Floral attraction and flower visitors of a subcanopy, tropical rainforest tree, Fontainea picrosperma

Abstract Flowering plants in tropical rainforests rely heavily on pollen vectors for successful reproduction. Research into pollination systems in tropical rainforests is dominated by canopy species, while subcanopy plant–pollinator interactions remain under‐represented. The microclimate beneath the rainforest canopy is characterized by low light levels and is markedly different from the canopy environment that receives more light energy. We studied the floral attractants and floral visitors of a dioecious, subcanopy tree, Fontainea picrosperma (Euphorbiaceae), in the Wet Tropics bioregion of northern Queensland, Australia. We found that wind pollination is rare and male and female flowers do not produce nectar. Female flowers are likely pollinated due to their perceptual similarity to pollen‐offering male flowers. Female flowers had the same scent profile as male flowers, and floral scent was an important floral attractant that acted to regulate pollinator behavior. The two most abundant scent compounds present in the floral bouquet were benzyl alcohol and 4‐oxoisophorone. These compounds are ubiquitous in nature and are known to attract a wide variety of insects. Both day‐time and night‐time pollinators contributed to successful pollen deposition on the stigma, and diurnal flower visitors were identified from several orders of insects including beetles, flies, predatory wasps, and thrips. Fontainea picrosperma is therefore likely to be pollinated by a diverse array of small insects. Synthesis. Our data indicate that F. picrosperma has a generalist, entomophilous pollination syndrome. The rainforest subcanopy is a distinctive environment characterized by low light levels, low or turbulent wind speeds, and relatively high humidity. Female flowers of F. picrosperma exhibit cost‐saving strategies by not producing nectar and mimicking the smell of reward‐offering male flowers. Insects opportunistically forage on or inhabit flowers, and pollination occurs from a pool of small insects with low energy requirements that are found beneath the rainforest canopy.


| INTRODUC TI ON
Many angiosperms in tropical rainforests exhibit breeding mechanisms that promote outcrossing and rely almost exclusively on animals for pollination Ollerton et al., 2011). Thus, pollination in rainforest ecosystems largely depends on animal abundance, distribution, and the flight behavior of flying insects (Aguilar et al., 2008;House, 1992House, , 1993Stacy et al., 1996). Mutualistic plant-pollinator interactions help to govern and maintain the rich biodiversity in tropical rainforests which are renowned for their high levels of tree species diversity and animal pollinator diversity (Boulter et al., 2006). Generalized pollination syndromes are common in tropical rainforests, presumably because generalization is favored in species that promote outcrossing and have no reproductive assurance Corlett, 2004;Vamosi et al., 2013).
Pollinator assemblages vary across tropical rainforest communities geographically and among vertical forest strata within rainforests. For example, trees beneath the canopy in Costa Rica receive more visits by pollinators such as hummingbirds and beetles than canopy trees. In contrast, canopy species receive more visits from medium-to large-sized bees (Bawa, 1990). Bee assemblages are also known to differ between strata in the Brazilian Tropical Atlantic Rainforest (Ramalho, 2004). Vertical strata are defined as understory which includes plant species <5 m in height; subcanopy that is comprised of trees >5 m tall but less than canopy stature at reproductive maturity; and canopy species which includes trees flowering at the top of the forest, including emergent trees (Kress & Beach, 1994). Temporal availability of floral resources as well as differences between plant species composition can result in differences in pollinator or floral visitor assemblages among vertical strata. Moreover, the rainforest environment below the canopy is markedly different to the forest canopy where the high light energy input results in greater resource availability and allows tree species to produce highly productive flowers (Appanah, 1991;Roubik, 1993). The total radiation reaching the understory in closed tropical rainforests is 2-3 percent of that reaching the canopy (Mabberley, 1992). In addition, the microclimate beneath the canopy has high relative humidity, stable diurnal air temperatures, and low wind speeds or turbulent wind patterns compared to the canopy (Corlett, 2004;Kato, 1996;New, 2018). These conditions can pose challenges for pollinating insects and can reduce the effectiveness of aerial pollen transport (Corlett, 2004;Kato, 1996).
Tree species that occupy the rainforest beneath the canopy often possess small, white or green flowers, with small amounts of pollen and/or nectar (Boulter et al., 2006;Renner & Fell, 1993;Machado & Lopes, 2004). The poor floral rewards offered potentially reflect the low light energy environment of the rainforest subcanopy as plant resource availability is affected by the amount of light energy available (Appanah, 1991;Bawa, Bullock, et al., 1985). In turn, rainforest, subcanopy plant species may be unable to support the energy requirements of large pollinators due to the energetic costs associated with flower and nectar production (Appanah, 1991;Borrell, 2005;Pyke, 1991). Scent, rather than visual attraction, may be an important attraction signal in the shaded rainforest beneath the canopy (Williams & Adam, 2010). However, generalizations between floral scent compounds and pollination by different functional groups of insects can be difficult to render because of the lack of literature available (Cordeiro et al., 2019;Raguso, 2008).
subcanopy tree endemic to the Australian Wet Tropics bioregion in upland tropical rainforests on the Atherton Tablelands, Queensland, Australia. Fontainea picrosperma's current distribution has been disrupted by anthropogenic habitat fragmentation primarily due to agricultural expansion, but also due to urban settlements. The Australian Wet Tropics makes a significant contribution to global biodiversity, harboring 16 of the 28 near-basal (or "primitive") lineages of flowering plants in addition to a high level of species endemism (Metcalfe & Ford, 2009). Despite this significance, there are few published studies of flower visitors in this bioregion and even fewer empirical assessments of successful pollinators (Boulter et al., 2009). Moreover, relatively few studies globally have focused on the pollination systems of plant species in the subcanopy, an environment that is markedly different to the rainforest canopy, and our understanding of plant-animal interactions in subcanopy rainforest species is sparse.
The pollination biology of F. picrosperma is also of significant commercial interest because its fruit contains the small molecule natural product, tigilanol tiglate (Boyle et al., 2014), which has recently been approved as a veterinary oncology product as a therapeutic for canine mast cell tumors (De Ridder et al., 2020) and is in clinical development as a local treatment of solid tumors in humans (Panizza et al., 2019). Tigilanol tiglate is not synthetically tractable and consequently relies on harvest of plantation-grown F. picrosperma fruit for its production. As apomixis (asexual reproduction) is not a significant means of reproduction in F. picrosperma (Grant et al., 2017), it is important to understand how pollen is delivered to Australian Wet Tropics, dioecious, EBC-46, Euphorbiaceae, floral scent, generalist pollination syndrome, tigilanol tiglate, understory the stigma. The aim of this study is to examine the floral attractants and flower visitors of a dioecious, subcanopy rainforest species, F. picrosperma. We asked the following questions: (a) What are the modes of pollen dispersal and deposition of F. picrosperma? (b) What are the main floral attractants in this subcanopy rainforest species? (c) What animals are attracted to the flowers in a subcanopy environment? Specifically, we ascertained whether wind is a potential pollination method, determined differences in the species' pollen deposition between nocturnal and diurnal pollinators, investigated floral attractants, and characterized the floral scent profile. We carried out flower observations, collected in-flower visitors, and used scent lures to trap insects attracted to the main compounds present in F. picrosperma's floral bouquet.

| Study species
Fontainea picrosperma is a subcanopy tree that grows up to 25 m tall (Jessup & Guymer, 1985). It is endemic to the mesophyll and notophyll vine forests on the Atherton Tablelands, north Queensland, Australia. Male and female individuals possess inflorescences with small, white flowers that have an unspecialized structure and an open access receptacle (Figure 1; see also Grant et al., 2017).

However, flowering phenology differs between sexes and individual
female flowers open for significantly longer than individual male flowers (Grant et al., 2017). Flowering occurs simultaneously between individuals of subpopulations from September to November.
The red drupaceous fruit (up to 3 cm diameter) ripen in December and January and are primarily dispersed by gravity. Natural stands of F. picrosperma therefore are not uniformly distributed; they form small, isolated but dense clumps (2-10 m intertree spacing) with ∼50:50 male:female ratios (Grant et al., 2017;Lamont et al., 2016).
Fontainea picrosperma has a geographically restricted natural range, and its distribution has been heavily influenced by both natural and anthropogenic habitat fragmentation (Lamont et al., 2016).
The locations of the populations included in this study were Boonjie and Evelyn Highlands as described in Lamont et al. (2016). The subpopulations of "Evelyn Highlands 1" and "Evelyn Highlands 2" are from the same refugial population approximately 3.5 km apart.
Boonjie and Evelyn Highlands are the two largest, continuous rainforest areas where F. picrosperma occurs. Data were collected over three flowering seasons in 2014, 2015, and 2016.

| Pollen dispersal and deposition
To determine if wind contributes to pollination, microscope slides smeared with a thin layer of Vaseline™ the size of a cover slip (300 mm 2 ) were hung from three male and three female flowering F. picrosperma trees (three slides per tree) in Evelyn Highlands 2. Slides were installed 1.5-3 m from the ground for 24 hr in the zone where flowers occur. New slides were installed each day for three replicate days. Once the slides were removed, the area of  Leica Microsystems) after staining with decolorized aniline blue (Kho & Bear, 1968;Shepherd, 2000).

| Floral attractants
Five female trees and four male trees were sampled for diurnal nectar secretion from the natural population, Evelyn Highlands 2. Individual flowers were tested for nectar rewards using a wash method following Morrant et al. (2008). This method is recommended for species with low floral nectar volumes and can measure nectar that accumulates from trichomal nectaries. Unopened buds were tagged and then bagged to exclude flower visitors for the duration of the experiment. One flower per tree was measured at various times on each day of sampling, when the flower was 0, 2, 3, and 5 days old. Each flower was placed into a vial containing 0.5 ml of distilled water. The sample was manually agitated for 5-10 min. The sugar concentration was measured using a hand-held BRIX refractometer (ATAGO PAL-S). This is a destructive method; therefore, a flower sampled at 5 days after opening, for example, remained bagged for the 5 days.
We identified volatile compounds emitted from F. picrosperma liage (e.g., Arimura et al., 2001;Grison et al., 1999). However, the short sampling time of SPME minimizes the collection of artifacts due to damage to the plant (Flamini et al., 2003). Six male and six female trees were used in the experiment. Flowers were sampled at different ages relative to opening (male 0, 1, 2, and 3 days; female 0, 1, 2, 3, 5, 7, 9, and 11 days). The differences in sampling ages between male and female flowers reflect the different male and female flower opening times. Three to six replicate flowers were sampled per flower age per sex (n = 23 males flowers and n = 24 female flowers in total). Flowers were removed between 0900 and 1500 hr and placed in a 10-ml septum cup vial to allow volatile odors to equilibrate for 1 hr at room temperature (22℃). A background control of the glasshouse and sample vials was also sampled each day of floral volatile testing. Volatiles from each sample were then collected using a solid-phase microextraction (SPME) holder (Augusto & Luiz Pires Valente, 2002). For each sample, the Supelco fiber coated with polydimethylsiloxane (PDMS, 100 µm) was manually inserted into the headspace chamber and exposed for 25 min at 22℃.
The helium carrier gas had a constant flow of 1 ml/min. The SPME fiber was manually inserted into the injection port fitted with a splitless liner at 200℃ and maintained for 2 min to allow the adsorbed volatiles to thermally desorb onto the GC column. The split ratio was shut from −0.5 to 2 min and then open at 30:1. The temperature program was operated at 40℃ for 2 min, ramping at 10℃/ min until 210℃ and holding for 1.5 min. The mass spectrometer analyzed a mass range from 40 to 250 (m/z), from 1 to 20.5 min at 70 eV.
Compounds were identified by comparison of (a) retention times to authentic standards, (b) Arithmetic Index (AI) (Adams, 2007), and (c) mass spectra of National Institute of Standards and Technology (NIST 08) Mass Spectral library.
Compounds present at similar abundance in the control samples were considered to be contaminants and were excluded from the analysis. Proportional abundance (relative amounts with respect to aggregate peak areas, excluding contaminants) of the individual constituents was calculated based on peak area measurements and expressed as a percent of the sum total peak area.

| Flower visitors
Two F. picrosperma populations, Evelyn Highlands 2 and Boonjie, were surveyed for diurnal flower visitors. Three male trees and three female trees for each population were each observed for a 10-min period every 2 hr (0820-1720 hr). Observations were conducted on sunny days with light winds during times conducive to diurnal insect activity. A total of 25 observation hours (5 days

| Statistical analysis
We tested for differences in the sum total of all detected compounds emitted between different flower ages within sex using a generalized linear model (GLM) with a negative binomial distribution and a log link function (package "MASS"; Venables & Ripley, 2002). Where differences were detected, we applied a Tukey's HSD test. For analyses of semiquantitative (i.e., relative amounts of scent compounds within a flower) differences in scent between sex, total peak area of compounds were fourth-root transformed before analysis. We calculated the Bray-Curtis similarity index and performed a permutational multivariate analysis of variance fitted to the distance matrices (PERMANOVA) (999 permutations) using the Adonis function (package "vegan"; Oksanen et al., 2016). We first analyzed the combined ages 0-3 days and then compared each age separately to test for differences between sex as the flower aged (package "pairwise.adonis"; Martinez Arbizu, 2019). Analyses were performed in R v.3.5.1 (R Development Core Team).
We tested for treatment (diurnal; nocturnal; open to flower visitors) effects on the number of pollen grains detected on a stigma using a GLM with a negative binomial distribution and a log link function to correct for overdispersion due to inflated zeros in the data (package "MASS"; Venables & Ripley, 2002). Differences between treatments were assessed using Tukey's HSD post hoc test (package "multcomp"; Hothorn et al., 2008). Analyses were performed in R (v.3.5.1).

Counts of individual morphospecies found in paired collision
and sticky traps were combined. We conducted a Mann-Whitney U test (SPSS v 24) to determine if there were significant differences (p < 0.05) between the number of individuals of each morphospecies in the control and scent lure traps. Only morphospecies with more than 10 individuals in the insect traps were included in the analysis (n = 29). Morphospecies with significant differences between treatment and control (n = 5) were identified to genus level. We also combined the number of individuals of each morphospecies within an order and tested for significant differences between treatments (p < 0.05, Mann-Whitney U test). Analyses were performed using IBM SPSS v 24.

| Pollination
To track the movement of pollen, we tested for the potential of wind pollination and the effectiveness of day and night pollinators.
Few airborne pollen grains were captured in the natural population (Table 1).
Pollination treatments affected pollen deposition in the natural population, Evelyn Highlands 2 ( Figure 2). Significantly more pollen grains were found on open (control) flowers than on flowers that were only exposed to nocturnal visitors (GLM, p = 0.005; Figure 2).
There was no significant difference in the number of pollen grains between open flowers and flowers that were exposed to visitors during the day (GLM, p = 0.373; Figure

| Floral attraction
There was no diurnal nectar secretion in F. picrosperma flowers. Nor Benzyl alcohol and 4-oxoisophorone (2,6,6-Trimethyl-2-cycloh exene-1,4-dione) were the two major scent compounds present in both male and female flowers ( Male flowers emitted significantly less scent at 0 day than male flowers at age 1, 2, or 3 days based on sum of total peak area (GLM, Note: Scent compounds are listed according to compound class. Mean relative percent of each compound is presented using data from flowers aged from 0 to 3 days. Standard error in parentheses. a Compounds were identified based on GC retention times and mass spectra of standard compounds purchased from Sigma-Aldrich. b Compounds were identified using the mass spectrum and AI (arithmetic retention index) (Adams, 2007). The remaining compounds were tentatively

| Flower visitors
Flowers were most frequently visited by insects from the orders Thysanoptera (thrips) and small (<1 cm in length) Coleoptera (beetles) ( Figure 5). More thrips were observed in the Boonjie subpopulation compared to the Evelyn Highlands 2 subpopulation. Thrips were not observed departing from the inflorescence during the day and therefore behaved as permanent inhabitants according to the criteria in Listabarth (1993). Beetles were often observed crawling on all parts of the inflorescence or chewing on flower parts including stigmas, anthers, and petals. The Lepidoptera observed in Evelyn Highlands 2 were all caterpillars, except one unidentified butterfly that briefly landed on the petals of a female flower and flew away without interacting with the stigma.
The most abundant insects captured within sampled inflorescences were thrips (Table 3). A range of small (<1 cm in length) "inflora" flower visitors from whole inflorescences were also present (Table 4). Aside from mosquitoes (n = 2), only a single specimen of each insect (n = 1) was collected in the flowers.
Five genera of insects were found in significantly greater numbers in the scent lure traps compared to the control traps in the Evelyn Highlands 1 population (p < 0.05, Mann-Whitney U test; Figure 6). Three were beetles including Aethyssius (Tenebrionidae: Alleculinae); Ictistygna (Anthicidae: Eurygeniinae); and another alleculine tentatively identified as Euomma. The other two genera were a minute (body length <1 mm) thrips parasitoid wasp that was either a species of Ceranisus or Thripobius (Eulophidae: Entedoninae) and Drosophila (Drosophilidae).
There were no significant differences between scent lure and control traps in any other taxa. When the number of individuals in each order was combined, there was a significant difference in the number of Lepidoptera between treatments (p = 0.034), with more present in the scent lure traps. There was no significant difference in any other order tested.

| D ISCUSS I ON
Our  (Knudsen et al., 2006). A wide variety of small insect taxa were observed visiting flowers of both sexes, and both day and night visitors contributed to pollen deposition on the stigmas.
Together with the unspecialized structure of F. picrosperma flowers (Grant et al., 2017), our data indicate that F. picrosperma has a generalist, entomophilous pollination syndrome.

| Floral attraction
No diurnal nectar secretion was found in F. picrosperma flowers. Our results suggest that pollinators are likely deceived by female flowers that offer no obvious reward due to their perceptual similarity to pollen-offering male flowers (Grant et al., 2017). Bakerian mimicry (where female flowers mimic male flowers and cheat pollinators out of a reward) can occur when females economize resources by not producing nectar (Renner, 2006;Thakar et al., 2003). Since nectar is costly in terms of seed production and photoassimilate production (Pyke, 1991;Southwick, 1984), the poor floral rewards offered potentially reflect the low light energy environment of the rainforest subcanopy (Appanah, 1991;Bawa, Bullock, et al., 1985).

F I G U R E 3
Emission patterns of the sum total of aggregate peak areas of all compounds across the age of flowers of Fontainea picrosperma in (a) male flowers and (b) female flowers Floral volatiles provide a way to attract pollinators, particularly in the shaded rainforest subcanopy where visual cues may be less important (Appanah, 1991;Knudsen et al., 1999;Koski, 2020;Williams & Adam, 2010). This occurs in other tropical species such as the understory palm Geonoma macrostachys (Borchsenius et al., 2016). Female flowers do not set fruit beyond 9 days (Grant et al., 2017), and the senescing stage of female flowers was characterized by an F I G U R E 4 Peak area of the detected floral volatile compounds from male and female Fontainea picrosperma flowers for each age 0-3 days. (a) 4-oxoisophorone, (b) Benzyl alcohol, (c) Methyl benzoate, (d) β-Ocimene (unknown isomer), (e) cis-3-Hexen-1-ol acetate, (f) Benzyl acetate, (g) Benzyl aldehyde, (h) 2,2,6-Trimethyl-1,4-cyclohexanedione, (i) Hexyl acetate, (j) Methyl salicylate, (k) β-Cyclocitral, (l) 1,1-Dimethyl-3-methylene-2-vinylcyclohexane, and (m) Geranyl isovalerate. For each compound, the graphs show the median (quartiles, minimum-maximum) of the total peak area (n = 5-6 for male flowers and n = 3-5 for female flowers) Buddleja, a genus well known to attract butterflies and other insects (Chen et al., 2012(Chen et al., , 2014. The compound also evokes antennal responses in moths, bees, and flies (Andersson, 2003;Guédot et al., 2008). Benzyl alcohol and methyl salicylate, also a constituent of F. picrosperma's floral emissions, are among the most common compounds in floral scent, occurring in 56 percent and 57 percent of families studied, respectively (Knudsen et al., 2006). Benzyl alcohol is known to attract Apis mellifera (honey bees; Dötterl & Vereecken, 2010 and references therein). Benzyl alcohol, as well as several of the other main floral compounds of F. picrosperma such as benzaldehyde, benzyl acetate, and methyl salicylate are known to dominate the floral scent profile of plants with nocturnal anthesis and are associated with moth, hawkmoth, nocturnal bee, nocturnal beetle, and other generalist insect pollination (Etl et al., 2016;Raguso et al., 1996). Other benzenoids present in the floral bouquet of F. picrosperma, such as benzaldehyde, also have a widespread distribution and are thought to be functionally attractive to pollinators (Schiestl, 2010). The monoterpene β-ocimene and the benzenoid methyl benzoate comprised a further 16 percent of the floral scent profile of F. picrosperma. β-ocimene is the most commonly encountered scent compound in generalist plant-pollinator interactions (Dobson, 2006), and cisβ-ocimene has been shown to stimulate foraging behavior in bees (Eltz et al., 2006;Gong et al., 2015). Methyl benzoate is also known to attract bees (Dudareva et al., 2000;Schiestl & Roubik, 2003) and stimulate electroantennogram responses in moth antennae (Raguso & Light, 1998).
These results are one of the few studies on a species of Euphorbiaceae, which remains a poorly sampled family with regard to floral chemistry (Knudsen et al., 2006). Nevertheless, this study shows that the floral bouquet of F. picrosperma contains compounds that have a widespread distribution and are known to attract a wide variety of both diurnal and nocturnal insect pollinators.

| Flower visitors
We determined that F. picrosperma flowers are visited by a diverse set of small insects. This is concordant with many other woody plants in Australian tropical and subtropical rainforest communities (Boulter et al., 2005;House, 1989;Webber et al., 2008;Williams & Adam, 1994;Worboys & Jackes, 2005) and with many species in rainforest communities elsewhere, for example, lowland dipterocarp forests (Kato, 1996;Momose et al., 1998). The floral traits of F. picrosperma suggest little adaptation to exploit specific pollinators (Grant et al., 2017). Both day and night visitors successfully deposited pollen on the stigma of female flowers. Small insect pollinators have low energy requirements and often move shorter distances compared to specialized insects, larger insects, or vertebrates that occur in the canopy (Appanah, 1991;Dick et al., 2008). This is supported by pollen gene flow data for F. picrosperma where pollen is found to disperse short distances (Grant et al., 2019).
Thrips were the most abundant insects observed and collected from both male and female F. picrosperma flowers. In other plant species, thrips inhibit the flowers and feed on the internal wall of the receptacle and pollen and contribute very little to cross pollination (Kondo et al., 2016). Their small body size means their movements can bypass the central stigma of female flowers (Irvine & Armstrong, 1990) and, combined with a weak flying ability, this could preclude thrips from efficient pollen transport because they are only able to carry small quantities of pollen. Fontainea picrosperma pollen grains are 40 µm in diameter (Grant et al., 2017), and pollen grain size in plants pollinated by thrips is usually <34 µm (Sakai, 2001). Thrips were the most common flower visitor observed in this study and by sheer numbers could contribute to pollination and successful reproduction in F. picrosperma. In addition, thrips may indirectly contribute to pollination by attracting predators or parasites. In this study, a specialist thrips parasitoid wasp, either a species of Ceranisus or Thripobius (Eulophidae), was significantly more attracted to the scent lure traps than the control traps with no scent. Other small parasitoid wasps including endoparasitoids of Coleoptera, Hemiptera, and Diptera (Austin et al., 2005) were found in flowers, and these orders of insects were observed visiting the flowers of both male and female trees.
Beetles (Coleoptera) were the second most commonly observed and collected visitors to F. picrosperma flowers. Beetles are thought to be critical in pollination in Australian tropical rainforests throughout the vertical strata and could pollinate up to one quarter of Australian tropical rainforest species (Irvine & Armstrong, 1990;Kitching et al., 2007;Wardhaugh et al., 2015;Webber et al., 2008;Worboys & Jackes, 2005). Thus far, no reliable pattern in floral scent chemistry unifies plants pollinated by tropical beetles (Dobson, 2006). Field observations suggest that the floral tissue provided some food reward targeted by chewing insects, which could cause incidental pollination. Beetles, for example, are known to visit flowers to feed on pollen, various floral tissues, and other floral exudates (Endress, 1996;Momose et al., 1998). Ictistygna sp.
(Anthicidae: Eurygeniinae) was the most abundant species captured in the scent lure traps. Adult anthicid beetles are omnivorous, being F I G U R E 6 Insects (identified to genus level) found in significantly larger numbers in the scent lure traps compared to the control traps with no scent lure (p < 0.05, Mann-Whitney U test) known to consume small arthropods and pollen and some species within the family are known predators of stink bugs (Hemiptera: Pentatomidae) (Athey et al., 2019) and Diaphania spp. (Lepidoptera: Crambidae) (Júnior et al., 2012), though little is known about eurygeniine biology (Lawrence & Ślipiński, 2013). The two other beetle species belonged to the Tenebrionidae, a family that characteristically feed on dead vegetable or animal matter and living plant tissue, although a few normally predacious species are known (Watt, 1974).
Tropical rainforests are rich in dipteran species, and flies are capable of transporting significant quantities of pollen, usually short distances (House, 1989;Weiss, 2001). Significantly more drosophilid flies were captured in the scent baited traps than in the control traps.
In other plant species studied where drosophilid flies were among the most abundant insect visitors, they carried little or no pollen and were inefficient or ineffective pollinators (Borchsenius et al., 2016). This study provides further evidence for the assertion that subcanopy or understory tropical rainforest species with flowers that lack morphological specialization are commonly pollinated by a diverse array of small insects (Appanah, 1991;Bawa, Bullock, et al., 1985;Devy & Davidar, 2006;Momose et al., 1998). Findings in our previous study of F. picrosperma determined that the species is pollen limited (Grant et al., 2017). Pollen transfer in F. picrosperma is predominantly confined to neighboring trees and larger male trees, with more intense floral displays have greater reproductive success than smaller males with less flowers (Grant et al., 2019). Thus, pollinators preferentially travel short distances between conspecific trees likely due to the opportunistic feeding patterns of small insects and lead to the short pollen dispersal distances that characterizes pollen mediated gene flow in F. picrosperma.

| CON CLUS ION
From our study, we determined that F. picrosperma is likely pollinated by an array of small insects, with thrips and beetles a common part of the flower visitor assemblage. This is congruent with observations from many other woody plants in Australian tropical and subtropical rainforests and in the subcanopy of these communities elsewhere where many plant species are pollinated by small, generalist insects.
The rainforest subcanopy environment is a distinctive microclimate for pollinators. In this low light, and therefore low energy environment, female flowers of F. picrosperma exhibit cost-saving strategies by not producing nectar and mimicking the smell of reward-offering male flowers. Male and female flowers displayed the same pattern of floral scent emission, characterized by a decrease in the amount of scent compounds produced as the flower aged. Scent therefore is likely to regulate pollinator behavior and be an important attractant to the species. The main scent constituents found in F. picrosperma are ubiquitous in plant floral bouquets and likely makes them indiscriminate to specific taxonomic groups. Our evidence indicates that pollination likely occurs from small insects with low energy requirements that opportunistically feed on or inhabit F. picrosperma flowers that offer poor rewards in the shaded subcanopy of the tropical rainforest.
This study improves our understanding of fine-scale ecological interactions within a poorly studied tropical rainforest community, the AWT, and contributes to the literature of flower visitors to subcanopy tree species which is an area under-represented compared to canopy tree species. Our focus on ecological aspects of F. picrosperma will also help to secure sustainable seed production for commercial manufacture of tigilanol tiglate. Pollinators are of critical importance in natural ecosystems (Garibaldi et al., 2013). Intact forests act as refugia and increase the presence of native pollinators by allowing them to complete their life cycle in a suitable habitat (Ricketts, 2004;Blanche & Cunningham, 2005;Garibaldi et al., 2011). Appreciating the value of the ecosystem services that wild pollinators provide to maintain biodiversity and to enhance fruit and seed production in agricultural crops reinforces the conservation value of remnant tropical rainforest vegetation in the heterogenic landscape where F. picrosperma persists.

ACK N OWLED G M ENTS
We would like to thank Tracey McMahon and Anastasia George for assistance with sample collection and analysis, and the Queensland Department of Environment and Heritage Protection for providing a Scientific Purpose permit. This work was supported by EcoBiotics Ltd and the University of the Sunshine Coast.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data used in this study will be openly available in the Dryad Digital Repository at https://doi.org/10.5061/dryad.2ngf1 vhnn.