Fruit bats and bat fruits: the evolution of fruit scent in relation to the foraging behaviour of bats in the New and Old World tropics

Authors


  • This article is dedicated to the memory of Elisabeth K. V. Kalko, who was passionate about bats and figs.

Correspondence author. E-mail: rhodgkison@hotmail.com

Summary

  1. Frugivory among bats (Chiroptera) has evolved independently in the New and Old World tropics: within the families Phyllostomidae and Pteropodidae, respectively. Bats from both families rely primarily on olfaction for the location of fruits. However, the influence of bats on the evolution of fruit scent is almost completely unknown.
  2. Using the genus Ficus as a model, the aims of this study were to explore the chemical composition of fruit scent in relation to two contrasting seed dispersal syndromes in Panama and Malaysia and to assess the influence of fruit scent on the foraging behaviour of neo- and palaeotropical fruit-eating bats (Artibeus jamaicensis and Cynopterus brachyotis, respectively). Two hypotheses were tested: (i) variation in fruit scent, between bat- and bird-dispersed figs, is independent of phylogeny and (ii) Old and New World fruit bats, which have evolved independently in each hemisphere, share the same olfactory preferences with respect to fruit scent.
  3. The fruit scents of bat- and bird-dispersed fig species were sampled in the field, using dynamic headspace adsorption techniques. New and Old World fruit bats were then captured and tested on natural fig fruit scents from both hemispheres.
  4. Chemical analyses, using gas chromatography (GC) and GC/mass spectrometry (MS), revealed a broad overlap in scent compounds between bat-dispersed fig species from both hemispheres. Their fruit scents were dominated by monoterpenes, which contrary to phylogenetic predictions, were completely absent from bird-dispersed species from both regions.
  5. The fruit scents of bat-dispersed figs were highly attractive to neotropical bats (A. jamaicensis) in behavioural experiments, whereas those of bird-dispersed figs were completely rejected. Neotropical bats (A. jamaicensis) exhibited a significant preference for fig fruit scents dominated by monoterpenes, independent of the geographical origin of the scent. Palaeotropical bats (C. brachyotis), by contrast, rejected monoterpene-rich fruit scents from the Neotropics.
  6. In a cluster analysis (which included additional, published data from the literature), the fruit scents of bat-dispersed figs were clumped by subgenus, with the exception of palaeotropical figs of the subgenus Sycomorus. C. brachyotis, from Malaysia, was the only fruit bat species that significantly preferred the fruit scents of Sycomorus figs that contained no monoterpenes.

Introduction

The evolution of fruit traits among fleshy-fruited plants remains a controversial subject in evolutionary biology, with two main competing hypotheses (Herrera 1992; Fischer & Chapman 1993; Mack 1993; Jordano 1995; Lomáscolo, Speranza & Kimball 2008). According to the adaptive hypothesis, patterns in the variation of fruit form can be largely attributed to the selective feeding preferences of mutualistic frugivores, which drive evolutionary change through the dispersal of seeds. This results in adaptations in fruit display known as ‘seed dispersal syndromes’, which match the foraging preferences of the main guilds of seed dispersers for each plant taxon (Janson 1983; Marshall 1983). For example, fruits that are conspicuously coloured, relative to the surrounding foliage, are generally attractive to visually orientated frugivores, such as birds and diurnal mammals (Schmidt, Schaefer & Winkler 2004; Schaefer, Schaefer & Vorobyev 2007). By contrast, the nonadaptive hypothesis suggests that variation in fruit form is largely influenced by phylogeny, as well as other random effects, and that frugivores simply learn to associate the fruits of different species of plants from cues that they are pre-adapted to recognize – in a process known as ‘ecological fitting’ (Janzen 1985). Although many studies have sought to test the relative importance of both mechanisms on the evolution of fruits (e.g. Jordano 1995; Lomáscolo, Speranza & Kimball 2008), to date both hypotheses have only been rigorously tested in relation to the evolution of visual cues, which are particularly important to the foraging behaviour of birds and diurnal mammals (Janson 1983; Schmidt, Schaefer & Winkler 2004; Schaefer, Schaefer & Vorobyev 2007; Lomáscolo & Schaefer 2010). By contrast, the selective influence of nocturnal seed-dispersing frugivores, which forage predominantly by olfaction (Luft, Curio & Tacud 2003; Hodgkison et al. 2007), has never been tested.

Old and New World fruit bats of the families Pteropodidae and Phyllostomidae, respectively, have been geographically isolated and phylogenetically independent throughout their adaptive radiation (Marshall 1983; Simmons & Geisler 1998; Teeling et al. 2000, 2005; Simmons et al. 2008). Both groups of bats are important seed dispersers for many plant taxa and show a distinct preference for scented, but visually inconspicuous fruits (van der Pijl 1957; Heithaus 1982; Marshall 1983; Fleming 1988). This reflects the importance of olfaction in the foraging ecology of both groups of bats (Rieger & Jakob 1988; Laska 1990a,b; Acharya, Roy & Krishna 1998; Thies, Kalko & Schitzler 1998; Luft, Curio & Tacud 2003; Korine & Kalko 2005; Hodgkison et al. 2007). However, the selective influence of bats on the evolution of specific chemical fruit traits is almost completely unknown (Hodgkison et al. 2007).

The plant genus Ficus (Moraceae) provides an excellent model to explore the chemical basis for the attraction of bats. With its pantropical distribution (Shanahan et al. 2001a), continuous fruiting patterns (Leighton & Leighton 1983; Terborgh 1986; Lambert & Marshall 1991) and high local species diversity in many tropical forests (Lambert 1987, 1989; Lambert & Marshall 1991; Shanahan & Compton 2001; Shanahan et al. 2001b; Hodgkison et al. 2003), this genus can provide many natural replicates within a single study area and can also provide an abundant supply of fruit throughout the year. Furthermore, contrasting patterns in the presentation of figs suggest that many species have specific adaptations for the attraction of nocturnal or diurnal frugivores – including pronounced differences in fruit scent (e.g. Kalko, Herre & Handley 1996; Shanahan & Compton 2001; Hodgkison et al. 2003; Borges, Bessière & Hossaert-Mckey 2008; Lomáscolo, Speranza & Kimball 2008; Lomáscolo et al. 2010; Borges et al. 2011). However, the phylogenetic component of this variation, as well as its ecological significance for the attraction of specific guilds of seed-dispersing frugivores, has never been tested.

Using the genus Ficus as a model, the aims of this study were to explore the chemical composition of fruit scent, in relation to two contrasting seed dispersal syndromes in Panama and Malaysia, and to assess the influence of fruit scent on the foraging behaviour of neo- and palaeotropical fruit-eating bats; Artibeus jamaicensis (Phyllostomidae) and Cynopterus brachyotis (Pteropodidae), respectively. Two hypotheses were tested: (i) variation in the composition of fruit scent, between bat- and bird-dispersed figs, is independent of phylogeny and (ii) Old and New World fruit bats, which have evolved independently in each hemisphere, share the same olfactory preferences, with respect to fruit scent.

Materials and methods

Study Sites and Species

The field work for this project was conducted, between February 2007 and April 2008, at two study sites in the Old and New World tropics at the following places: Kuala Lompat (3º43′N, 102º17′E), within the Krau Wildlife Reserve, Pahang, Peninsular Malaysia; and on Barro Colorado Island, Panama (9º09′N, 79º51′W). Fig species, and their attendant seed dispersers, have been extensively studied at both sites, with several long-term projects on frugivory by birds (Lambert 1987, 1989; Lambert & Marshall 1991), bats (Bonaccorso 1979; Kalko, Herre & Handley 1996; Korine, Kalko & Herre 2000; Wendeln, Runkle & Kalko 2000; Hodgkison et al. 2003; Korine & Kalko 2005), primates (Chivers 1977; Raemaekers 1984), squirrels (Payne 1979) and other mammals (Lambert 1990; Hodgkison et al. 2003). Thus, both sites are extremely well suited for comparative studies on the fruit characteristics of bird- and bat-dispersed figs and the foraging behaviour of Old and New World fruit bats.

The Krau Wildlife Reserve consists of approximately 500 km2 of old-growth forest, which rises from 50 m above sea level (a.s.l) at Kuala Lompat, to over 2000 m a.s.l at the summit of Gunung Benom. The vegetation at Kuala Lompat can be described as lowland evergreen mixed dipterocarp forest, but is unusually rich in large Leguminosae (Raemaekers, Aldrich-Blake & Payne 1980). The site is also extremely rich in strangler and free-standing figs (Ficus spp.), with 41 species reported to date – at least 25 of which produce fruits consumed and dispersed by birds and 10 by bats, with very little overlap (Lambert 1987, 1989; Lambert & Marshall 1991; Hodgkison et al. 2003).

Our New World study site, on Barro Colorado Island (BCI), is situated on Gatun Lake, along the Panama Canal, and is approximately 15·6 km2 in area. The vegetation on BCI is semideciduous tropical lowland rain forest, with an extremely high abundance of figs (Foster & Brokaw 1982; Leigh, Rand & Windsor 1982). Seventeen species of figs (Ficus spp.) have been reported to date, 14 of which produce fruits predominantly consumed and dispersed by bats and three by birds, with very little overlap (Kalko, Herre & Handley 1996; Korine, Kalko & Herre 2000).

In both study areas, fig species consumed and dispersed by birds are generally characterized by visually conspicuous fruit displays, produced as a result of distinct colour changes upon ripening – often following the transition from green to red (Lambert 1987, 1989; Kalko, Herre & Handley 1996; Korine, Kalko & Herre 2000). Fig species dispersed by bats, by contrast, tend to be visually less conspicuous to humans (typically shades of green, yellow-green, or dark red-brown), but are characterized by the production of distinctive fruit scents when ripe (Kalko, Herre & Handley 1996; Korine, Kalko & Herre 2000; Hodgkison et al. 2003).

Ten fig species (from three subgenera and five sections) were selected for sampling in the present study. The characteristics of these species, including nomenclature, growth form, fruit colour, mode of presentation and primary dispersers are summarized in Table 1. According to recent molecular and morphological phylogenies, the subgenus Pharmacosycea is considered basal within the genus, whereas Sycomorus and Urostigma are terminal (Weiblen 2000; Jousselin, Rasplus & Kjellberg 2003). The fruit scent of one bat-dispersed species from the study is known to be dominated by short-chain fatty-acid-derived esters, alcohols and ketones – the most notable of which include 2-heptyl acetate, 2-heptanol and 2-heptanone (Hodgkison et al. 2007).

Table 1. Scent samples were collected from the ripe fruits of 10 species of Ficus (from three subgenera and five sections) in a) Malaysia and b) Panama. Three species are predominantly dispersed by birds and primates and seven by bats. All bird-dispersed species were monoecious, hemiepiphytic stranglers with red ripening, axillary fruits. Bat-dispersed species were either hemiepiphytic stranglers with axillary (monoecious) fruits or free-standing trees with axillary (monoecious) or caulicarpic/geocarpic (dioecious) fruits with variable colours when ripe. Samples from dioecious figs were collected only from female, seed-bearing fruits
Ficus speciesSub-genusSectionPrimary dispersersRipe colourGrowth formPresentation
a) Kuala Lompat, Malaysia
 F. hispida Linn. F.SycomorusSycocarpusBatsYellow-greenTreeCauli/geocarpic
 F. variegata Bl.SycomorusSycomorusBatsDeep redTreeCaulicarpic
 F. cucurbitina KingUrostigmaConosyceaBirds/primatesOrange-redHemiepiphyteAxillary
 F. dubia KingUrostigmaConosyceaBirds/primatesDeep redHemiepiphyteAxillary
b) Barro Colorado Island, Panama
 F. nymphiifolia P. MillerUrostigmaAmericanaBatsGreenHemiepiphyteAxillary
 F. obtusifolia KunthUrostigmaAmericanaBatsGreenHemiepiphyteAxillary
 F. popenoei StandleyUrostigmaAmericanaBatsGreen-redHemiepiphyteAxillary
 F. trigonata L.UrostigmaAmericanaBatsGreenHemiepiphyteAxillary
 F. costaricana (Liebm.) Miq.UrostigmaAmericanaBirds/primatesRedHemiepiphyteAxillary
 F. maxima P. MillerPharmacosyceaPharmacosyceaBatsGreenTreeAxillary

The fruit bat species selected for behavioural experiments in this study were C. brachyotis (Müller) at Kuala Lompat and A.jamaicensis Leach on BCI. Both bat species are known to consume the fruits of all bat-dispersed fig species that were selected for scent sampling in each study site (Kalko, Herre & Handley 1996; Hodgkison et al. 2003).

Fruit Scent Sampling

Fruit scents of bat- and bird-dispersed figs were sampled in the field using dynamic headspace adsorption techniques (Dobson et al. 2005; Hodgkison et al. 2007). Ripe fruits from each species were removed from plants and placed inside glass sample chambers – all of which were connected, by equal lengths of silicone tubing, to a single battery-operated pump (SKC, type 224). Air drawn into each glass sample chamber, through a single inlet, was filtered and cleaned of atmospheric pollutants by a cylindrical borosilicate glass cartridge packed with activated charcoal (Orbo-32, Supelco: diameter x length = 8 mm × 110 mm with 600 mg Activated coconut charcoal) and plugged with silanized glass wool (Raguso & Pellmyr 1998). Air drawn out of each chamber, through a single outlet, was sampled for volatile fruit scent compounds, by a sorbent tube containing 5 mg of Super Q (Alltech Associates Inc., Deerfield, IL, USA). Both the sorbent tubes and charcoal filters were connected to the glass sample chambers by short lengths of Teflon tubing.

In total, four glass sample chambers were used to collect fruit scents during each sampling session. This allowed up to three samples to be collected simultaneously, along with a blank control. Each sampling session was conducted at night (between 19:30 and 05:00 h) with a flow rate of 100 mL min−1 per chamber and a total sampling time of 4 h. For both quantitative analyses and behavioural experiments, scent samples were collected simultaneously from two fruits per sample chamber. However, due to their small size compared with the other fig species sampled in this study (ca. 1 g per fruit compared with ca. 10 g), scent samples of Ficus costaricana were collected from 25 fruits (ca. 25 g) per chamber.

Samples were eluted using 0·150 mL of 5 : 1 pentane/acetone and stored in borosilicate glass vials at −18 °C. Sorbent tubes were cleaned between samples using three solvents: dichloromethane (LiChrosolv, Merck, Darmstadt, Germany), diethyl ether (ChromaSolv, Merck, Darmstadt, Germany) and pentane (SupraSolv, Merck, Darmstadt, Germany).

Chemical Analysis of Fruit Scent Compounds

For quantitative analyses, 0·1 μg of n-octadecane was added as an internal standard to each of the eluted fruit scent samples collected by dynamic headspace adsorption (see above). All samples were then analysed using a HP 5890 Series II gas chromatograph (Hewlett-Packard, Palo Alto, CA, USA), equipped with a DB 5 capillary column (30 m × 0·25 mm i. d., J & W Scientific) using hydrogen as the carrier gas (2 mL min−1 constant flow). One μL of each sample was injected splitless at 40 °C. After 1 min, the split valve was opened and the temperature increased to 300 °C at a rate of 4 °C min−1.

Gas chromatography/mass spectrometry analyses were carried out on an Agilent Technologies 7890A Series GC connected to an Agilent Technologies 5975 °C mass selective detector fitted with a HP5-MS fused-silica column (29·9 m, 0·25 mm i. d., 0·25 μm thick film, Agilent Technologies). Mass spectra (70 eV) were recorded in full scan mode. Retention indices were calculated from a homologue series of n-alkanes. Compounds were identified by their mass spectra and gas chromatographic retention indices, which were then compared with those of reference compounds. Certain sesquiterpenes were identified by comparing their mass spectra and retention indices with those reported in a critically evaluated data base (Joulain & König 1998).

Enantiomers of linalool were separated on a Hydrodex-6-TBDMS chiral phase column (Macherey & Nagel, 35 m, internal diameter 0·25 mm, phase thickness 0·25 μm) using a gas chromatograph GC Top 8000 (ThermoQuest, Thermo Electron, Waltham, MA, USA). Hydrogen was used as the carrier gas. The oven temperature was set at 50 °C for 5 min and then raised to 210 °C at a rate of 1 °C min−1.

Behavioural Experiments on Bats

Behavioural experiments were used to test the effect of eight natural fruit scents on the foraging behaviour of captive bats (A. jamaicensis and C. brachyotis). The fruit scents selected for testing included four bat-dispersed fig species (Ficus maxima, Ficus nymphiifolia, Ficus obtusifolia, Ficus trigonata) and one bird-dispersed fig species from Panama (F. costaricana) and two bat-dispersed species from Malaysia (Ficus hispida and Ficus variegata). Bats from Panama (A. jamaicensis) were tested on the fruit scents of two bat-dispersed species from Malaysia (F. hispida and F. variegata) as well as two bat-dispersed fig species (F. obtusifolia, F. trigonata) and one bird-dispersed species (F. costaricana) of the same sub-genus and section from Panama. Bats from Malaysia (C. brachyotis) were tested on the fruit scents of two bat-dispersed species, from separate subgenera, from Panama (F. maxima and F. nymphiifolia), as well as F. variegata.

Behavioural experiments, in both study sites, were conducted within large flight cages (3 × 3 × 2·5 m3 – width x length x height), using newly captured, untrained bats. Bats were obtained from the field using mist nets (Kunz, Hodgkison & Weise 2009), with one individual retained for bioassays during each night of sampling. All newly captured bats were introduced into the flight cage on the night of capture, and then left undisturbed, with an abundant supply of food, until the following day. Immediately prior to dusk, the following evening, any uneaten fruits were removed from the flight cage, to prevent bats from becoming distracted during the experiments.

Each behavioural experiment consisted of a two-choice fruit selection trial, with no reward, in which each bat was presented with two unripe fruits: one of which was scented with a natural headspace sample (that was collected using the techniques described above); the other of which was unmanipulated and so provided a naturally unscented control (see Hodgkison et al. 2007 for details). The identity of the dummy fruits was determined by the availability of unripe figs in the field. The sequence in which the natural fruit scents were presented to the bats during the experiments was randomized for each of the individuals tested. During each experiment, the behavioural reactions of the bats were recorded on videotape, using the ‘Nightshot’ facility of a Sony digital video camera (Handycam DCR-PC350E). The effect of each treatment was quantified by the presence or absence of a single feeding attempt, which was defined as an attempt by a bat to bite and remove a fruit. To motivate the bats throughout the experiments, each bioassay was preceded by a feeding trial in which all bats were rewarded with a portion of ripe fruit. To avoid the introduction of foraging bias, individual bats were not rewarded with the same fig species that were being used in subsequent scent experiments. Each experiment lasted for 20 min, and in most cases, all experiments were performed in a single night (unless interrupted by heavy rain). All fruits were presented to the bats using wires to fix them to metal or wooden poles that formed the outer structure of the cage. To prevent the bats from learning the spatial arrangements inside the cage, all fruits were presented to the bats in different locations during each successive trial (Nolte & Mason 1998). The only artificial light source was an infrared beam emitted by the camera. To ensure sufficient illumination, all treatment and control samples, in the two-choice feeding trials, were separated by a distance of ca. 30 cm.

Once the bioassays were completed, each bat was recaptured in the flight cage using a hoop net, marked with an individually numbered ball-chain necklace (Ball-Chain Manufacturing Company, Mount Vernon, NY, USA) and then released at the site of capture (Kunz & Weise 2009). This ensured that no individual bat was exposed to a single bioassay on more than one occasion.

Data Analysis

The statistical significance of differences in scent composition, between bat- and bird-dispersed figs, was tested with analysis of similarity (Anosim), via 999 randomized permutations using Primer-E 6. This analysis was based on the relative amounts of all 81 compounds that occurred in the combined fruit scents of all nine species. To reduce the influence of highly dominant compounds, the data were square root–transformed prior to analysis.

The results of the behavioural experiments were arranged on 2 × n contingency tables to express the frequency of feeding attempts in relation to n treatments (i.e. feeding attempt vs. no feeding attempt × n treatments). These results were then analysed with the G test of independence (Sokal & Rohlf 1995).

Results

Chemical Analysis of Fruit Scent Compounds

Seventy-nine compounds were identified in the combined ripe fruit scents of all nine fig species investigated during this study, including the following: saturated hydrocarbons, unsaturated hydrocarbons, aromatic hydrocarbons, monoterpenes and sesquiterpenes. A further two compounds were also recognized, but were not identified to compound class (see Appendix S1, Supporting information). The five most dominant scent compounds of each species are shown in Table 2.

Table 2. The five most dominant compounds, in four compound classes (a–d), present in the ripe fruit scents of three bird- and six bat-dispersed fig species from Panama and Malaysia
Compound class/nameRRIBird-figsBat-figs
Fico Ficu Fidu Finy Fiob Fipo Fitr Fima Fiva
  1. Compounds are ranked 1–5 in order of decreasing dominance. Ficus species include the following: F. costaricana (Fico), F. cucurbitina (Ficu), F. dubia (Fidu), F. nymphiifolia (finy), F. obtusifolia (Fiob), F. popenoei (Fipo), F. trigonata (Fitr), F. maxima (Fima) and F. variegata (Fiva).

a) Aromatic compounds
Methyl benzoate1093222
Ethyl benzoate117543
Methyl anthranilate1338 4
b) Monoterpenes
α-Pinene9282
β-Pinene9694
β-Myrcene986 5
(Z)-β-Ocimene10345
(E)-β-Ocimene1046111111
(R)-(-)-Linalool11012
 (E,E)-2,6-Dimethyl-1,3,5,7-octatetraene11203
 allo-Ocimene11273354
c) Sesquiterpenes
α-Copaene1374232
β-Copaene142733
β-Caryophyllene142811154
 iso-Germacrene143855
Sesquiterpene E147144
Germacrene D14762
Bicyclogermacrene149043
 (E,E)-α-Farnesene151153
(E)-Nerolidol156345
Sesquiterpene 215732
d) Aliphatic hydrocarbons
1-Tetradecene13985

The ripe fruit scents of all six bat-dispersed fig species, from both Malaysia and Panama, were dominated by monoterpenes; most notably (E)-β-ocimene (Table 3). Indeed, monoterpenes accounted for between 50 and 91% of the total scent production in each species (Table 3). Two other compound classes (aromatic hydrocarbons and sesquiterpenes) were also prominent in the fruit scents of certain species; but, overall, their contributions were highly variable (Table 3). In contrast, the ripe fruit scents of bird-dispersed figs were almost totally dominated by sesquiterpenes, the most notable of which were β-caryophyllene and α-copaene (Table 3).

Table 3. The proportional contributions of four compound classes to the fruit scents of six bat- and three bird-dispersed fig species from Panama and Malaysia
Compound class (mean ± SD)
Ficus species n Aromatic compoundsMonoterpenesSesquiterpenesOthersa
  1. a

    Including aliphatic hydrocarbons.

a) Bat-dispersed figs
F. obtusifolia 540·12 ± 12·3750·31 ± 14·907·21 ± 2·832·36 ± 0·58
F. maxima 56·26 ± 4·9852·35 ± 19·5139·63 ± 15·441·76 ± 1·05
F. nymphiifolia 531·84 ± 7·2562·62 ± 6·335·34 ± 2·260·19 ± 0·07
F. popenoei 22·10 ± 0·2668·13 ± 8·4529·32 ± 8·370·00 ± 0·00
F. trigonata 537·39 ± 6·1052·58 ± 9·479·90 ± 3·660·13 ± 0·10
F. variegata 51·93 ± 0·9190·71 ± 3·957·36 ± 3·210·00 ± 0·00
b) Bird-dispersed figs
F. costaricana 20·00 ± 0·000·00 ± 0·0098·12 ± 2·231·88 ± 2·23
F. cucurbitina 20·00 ± 0·003·81 ± 1·1180·57 ± 1·6015·62 ± 0·49
F. dubia 20·57 ± 0·420·00 ± 0·0099·43 ± 0·420·00 ± 0·00

A cluster analysis, based on the degree of similarity (Bray–Curtis) in chemical composition, revealed that the fruit scents of figs formed two statistically significant clusters according to seed disperser (Anosim Global R = 0·83, < 0·001) (Fig. 1). Bat-dispersed fruit scents also clustered by subgenus and species, with the exception of F. obstusifolia and F. trigonata (Fig. 1).

Figure 1.

Similarity (Bray–Curtis) in the fruit scents of bat- and bird-dispersed fig species from Panama and Malaysia, based on the relative proportions of 81 compounds present in the combined scents of nine species (see Appendix S1, Supporting information). Fruit scents cluster primarily by disperser (Anosim Global R = 0·83, < 0·001). Bat-dispersed species also cluster by subgenus and species, with the exception of Ficus obstusifolia and Ficus trigonata.

Behavioural Experiments on Bats

The ripe fruit scents of bat-dispersed figs, from Panama and Malaysia, successfully induced feeding attempts in the neotropical fruit bat A. jamaicensis, with reaction rates of between 40 and 80% (Table 4). No significant difference in response rate was recorded among three fig species, from two separate subgenera, with chemically similar odour bouquets (F. obtusifolia, F. trigonata and F. variegata); with response rates of 65–80% (Table 4). However, significant variation, among bat-dispersed figs, was recorded with the inclusion of F. hispida, a bat-dispersed fig from Malaysia. The fruit scent of F. costaricana, a bird-dispersed fig from Panama, induced significantly fewer feeding attempts among neotropical bats, than two bat-dispersed figs from the same subgenus and section (F. obtusifolia and F. trigonata, Table 4).

Table 4. Observations on the foraging behaviour of a) Neotropical fruit bats (Artibeus jamaicensis) and b) Palaeotropical fruit bats (C. brachyotis) in response to the natural fruit scents of bat and bird-dispersed figs from Panama and Malaysia. In the case of A. jamaicensis, the frequency of feeding attempts varied significantly between bat-dispersed fig species from Panama and Malaysia (G test of independence, = 8·26, d.f. = 3, = 0·04), but not between species with chemically similar fruit scents (i.e. F. obtusifolia, F. trigonata and F. variegata:= 1·18, d.f. = 2, = 0·55). Feeding rates of A. jamaicensis also varied significantly between predominantly bat- and bird-dispersed species within the same Neotropical section (Americana) of the subgenus Urostigma (= 40·60, d.f. = 2, < 0·001). In the case of C. brachyotis, the frequency of feeding attempts varied significantly between bat-dispersed fig species from Malaysia and Panama, despite strong similarities in chemical composition (G test of independence, = 44·99, d.f. = 2, ≤ 0·001)
Ficus speciesFicus subgenusOrigin of fig fruit scentMain disperser of fig speciesNumber of bats tested (female : male)Number of bats that attempted to feedFeeding attempt frequency (%)
a) Artibeus jamaicensis (Panama)
F. obtusifolia UrostigmaPanamaBats20 (10 : 10)1680
F. trigonata UrostigmaPanamaBats20 (10 : 10)1575
F. costaricana UrostigmaPanamaBirds20 (10 : 10)00
F. variegata SycomorusMalaysiaBats20 (10 : 10)1365
F. hispida SycomorusMalaysiaBats20 (7 : 13)840
Pentane (control)20 (10 : 10)00
b) Cynopterus brachyotis (Malaysia)
F. maxima PharmocosyceaPanamaBats20 (10 : 10)00
F. nymphiifolia UrostigmaPanamaBats20 (10 : 10)00
F. variegata SycomorusMalaysiaBats20 (10 : 10)1575
Pentane (control)20 (10 : 10)00

The ripe fruit scents of bat-dispersed figs from Panama (F. maxima and Fnymphiifolia) failed to induce a single feeding attempt in the palaeotropical fruit bat C. brachyotis, despite obvious similarities in fruit scent (Fig. 1). In contrast, 75% of the 20 individuals that were tested responded to the fruit scent of F. variegata from Malaysia (Table 4).

During the bioassays, 17% of the bats (20% in Panama and 13% in Malaysia) refused to feed in captivity and were released without study.

Discussion

The ripe fruit scents of neo- and palaeotropical bat-dispersed figs were chemically similar, despite broad differences in phylogeny. The bouquets of all six species were dominated by monoterpenes, most notably (E)-β-ocimene, which together comprised 50–90% of the total scent production in each species. In contrast, the ripe fruit scents of bird-dispersed figs, from both hemispheres, were almost completely comprised of sesquiterpenes, despite strong phylogenetic affinities with their bat-dispersed congeners (Jackson et al. 2008). Furthermore, the functional significance of this variation was clearly demonstrated by behavioural experiments on bats (A. jamaicensis). When exposed to the natural fruit scents of bat-dispersed figs from Panama, up to 80% of the bats that were tested were induced to feed, whereas the natural fruit scents of bird-dispersed figs (F. costaricana) were completely rejected. Thus, variation in fruit scent was independent of phylogeny, and only monoterpene-dominated fruit scents were attractive to bats.

However, a strong phylogenetic component to the variation in fruit scent chemistry was revealed at a finer ecological scale among fig species dispersed by bats. In a cluster analysis, the fruit scents of most bat-dispersed figs were clustered by subgenus and species, independent of their geographical origins (and, thus, the phylogeny of their fruit bat dispersers). The subgenus Pharmacosycea, which is considered basal within the genus (Weiblen 2000; Jousselin, Rasplus & Kjellberg 2003) also had the most chemically dissimilar fruit scent. Whereas two species within the subgenus Urostigma (F. obtusifolia and F. trigonata), which were unresolved in a recent molecular phylogenetic analysis (Jackson et al. 2008), were partially unresolved in relation to fruit scent chemistry. It is also interesting to note that the most distantly related fig species, with the most chemically contrasting fruit scents, actually occurred sympatrically within the same study area and, furthermore, shared the same guild of fruit bat dispersers - F. maxima (Pharmacosycea); and F. obtusifolia and F. trigonata (Urostigma). Thus, variation in the fruit scent of bat-dispersed figs appears to be largely phylogenetic in structure despite the geographical isolation and independent evolution of their Neo- and Palaeotropical fruit bat dispersers (Marshall 1983; Simmons & Geisler 1998; Teeling et al. 2000, 2005; Simmons et al. 2008). However, a more complex pattern is revealed with the addition of further data from the Old World tropics (Hodgkison et al. 2007; Borges, Bessière & Hossaert-Mckey 2008; Borges et al. 2011).

In contrast to the monoterpene-dominated fruit scents revealed in the present study, previous studies in the Old World tropics have revealed other bat-dispersed fig species (of the subgenus Sycomorus) with strikingly different fruit scents (Hodgkison et al. 2007; Borges, Bessière & Hossaert-Mckey 2008; Borges et al. 2011). In all three studies, the fruit scents of bat-dispersed figs were dominated by fatty-acid-derived compounds (including alcohols, ketones and esters), and in some cases (e.g. Ficus hispida), monoterpenes were completely lacking. Indeed, if published scent data from two such species (F. hispida and Ficus scortechinii) are added to the cluster analysis of the present study (data from Hodgkison et al. 2007), they reveal that bat-dispersed fig species from the Old World tropics (subgenus Sycomorus) form a completely separate cluster (Fig. 2). Thus, these results suggest that, in contrast to their Neotropical counterparts, C. brachyotis (and possibly other fruit bat species that consume the fruits of Sycomorus figs) may have no obvious preference for fruit scents dominated by monoterpenes.

Figure 2.

Similarity (Bray–Curtis) in the fruit scents bat- and bird-dispersed fig species from Panama and Malaysia, based on the relative proportions of 101 compounds present in the combined scents of 11 species. The fruit scents of Palaeotropical bat-dispersed figs (subgenus Sycomorus) do not cluster according to disperser. *Additional data from Hodgkison et al. 2007.

The results of behavioural experiments on bats in Panama and Malaysia support this hypothesis. When exposed to the natural fruit scents of bat-dispersed figs from both biogeographical regions, the neotropical fruit bat, A. jamaicensis, responded positively to all monoterpene-dominated fruit scents, independent of their phylogeny and geographical origins. In contrast, the Old World fruit bat, C. brachyotis, which naturally consumes the fruits of Sycomorus figs, responded positively to the fruit scent of F. variegata from Malaysia, but completely rejected two similar monoterpene-rich fruit scents from Panama. However, when New World bats were exposed to the natural fruit scent of F. hispida (an Old World fig scent completely lacking in monoterpenes – Hodgkison et al. 2007), their response rates were significantly lower. Thus, A. jamaicensis has a significant preference for monoterpenes, which was not observed in Palaeotropics (at least in the case of C. brachyotis). As most fig species attract several species of bats (e.g. Kalko, Herre & Handley 1996; Hodgkison et al. 2003), this suggests that fruit bat species with similar diets are also likely to display similar olfactory foraging preferences. Thus, the lack of specialized foraging preference displayed by C. brachyotis suggests that the evolutionary associations between bats and figs could be generally more diffuse in the Old World tropics, reflecting the greater dietary overlap between bats and other seed-dispersing frugivore groups within the region (Fleming 2005).

It is interesting to note that several prominent compounds in the fruit scent of F. hispida (e.g. pentyl acetate) are also detected at very low concentrations by rats (Laska 1990a). There is little doubt that rats (Muridae), and other small mammals, such as tree shrews (Tupaiidae) and civets (Viveridae), are important, albeit opportunistic frugivores in many Old World tropical forests, and yet their contributions to seed dispersal are almost completely unknown. Nevertheless, several species of rats, as well as other terrestrial mammals, are known to feed on figs, including the fruits of F. hispida (Shanahan et al. 2001a; K. Wells pers. comm.; RH pers. obs.). Thus, the contribution of nonvolant mammals to seed dispersal could be greater than previously thought. Furthermore, several species of figs in the Old World subgenus Sycomorus (including F. hispida and F. scotechinii) are at least partially geocarpic, producing fruits directly from the base of the tree, or from long runners that trail along the ground. This specialized mode of presentation, which is unlikely to be associated with bats, could provide further evidence that small, predominantly terrestrial mammals are important seed dispersers for many, so called ‘bat figs’. Therefore, the fruity scents, associated with certain caulicarpic and geocarpic figs in the Palaeotropics, could have evolved as a result of the selective feeding behaviour of a wide range of frugivores. It is also interesting to note that the fruits of at least one exclusively geocarpic species of palaeotropical fig (F. semicordata) were rejected by captive bats (C. brachyotis) despite their fruity scent (R.H. pers. obs.). However, the dominant compounds in the fruit scent of this, and other exclusively geocarpic fig species, are currently unknown.

Clearly, further research is still required to investigate the function of fruit scent in relation to the foraging behaviour of fruit bats and other major groups of seed-dispersing frugivores. However, the results of this study have demonstrated that variation in fruit scent between bird- and bat-dispersed figs is independent of phylogeny and that the fruit scents of bat-dispersed figs and the olfactory foraging preferences of fruit-eating bats (i.e. C. brachyotis vs. A. jamaicensis) are more varied in the Palaeotropics.

Acknowledgements

We thank the Smithsonian Tropical Research Institute (Panama) for their support throughout this study and for extending their hospitality to Sharon Balding and Felix Hodgkison. We also thank the Department of Wildlife and National Parks (Malaysia) and the Economic Planning Unit of the Prime Minister's Department (Malaysia) for granting us permission to work in Malaysia. Finally, we gratefully acknowledge the valuable contributions of the following individuals: Ahmad Bin Dagu, for field assistance and logistical support in Malaysia; Sonja Gaessler, Larissa Albrecht and Katrin Petschl, for field assistance on BCI; Allen Herre, Adalberto Gómez, Nelida Gómez and Charlotte Jandér, for sharing their knowledge of figs on BCI; Konstans Wells, for sharing information on small mammals in S.E. Asia; and Ingrid Dillon and Andrea Weiss, for administrative support and laboratory assistance at Ulm University. This research adhered to the Association of Animal Behaviour Guidelines for the Use of Animals in Research as well as the Smithsonian Tropical Research Institute's Protocol for the Humane Use of Live Vertebrates. We gratefully acknowledge the valuable contributions of three anonymous referees. The project was funded by a grant from the Lubee Bat Conservancy to THK and RH and by funds from the Deutsche Forschungsgemeinschaft (DFG, Germany) to EKVK and MA (KA 124 8-1). AZ and WAWM were funded by MOHE-UKMTOPDOWN (Grant No. ST08FRGS00032010) and MOA (Grant No. 050102SF1041).

Ancillary