How widespread is pollination by sexual deception of fungus gnats in Pterostylis (Orchidaceae)?

Abstract

Pollination by sexual deception has evolved multiple times in the Orchidaceae, with most known cases involving male Hymenoptera as pollinators. The diverse Australasian orchid genus Pterostylis, characterized by elaborate trap flowers, contains some species pollinated by sexual deception of fungus gnats (Diptera). However, there is considerable variation in floral morphology, suggesting that additional pollination strategies or pollen vectors may be involved. Here, we test the hypothesis that sexual deception of male Diptera is taxonomically widespread by investigating the pollination systems across a representative subset spanning nine out of 10 sections and 18 Pterostylis species. We confirm four new cases of pollination by sexual deception of male fungus gnats (families Mycetophilidae, Keroplatidae, and Sciaridae) and accrued evidence for three further cases. Each of these orchids was pollinated by a single species of fungus gnat, with two species exploiting the same pollinator. Unexpectedly, we observed insect feeding behaviour on two species pollinated by sciarid gnats and phorid flies, respectively, with trace levels of sucrose detected where feeding was observed. Our results show that the sexual deception of male fungus gnats is likely to be the dominant mode of pollination in Pterostylis, although other poorly understood pollination strategies are also present.

INTRODUCTION

A key feature of the diverse plant family Orchidaceae is an abundance of highly specialized plant–pollinator interactions (Johnson and Steiner 2003, Tremblay et al. 2005, Phillips et al. 2020, Ackerman et al. 2023). One of the most highly specialized pollination strategies is that of sexual deception, in which flowers exploit the innate sexual preferences of male insects to achieve pollination (Schiestl 2005, Peakall 2023). Long-range pollinator attraction is achieved by the chemical mimicry of female insect sex pheromones (Schiestl et al. 1999, 2003, Bohman et al. 2016), with the consequence that sexually deceptive orchids typically only have one pollinator species (Peakall et al. 2010, Phillips et al. 2017, Paulus 2018).

Sexual deception is almost entirely restricted to the Orchidaceae, where it is known from 20 genera and several hundred species spanning Australasia, Europe and the southern Mediterranean, South Africa, Central America, and South America (Gaskett 2011, Bohman et al. 2016, Peakall 2023). At present, the only two documented cases of sexual deception outside of the Orchidaceae are Gorteria diffusa (Asteraceae) in southern Africa (Ellis and Johnson 2010) and Iris paradoxa (Iridaceae) in Europe (Vereecken et al. 2012). Based on current knowledge (see reviews in Gaskett 2011, Peakall et al. 2020, Peakall 2023), male Hymenoptera are the most frequent documented pollinators of sexually deceptive plants, but male Diptera (Blanco and Barboza 2005, Ellis and Johnson 2010, Phillips et al. 2014b, Martel et al. 2016, Reiter et al. 2019b) and Coleoptera (Wakamura et al. 2020, Cohen et al. 2021) are increasingly being found as pollinators in sexually deceptive systems.

Sexually deceptive orchids typically have dull coloured reddish or green flowers and an ‘insectiform’ labellum that exhibits highly modified structures such as hairs, hinges to aid with positioning or entrapment of the pollinator, and labellum appendages (see Johnson and Schiestl 2016, Peakall 2023 for photographs). Among some sexually deceptive orchids, the labellum traits may be partly correlated with the type of pollinators exploited. For example, unrelated Ophrys speculum (Europe), Bipinulla penicillata (South America), and Calochilus spp. (Australia) exploit male scoliid wasps as pollinators and all have long hairs on the labellum margin (Ciotek et al. 2006, Johnson and Schiestl 2016, Peakall 2023). Meanwhile, Pterostylis cycnocephala (Australia) and Lepanthes glicensteinii (Central America) both exploit male sciarid fungus gnats (Diptera) and have intricate labellum appendages that the males attempt to copulate with (Blanco and Barboza 2005, Hayashi et al. 2022). By contrast, some species have contrasting labellum shapes, yet exploit the same family of insects. For example, sexually deceptive Drakaea, Chiloglottis, Caladenia, and Paracaleana from the Australasian tribe Diurideae display markedly different floral morphologies yet all are pollinated by male thynnine wasps (Thynnoidea: Thynnidae) (Stoutamire 1974, Phillips et al. 2013, De Jager and Peakall 2016).

With ~300 described species, Pterostylis R.Br. (Orchidoideae: Cranichideae) sits alongside Caladenia (Orchidoideae: Diurideae) as the two most diverse terrestrial orchid genera in Australasia (Clements et al. 2011, Backhouse et al. 2019). Yet unlike Caladenia, there has been surprisingly little formal research on Pterostylis pollination (although see Phillips et al. 2014b, Reiter et al. 2019b, Hayashi et al. 2021, 2022, Bower 2024). All Pterostylis species share some unusual floral traits (Figs 1–4), including the galea that encloses the column (Jones 2021). This hooded structure is formed by the enlarged dorsal sepal and the petals, which overlap to form a chamber (Fig. 1). In some species, upswept lateral sepals (which are fused at their base) form the front of the galea (e.g. Fig. 2A–C), while in other species they protrude downwards (e.g. Fig. 2D–J). In addition, all species have a motile labellum, which is triggered by movement and temporarily traps visiting insects within the galea, and an elongate column with apical wings near the anther that forms a tight ‘escape’ passage where pollen transfer and pollinia removal occur (Cheeseman 1872, Sargent 1909, Jones and Clements 2002a, Jones 2021) (Figs 1–4). The typically green or dull reddish, hooded flowers give rise to the common name of ‘greenhoods’ or ‘rustyhoods’, respectively. However, beyond the characteristic dull coloured flowers there is remarkable variation in floral morphology across the genus, with species featuring variously upswept or deflexed lateral sepals, and a diversity of labellum traits including hairy and triangular, pale with prominent blackish appendages or elongate with long hairs (Figs 1–4).

Early observations of small flies (Diptera) inside Pterostylis flowers led to the expectation of Dipteran pollination (Cheeseman 1872, Coleman 1934, Hyett 1960). Sargent (1909, 1934) observed the pollination process in P. vittata, reporting that a ‘small gnat’ visited the flower, and become entrapped before removing pollen during its escape from the galea. Female mosquitoes (Diptera; Culicidae) and an unknown fly were reported as pollinators of P. procera in Queensland, although the pollination mechanism was not elucidated (Bartareau and Jackes 1994). Additional observations and photographs of male fungus gnats (Keroplatidae, Mycetophilidae, and Sciaridae) visiting Pterostylis flowers and carrying orchid pollinia suggest these Diptera may be important pollinators of Pterostylis (e.g. (Bates 1989, Kuiter 2020)).

More than 100 years after Sargent first observed the pollination process, Phillips et al. (2014b) were the first to confirm the pollination mechanism in a Pterostylis species. They demonstrated that pollination of P. orbiculata (these populations were previously classified as P. sanguinea) occurs by the sexual deception of males of a single undescribed species of fungus gnat (Mycomya sp., Mycetophilidae). The males often show probing copulatory behaviour on the labellum, leading to the labellum triggering and closing into the galea, temporarily entrapping the insect in the flower. Pollen deposition and removal occurs as the fungus gnats escape from the flower via the reproductive structures in the hood of the flower. Subsequent work on the chemistry underpinning pollinator attraction in P. orbiculata revealed that the orchid labellum produces an unusual long-chain triene and other hydrocarbons, which mimic the sex pheromone of the female fungus gnat (Hayashi et al. 2021). A further five Australian species of Pterostylis are now confirmed to sexually deceive male fungus gnats for pollination, with three families of fungus gnats exploited (Mycetophilidae, Keroplatidae and Sciaridae) (Reiter et al. 2019b, Hayashi et al. 2022).

Meanwhile in New Zealand, males of three fungus gnat (Mycetophilidae) species were captured at sticky traps placed near three Pterostylis species, with each species only carrying the pollinia of one orchid species (Thalwitzer et al. 2018). In one of the species, P. irsoniana, a male fungus gnat was also observed displaying abdomen curling on the labellum, suggesting that pollination by sexual deception may be occurring in these species (Thalwitzer et al. 2018). This collective evidence indicates that pollination by the sexual deception of fungus gnats is widespread in Pterostylis. However, this hypothesis is yet to be evaluated systematically across the genus.

The paucity of research on pollination in Pterostylis, especially given the diversity in species number and floral form, presents a significant challenge to our ecological and evolutionary understanding of the genus and is likely to hamper the effective conservation of threatened species. A total of 28 Pterostylis species are listed as nationally threatened in Australia under the Environment Protection and Biodiversity Conservation Act 1999 (EPBC Act) and a further four species are classified as threatened in New Zealand (De Lange et al. 2017). Knowledge of the ecological requirements of these threatened species, including their pollinators and mycorrhizal associations, will enable more informed management to improve their conservation outlook (Reiter et al. 2016, Phillips et al. 2020).

The overall aim of the study was to test the hypothesis that sexual deception is widespread across the different groups and floral morphologies in Pterostylis. Therefore, representatives from nine of the 10 sections within Pterostylis sensu lato (Janes and Duretto 2010) were included in the study, including three threatened species. Following Phillips et al. (2014b), the criteria used as evidence for pollination by sexual deception included observations of pre-mating, pre-copulatory, and attempted copulation behaviours by specific male pollinators. Conversely, evidence for pollination by other strategies included observations of feeding behaviour and the involvement of females as pollinators.

MATERIALS AND METHODS

Study species

A total of 18 species of Pterostylis were investigated, including representatives from both subgenera and nine of 10 sections (Fig. 2, Supplementary Table S1 in the Supporting Information). Seven species were included from subgenus Pterostylis, comprising two species from section Pterostylis (P. acuminata R.Br. and P. cucullata R.Br.), one species from section Parviflorae (P. furva (D.L.Jones) D.L.Jones), and four species from section Foliosae (P. ampliata (D.L.Jones) D.L.Jones, P. crispula (D.L.Jones & C.J.French) D.L.Jones & C.J.French, P. ectypha (D.L.Jones & C.J.French) D.L.Jones & C.J.French, and P. laxa Blackmore). Eleven species were included from subgenus Oligochaetochilus, comprising three species from section Oligochaetochilus (P. cheraphila D.L.Jones & M.A.Clem., P. pusilla R.S.Rogers, and P. valida (Nicholls) D.L.Jones), three species from section Squamatae (P. jonesii G.N.Backh., P. longifolia R.Br. and P. umbrina (D.L.Jones) G.N.Backh.), two species from section Catochilus (P. barbata Lindl. and P. extensa (D.L.Jones) D.L.Jones), and one species each from sections Pharochilum (P. daintreeana F.Muell. ex Benth.), Stamnorchis (P. recurva Benth.), and Urochilus (P. sargentii C.R.P.Andrews).

Study system

Pterostylis (Orchidoideae: Cranichideae) are found throughout southern and eastern Australia and New Zealand, with a few additional species in New Caledonia, Papua New Guinea, Indonesia, and East Timor (Jones and Clements 2002a, Clements et al. 2011, Jones 2021). In Australia alone, there are 254 described species and an estimated 133 undescribed taxa as of 2019 (Backhouse et al. 2019).

The floral morphology varies markedly across Pterostylis (see Figs 1–4), which has led some authors to split the genus into between 11 and 16 genera (Szlachetko 2001, Jones and Clements 2002b, Clements et al. 2011). However, this taxonomy has not been accepted by the Council of Heads of Australasian Herbaria (APNI 2024). Pterostylis is recognized as containing two subgenera, Pterostylis and Oligochaetochilus, which further comprise three and seven sections, respectively (Janes and Duretto 2010, Janes et al. 2010) (Fig. 2). Sexual deception has been confirmed to operate in both subgenera and in the sections Pterostylis, Oligochaetochilus, Urochilus, and Hymenochilus (Phillips et al. 2014b, Reiter et al. 2019b, Hayashi et al. 2022).

To our knowledge there are no documented cases of obligate or fail-safe self-pollination in the genus, with all Pterostylis species predicted to be reliant on insect vectors for pollination, although this remains to be systematically tested. Published hand-pollination and pollinator exclusion experiments have confirmed that P. brumalis, P. oliveri, P. irsoniana, P. alobula, P. patens, P. cycnocephala, and P. orbiculata are self-compatible but reliant on pollinators for natural fruit set (Lehnebach et al. 2005, Phillips et al. 2014b, Bodley et al. 2016, Thalwitzer et al. 2018, Hayashi et al. 2022). Furthermore, none of the 43 Pterostylis species cultivated by the Orchid Conservation Program at the Royal Botanic Gardens Victoria, including study species P. cucullata, P. cheraphila, and P. valida, set seed in the shade house without hand pollination (Reiter 2024 Unpublished). Most species are described as being scentless to humans, with the exception of species in the section Parviflorae (Fig. 2), which are described as smelling like semen or fish (Jones 2021).

Pollinator observations

The pollinator baiting method, whereby picked flowers are experimentally moved through the landscape as ‘baits’ to attract pollinators (Stoutamire 1974, Peakall 1990), was used as the primary tool for the study of pollinator behaviour and pollinator attraction in this study. Pollinator baiting is known to be highly effective as an experimental tool for studying sexually deceived bee, wasp, and fungus gnat orchid pollinators (e.g. Phillips et al. 2014b, Bohman et al. 2016, Reiter et al. 2019b, Peakall et al. 2020). If multiple inflorescences are used, pollinator baiting can also be used to attract and study pollinators in some non-sexually deceptive systems as well (Reiter et al. 2018, Scaccabarozzi et al. 2018).

In brief, one to five picked inflorescences were held upright in a rack, with the base of each inflorescence kept in water. To minimize the impact of picking flowers of the threatened species P. cheraphila, P. cucullata, and P. valida, cultivated plants in pots from the living collection at the Royal Botanic Gardens Victoria were used for these species. Bait flowers were presented at each location for 5–30-min trials (typically 10 min, depending on the orchid species and the abundance and frequency of pollinator responses). On some occasions where there were numerous insect responses, trials were extended up to a maximum of 90 min. For P. furva, which did not appear to sexually attract insects, a combination of shorter (10–15 min) and longer (60–90 min) trials were used. All responses from insects during trials were recorded, including the time to response from the beginning of the trial. These trials were then repeated at different locations throughout the landscape in a variety of different microhabitats (e.g. at the base of trees, in open areas, near shrubs, in areas with deep leaf litter, near creeks or watercourses). The total observation minutes for each species during baiting trials ranged from 270 min to 1950 min (see Supplementary Table S1 in the Supporting Information for details). Inflorescences were kept cool (~5–10°C) when not in use. Representative specimens of bait flowers have been lodged as vouchers at the Australian National Herbarium in Canberra (CNBR).

Following observations of both potential courtship behaviour and feeding behaviour at bait flowers of P. crispula, observations at natural flowers were also made to confirm whether similar behaviour occurs in situ. These observations were conducted between 8:30 a.m. and 5:30 p.m. (3 hours of continuous observation per day while varying the time of day).

Field studies were conducted from May 2017 to June 2020 across southern Australia [Australian Capital Territory (ACT), New South Wales (NSW), Victoria (Vic), and Western Australia (WA)], with a combined total of >288 hours of observation (see Supplementary Table S1 in the Supporting Information). Pollinator baiting was conducted within the natural range of the study species, with the exception of 16 trials with P. furva conducted around Canberra in the ACT, ~50 to 80 km outside the natural range of P. furva.

Confirmation of pollination

Confirmation of pollinator status for a given insect species requires observing pollen uptake, transport of pollen between flowers, and deposition of pollen on the stigma (Adams and Lawson 1993). Insects were therefore classified as pollinators after they were: (i) observed removing pollinia, (ii) observed arriving at flowers carrying pollinia, and (iii) observed subsequently depositing pollen. In some Pterostylis species (e.g. P. crispula and P. furva) the stigma is hidden inside the hooded flower, and it is not possible to observe the stigma or trap mechanism without damaging the flower. In such cases, pollen deposition was deemed to have been possible if insects carrying pollinia entered the galea for an extended period of time (>30 seconds). Following Adams and Lawson (1993), insects that were observed removing pollinia, but were not observed depositing pollen, were classified as ‘likely pollinators’. Pterostylis species where fewer than 10 responses were observed, or where pollinia uptake or deposition were not observed, were considered to have insufficient data to confirm pollinators or likely pollinators.

Insect identification and sex

Fungus gnats (Mycetophilidae and Keroplatidae) were identified to genus level using available keys for Australian fungus gnats (Malloch 1928, Tonnoir 1929) and corroborated using more comprehensive keys to Nearctic fungus gnats (Vockeroth 1981). For the family Sciaridae, Adam Broadley (Science and Surveillance Group, Department of Agriculture, Fisheries and Forestry, Australian Government) created permanent slide mounts and identified fungus gnats by morphological comparison with the Skuse types and other determined specimens (Broadley et al. 2016). Dagger flies (Empididae) were identified to genus level using keys to Nearctic Diptera (Steyskal and Knutson 1981) as comprehensive keys to Australian fauna do not exist. Identification of scuttle flies (Phoridae) is challenging, and these were not identified beyond family level with the exception of individuals implicated in pollination that were tentatively identified to genus level using the key in Disney (1994). Diptera collected during this project are held at the Australian National University pending further investigations into their taxonomy. Vouchers will be lodged at the Australian National Insect Collection at the conclusion of these studies.

Where possible, the sex of insect visitors was recorded in the field. In general, this was possible for larger Mycetophilidae and Keroplatidae fungus gnats with close observation of the size and shape of the abdomen in the field. However, the Sciaridae and Phoridae in this study were too small to sex in the field, and sex determination required detailed examination of the genitalia of captured specimens in the laboratory.

DNA barcoding

Species-level identification of fungus gnats and phorids is challenging. Moreover, most Australian fungus gnat and phorid species are undescribed. DNA barcoding of the mitochondrial COI region is often used to establish species-level distinctions in insects, including recent studies of fungus gnats (Phillips et al. 2014b, Shin et al. 2014, Põldmaa et al. 2015, Hayashi et al. 2021). Therefore, DNA barcoding was used to estimate the number of pollinator species for a given orchid species.

DNA extraction was performed following the salt extraction method detailed in Hayashi et al. (2021), modified from Bruford et al. (1998). The resulting DNA was amplified and sequenced using the primers and methods of Hayashi et al. (2021) using a touchdown protocol modified from Griffiths et al. (2011). The final product was sequenced in both forward and reverse directions using the original PCR primers on an ABI 3130xl Genetic Analyser.

Sequences were edited and aligned using Geneious v.9.1.8 (Biomatters Ltd) and phylogenetic analyses were performed in MEGA X v.10.154 (Kumar et al. 2018) using a maximum likelihood analysis with 1000 bootstrap replicates and a GTR + I + G substitution model. Sequences from each family of insect were analysed separately, as the COI region is not useful for elucidating deeper phylogenetic relationships. Intra- and interspecific genetic distance were calculated for the families Mycetophilidae, Sciaridae, and Keroplatidae using the Kimura-2 Parameter (K2P) genetic distance in MEGA. Genetic distances were not calculated for the Phoridae due to the small number of sequences within each putative species.

Some relevant COI fungus gnat pollinator sequences were previously published in Hayashi et al. (2021) and Hayashi et al. (2022). Representative sequences of each new taxon identified in this study can be found at GenBank (OM202518–OM202545; see Supplementary Figs S1–S4 in the Supporting Information for phylogenetic trees). In addition, the BLAST (National Centre for Biotechnology Information) and BOLD (Barcoding of Life Data Systems) databases were searched for matching sequences of >98% similarity (Supplementary Table S2 in the Supporting Information). A cut-off of 2% (K2P) divergence was used to define species-level differences, in accordance with generally accepted thresholds in Diptera (Barcode of Life Data Systems) (Meier et al. 2008).

Insect behaviour

The behaviour of all insects visiting Pterostylis flowers was recorded and classified into two types: feeding or sexual. Feeding behaviour was defined by the repeated probing of the flower with mouthparts by male or female insects in the absence of any sexual behaviour. Sexual behaviour was defined by evidence of male courtship behaviour and/or attempted copulation with the flower (pseudocopulation). Sexual behaviour was further classified into the following categories of presumed increasing sexual response: (i) landing on the flower only (i.e. no courtship or sexual behaviour), (ii) wing fanning (repeated wing movements), and (iii) pre-copulatory or copulatory behaviour (abdomen curling, exserted genitalia, probing of the orchid with genitalia, pseudocopulation).

Although abdomen curling and probing of the flower with genitalia can be confidently considered sexual behaviours (Phillips et al. 2014b), the lack of published information on fungus gnat mating behaviour (particularly Mycetophilidae and Keroplatidae fungus gnats) poses difficulties for unequivocally defining the behaviours associated with fungus gnat courtship. Wing fanning is a courtship or pre-mating behaviour in many insects (Wicker-Thomas 2007), including fungus gnats (Alberts et al. 1981, Frank and Dettner 2008, Andreadis et al. 2015). However, wing fanning behaviour is typically poorly defined and the behaviour of many insects outside of courtship situations (e.g. feeding, searching for food) is poorly known, making it difficult to unambiguously classify wing movements on flowers as sexual behaviour, even in cases where wing fanning occurs during courtship. Therefore, we do not consider wing fanning of male fungus gnats on orchids as confirmation of sexual behaviour, unless it was followed by abdomen curling or attempted copulation. Nevertheless, wing fanning may serve to distinguish between merely landing on a surface and resting, compared to searching behaviour. In this study, wing fanning was defined as repeated wing movements by fungus gnats, either slow and sustained, rapid and fluttery, or stop–start, performed while stationary or while walking or rapidly running on the orchid flower.

Patterns of sexual behaviour

To test whether there were differences between species in the proportion of responding insects exhibiting sexual behaviour, pairwise Fishers exact tests with Bonferroni’s correction were performed on the total number of responding individuals displaying sexual behaviour (defined as abdomen curling, exserted genitalia, probing of the orchid with genitalia, or pseudocopulation) and non-sexual behaviour for each species, with the null hypothesis that the proportions of sexual behaviour are equal between species. To test whether there were differences in sexual behaviour between the subgenera, a one-way ANOVA was performed on the mean percent of responses for each species displaying sexual behaviour, using data for P. acuminata, P. cheraphila, P. longifolia, P. jonesii (this study), P. cycnocephala (Hayashi et al. 2022), P. basaltica, P. boormanii, P. curta, P. nutans (Reiter et al. 2019b), and P. orbiculata (Phillips et al. 2014b).

All statistical analyses were performed in R version 3.6.2 (Team 2019). The pairwise Fisher’s exact tests were performed using the ‘fisher.multcomp’ function in the R package RVAideMemoire (Hervé 2020) and the one-way ANOVA was performed with the function ‘aov’ in base R.

Natural fruit set

The proportion of flowers setting fruit was measured in a representative subset of nine species spanning both subgenera and seven of 10 sections, across a total of 34 populations. Individual plants within a population were counted during the flowering period by placing tags near each plant, or by running a transect through the population and measuring the distance of each plant along the transect. After flowering had finished, plants were revisited and flowers with a swollen and turgid or dehisced seed pod were counted as having set fruit. Fruit set was calculated as the percentage of all flowers within a population setting fruit, not including flowers that were damaged or grazed. For species with multiple flowers on each inflorescence, the percentage of inflorescences that set at least one fruit was also calculated.

Sugar analysis

Preliminary trace level sugar sampling and analysis was conducted for the two species in which feeding behaviour was observed: the sciarid pollinated P. crispula and phorid pollinated P. furva. In both species, the feeding behaviour was at the entrance to the flower, either at the point where the conjoined lateral sepals (synsepalum) separate into the left and right sepal tips (in P. crispula, Fig. 3F), or at the point where the sepal tips curve upwards and join the galea (in P. furva, Fig. 3C). Our sugar sampling focussed on this feeding point, with ‘control’ samples taken from the top of the galea.

The sampling, derivatization, and gas chromatography with mass spectrometry (GC–MS) analysis procedures followed our previous method (Reiter et al. 2018, Phillips et al. 2023a), with minor modifications as outlined in detail in the Appendix in the Supporting Information. In brief, sugar sampling involved dissolving any surface sugars on relevant flower parts into a drop of an aqueous solution of Ribitol (as an internal sugar standard). Samples were then stored frozen. Subsequently, derivatization via a method adapted from Lisec et al. (2006) was performed before GC–MS analysis and quantitation.

RESULTS

Pollinator baiting and DNA barcoding

A total of 1380 insects (all Diptera) were attracted to the bait flowers of the 18 test species, including confirmed pollinators for six species and likely pollinators for a further three species (Table 2 and Supplementary Table S1 in the Supporting Information). A total of 35 Diptera were observed removing pollinia and 51 Diptera were observed arriving at Pterostylis flowers carrying orchid pollinia (Supplementary Table S1 in the Supporting Information). In all cases, the pollinia were attached to the dorsal side of the insect’s thorax. Feeding behaviour of Diptera was observed in two orchid species in subgenus Pterostylis (P. crispula and P. furva) while sexual behaviour of insects was observed in the remaining seven orchid species with confirmed or likely pollinators (Tables 2, 3). Various other small insects were occasionally observed crawling on to, or landing on, bait flowers but these did not appear to be attracted to flowers and were never observed entering the flower or becoming trapped (see Supplementary Table S1 in the Supporting Information). No other Pterostylis species with similarly sized pollinia were observed flowering at baiting sites during the study period, with the minor exceptions noted in the footnote of Table 2.

A total of 193 insects were captured on Pterostylis flowers and COI DNA sequences were obtained for 168 of these samples (all Diptera) (see Supplementary Figs S1–S4 in the Supporting Information). A further two individuals belonging to Xenoplatyura conformis (Keroplatidae) were included from a previous study (Reiter et al. 2019b), to test whether these represented the same species as found in the present study. The intraspecific maximum K2P distances within putative species were: 1.8 (Mycetophilidae), 0.6 (Sciaridae), and 0.2 (Keroplatidae), while interspecific minimum K2P distances among putative species were: 11.3 (Mycetophilidae), 6.6 (Sciaridae), and 12.8 (Keroplatidae), respectively.

Species with feeding behaviour: P. crispula

Austrosciara spp. (Sciaridae) were attracted to P. crispula flowers and were observed to remove pollinia and deposit pollen (see Table 2, Fig. 3E, F, Supplementary Video S1 in the Supporting Information). A total of 405 responses from Sciaridae were observed landing on P. crispula flowers, including 245 responses during baiting trials and 160 responses during in situ observations. Results from DNA barcoding revealed three putative species regularly visiting P. crispula flowers (see Supplementary Fig. S3 in the Supporting Information for details). Austrosciara sp. B (n = 10) was only captured at flowers sourced from Carrabungup Nature Reserve on the Swan Coastal Plain (baiting conducted at Carrabungup and various sites along the Darling Range), while Austrosciara spp. A and C (n = 6, n = 13, respectively) were only captured at flowers sourced from the Darling Range (baiting conducted at various sites along the Darling Range). Interestingly, Austrosciara sp. A (n = 4) was captured at P. ectypha flowers while baiting at Carrabungup, confirming that Austrosciara sp. A was present at Carrabungup. However, no Austrosciara sp. A were captured at P. crispula (Carrabungup) flowers while baiting at Carrabungup, indicating specific pollinator attraction.

Of the 42 gnats captured in total to P. crispula, 38 were male (90.4%) and four were female. As most visiting gnats appeared to be males, possible females were targeted for capture and thus the percentage of individuals responding to flowers that are female is likely to be lower than presented here. In total, two males were observed arriving at flowers carrying orchid pollinia, and nine were observed removing orchid pollinia. Two females were observed carrying orchid pollinia, one each at Carrabungup Nature Reserve and John Forrest National Park, but none were observed removing pollinia. Female gnat behaviour on the flower included landing and resting (n = 1) or feeding at the flower entrance (n = 2), while one female was recorded wing fanning before feeding at the entrance to the flower and entering.

Of the responding gnats that landed on P. crispula flowers, 90.1% (n = 365) performed wing fanning behaviour but none displayed confirmed sexual behaviour toward the flower. The proportion of responding gnats displaying wing fanning behaviour was significantly higher in baiting trials (94.3%, n = 231) than in situ observations (83.8%, n = 134) (Fishers exact test, P = .001), although in both cases wing fanning was observed in most responses. Feeding behaviour, characterized by the gnat probing the surface of the flower with its mouthparts, was observed in 33.1% of responses (n = 134). Prolonged feeding behaviour (>1 s) was always performed at the entrance to the flower, where a small, dark green appendage protrudes inside the flower (see Fig. 3F). While feeding at the entrance of the flower, gnats often proceeded to enter or fall into the flower, a process required for entrapment and subsequent pollination. Most instances of a gnat entering the flower (81.9%, n = 95), were associated with the gnat first feeding at the entrance. There was no significant difference between baiting trials and in situ observations in the proportion of responses that displayed feeding behaviour (Fishers exact test, P = .23) or entered the flower (Fishers exact test, P = .37).

Species with feeding behaviour: P. furva

No fungus gnats were observed on P. furva flowers; instead, 134 responses from Phoridae (Diptera) were observed. Three phorids were observed removing pollinia, with two of these captured and later designated as Phoridae sp. G based on DNA barcoding data. One individual, later captured and designated as Phoridae sp. C, was observed arriving at P. furva flowers carrying pollinia and subsequently depositing pollen (Fig. 3, Table 2, Supplementary Fig. S4, Supplementary Video S2 in the Supporting Information). Based on their morphology when observed under a dissecting microscope, Phoridae sp. G was tentatively identified as a Megaselia sp., while Phoridae sp. C was tentatively identified as a Woodiphora sp. The two taxa were indistinguishable in size, colour, and behaviour in the field.

A total of 23 phorids (representing seven taxa) were captured at P. furva flowers in their natural range. Given the number of taxa attracted, baiting was also conducted in the Canberra region ~50–80 km outside the natural range of P. furva, to investigate whether there were local differences in the species of insects responding. A further 10 phorids were captured in the Canberra region, representing eight taxa, including six taxa not captured in the natural range of P. furva. The responding Phoridae were predominately female (18 female, one male; sex determined in a subset of 19 individuals).

No wing fanning or sexual behaviour was observed in the insects that responded to P. furva. However, feeding behaviour was observed in 26.5% (n = 30) of responding individuals that landed on flowers. Prolonged feeding behaviour (>1 s) was always performed at the entrance to the flower, particularly at the point where the sepal tips curve upwards and join the galea. In some cases, minute droplets of exudate were observed at this feeding point (see Fig. 3C). While feeding at the entrance of the flower, the phorids often proceeded to enter the galea, a process required for entrapment and subsequent pollination. In 78.1% (n = 25) of responses where a phorid entered the flower, the phorid first displayed feeding behaviour before entering.

Species with feeding behaviour: sugar analysis

The outcomes of the sugar analysis in P. crispula (n = 6) and P. furva (n = 3) are summarized in full in the Supplementary Appendix Table 1 in the Supporting Information. An average of 3.7 ± 1.1 µg (mean ± SE) of sucrose was detected at the appendage in P. crispula, while a comparable level of 2.6 ± 1.9 µg of sucrose was detected at the entrance in P. furva. Although representing trace levels, the amounts of sucrose detected at these points of feeding behaviour were >100 times greater than the negligible levels of sucrose detected in the ‘control’ samples taken. While sucrose dominated, minute levels of glucose and fructose were also detected, with the average percentage of sucrose being 92% in P. crispula and 69% in P. furva.

Sexually deceptive species

Four Pterostylis species spanning both subgenera and three sections were confirmed to be pollinated by sexual deception: P. acuminata (Pterostylis: Pterostylis), P. cheraphila (Oligochaetochilus: Oligochaetochilus), and P. jonesii and P. longifolia (both Oligochaetochilus: Squamatae) (see Table 2, Figs 2, 3, Supplementary Videos S3–S6 in the Supporting Information). Each species of orchid only attracted males of a single species of fungus gnat (Table 2, Supplementary Figs S1, S2 in the Supporting Information), which were observed to attempt copulation with the labellum, remove pollinia, and deposit pollen. Three of the four species were pollinated by Mycomya (Mycetophilidae) fungus gnats: P. acuminata and P. longifolia were both pollinated by Mycomya sp. A while P. jonesii was pollinated by Mycomya sp. E (Supplementary Fig. S1 in the Supporting Information). Pterostylis cheraphila was pollinated by Xenoplatyura conformis (Keroplatidae) (Supplementary Fig. S2 in the Supporting Information).

Additional observations of sexual behaviour

Sexual behaviour from likely pollinators was observed in a further three Pterostylis species spanning both subgenera and three sections: P. cucullata, P. umbrina, and P. sargentii (see Table 2, Fig. 4, Supplementary Videos S7, S8 in the Supporting Information). In all three species, males of a single species of Mycetophilidae were attracted (see Supplementary Fig. S1 in the Supporting Information). In each species, male fungus gnats were observed removing pollinia and arriving at bait flowers carrying pollinia but were not confirmed to deposit pollen. For example, despite observing 368 fungus gnat responses to P. cucullata, including 157 that landed on bait flowers, only two gnats were observed removing pollinia and none were observed depositing pollen. Six gnats were observed visiting P. cucullata flowers carrying pollinia, although none of these entered the galea. However, the rate of entrance of gnats to the galea and sexual behaviour was low: 26.1% (n = 41) of responses that landed on flowers entered the galea, and only 7.6% (n = 12) displayed sexual behaviour. In addition, the labellum of bait flowers of P. cucullata became rapidly desensitized during transport between bait sites, and only 7.3% (n = 3) of gnats entering the galea triggered the labellum and became trapped.

Species with insufficient data

There were insufficient data in nine species to confirm pollinators or likely pollinators (Supplementary Table S1 in the Supporting Information). However, although fewer than 10 responses were observed for three species (P. ectypha, P. daintreeana, P. barbata), a Dipteran species was observed removing pollinia in each case. Four species (P. ampliata, P. laxa, P. pusilla, P. valida) attracted >10 responses from Diptera but no insects were entrapped nor observed removing or depositing pollinia, although insects were observed arriving at bait flowers carrying orchid pollinia in P. ampliata, P. pusilla, and P. valida (Supplementary Table S1 in the Supporting Information). In most cases these responses involved weak approaches without landing on the flower. However, in P. ampliata, sexual behaviour with the labellum was observed in 36 of 91 responses from male Tetragoneura sp. A (Mycetophilidae), although these gnats were too small to trigger the labellum and therefore could not act as pollinators. In all cases, responding insects were fungus gnats (Diptera; Sciaroidea) except for P. extensa and P. barbata (both in section Catochilus). In P. extensa, three male Empis sp. (Diptera; Empididae) were observed arriving at P. extensa flowers carrying dead flies (Muscidae). One male Empis sp. was found alive inside a P. barbata flower and was observed depositing pollen from another flower on the stigmatic surface and removing pollinia on its escape from the flower.

Rates of sexual behaviour

Despite belonging to different subgenera, P. acuminata and P. longifolia were both pollinated by Mycomya sp. A (Mycetophilidae). However, male gnats displayed significantly lower proportions of sexual behaviour (i.e. abdomen curling and attempted copulation) on P. longifolia flowers than P. acuminata flowers (Fishers exact test, P < .001; Supplementary Table S3 in the Supporting Information). Pterostylis cheraphila, P. basaltica, and P. boormanii (section Oligochaetochilus) were all pollinated by the keroplatid gnat Xenoplatyura conformis (this study, Reiter et al. 2019b), although the orchid species do not overlap in geographic range. Male gnats displayed similar proportions of sexual behaviour on P. cheraphila and P. basaltica flowers (P = 1; Table 3, Supplementary Table S3 in the Supporting Information) but displayed significantly lower proportions of sexual behaviour on P. boormanii flowers (P = .002; Table 3, Supplementary Table S3 in the Supporting Information).

Across all confirmed sexually deceptive species (see Table 3), mean rates of sexual behaviour were not significantly different between subgenus Pterostylis (labellum enclosed within galea) and subgenus Oligochaetochilus (labellum outside of galea) (one-way ANOVA, F(1,8) = 0.99, P = .35).

Fruit set

Across the nine species where fruit set was quantified (Table 4), the average fruit set within a species varied from 0.8% in P. umbrina to 57.5% in P. crispula, with an average across species of 23.7% ± 6.7% (mean ± SE). In some species, fruit set was highly variable from site to site (e.g. 13%–77% in P. longifolia, 6%–58% in P. cucullata, 10%–95% in P. crispula). For species with multiple flowers per inflorescence, the average percent of inflorescences that set at least one fruit was also variable, ranging from 1.5% in P. umbrina to 84.9% in P. furva, with an average across species of 35.5 ± 12.0%.

Average fruit set within the subgenus Pterostylis was 33.0% ± 11.2% (n = 4 species) compared with 16.2% ± 7.4% (n = 5 species) in subgenus Oligochaetochilus. Across the confirmed and likely sexually deceptive species (P. cheraphila, P. cucullata, P. longifolia, P. nutans, P. sargentii, and P. umbrina), fruit set averaged 17.8% ± 6.2%, while across the species in which feeding behaviour was observed (P. crispula and P. furva) fruit set averaged 50.5%.

DISCUSSION

Here, we confirm pollinators for six species of Pterostylis, including four sexually deceptive species (P. acuminata, P. cheraphila, P. jonesii, and P. longifolia) and two putative food-rewarding species (P. crispula and P. furva). Two of the confirmed sexually deceptive species (P. jonesii and P. longifolia) are in section Squamatae, where sexual deception has not been previously confirmed. Sexual behaviour from likely pollinators was observed in a further three species. Fungus gnats (Diptera: Sciaroidea) were the only insects regularly observed visiting Pterostylis flowers, except for P. furva, which instead attracted phorid flies.

Evidence for feeding behaviour

In our survey, two species of Pterostylis were pollinated by Diptera that displayed feeding behaviour at the entrance to flowers before entering the galea and becoming trapped. In P. crispula, three species of Austrosciara sp. fungus gnats (Sciaridae) acted as pollinators. In P. furva, seven species of Phoridae were attracted to flowers but only two were observed with orchid pollinia. In both orchid species, feeding behaviour at the entrance to the flowers appears to be associated with floral visitors subsequently entering the flower. Of the visitors that entered P. crispula and P. furva flowers, a process required for pollination, >81% and 78%, respectively, first displayed feeding behaviour before entering.

The discovery of food foraging behaviour in P. crispula and P. furva flowers was unexpected, as Pterostylis flowers have long been considered to be rewardless (Adams and Lawson 1993, Bernhardt 1995, Jones and Clements 2002a, Jones 2021). To the best of our knowledge, this is the first confirmation of feeding behaviour as a component of the pollination strategy of any Pterostylis. Trace level analysis detected low levels of sucrose and traces of glucose and fructose from samples taken at the flower entrance. These samples were from the sites of release of the minute droplets of exudate observed in P. furva (Fig. 3C) and the shiny green appendage of P. crispula (Fig. 3F).

The finding that sucrose was the dominant sugar in both P. crispula and P. furva matches the general pattern reported for the Orchidaceae, where some 80% of the investigated orchid species produce sucrose rich nectar. On the other hand, Dipteran pollinators have long been reported to prefer hexose rich nectar, although cases of pollination by Diptera involving sucrose rich nectar are also known (see mini-review in Brzosko and Mirski 2021). In Tolmiea menziesii (Saxifragaceae), where the large fungus gnat, Gnoriste megarrhina (Mycetophilidae) is the main but not exclusive pollinator, the 0.1–0.2 µL volumes of nectar per flower were reported to be unusually dilute with a maximum of 12% sucrose equivalents (Goldblatt et al. 2004).

To the best of our knowledge, trace level sugar analysis by LC–MS or via derivatization and GC–MS, as used in this study, has not been reported for other Dipteran pollinated species. However, a consistent finding across nectar-rewarding fungus gnat pollinated species are minute volumes of nectar, spanning in volume from insufficient to be sampled to a maximum of just 2 µL (Goldblatt et al. 2004, Okuyama et al. 2004, Mochizuki and Kawakita 2018). Given such tiny nectar volumes, the application of trace level sugar analysis in future studies of plants pollinated by small Diptera is certain to prove informative.

Mechanism of pollinator attraction in P. crispula

Despite exhibiting feeding behaviour on the flower, the mechanism for long-distance pollinator attraction in P. crispula remains unclear. The frequent wing fanning and strong bias towards males suggests that long-distance attraction is not necessarily in response to cues associated with food. Specialized pollination systems with an unexplained male bias have been observed in orchids and other plants (Johnson and Steiner 1994, Suetsugu and Sueyoshi 2018, Reiter et al. 2019a, Sugiura and Maehara 2019, Rupp et al. 2021). For example, the Australian orchid Caladenia drummondii attracts males of a single species of Calopompilus (Hymenoptera; Pompilidae) that do not display sexual behaviour, although the mechanism of pollinator attraction remains unresolved (Phillips et al. 2021).

There are several possible explanations for the male bias observed in the fungus gnat pollinators of P. crispula: (i) males emerge before females; (ii) male-biased sex ratio in gnat populations; (iii) different dietary preferences between males and females; and (iv) sexual attraction of males before switching to food foraging behaviour on the flower (e.g. Phillips et al. 2023). The strong male bias was observed across a 3-week period in multiple years at multiple sites, suggesting male emergence prior to females is unlikely to fully account for the observed male bias. Both male- and female-biased sex ratios have been observed in other Sciaridae (Eberhard and Flores 2002, Nielsen and Nielsen 2004, Cheng et al. 2017), although this does not explain the wing fanning behaviour and highly specific attraction. Both sexes of Sciaridae have been observed seeking nectar on flowers (Ackerman and Mesler 1979, Barbosa et al. 2009, Heiduk et al. 2021), but wing fanning behaviour and male-biased sex ratios have not been reported in these systems.

The strong male bias, high specificity, rapid attraction to bait flowers and possible courtship behaviour (wing fanning) suggests that attraction of males by sexual cues is a possibility. Rapid wing fanning and frantic searching behaviour is characteristic of male Sciaridae courtship (Alberts et al. 1981, Andreadis et al. 2015, Uddin et al. 2016). However, the lack of attempted copulation on flowers, the observation of two pollen-carrying females individuals, and evidence of wing fanning on the flower by one female individual, challenges the notion of a pollination system based on sexual behaviour alone. Potential sexual cues from P. crispula flower may be too weak to elicit copulatory behaviour from male gnats, while putative food rewards could also attract low numbers of female gnats irrespective of sexual cues. For example, in some populations of the food-rewarding, sexually deceptive daisy Gorteria diffusa (Asteraceae), both male and female bee-flies are attracted to flowers and feed on nectar and pollen, but only males attempt to copulate with black spots on ray florets (Ellis and Johnson 2010). Meanwhile, in some sexually deceptive Caladenia, male thynnine wasps are attracted to flowers by sexual chemical cues and they regularly attempt copulation with the distal sepal tips of the flower. However, an infrequent switch to feeding behaviour on the labellum is required to bring them into contact with the column where pollination occurs (Phillips et al. 2023). Further research is needed to tease apart the method of pollinator attraction in P. crispula. Convincing evidence for sexual attraction will require the chemical characterization of floral odour in P. crispula and the female sex pheromones in Austrosciara spp. gnats. Although obtaining such chemical knowledge has been recently achieved for P. orbiculata and its Mycomya fungus gnat pollinator species (Hayashi et al. 2021), this is a challenging task.

Mechanism of pollinator attraction in P. furva

Phoridae were the only insects regularly observed visiting P. furva flowers in our study, providing evidence for the first documented pollination system in Australia involving specialization on phorids. The family Phoridae is cosmopolitan and highly diverse, with >4000 described species (Pape et al. 2011). However, the Australian phorid fauna remains poorly documented, with only 91 species described, compared to 340 species in the British Isles alone (Disney 2011). Unlike other Pterostylis, which tend to be pollinated by one or a few insect species, seven phorid taxa were attracted to P. furva flowers as well as an additional six taxa outside the natural range of P. furva, suggesting pollinator attraction is through cues that are attractive to a range of species.

Pterostylis furva flowers along with other members of the Parviflorae, are unusual among Pterostylis species in having floral odours that are detectable to the human nose. The strong ‘semen-like’ smell of P. furva (Egan et al. 2020, Jones 2021), although uncommon, has been reported in other plants. In Stemona japonica (Stemonaceae), the rewardless, semen-smelling flowers are pollinated almost entirely by Antherigona flies (Diptera; Muscidae) (Chen et al. 2015), while various Diptera were observed visiting Photinia serrulata (Rosaceae) and Castanopsis sclerophylla (Fagaceae) (Zhang et al. 2018). Chen et al. (2015) hypothesized that S. japonica may be a carrion mimic, as the semen odour is from 1-pyrroline, a well-known component of decaying vertebrate carcasses which is produced via the oxidation of putrescine (Durlu-Özkaya et al. 2001, Paczkowski and Schütz 2011). Other examples of semen-smelling flowers are found in the families Apocynaceae (Shuttleworth 2016) and Berberidaceae (Kaiser 2006).

Carrion mimicry has been documented in a wide range of plant families (Jürgens and Shuttleworth 2016), including the Orchidaceae (Humeau et al. 2011, van der Niet et al. 2011). Carrion mimicking flowers are often dark red, brown, or purple, sometimes with hairs or appendages, and have a strong odour often dominated by oligosulfides, nitrogenous compounds, aliphatic acids, and p-cresol (Jürgens et al. 2013, Jürgens and Shuttleworth 2016). Given the semen-like odour from P. furva flowers, attraction of Phoridae, and dark green-reddish flowers, it is possible that this species is also a carrion mimic. Bioassays comparing insect visitors to P. furva flowers and potential food sources of phorids (e.g. decaying vertebrate carcasses) (Disney 1994, van der Niet et al. 2011), and investigations into the floral odour emitted by P. furva flowers, may shed light on the method of pollinator attraction.

Evidence for sexual deception

We have confirmed four new cases of pollination by sexual deception of male fungus gnats in Pterostylis (P. acuminata, P. cheraphila, P. jonesii, and P. longifolia). Male fungus gnats were regularly observed attempting to copulate with the labellum of all four species and were observed removing and depositing pollen. Our observations for Pterostylis jonesii and P. longifolia are the first confirmation of pollination by sexual deception in Pterostylis section Squamatae. Pterostylis acuminata (section Pterostylis) and P. cheraphila (section Oligochaetochilus) are in sections that have previously been confirmed to contain sexually deceptive species (Reiter et al. 2019b). Our results bring the total number of confirmed sexually deceptive Pterostylis species to 13 (Table 3) (Phillips et al. 2014b, Reiter et al. 2019b, Hayashi et al. 2022). Collectively, these cases span both subgenera of Pterostylis and five of nine sections.

In addition to the four confirmed cases of pollination of Pterostylis by sexual deception of gnats, we also found evidence of sexual behaviour by fungus gnats on the flowers of other Pterostylis species. Our observations of male only attraction, sexual behaviour, and removal of orchid pollinia suggest that P. cucullata, P. umbrina, and P. sargentii are probably pollinated by sexual deception of male fungus gnats. However, we did not observe pollen deposition in these three species, and additional observations are required to confirm the pollinators for these species.

How widespread is sexual deception likely to be in Pterostylis?

On the basis of the results presented here, and previous studies (Phillips et al. 2014b, Reiter et al. 2019b, Hayashi et al. 2022), we predict that sexual deception is likely to be the main pollination strategy in sections Pterostylis (56 described species), Squamatae (30 species), Hymenochilus (16 species), Oligochaetochilus (63 species), and Urochilus (eight species). Floral morphology is conserved within each section, suggesting that species within each section are likely to use the same pollination strategy and family of pollinator. If our prediction holds true, most of these 173 species of Pterostylis may be sexually deceptive.

It is also likely that sexual deception is present in section Catochilus (20 species), as male Empis dagger flies (Diptera; Empididae) have been observed carrying orchid pollinia and presenting dead flies to P. plumosa flowers (Kuiter 2017). The presentation of dead prey by male Empis flies can be considered evidence of courtship behaviour, as the presentation of nuptial gifts to females is crucial in stimulating copulation in Empis (Steyskal and Knutson 1981). Our own observations are consistent with Empis attraction in P. barbata and P. extensa, although further research is needed to confirm the pollinator species involved and pollination mechanism.

In section Foliosae (72 species), evidence for sexual deception is equivocal. It is possible that some species may be sexually deceptive, based on observations of sexual behaviour from a non-pollinating gnat in P. ampliata (Supplementary Table S1 in the Supporting Information) and observations of sexual behaviour from male fungus gnats on P. grandiflora (Kuiter 2020). Meanwhile, it is unclear what method of attraction is involved in P. crispula, although related species (colloquially known as ‘snail orchids’) are also likely to be pollinated by Sciaridae. For example, male Sciaridae have been observed visiting flowers and removing pollinia of the snail orchid P. nana in Victoria (Kuiter 2020), and four male Austrosciara sp. A were observed visiting P. ectypha, including one that removed pollinia.

In contrast to most Pterostylis species, we predict that the 22 species in the section Parviflorae (including P. furva) are likely to use a pollination strategy other than sexual deception. We have no data on insect behaviour for the monotypic sections Stamnorchis and Pharochilum.

Fruit set in Pterostylis

The average fruit set (percentage of all flowers in a population that set fruit) across the nine Pterostylis species in this study was 23.7% ± 6.7% (mean ± SE), which is lower than the global average for temperate orchids (34.6% ± 2.3%) but higher than the global average for fly-pollinated orchids (11.9% ± 2.9%) (Tremblay et al. 2005). In our study, for species that were either confirmed or likely sexually deceptive, fruit set averaged 17.8% ± 6.2%. With the inclusion of fruit set data from the sexually deceptive P. orbiculata (Phillips et al. 2014b). and P. cycnocephala (Hayashi et al. 2022), fruit set averaged 21.3% ± 5.1%. This is similar to the global average of ~20% fruit set reported for sexually deceptive orchids (Gaskett 2011). Across P. crispula and P. furva, which have detectable floral sugar, the higher average of ~50% matches the general pattern of higher rates of fruit set in rewarding orchids, when compared with deceptive orchids (Tremblay et al. 2005).

In some species in our study (e.g. P. longifolia, P. crispula, P. cucullata) fruit set was highly variable between populations. Fruit set in two New Zealand species has been shown to vary greatly between populations (4.4%–47%, P. irsoniana) and years (0%–80%, P. oliveri) (Thalwitzer et al. 2018). Fungus gnats can be locally abundant in moist environments (Colless and McAlpine 1970, Jakovlev 2012, Matile 2016), which in turn may lead to high fruit set for orchids growing nearby. Nonetheless, a patchy distribution of fungus gnats in the landscape may lead to high variability in fruit set, particularly as many Pterostylis species are likely to rely on one or few species of pollinator. Pronounced variability in fruit set between populations has also been observed in some other genera of sexually deceptive orchids (e.g. Phillips et al. 2014a). There is also evidence for temporal variation within sites. For example, fruit set in the thynnine wasp-pollinated Caladenia tentaculata varied from 12% to 82% across years (Peakall and Beattie 1996) and in C. xanthochila varied from 0% to 17.2% across 16 years (Reiter et al. 2023).

Implications for threatened Pterostylis species

In Australia, Pterostylis ranks as the third-most threatened orchid genus in terms of number of species listed under federal threatened species legislation, behind Caladenia and Prasophyllum (Wraith and Pickering 2019). At present, of the 254 described Pterostylis species (Backhouse et al. 2019), 28 species (9.5%) are listed as nationally threatened under the Australian Environment Protection and Biodiversity Conservation Act 1999 (EPBC Act), including our study species P. cheraphila, P. cucullata, and P. valida. It is of interest that we found evidence for sexual deception in P. cheraphila and P. cucullata (to be confirmed in the latter). On the basis of these new findings and previous studies (Phillips et al. 2014b, Reiter et al. 2019b, Hayashi et al. 2022), we predict that 21 (75%) out of the 28 nationally threatened Pterostylis are likely to be sexually deceptive and therefore reliant on just one or few species of fungus gnat pollinator.

Our results show that four Dipteran families (Mycetophilidae, Keroplatidae, Sciaridae, and Phoridae), are key pollinators of Pterostylis. However, in Australia these are ‘orphan’ Dipteran families, where even basic taxonomic and ecological research is almost entirely lacking (Austin et al. 2004). For example, <30% of the mycetophilids lodged at the Australian National Insect Collection were estimated to belong to described species (Colless 1970). Furthermore, since that observation of 50 years ago, there has been no further taxonomic research on the group. Indeed, only four species of Mycomya are described in Australia (Skuse 1888, Tonnoir 1929), yet six taxa in this genus were collected at Pterostylis flowers as part of this study alone. Resolving the taxonomic diversity of fungus gnats will provide a critical starting point to understanding their distribution and ecological requirements.

Although there has been very little research in Australia, fungus gnats are often associated with moist environments and areas of abundant leaf litter and organic matter (Colless and McAlpine 1970, Jakovlev 2012, Matile 2016). In the context of a drying and warming climate across much of southern Australia (BOM 2024), it is clear that further research is urgently needed to understand their response to drought, increased temperature and bushfires, and the implications this may have for the pollination and therefore future survival of Pterostylis species.

We found evidence for pollinator sharing between P. cheraphila (this study), P. basaltica and P. boormanii (Reiter et al. 2019b), which are all pollinated by Xenoplatyura conformis. In addition, we observed responses of X. conformis to flowers of the threatened species P. valida, although we did not observe any pollen removal or deposition, leaving it unclear whether X. conformis is a pollinator of this orchid. Given its status as a pollinator (or potential pollinator) of several threatened Pterostylis species, it would be desirable to better understand the ecology and distribution of X. conformis, although collections span from western Victoria (Reiter et al. 2019b) (this study) through NSW (Skuse 1888) to northern Queensland (Malloch 1928), suggesting it is a widespread eastern Australian species. As such, it may well be involved in the pollination of additional Pterostylis species. The potential for hybridization due to pollinator sharing should be carefully considered before reintroductions or assisted migrations are used in these Pterostylis species and potentially other species in section Oligochaetochilus.

CONCLUSIONS

Sexual deception has now been confirmed in 13 Pterostylis species representing a wide range of floral morphologies, indicating this pollination strategy is likely to be the dominant, albeit not the only, mode of pollination within the genus. This knowledge has important implications for conservation management planning for threatened Pterostylis species and raises the urgent need for more research to help us better understand fungus gnat diversity and ecology in Australia. As the only plant genus involving sexual deception of three families of Diptera, Pterostylis offers a unique avenue to explore the evolution of pollination by sexual deception in orchids, where until now our evolutionary understanding has been largely based on studies of Hymenopteran cases.

Acknowledgements

We thank Alice McCarthy, Ron Fauntleroy, Allan Aitken, Ross and Carolyn Jones, and Matthew Mullaney for providing fieldwork assistance; Adam Broadley and Wen Lee for assisting with the identification of Sciaridae and Phoridae, respectively; Yasushi Hayashi for translating fungus gnat papers from Japanese; James Perkins for assistance with the sugars analysis and quantitation. Finally, thanks to all those that provided valuable information on field sites. Photographs and video by T.H.

Conflicts of interest

The authors declare no conflicts of interest.

Declaration of funding

This work was supported by funds to T.H. from the Australian Orchid Foundation, Holsworth Wildlife Research Endowment, Ecological Society of Australia, Australian Government Research Training Program, and the Research School of Biology at The Australian National University.

Data availability

The new DNA sequence data underlying this article are available in the GenBank Nucleotide Database at https://www.ncbi.nlm.nih.gov/genbank/ (OM202518-545). Other data underlying this article will be shared on request to the corresponding author.

References