The chemical ecology of seed dispersal in monoecious and dioecious figs

Authors


*Correspondence author. E-mail: renee@ces.iisc.ernet.in

Summary

  • 1In the nursery pollination system of figs (Ficus, Moraceae), flower-bearing receptacles called syconia breed pollinating wasps and are units of both pollination and seed dispersal. Pollinators and mammalian seed dispersers are attracted to syconia by volatile organic compounds (VOCs). In monoecious figs, syconia produce both wasps and seeds, while in (gyno)dioecious figs, male (gall) fig trees produce wasps and female (seed) fig trees produce seeds.
  • 2VOCs were collected using dynamic headspace adsorption methods on freshly collected figs from different trees using Super Q® collection traps. VOC profiles were determined using gas chromatography–mass spectrometry (GC–MS).
  • 3The VOC profile of receptive and dispersal phase figs were clearly different only in the dioecious mammal-dispersed Ficus hispida but not in dioecious bird-dispersed F. exasperata and monoecious bird-dispersed F. tsjahela.
  • 4The VOC profile of dispersal phase female figs was clearly different from that of male figs only in F. hispida but not in F. exasperata, as predicted from the phenology of syconium production which only in F. hispida overlaps between male and female trees. Greater difference in VOC profile in F. hispida might ensure preferential removal of seed figs by dispersal agents when gall figs are simultaneously available.
  • 5The VOC profile of only mammal-dispersed female figs of F. hispida had high levels of fatty acid derivatives such as amyl-acetates and 2-heptanone, while monoterpenes, sesquiterpenes and shikimic acid derivatives were predominant in the other syconial types. A bird- and mammal-repellent compound methyl anthranilate occurred only in gall figs of both dioecious species, as expected, since gall figs containing wasp pollinators should not be consumed by dispersal agents.

Introduction

Considerable research has been done on volatile organic compounds (VOCs) in floral scents (Knudsen, Tollsten & Bergström 1993; Grison-Pigéet al. 2001; Knudsen et al. 2006), and specific compounds within these bouquets have also been implicated in pollinator attraction (Ayasse et al. 2003; Ashman et al. 2005). However, similar research on volatiles responsible for attracting dispersers to ripened fruit has been scarce (Dudareva et al. 2006) with some research on bats (Mikich et al. 2003; Sánchez et al. 2006) and primates (Laska et al. 2006). Furthermore, the difference in volatile production between flowers and fruit of the same species in relation to the attraction of pollinators or dispersers has not been investigated, largely due to the research priorities of investigators who are either pollination or seed dispersal specialists. In the Ficus (Moraceae) system, flowers are borne inside the syconium, which will ripen to form a fig. This system is unique because the syconium is the unit of attraction for both pollinators and seed dispersers. The highly species-specific pollinating fig wasps use VOCs produced by figs to find pollen-receptive figs (Hossaert-McKey, Gibernau & Frey 1994; Grison-Pigé, Bessière & Hossaert-McKey 2002a; Grison-Pigéet al. 2002b). Because the fig system is a nursery pollination system (Anstett, Hossaert-McKey & Kjellberg 1997), VOC signatures must vary with the fig developmental cycle, so that pollinators and seed dispersers are attracted to the fig during appropriate time windows. In dioecious fig species, moreover, in which pollinators can only breed within gall figs produced by male trees, and seeds are only produced within seed figs on female trees (Verkerke 1989), there should be selection on gall and seed figs to be indistinguishable from each other at the pollination stage (Grafen & Godfray 1991; Patel et al. 1995), and for dispersal agents to be attracted to seed figs rather than gall figs in the dispersal stage (Corner 1978; Lambert 1992; Patel & McKey 1998; Dumont, Weiblen & Winkelmann 2004). Although the pollinators of fig species are closely related to each other, fig seed dispersers are phyletically diverse, including a huge range of birds varying in size and many other traits (Compton, Craig & Waters 1996; Korine, Kalko & Herre 2000; Shanahan et al. 2001), as well as mammals including bats, rodents such as sciurids and primates (Shanahan et al. 2001).

These features of the fig pollination and seed dispersal system present unique constraints on the chemical ecology of these mutualisms. In this paper, we present the first data comparing differences between the VOC signatures of the receptive (pre-pollination) and dispersal phase of three fig species (one monoecious and two dioecious) and examine the dispersal phase VOCs in greater detail. Based on the ecology of figs and their dispersal agents, we made the following predictions:

  • 1There should be a distinct shift in volatile signatures of syconia between the receptive (pre-pollination) stage and the post-wasp dispersal (seed dispersal) stage, so that overlap in volatile signatures is minimal.
  • 2Within a dioecious species, the volatile signatures of ripe seed figs should be distinguishable from that of gall figs.
  • 3Mammal-dispersed figs should have a volatile signature distinguishable from that of bird-dispersed figs, since mammals and birds have different olfactory sensitivities. Seed figs that are mammal-dispersed should produce volatile signatures of higher scent intensity (greater volatile quantities) compared to gall figs of the same species, or to seed figs of bird-dispersed species.

Materials and methods

study sites and species

The study was conducted at Agumbe, Shimoga District (13°30′N, 75°5′E), and at Kudremukh, Chikmagalur District (13°15′N, 75°7′E) of Karnataka State in the Western Ghats of India. The sites are separated by 75 km. Volatiles were harvested from three sympatric species: two (gyno)dioecious (Ficus hispida L.: section Sycocarpus, and Ficus exasperata Vahl: section Sycidium) and one monoecious species (Ficus tsjahela Burman: section Urostigma). Ficus hispida, a short free-standing fig usually found along streams, bears caulicarpous syconia that are yellowish-green when ripe [fig diameter; female: mean = 23·9 mm (SE = 0·3, n = 40 figs); male: 25·3 (0·7, n = 40)]. Geocarpic syconia are also present on stolons that emerge from the base of the trunk and are recumbent on the ground. Ficus exasperata, a medium-sized free-standing fig bears axillary syconia that are red when ripe [fig diameter; female: mean = 19·7 mm (SE = 0·4, n = 50 figs); male: 22·4 (0·4, n = 50)]. Ficus tsjahela, a tall free-standing fig bears ramicarpous syconia that are creamy-beige when ripe [fig diameter; mean = 5·0 mm (SE = 0·1, n = 20 figs)]. The sampled fig trees occurred within a small area (c. 5–10 km2) at each site and belonged to the same population at each site.

voc collection and identification

VOCs from fig syconia were collected by dynamic headspace adsorption methods as described in Grison, Edwards & Hossaert-McKey (1999). Freshly harvested receptive or dispersal phase figs from each single tree were enclosed in a polyethylene terepthalate (Nalophan®) bag (Kalle Nalo GmbH, Wursthüllen, Germany) through which a constant airflow was maintained by two micropumps (incoming flow rate: 400 mL min−1 and outgoing flow rate: 300 mL min−1) over an Alltech Super Q® volatile collection trap (VCT; ARS Inc., Gainesville, FL) for 3 h per sample. Incoming air was cleaned using activated charcoal filters. For F. tsjahela, since the branches of this species are also strongly aromatic (unlike those of the other two species), we also extracted VOCs from branches at receptive and dispersal stages. Controls (empty bags) were run at the same time to determine any ambient airborne contamination. VCTs stored at −20 °C until analysis were later eluted with 150 µL dichloromethane to which 20 µL of a 200-ng µL−1 solution of nonane and dodecane were added as internal standards. The eluates were analysed in a CP-3800 (Varian Inc., Palo Alto, CA) gas chromatograph with an FID detector coupled with a Saturn 2000 mass spectrometer (Varian Inc.) in the split mode at 1 : 4 split ratio. For both gas chromatography (GC) and mass spectrometry (MS), a CP-SIL low bleed Varian column (30 m, ID 0·25 mm, film thickness 0·25 µm), and helium carrier gas (at 1 mL min−1) were used. The temperature programme for the analysis was: 50 °C for 3 min, ramped by 3 °C min−1 to 100 °C, by 2·7 °C min-1 to 140 °C, by 2·4 °C min−1 to 180 °C and by 6 °C min−1 to 250 °C. Compound identification was based on comparison of mass spectra with the NIST 98 MS library and on retention indices reported in literature.

Each sample reported in this paper was collected from a different tree. Each tree was sampled only for a particular phenophase. The numbers of figs from which volatiles were extracted varied for each sample depending on availability of the phenophase on the tree. The numbers of figs from which volatiles were extracted for each sample were as following: Receptive phase: F. hispida: mean = 117·3 figs (SE = 10·1, n = 30 trees), F. exasperata: 230·5 (18·5, n = 33), F. tsjahela: 604·3 (200·4, n = 6); Dispersal phase: F. hispida: mean = 37·1 figs (SE = 3·7, n = 16 trees), F. exasperata: 96·5 (9·5, n = 24), F. tsjahela: 618·6 (144·0, n = 12). Therefore, the VOC signature is represented as percentages of each volatile in the sample. Absolute quantities of VOCs per sample were calculated on a per fig and per hour basis (ng fig−1 3 h−1), for comparison across species and phenophases.

data analysis

Principal component analysis (PCA, covariance matrix, StatBox v6·3) was used to compare patterns of scent composition between the different stages and sexes for each species, or among species for the dispersal stage. For these multivariate analyses, we used the relative amounts (percentages) of the compounds. We also estimated the total quantities of volatiles produced by figs using calibrated quantities of the two internal standards. We then tested the effect of sex and stage on the intensity (i.e. quantity) of VOC emission for each species using anovas (on log-transformed data when required) (proc glm, type 3, sas v9) followed by a multiple comparison of means (LSMEANS with Tukey–Kramer multiple comparison tests) when appropriate. Finally, for each species, we investigated stage and sex effects on the overall abundance of the compounds using a multivariate analysis of variance (manova, proc glm, sas v9), followed by univariate (i.e. sequential analyses) on each dependent variable (quantity of each VOC) to test which compounds contribute to the overall significance in this analysis (Stevens 1992).

Results

ficus hispida

The volatile signature of the receptive phase figs had, on average, 36·0 compounds (SE = 1·4, n = 30 trees) while that of the dispersal phase had 25·7 compounds (1·4, n = 15 trees). A PCA performed using only VOCs that occurred at least once in a sample at > 5% levels (33 compounds) showed clear separation of receptive and dispersal phase figs (Fig. 1a). Moreover, seed (female) and gall (male) figs were also clearly separated in the dispersal phase (Fig. 1a). The first two components explained 67% of the variance. Similar results were obtained with VOCs at > 1% levels. Several VOCs were responsible for this separation (Fig. 1b). The VOCs that were largely responsible for the separation of female dispersal figs in this PCA analysis were 2-amyl acetate, 2-heptanol, n-amyl acetate and 2-heptyl acetate. Dispersal phase male figs were characterized by indole, α-trans-bergamotene and methyl anthranilate. For these dispersal figs, we had sufficient samples from two calendar years to examine a year effect, and we found none (analysis not shown), indicating that the VOC difference between sexes was stable across time.

Figure 1.

Volatile profiles of Ficus hispida. (a) PCA analysis of VOCs produced by male and female figs in receptive and dispersal phases. (b) Factor loadings of VOCs in the above PCA analysis.

We found a significant effect of sex and stage on the total number of compounds emitted by figs of both sexes and between figs at receptive and dispersal stages while the interaction effect was non-significant (Table 1). Figs of both sexes produced more compounds at receptive stage than at dispersal stage [females: mean = 38·8 compounds (SE = 1·8, n = 16 trees), and 28·8 (2·1, n = 6); males: mean = 32·7 compounds (SE = 1·9, n = 14 trees), and 23·7 (1·7, n = 9); LSMEANS P < 0·01]. Using the total quantity of VOCs emitted by the sexes in receptive and dispersal stages (log-transformation employed to meet anova requirements for normality), we also found significant effects of sex and of stage, as well as a significant interaction effect between sex and stage in scent quantity (Table 2). These effects were due to the large quantities of VOCs emitted by female figs in the dispersal stage [mean = 58·1 ng fig−1 3 h−1 (SE = 7·3, n = 6 trees)] which were significantly greater (P < 0·0001 in all cases) than those emitted by female figs in the receptive stage [mean = 2·5 ng fig−1 3 h−1 (0·7, n = 16)], males in the receptive stage [mean = 2·9 ng fig−1 3 h−1 (0·6, n = 14)] and males in the dispersal stage [mean = 2·7 ng fig−1 3 h−1 (0·6, n = 9)].

Table 1.  Results of analysis of variance on number of compounds produced by male and female figs in receptive and dispersal stages
SpeciesEffectFdfP
Ficus hispidaSex7·331, 440·010
Stage20·811, 44< 0·0001
Sex × stage0·061, 44NS
Ficus exasperataSex2·931, 520·093
Stage34·561, 52< 0·0001
Sex × stage16·561, 520·0002
Ficus tsjahelaStage1·491, 17NS
Year1·672, 17NS
All three speciesSpecies2·362, 115< 0·0001
Stage0·211, 115< 0·0001
Species × stage12·612, 115NS
Table 2.  Results of analysis of variance on total quantity (ng fig−1 3 h−1) of different compounds produced by male and female figs in receptive and dispersal stages (log quantities used in analysis)
SpeciesEffectFdfP
Ficus hispidaSex16·891, 440·0002
Stage34·481, 44< 0·0001
Sex × stage28·471, 44< 0·0001
Ficus exasperataSex0·731, 52NS
Stage5·721, 520·02
Sex × stage0·571, 52NS
Ficus tsjahelaStage0·011, 17NS
Year1·302, 17NS
All three speciesSpecies2·362, 1150·099
Stage0·211, 115NS
Species × stage12·612, 115< 0·0001

To investigate the contribution of some specific VOCs to these results, we selected those compounds with mean quantities > 0·04 ng fig−1 3 h−1 for receptive figs and > 0·14 ng fig−1 3 h−1 for dispersal figs, and performed a manova which showed a significant effect of stage (Wilks’λ = 0·0044, F26,16 = 140·41, P < 0·0001) and of sex (Wilks’λ = 0·0047, F26,16 = 129·05, P < 0·0001) as well as an interaction effect of stage and sex (Wilks’λ = 0·0045, F26,16 = 135·88, P < 0·0001). From the univariate sequential anova analysis on these VOCs included in the manova procedure, we found that several fatty acid derivatives such as 2-amyl acetate, n-amyl acetate, 2-heptanol, 2-heptanone, 2-heptyl acetate and 2-nonyl acetate were significantly higher in the female dispersal phase than in other combinations of sex and phase (Table 3), as were compounds such as β-elemene, β-caryophyllene and α-humulene, while indole was higher in male dispersal figs (Table 3). Methyl anthranilate was absent in the female dispersal phase (Table 3, see Appendix S1 in Supplementary material).

Table 3. manova on quantities of compounds (ng fig−1 3 h−1) for the three Ficus species, followed by univariate sequential anova analysis on some relevant compounds
Compoundmanova effectRemarks (results of Tukey–Kramer tests)
SexStageSex × stage
  1. * P < 0·05, **P < 0·001, ***P < 0.0001.

Ficus hispida
Overall manova********* 
2-amyl acetate****** Higher in female dispersal
2-heptanol**** 
2-heptanone****** 
n-amyl acetate** 
2-heptyl acetate****** 
2-nonyl acetate****** 
β-elemene**** 
β-caryophyllene***** 
α-humulene**** 
Methyl anthranilate*** Absent in female dispersal
Indole** Higher in male dispersal
cis-linalool oxide *NS  
Cyclolinalone*NS  
α-copaene*NS  
α-trans-bergamotene**NS  
Germacrene D*NS  
δ-cadinene*NS  
Ficus exasperata
Overall manova**NS 
β-caryophyllene*NSNS 
α-copaene*NSNS 
Germacrene D*NSNS 
α-humulene*NSNS 
(Z)-ocimene***NS 
MyrceneNS*** 
p-cymeneNSNS* 
γ-terpineneNSNS** 
Ficus tsjahela
Overall manova NS  

ficus exasperata

The volatile signature of the receptive phase figs had, on average, 37·9 compounds (SE = 1·1, n = 33 trees) while that of the dispersal phase had 29·4 compounds (1·8, n = 24). A PCA performed using only VOCs that occurred at least once in a sample at > 5% levels showed no clear separation of receptive and dispersal phase figs (Fig. 2). A PCA using only compounds > 1% gave similar results. There was also no striking separation of male (gall) and female (seed) figs like that found for F. hispida. Here also, as in F. hispida, methyl anthranilate was present in gall figs but absent in seed figs (Supplementary Appendix S1).

Figure 2.

Volatile profiles of Ficus exasperata. PCA analysis of VOCs produced by male and female figs in receptive and dispersal phases.

Unlike for F. hispida, we found no effect of sex in the total number of compounds produced. However, the effect of stage and the interaction effect between sex and stage were significant (Table 1). Males produced the same number of compounds at receptive and dispersal phase [mean = 35·2 compounds (SE = 1·4, n = 16 trees); and 31·8 (1·7, n = 17)], while females produced more compounds in receptive [mean = 39·6 compounds (1·5, n = 14 trees)] compared to dispersal phase [mean = 20·8 (3·0, n = 6); LSMEANS P < 0·01]. We found no effect of sex in the total quantity (log-transformed) of VOCs (data only from Agumbe), but a significant effect of stage, and a non-significant interaction effect between sex and stage in scent intensity (Table 2). The effect of stage was due to the larger quantities of VOCs produced by male figs in the receptive stage [mean = 0·6 ng fig−1 3 h−1 (SE = 0·1, n = 16 trees)] which were significantly different (P < 0·05) from that of males in the dispersal stage [mean = 0·2 ng fig−1 3 h−1 (0·02, n = 17)] but not significantly different from that of females in the receptive stage [mean = 0·3 ng fig−1 3 h−1 (0·1, n = 14)], and females in the dispersal stage [mean = 0·2 ng fig−1 3 h−1 (0·1, n = 6)].

To investigate the contribution of some specific VOCs to these results, we selected VOCs with mean quantities > 0·05 ng fig−1 3 h−1 for both receptive and dispersal figs and using a manova we found a significant effect of stage (Wilks’λ = 0·38, F20,30 = 2·48, P = 0·012) and of sex (Wilks’λ = 0·43, F20,30 = 1·98, P = 0·04), but a non-significant interaction between sex and stage. From the univariate sequential anova analysis on these VOCs included in the manova procedure, we found that only myrcene was emitted in greater concentration in the male dispersal phase, while (Z)-ocimene was significantly greater in the receptive phase (Table 3). Myrcene, p-cymene, and γ-terpinene showed significant interactions between sex and stage (Table 3).

ficus tsjahela

The volatile signature of the receptive phase had, on average, 54·3 compounds (SE = 2·7, n = 6 trees) while that of the dispersal phase had 49·2 compounds (2·7, n = 12). A PCA performed using only VOCs that occurred at least once in a sample at > 3% levels showed no clear separation of receptive and dispersal phase figs (Fig. 3a); some VOCs such as β-caryophyllene and α-humulene seemed to be more associated with receptive figs (Fig. 3b). In the PCA analysis, we found no effect on the results if we included or removed the data of the aromatic branches. This indicates that the branches cannot be separated from the aroma of the figs at the corresponding stages (Fig. 3a) and also that figs in receptive or dispersal stages are themselves indistinguishable, at least by this analysis.

Figure 3.

Volatile profiles of Ficus tsjahela. (a) PCA analysis of VOCs produced by male and female figs in receptive and dispersal phases, as well as branches collected during receptive and dispersal phases. (b) Factor loadings of VOCs in the above PCA analysis. NI, not identified compound.

There was no difference between receptive and dispersal stages in the number of VOCs emitted by figs [Table 1; receptive phase: mean = 54·3 VOCs (SE = 2·7, n = 6); dispersal phase: mean = 49·2 VOCs (2·7, n = 12)]. There was also no significant difference in the total quantity of VOCs produced between receptive and dispersal stages (Table 2), and no effect of the year of collection [Tables 1 and 2; receptive figs: mean = 5·6 ng fig−1 3 h−1 (SE = 2·6, n = 6); dispersal figs: mean = 3·1 ng fig−1 3 h−1 (1·0, n = 12)]. Seven VOCs were most important in the signature: α-pinene, (E)-ocimene and β-caryophyllene were present in > 10% quantity on average, while β-pinene, 4-ethyl anisole, α-copaene and germacrene D were present in > 3% quantity on average (Fig. 3b, Supplementary Appendix S1). However, there was no effect of stage on VOC quantities (manova: Wilks’λ = 0·54, F6,11 = 1·6, P = 0·24).

Comparison of dispersal phase volatiles of the three species

We attempted to determine whether the volatile signatures of the dispersal stages of the three species could be separated using a PCA analysis. For this, we employed only VOCs that occurred at least once in a sample at > 5% levels (43 compounds). In this combined analysis, we found three distinct clusters: gall figs of F. hispida, seed figs of F. hispida, and a combined cluster of dispersal figs of F. exasperata and F. tsjahela (Fig. 4a). Furthermore, an anova on the scores of these individuals on the three main principal components showed that Principal Component 1 (35% of variance) clearly separated gall and seed figs of F. hispida from the others (F4,44 = 134·73, P < 0·0001) (Fig. 4a,b), while Principal Component 2 (22% of variance) separated male F. hispida from the rest (F4,44 = 43·44, P < 0·0001) and Principal Component 3 (8% of variance) separated F. tsjahela from the rest (F4,44 = 24·37, P < 0·0001).

Figure 4.

Volatile profiles of all Ficus species examined together. (a) PCA analysis of dispersal phase VOCs of male and female figs of Ficus hispida and F. exasperata, and of monoecious F. tsjahela. (b) Factor loadings of VOCs in the above PCA analysis.

There was a significant effect of species and of stage on the total number of VOCs produced, but a non-significant interaction between species and stage (Table 1). However, for the total quantity of volatiles produced, there was a non-significant effect of species and stage but a significant interaction effect between species and stage (Table 2). This was due mainly to the large amounts of compounds produced by female seed figs of F. hispida (Fig. 5a,b) compared to those produced by other figs. An analysis of quantities of compounds produced by each dioecious species, conducted to understand the interaction effect of sex, stage and species, showed that only seed figs of F. hispida were significantly different from the other categories (LSMEANS, Tukey–Kramer multiple comparisons, P < 0·0001). Fatty acid derivatives were higher both in percent as well as in absolute concentration only in seed figs of F. hispida relative to all other figs while shikimic acid derivatives, monoterpenes and sesquiterpenes characterized the other syconial stages (Fig. 5a,b).

Figure 5.

(a) Mean percent VOCs in different classes in the different Ficus species and sexes. The relative composition of fatty acid derivatives is detailed (pie chart) for female dispersal figs of F. hispida. (b) Mean quantity of VOCs (ng fig−1 3 h−1) in different classes in the different Ficus species and sexes.

Discussion

We have shown that, within species, volatile signatures of figs vary with stage of development of the fig as well as with sex. Volatile signatures are also species-specific. Moreover, these differences and similarities can be put into the context of the particular biology of the fig species, especially whether the species is dispersed by birds, mammals or both, and whether the species, if dioecious, has its male and female syconia available simultaneously or at different times.

volatile signatures of receptive vs. dispersal phases

The fig is a climacteric fruit wherein sharp rises in respiration and ethylene production within the syconium precede the ripening process (Marei & Crane 1971) and influence release of pollinating wasps from the syconium (Galil, Zeroni & Bar Shalom 1973). Thus, differences between the VOCs of receptive and dispersal stages of figs are expected. Ficus hispida, which is a mammal-dispersed fig (Tang et al. 2007) showed clear separation between the volatile signatures of receptive and dispersal stages. However, in the largely bird-dispersed F. exasperata (Whitney et al. 1998) and F. tsjahela (R.M. Borges, personal observation) there was no such clear separation. The reasons for such a difference between the species are not yet clear. Furthermore, since these are the first data on comparisons between volatile signatures of figs in the receptive and dispersal stages, we are unable to determine whether this is a general difference between mammal- and bird-dispersed figs in receptive and dispersal syconia. That figs do differ in VOC profile even between pre-pollination and post-pollination stages is, however, known for one of our studied species, F. hispida (Proffit et al. in press).

volatile signatures of the dispersal phase of seed figs vs. gall figs

The VOC signature of the dispersal phase of the seed figs of F. hispida was significantly different from that of the gall figs, as predicted (Figs. 1 and 5). However, this prediction was not borne out in F. exasperata, in which the signatures of seed figs and gall figs were not as clearly separated (Figs 2 and 5). How might these differences between the two dioecious species be explained? The phenology of syconium production in F. hispida differs from that of F. exasperata (Patel 1996). In F. hispida, receptive or dispersal stage figs are available simultaneously on trees of both sexes in the population (synchrony between sexes) while in F. exasperata, there is low overlap between syconia production in the sexes (asynchrony between sexes) (Patel 1996). There should be selection on seed figs to be different from gall figs such that only seed figs are consumed (Corner 1978; Lambert 1992; Patel & McKey 1998; Dumont et al. 2004). However, selection for this difference is expected to be greater in dioecious species that produce seed and gall figs simultaneously in the same season (Lambert 1992). Thus, there should be less selection pressure on F. exasperata than on F. hispida in the dispersal phase to make the gall fig volatile signature different from that of seed figs. From the perspective of attraction using volatiles, we suggest that one likely response to such selection could be the evolution of a volatile signature in seed figs of F. hispida that is more attractive to mammalian dispersers, since frugivorous birds are known to be more responsive to visual cues (Sallabanks 1993). The ripe figs of F. exasperata are bright red in colour in contrast with the dull colour of the ripe figs of F. hispida. Sex differences in VOC signature could also result from other differences in chemical composition reflecting the higher nutrient quality of seed compared to gall figs, as was found for F. pungens, in which seed figs had higher levels of soluble carbohydrates and fat and lower fibre contents compared to gall figs (Dumont et al. 2004). Such differences could lead to the higher levels of fatty acid derivatives found in the VOC signature of the seed figs of F. hispida (Fig. 5a,b). Since mammals, including bats, are sensitive to such compounds which include aliphatic esters and alcohols (Laska 1990; Laska & Seibt 2002a, b), incorporating these compounds in VOC signatures could be a strategy to target mammals as preferred dispersers for such figs. Furthermore, F. hispida is cauliflorous and geocarpic, and therefore, more likely to be preferentially consumed by bats and other mammals such as rodents (Lambert & Marshall 1991).

Mammalian dispersers of figs usually occur at lower densities compared to birds (Compton et al. 1996; Shanahan & Compton 2001). This could result in selection for a stronger VOC signal, especially in seed figs, of bat- or rodent-dispersed fig species. Consistent with this prediction is the fact that in our study, the highest concentration of the VOC signal was found in the dispersal-phase seed figs of F. hispida (Fig. 5b). Bats are indeed effective seed dispersers of F. hispida in south Asia (Tang et al. 2007), and the continuously available steady-state fruiting of F. hispida, in which only few ripened fruit are available at any one time, attracts solitary bat species (Corlett 2006). Furthermore, there is evidence that the olfactory detection threshold of many mammals including bats, primates and rats, is negatively correlated with carbon chain length of the compounds (Laska, Seibt & Weber 2000), which means that the longer the carbon chain length, the better the detection at lower quantities. This also means that the higher the volatility of the compound (i.e. in approximate terms, the fewer carbon atoms it possesses), the higher the olfactory detection threshold. This suggests that F. hispida, which produces a few mature seed figs each night for dispersers, will have a higher probability of attracting low-density mammalian dispersers if it produces compounds of higher volatility in higher amounts and compounds with lower volatility in lower amounts, in order to facilitate detection. Since volatility of a compound can be approximated by its retention time in the GC analysis, we grouped volatiles according to their retention times (in 5-min intervals) and determined their representation in the signals of the various species and sexes (Fig. 6a,b). As predicted, the seed figs of F. hispida produced the highest percents and quantities of compounds with the highest volatility, while in the other figs, the opposite pattern was observed (Fig. 6a,b).

Figure 6.

(a) Mean percent volatiles in 5-min intervals of retention time in the GC analysis in the different Ficus species and sexes. (b) Mean quantity (ng fig−1 3 h−1) in 5-min intervals of retention time in the GC analysis in the different Ficus species and sexes.

volatile signatures of mammal-dispersed f. hispida vs. bird-dispersed f. exasperata and f. tsjahela

Of the various mammal dispersers of fig seeds (Shanahan et al. 2001), bats are the most important in terms of numbers as well as the distances to which fig seeds are carried (Lambert 1990; Kalko, Herre & Handley 1996; Shilton et al. 1999; Korine et al. 2000). However, figs appear to be one group in the Asian tropics in which specialist fig feeders have evolved among birds. For example, some fruit pigeons (Treron), some barbets (Megalaima) and one hornbill are specialist fig consumers that travel large distances in search of fig crops (Lambert & Marshall 1991; Shanahan et al. 2001). Furthermore, if plants evolve to utilize the services of specialist fig consumers such as Treron, the lipid (fatty acid) content of these fruits should be lower since some birds such as Treron lack a gall bladder (Garrod 1874). Treron phoenicoptera and Megalaima (M. zeylanica and M. rubricapilla) are important visitors to both F. exasperata and F. tsjahela at our study sites (R.M. Borges, personal observation).

Since bird-dispersed figs seem to have no obvious odour associated with them (Compton et al. 1996; Kalko et al. 1996), it appears that birds use other cues such as visual cues to locate ripe figs. However, with the exception of our present study, no investigation of volatiles produced by bird-dispersed figs has been conducted. Our study has shown that even bird-dispersed figs such as those of F. exasperata and F. tsjahela produce volatiles, but that these volatiles are substantially different in quality and quantity from those produced by mammal-dispersed figs (Fig. 4a). Furthermore, the volatiles produced by gall figs of the continuously fruiting mammal-dispersed F. hispida were also indistinguishable from those of bird-dispersed figs. Although birds use olfaction in social behaviour (Hagelin, Jones & Rasmussen 2003), foraging (Bonadonna et al. 2006), homing (Wallraff 1990) and self-medication (Petit et al. 2002), the role of dispersal-phase VOCs in attracting or repelling birds is completely unexplored. We found methyl anthranilate [a bird- and mammal-repellent compound (Mason, Adams & Clark 1989; Clark & Mason 1993; Nolte, Mason & Clark 1993)] in the gall figs of both F. hispida (in very high quantities) and F. exasperata but none in the seed figs of either species. Figs are thus possibly using specific compounds to repel dispersal agents from gall figs and to facilitate preferential consumption of seed figs, as earlier predicted.

Research on the salience of cues or signals that can evolve to facilitate communication between plants and vertebrates is growing. However, this research has largely focused on mammals rather than on birds. For example, 2-heptanone which was found in our study to be characteristic of seed figs of F. hispida, is associated with oestrus in mammals such as rodents (Schwende, Wiesler & Novotny 1984). A generalist mammal such as the rat, Rattus norvegicus, is also highly sensitive to the banana-like odour of esters such as amyl acetate (Alberts & May 1980). 2-heptanone and amyl acetates constituted 68·7% of the VOC profile of seed figs of F. hispida (Fig. 5a). Bats have been found to be attracted to fruity odours (esters, alcohols) (Laska 1990), although they avoid higher concentrations of ethanol (Sánchez et al. 2006). As in other mammals in which innate or acquired olfactory sensitivity is context-dependent (Laska et al. 2000), bats may also differ in their preference for volatile cues based on their food preferences (Mikich et al. 2003). There may also be region-specific olfactory adaptations of taxa as found for flower-visiting bats (von Helversen, Winkler & Bestmann 2000; Pettersson, Ervik & Knudsen 2004). However, the frequency of occurrence of compounds in the environment of a species may not necessarily determine olfactory sensitivity in that species to those compounds (Laska et al. 2006). It is clear that research on olfactory adaptations is in its infancy (Laska et al. 2000), and much more needs to be determined about the chemical environment of species and the salience of potential communication signals. Thus, just as mammal- and bird-dispersed fruit differ in sugar types based on differences in disperser physiology and nutritional ecology (Martínez del Rio & Restrepo 1993), so also volatiles may differ between bird- and mammal-dispersed fruit, as we found for F. hispida, F. exasperata and F. tsjahela. More comparative data performed in the context of phylogenetic contrasts are needed to validate this prediction.

Acknowledgements

This research was funded by grants: (CEFIPRA 2609-1) to Renee M. Borges and Martine Hossaert-McKey, and GDR Ecologie Chimique (CNRS) to Martine Hossaert-McKey and Jean-Marie Bessière. We thank Saveer Ahmed, Mahua Ghara, Pooja Muralidharan, Magali Proffit, Rakhi Rai, Yuvaraj Ranganathan, and Catherine Soler for their help in fieldwork and data analysis. We thank the Karnataka Forest Department for research permission. We thank Finn Kjellberg, Doyle McKey and James Cresswell for comments on the manuscript.

Ancillary