Reproductive resource partitioning in two sympatric Goniothalamus species (Annonaceae) from Borneo: floral biology, pollinator trapping and plant breeding system

The floral phenology, pollination ecology and breeding systems of two sympatric early-divergent angiosperms, Goniothalamus tapisoides and G. suaveolens (Annonaceae) are compared. The flowers are protogynous and morphologically similar, with anthesis over 23–25 h. Both species are predominantly xenogamous and pollinated by small beetles: G. tapisoides mainly by Curculionidae and G. suaveolens mainly by Nitidulidae. Coevolution and reproductive resource partitioning, reducing interspecific pollen transfer, is achieved by temporal isolation, due to contrasting floral phenologies; and ethological isolation, due to contrasting floral scents that contain attractants specific to the two beetle families. Analysis of floral scents revealed three volatiles (3-methylbutyl acetate, ethyl hexanoate and 2-phenylethanol) that are known to be nitidulid attractants in the floral scent of G. suaveolens, but absent from that of G. tapisoides. An effective pollinator trapping mechanism is demonstrated for both species, representing the first such report for the family. Trapping is achieved by the compression of the outer petals against the apertures between the inner petals. This trapping mechanism is likely to be a key evolutionary innovation for Goniothalamus, increasing pollination efficiency by increasing pollen loading on beetles during the staminate phase, promoting effective interfloral pollinator movements, and increasing seed-set by enabling rapid turn-over of flowers.

Reproductive interference can be defined as the fitness cost resulting from sexual interactions between different species [1][2][3] . This is commonly manifested in plants as interspecific pollen transfer between sympatric species, and can lower reproductive efficiency due to competition for pollinators, pollen wastage, clogging of the stigma surface with alien pollen (hence impeding successful germination of compatible pollen), and the formation of sterile or poorly adapted hybrids [4][5][6] . Although many studies have focused on the deleterious consequences arising from reproductive interference [7][8][9] , other research has suggested that the increased flowering density observed in such species can be beneficial as it increases pollinator abundance 10 .
Various reproductive isolation mechanisms, including mechanical and ethological isolation, are known to reduce interspecific pollen transfer in sympatric species 10,11 . Mechanical isolation is generally achieved by structural adaptations to the flowers of competing species to ensure the deposition/retrieval of pollen from different parts of the body of the shared pollinator 2,6,12 , or through the evolution of species-or genus-specific floral morphologies that are adapted for different pollinator guilds 11,13 . Ethological isolation is commonly achieved following adaptations to floral shape, pigmentation and scent 11 : although this can be a direct consequence of mechanical isolation 11 , it can also be due to changes in floral scent chemistry that favour different pollinators [14][15][16] , and the 'flower constancy' behaviour of some pollinators in which the pollinator makes repeated visits to the same plant population 8,17 . Reproductive isolation is widely believed to be a major driver of floral evolution, enhancing pollination efficiency by reducing interspecific pollen transfer and hence maximising resource partitioning 2,10,18 .
Many studies of reproductive interference and the evolution of reproductive isolation between sympatric species have focused on taxa with structurally complex flowers that maintain a high degree of pollinator specificity.   with average maximum and minimum annual temperatures of 32.1 °C and 23.8 °C, respectively, and with an average annual rainfall of 2770 mm 31,32 .
All field observations were conducted in well preserved localities, largely free from anthropogenic disturbance. Goniothalamus tapisoides was studied at four sites, viz.: (1)   suaveolens were tagged and monitored during April and May in both years. Detailed timing of the onset and cessation of floral sexual stages were recorded over 10 days at 1-hr intervals beginning immediately before the onset of the pistillate phase and continuing until the end of the staminate phase. The following features were monitored throughout anthesis: changes in floral organ colour, presence of stigmatic exudate, movement of petals and associated opening and closing of the apertures in the floral chamber, and wilting and abscission of floral organs. Stigmatic receptivity were assessed using 3% hydrogen peroxide (H 2 O 2 ) solution, in which the formation of gas bubbles is indicative of peroxidase enzyme activity 33 . Stigmatic receptivity was noted to be associated with visible changes in the presence of stigmatic exudate and stigmatic colour change, providing a logistically easier indicator of receptivity. The onset of the staminate phase was indicated by anther dehiscence.
Flower-level phenological data were assessed for multiple flowers in each individual tree, enabling the phenological pattern at the individual plant level (the number of pistillate-phase and staminate-phase flowers) to be determined and hence, by concatenating the data from ten tagged individuals of each species, enabling population-level phenological patterns to be deduced. This information enables an assessment of intra-and inter-individual floral synchrony. The assessment whether floral visitors were effective pollinators was based on the following criteria: relative visitation rate; coincidence of visits with the duration of the pistillate and staminate floral phases; attachment of pollen grains to the insect; and evidence of interfloral movement 34 . Since Goniothalamus species are known to be protogynous, the lattermost criterion was indicated by observing pollen-laden insects in pistillate-phase flowers.
Floral thermogenesis and scent chemistry. Temperatures inside the pollination chamber of five flowers of each Goniothalamus species were measured using a digital temperature data logger (Testo 176-T4, Testo, New Jersey) with type-K thermocouples (± 0.3 °C accuracy). Measurements were taken at 5-min intervals, beginning before the onset of the pistillate phase and continuing beyond the end of the staminate phase. Ambient temperatures immediately outside the pollination chamber were recorded simultaneously using the same data logger and thermocouples after cross-calibration.
Floral scents were sampled separately using immature flowers and pistillate-, interim-and staminate-phase anthetic flowers. The flowers were collected and sealed in previously unused polypropylene bags, and a solid-phase microextraction (SPME) fibre with a 65-μ m divinylbenzene/polydimethylsiloxane coating inserted for at least 2 hours to adsorb volatile compounds. Air was also sampled from an empty polypropylene bag as a control, enabling any volatiles identified from the sample to be excluded from the floral scent analyses. Gas chromatography-mass spectrometry (GC-MS) was used to analyse the SPME fibres, using an Agilent 6890N gas chromatograph (Agilent Technologies, Palo Alto, California) coupled with an Agilent 5973 mass selective detector with a 30 × 0.255 mm i.d. DB-WAX capillary column and a 0.25 mm film (J & W Scientific, Folsom, California), with helium as the carrier gas. The SPME fibre was inserted into the inlet for 1 minute at 250 °C to allow vaporization of volatiles; the oven was then maintained at 50 °C for 5 min, and subsequently increased at 5 °C per minute until reaching 230 °C. Verification of chemicals was achieved by comparing retention times of selected pure compounds.
The NIST 2014 MS library bundle (National Institute of Standards and Technology, Gaithersburg, Maryland) was used to determine compound identity, although compounds with estimated identity likelihoods < 80% were considered as unknown. Chemicals that were recorded in different floral phases (non-receptive, pistillate, interim and staminate) were listed after excluding those found in the controls. The volatiles regarded as insect attractants were then listed. PCR products were added to 1.8% agarose gels made with SYBR Safe DNA gel stain (Invitrogen, Hong Kong). PCR products were then transferred to the gel bathed in × 1 TAE buffer and run at 85 V for 65 min. A 100-bp DNA ladder (Fermentas, Hong Kong) was added for determining loci for each primer by comparing fragment sizes. The gels were then exposed under UV and bands recorded using a UVP gel documentation system. Bands were recorded as either present or absent, and the data matrix analysed using POPGENE ver. 1.32 35 with the assumption that the populations are in Hardy-Weinberg equilibrium at all loci.

Results
Floral phenology. Goniothalamus tapisoides and G. suaveolens are both clearly protogynous, with anthesis of similar duration (23 h and 25 h, respectively), extending over two days (Fig. 4). Growth of the outer petals exposes the basal apertures between the inner petal 'claws' , allowing pollinators to enter the otherwise enclosed floral chamber (a in Fig. 4). The pistillate phase begins in the morning with the formation of sticky stigmatic exudate (b in Fig. 4), associated with a sweet fruity scent, and continues until the early evening (e in Fig. 4). The apertures become blocked by the outer petals (trapping any potential pollinators) prior to the end of the pistillate phase (around dusk of the same day; d in Fig. 4). The apertures are closed as the anthers dehisce at the start of the staminate phase (f in Fig. 4) and remain closed until dawn the next day when the corolla abscises (g in Fig. 4). There is clear evidence of a non-sexual interim phase separating the pistillate and staminate phases of both species. There is no evidence of flowering synchrony in either species.
The floral phenological changes are described in detail below, highlighting the differences between the two species: Stage I: Pre-receptive phase. The outer petals change from red-tinged to cream in G. tapisoides (Fig. 2a,b) and from green to yellow in G. suaveolens (Fig. 3a,b). The outer petals are initially compressed against the inner petal dome (Figs 2a and 3b), blocking the chamber apertures and hence preventing pollinators from accessing the floral reproductive organs. The outer petals subsequently begin to rise, exposing the apertures (Figs 2b and 3c) from 0300 hours in G. tapisoides and 0600 hours in G. suaveolens (a in Fig. 4).
Stage II: Pistillate phase (duration: 13 h in G. tapisoides; 9 h in G. suaveolens). The stigmas become receptive in the morning (0600 hours in G. tapisoides, 0900 hours in G. suaveolens; b in Fig. 4), as indicated by the formation of sticky stigmatic exudate (Figs 2e and 3e), which remains visible until that evening (1900 hours in G. tapisoides, 1800 hours in G. suaveolens; e in Fig. 4). The flowers emit distinctive floral odours (a sweet watermelon-like scent in G. tapisoides, and a strong banana-like scent in G. suaveolens) throughout this stage. The colour of the inner petals intensifies (from paler to darker brown) from 0900 hours in G. tapisoides (Fig. 2c), with the outer petals becoming brown after 1800 hours (Fig. 2f); no significant colour change was observed in G. suaveolens flowers,  (Fig. 5a,e). Subsequent growth of the outer petals closes these apertures from 1600 hours in G. tapisoides (Fig. 2f), and from 1700 hours in G. suaveolens (Fig. 3g; d in Fig. 4), effectively trapping pollinators inside the pollination chamber (Fig. 5f).
Stage III: Interim phase (duration: 3 h in G. tapisoides; 9 h in G. suaveolens). The stigmatic exudate begins to dry from 1900 hours in G. tapisoides and 1800 hours in G. suaveolens (e in Fig. 4). This drying is associated with browning of the carpels (Figs 2g,h, and 3f) and is indicative of the end of the pistillate phase: since the anthers have yet to dehisce it is clear that the flower is not sexually functional during the interim phase.
Stage IV: Staminate phase (duration: 7 h in both species). The onset of this phase is indicated by anther dehiscence (Figs 2i and 3h), which begins in the late evening (2200 hours) of the first day in G. tapisoides and in the early morning (0300 hours) of the second day in G. suaveolens (f in Fig. 4). The stamens partially abscise later, remaining suspended from the receptacle by extended tracheary elements (evident in Fig. 2j). The floral chamber remains tightly closed throughout, with the pollinators trapped inside. The staminate phase ends in the early  Fig. 4). Thirteen different floral visitors belonging to five orders were recorded for G. tapisoides. Four of these species were small beetles that were retrieved from within the floral chamber (Table 1): three small curculionid beetles (Endaeus spp. 1-3: Supplementary Fig. S1a,b,d); and a small nitidulid beetle, tentatively identified as 'cf. Carpophilus' (Supplementary Fig. S1e). The other floral visitors were only observed outside the floral chamber, and included two rarely observed unidentified beetle species (both larger than the apertures in the floral chamber and therefore unable to access the floral reproductive organs; one individual observed per species), as well as ants (four species; a few individuals observed), and single observations each of fly, spider and cockroach species. Amongst these floral visitors, only the curculionid and nitidulid beetles are potential pollinators of G. tapisoides since they were retrieved from within the floral chamber and observed to pass through the apertures in the chamber (Fig. 5e).
The two most commonly observed beetle pollinators of G. tapisoides were Endaeus spp. 1 and 2 (Table 1). Endaeus sp. 1 was observed to feed on stigmatic exudate during the pistillate phase (Fig. 5b). Pollen grains were observed on the bodies of Endaeus sp. 2 ( Supplementary Fig. S1c), retrieved from within the floral chamber of pistillate-phase flowers. Endaeus spp. 2 and 3 were observed copulating on the outer petals of pistillate-phase flowers (Fig. 5c,d). The 'cf. Carpophilus' beetle was observed to be active inside the floral chamber during the pistillate phase. The maximum number of beetle individuals found in a closed chamber was three.
Seven different floral visitors belonging to two orders were associated with G. suaveolens. Five of these species were beetles (Table 1), including three small curculionids (Endaeus spp. 1 and 2: Supplementary Fig. S1a,b; and a species tentatively classified as 'cf. Endaeus': Supplementary Fig. S1f), and two small nitidulids (Carpophilus cf. marginellus Motschulsky: Supplementary Fig. S1g; and Carpophilus (Ecnomorphus) cf. dilutus Motschulsky: Supplementary Fig. S1h). In addition to these five beetle species, which were all observed inside the floral chamber, two Drosophila species (Diptera) were observed to visit G. suaveolens flowers, although they did not enter the floral chamber. Goniothalamus tapisoides and G. suaveolens therefore share two potential pollinators: Endaeus spp. 1 and 2.
Carpophilus cf. marginellus (Fig. 5h) was the most abundant species observed inside G. suaveolens flowers, with up to 18 individuals in a single floral chamber; despite the frequency of these visits, there was no evidence of pollen grain attachment. The rarely encountered 'cf. Endaeus' species was observed to probe between stigmas and to bore holes into the receptacle after the end of the staminate phase and petal abscission (Fig. 5i). This species was notably seen carrying pollen grains (Fig. 5i). Endaeus sp. 1 was found near the apertures of pistillate-phase flowers (Fig. 5g) and also inside floral chambers.
Floral thermogenesis and scent chemistry. Comparison of internal floral and ambient temperatures revealed no evidence of thermogenesis or heat retention in either species. Goniothalamus tapisoides flowers emit a weak watermelon-like scent during the pistillate phase, but become almost odourless during the interim and staminate phases. A total of 41volatile compounds were identified from the scents of mature (pistillate-, interimand staminate-phase) and non-receptive flowers of G. tapisoides (Supplementary Table S1). Mature flowers of G. suaveolens emit a strong sweet banana scent, which is most intense during the pistillate phase. A total of 38 volatile compounds were recorded (Supplementary Table S2).

Discussion
Floral biology, pollination ecology and breeding system. Goniothalamus tapisoides flowers are protogynous with 23-h anthesis, beginning in the early morning (0600 hours) and ending the following morning (0500 hours). As with all hermaphroditic-flowered Annonaceae species 36 , the flowers are protogynous. There is evidence of a 3-h non-sexual interim phase separating the pistillate and staminate phases, which would effectively preclude autogamy. Although pistillate/staminate-phase floral synchrony (in which flowers on an individual are synchronised in order to avoid maturation on consecutive days, hence avoiding geitonogamy) is likely to be widespread in Annonaceae 37 , there is no evidence for such synchrony in G. tapisoides: 25% of flowers are borne on the same tree on consecutive days. The likelihood of geitonogamy occurring in this species is minimised, however, by the limited number of mature flowers borne on the same individual on consecutive days, and it seems probable that G. tapisoides is predominantly xenogamous. This conclusion is supported by the results of the ISSR analysis: within-population gene diversity (H S ) contributes c. 90% of total genetic diversity ( Table 2)    Endaeus beetles are likely to be attracted to G. tapisoides flowers by the watermelon-like scent that is emitted during the pistillate phase. GC-MS analysis of the scent revealed four volatiles that have been implicated as chemical signals (attractants, allomones, kairomones or pheromones) for Coleoptera. Three compounds (1,3,3-Trimethyl-2-oxabicyclo[2.2.2]octane, phenylmethanol and phenol) were found to be semiochemicals for Curculionidae [39][40][41][42][43] (Table 3), which are the main pollinators of G. tapisoides (although it should be noted that 1,3,3-trimethyl-2-oxabicyclo[2.2.2]octane was only recorded from staminate-phase flowers, and is therefore unlikely to function as an attractant in G. tapisoides). The attractant phenylmethanol, which is only produced during the pistillate phase, has a sweet aroma, previously recorded in papaya fruits 44 and is therefore likely to contribute to the fruity component of the floral scent. The scent dissipates towards the end of the pistillate phase, but by that stage the apertures in the floral chamber are obstructed by the movement of the outer petals, effectively trapping the beetles within the chamber (discussed below). The beetles are initially likely to be rewarded by stigmatic exudate: Endaeus sp. 2 was observed to feed on exudate, supporting previous reports for Endaeus on flowers of other Annonaceae species 26 . This food source would become unavailable to beetles during the staminate phase, but again it is noteworthy that the beetles are trapped within the floral chamber at this stage.
Goniothalamus suaveolens flowers are similarly protogynous with 25-h anthesis, from 0900 hours until 1000 hours the following morning. As with G. tapisoides, the pistillate and staminate phases in G. suaveolens flowers are separated by a non-sexual interim phase; this phase is considerably longer than that of G. tapisoides, extending over 9 h, but similarly functions to preclude autogamy. Pistillate/staminate-phase floral synchrony was not observed (17% of flowers were borne on the same tree on consecutive days), and hence geitonogamy is possible. As with G. tapisoides, however, the likelihood of geitonogamy is reduced by the limited number of mature flowers borne on the same individual on consecutive days. The conclusion that G. suaveolens is predominantly xenogamous is also be supported by the ISSR data: within-population gene diversity (H S ) contributes c. 75% of the total diversity (Table 2). Although the coefficient of genetic differentiation between populations (G ST ) in G. suaveolens was 0.27 (typical for mixed-mating species), the level of gene flow between populations (N m = 1.35) was high: average N m values for mixed-mating and outcrossing species are 0.727 and 1.154, respectively 45 . Evidence for outcrossing is also provided by the gene identity (I) and genetic distance (D) values which suggest that the three populations are genetically similar despite the considerable distance between populations and that they may have originated from a common ancestral population.   All five species of beetle retrieved from within the floral chamber of G. suaveolens are potential pollinators. Although two of these species are shared with G. tapisoides (Endaeus spp. 1 and 2), these species are more likely to be the effective pollinators of G. tapisoides than G. suaveolens because of their very infrequent visits to the latter. Carpophilus cf. marginellus (Nitidulidae), however, was likely to be an effective pollinator of G. suaveolens given its frequent visits to mature flowers. Six volatiles identified from the floral scent of G. suaveolens are reported to be semiochemicals for Coleoptera, with four regarded as specific attractants to Nitidulidae or Curculionidae (Table 3). Among these, 3-methylbutyl acetate, ethyl hexanoate and 2-phenylethanol, which are reported as attractants or kairomones to Nitidulidae [46][47][48] and were only released by G. suaveolens flowers during the pistillate phase, are noticeably absent from the floral scent of G. tapisoides: this may explain the infrequency of nitidulid visits to G. tapisoides flowers. 2-phenylethanol has been reported as a kairomone for Carpophilus 48 , which is the genus identified as the major pollinator for G. suaveolens. Several curculionid attractants are present in the floral scents of both Goniothalamus species, possibly explaining visits by Endaeus spp. 1 and 2 to both. It is also noteworthy that 3-methylbutyl acetate and ethyl hexanoate are aliphatic esters associated with fruity odours [49][50][51] : the former, which is a key component of banana scent 50 , was only recorded during the pistillate phase, when the banana-like scent was evident.
Pollinator trapping. The pollination chamber in Goniothalamus flowers is formed by apically connivent and postgenitally fused inner petals, with small apertures located between the basal 'claws' of the inner petals that are sometimes blocked by the outer petals that alternate with the inner petals. This type of chamber ('type III' 52 ) is relatively common in Annonaceae, and is hypothesised to have evolved independently in several lineages 52 . Floral chambers are believed to serve several functions in the family, including maintenance of a micro-environment optimized for pollinators (e.g., retaining floral heat) and providing a protected feeding, copulation and/or ovipositing site. Although essentially closed, the floral chamber generally does not prevent the arrival or departure of floral visitors 37 .
We present evidence of a 'trapping' mechanism that effectively prevents beetles from leaving the flower once the floral chamber has closed: this is the first time that such a mechanism has been reported in Annonaceae. Pollinator trapping in G. tapisoides and G. suaveolens occurs before the end of the pistillate phase (d in Fig. 4): the beetle pollinators therefore cannot leave the flower until petal abscission (g in Fig. 4), after the end of the staminate phase. Trapping in Goniothalamus is achieved by growth of the outer petals so that the adaxial surface of each petal is compressed against the apertures in the inner petal dome: the margins of the inner petal claws are glabrous (Fig. 5e) and are therefore able to closely compress against the corresponding glabrous region on the adaxial surface of the outer petals (Fig. 2k). The margins of the inner petal claws are often extended to form a thin extended flange in Goniothalamus tapisoides ( Fig. 2l; see also G. tomentosus 53 ), enhancing the intimacy of the connection. Field observations of G. tapisoides and G. suaveolens suggest that the trapping mechanism is particularly effective in the former due to the presence of an extended flange and the additional growth of the outer petals, further blocking the apertures. It is noticeable that the pollination chamber in G. tapisoides cannot easily be opened by artificially raising the outer petals; the trap in G. suaveolens flowers, in contrast, is looser. It seems unlikely that trapped beetles would be able to leave closed chambers in either species, and this was never observed during our field observations. Based on morphological assessment of other species in the genus, it appears that this trapping mechanism is common to all species and is likely to be a key evolutionary innovation of considerable functional significance.
Anthesis in Annonaceae normally extends over three days (36-50 h): e.g., 36 h in Huberantha korinti (as 'Polyalthia korinti') 22 and Monoon coffeoides (as 'Polyalthia coffeoides') 22 ; 39 h in Mitrephora heyeana 54 57 . In these cases, the timing and duration of the pistillate phase on the first day of anthesis is similar to that of the generally staminate phase the next day, and as a result pistillate-and staminate-phase flowers can co-exist without significantly encouraging geitonogamy: pollen-laden beetles departing from post-staminate-phase flowers (in the late evening) would not be attracted to other flowers entering the pistillate phase until early the following morning, and this time difference would enhance opportunities for pollinator movement between different trees, albeit subject to attrition costs associated with pollinator loss.
The two Goniothalamus species studied here have compressed anthesis (23 h in G. tapisoides; 25 h in G. suaveolens); similarly abbreviated anthesis has been recorded in Dasymaschalon (26 h) 37 . Anthesis commonly extends over two days in early-divergent angiosperms, with the pistillate phase usually starting in the morning and the staminate phase ending in the morning of the next day 59 . Abbreviated anthesis may provide a selective advantage by increasing the turnover rate of flowers and hence promoting fruitset. One of the consequences of compressed anthesis, however, is that the different sexual phases of the flowers will occur at different times of the day (Fig. 4). Interfloral movement of beetles between staminate-and pistillate-phase flowers is optimized, however, by delaying beetle departure from flowers until the onset of the pistillate phase in other flowers. Pollinator trapping in species with short anthesis not only enhances pollination efficiency by increasing the likelihood of interfloral pollen transfer, but also increases the time during which the pollinators inside staminate-phase flowers can inadvertently collect pollen. Since the pollinators become trapped within flowers prior to the end of the pistillate phase, there is no selective advantage for the plant in continuing to provide pollinator attractants or rewards: the plant is therefore able to limit its reproductive investment by terminating floral scent production much earlier than is typical in the family. Pollinator trapping is widespread in the genus, including G. malayanus Hook. f. & Thomson, G. parallelivenius Ridl. and G. velutinus (pers. observ.).
Scientific RepoRts | 6:35674 | DOI: 10.1038/srep35674 Coevolution and partitioning of pollination resources. This is the first study focusing on the evolution of reproductive isolation between sympatric species of early-divergent angiosperm. We provide evidence that supports the hypothesis that pollinator resources are partitioned among two sympatric congeners, Goniothalamus suaveolens and G. tapisoides. Partitioning is mainly achieved by ethological and temporal isolation, both of which are likely to be effective in reducing interspecific pollen transfer.
Ethological isolation between these species can be inferred from the chemical analysis of floral scents. The flowers of these species emitted distinctive scents: watermelon-like in the case of G. tapisoides, and banana-like in G. suaveolens. Although these scents were shown to include many volatiles that are generalist coleopteran attractants, the GC-MS analysis also provides evidence of specific specialisation for curculionid and nitidulid beetles (Table 3), which are both frequent floral visitors to the two species (Table 1). Goniothalamus suaveolens flowers, which are primarily pollinated by Carpophilus beetles (Nitidulidae), contains three volatiles that have been identified as kairomone or attractants for nitidulids (Table 3), viz. 3-methylbutyl acetate 60 , ethyl hexanoate 47 and 2-phenylethanol 47,48 . In marked contrast, the floral scent of G. tapisoides did not contain volatiles that have previously been linked specifically with nitidulids ( Table 3). The effective pollinators of G. tapisoides are likely to be Endaeus beetles (Curculionidae), and chemical analysis of the floral scent revealed three volatiles (1,3,3-trimethyl-2-oxabicyclo[2.2.2]octane, phenylmethanol, and phenol) that have previously been identified as semiochemicals for curculionids [39][40][41][42][43] (Table 3).
Flower-level phenology also contributes to reproductive resource partitioning among these species. Goniothalamus tapisoides flowers are sexually functional for 23 h from 0600 hours, whereas G. suaveolens flowers are functional for 25 h from 0900 hours. The floral phenology of the two species are separated by several hours: pollen-laden beetles leaving G. tapisoides flowers at the end of the staminate phase around 0500 hours (g in Fig. 4a) would be attracted to pistillate-phase flowers that become functional shortly after (from 0600 hours; b in Fig. 4a); in G. suaveolens, however, the pollen-laden beetles leaving post-staminate flowers around 1000 hours (g in Fig. 4b) would be attracted to pistillate-phase flowers that are already functional (0900 hours; b in Fig. 4b).
Since the beetles presumably respond immediately to other attractants, it is likely that beetles would move between conspecific flowers rather than moving between different species.
These two sympatric and phylogenetically closely related Goniothalamus species therefore show reproductive resource partitioning as a result of within-flower phenological differences (in which anthesis is separated by a few hours) and the production of floral scents that are specific to different beetle families. Another key observation is that these two Goniothalamus species possess abbreviated anthesis of 23-25 h, correlated with a previously undescribed trapping mechanism. This mechanism strengthens pollination efficiency by increasing the likelihood of interfloral pollen transfer and the time that beetles are able to collect pollen during the staminate phase.