent in other Phyllantheae plants, including those with and without Epicephala pollination, should therefore provide new insights into how olfactory signal coevolution may have facilitated the origin of obligate mutualisms and shaped the subsequent co-diversification of the plants and pollinators.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

We thank Yuichi Kameda for assistance in collecting Breynia vitis-idaea fruits. Jette Knudsen generously provided authentic standards for GC-MS, GC-EAD and behavioural analyses. We also thank Amami Wildlife Conservation Center for logistic support during the fieldwork, and three anonymous reviewers for their constructive comments on an earlier version of the manuscript. Financial support for this work was provided by the Japan Society for the Promotion of Science, the Royal Physiographic Society in Lund, the Scandinavia-Japan Sasakawa Foundation, and the Swedish Royal Academy of Sciences.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Table S1 Study sites, dates and sample sizes of Breynia vitis-idaea headspace collections in the present study

Table S2 Loading of the first two principal components of the 19 compounds identified in headspace samples of Breynia vitis-idaea used in the first principal component analysis

Table S3 Loading of the first two principal components of the four antennal-active compounds in headspace samples of Breynia vitis-idaea used in the second principal component analysis

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

FilenameFormatSizeDescription
NPH_3227_sm_tS1-S3.doc66KSupporting info item