Erschienen in: Biological Journal of the Linnean Society ; 119 (2016), 2. - S. 283-298 https://dx.doi.org/10.1111/bij.12841

Behaviour of sexually deceived ichneumonid wasps and its implications for pollination in ()

ALYSSA M. WEINSTEIN1,2,3*, BELINDA J. DAVIS2,3, MYLES H. M. MENZ3,4, KINGSLEY W. DIXON2,3,5 and RYAN D. PHILLIPS1,2,3

1Evolution, Ecology and Genetics, Research School of Biology, The Australian National University, Canberra, ACT, 0200, Australia 2Kings Park and Botanic Garden, The Botanic Gardens and Parks Authority, Fraser Avenue, West Perth, WA, 6005, Australia 3School of Biology, The University of , Crawley, WA, 6009, Australia 4Institute of Ecology and Evolution, University of Bern, Baltzerstrasse 6, 3012, Bern, Switzerland 5Department of Agriculture and Environment, Curtin University, Bentley, WA, 6102, Australia

Pollination via sexual deception is hypothesised to be associated with more frequent outcrossing and greater pollen dispersal distances than strategies involving food-foraging behaviour. In this study, we investigated the behaviour and movement distances of Lissopimpla excelsa (: ), and their implications for the pollination of the sexually deceptive Cryptostylis ovata (Orchidaceae). Pollinator observations revealed that while L. excelsa will alight on multiple flowers within a single visit to a patch of orchids, the frequency of attempted copulation decreases with successive visits, suggesting that pollinator learning may inhibit within- patch pollen transfer. Mark-recapture demonstrated that 25% of wasps revisited inflorescences within a day and 50% revisited within a week. Despite the apparent site fidelity of some individuals, L. excelsa often move over large distances (maximum = 625 m), and are capable of dispersing pollen between patches. To resolve the consequences of pollination by sexual deception of ichneumonids, we compared our results with those from studies of other sexually deceptive systems. While pollination rates were comparable with other sexually deceptive orchids, L. excelsa showed high rates of column contact and moved over large distances relative to other sexually deceived pollinators. Among sexually deceptive orchids in general, the frequency of column contact was not correlated either with the frequency of pseudocopulation or with pollination rate. These results suggest that the consequences of pollination by sexual deception may vary extensively between plant taxa due to variation in floral traits, and behavioural differences between pollinator groups. © 2016 The Linnean Society of London, Biological Journal of the Linnean Society, 2016, 119, 283–298.

KEYWORDS: Cryptostylis ovata – Lissopimpla excelsa – mark-recapture – pollen dispersal – pollina- tion – pollinator learning – sexual deception.

INTRODUCTION orchids, which effect pollination through floral mimi- cry of female (Coleman, 1927b; Kullen- Deceived pollinators typically avoid repeat visits to berg, 1961; Stoutamire, 1974; Schiestl et al.,1999; rewardless , potentially leading to higher Gaskett & Herberstein, 2010), outcrossing and pollen rates of outcrossing and greater offspring fitness dispersal may be greater because the insects are (Dressler, 1981; Peakall & Beattie, 1996; Johnson searching for mates rather than moving between & Nilsson, 1999; Jersakov a, Johnson & Kindlmann, nearby plants while foraging (Peakall & Beattie, 2006; Ellis & Johnson, 2010). In sexually deceptive 1996; Peakall & Schiestl, 2004). These predictions have received some empirical support, with high *Corresponding author. E-mail: [email protected] outcrossing rates reported for the thynnine-pollinated

283

Konstanzer Online-Publikations-System (KOPS) URL: http://nbn-resolving.de/urn:nbn:de:bsz:352-2-1mamw2p6kh81d9 284 sexually deceptive orchids Caladenia tentaculata attraction to Cryptostylis flowers (Coleman, 1930a; (Peakall & Beattie, 1996), Chiloglottis valida, and Schiestl, Peakall & Mant, 2004), recent evidence sug- Chiloglottis aff. jeansii (Whitehead, Linde & Peakall, gests that colour mimicry but not shape mimicry 2015). In contrast, in the ant-pollinated Leporella may also be important for pollinator sexual attrac- fimbriata, outcrossing is low due to both extensive tion or detection of the flower upon approach (Gas- pollen transfer within large clonal patches and kett & Herberstein, 2010; Gaskett, 2012). occasional revisitation to flowers (Peakall & James, At present, the courtship and mate searching 1989). behaviour of L. excelsa and its consequences for the In sexually deceptive orchids the primary trait pollination of Cryptostylis remains poorly known. In involved in pollinator attraction is chemical mimicry the majority of studied ichneumonid species, male of the sex pheromones of female insects (Kullen- courtship behaviour involves antennal vibration and berg, 1961; Schiestl et al., 1999, 2003; Bohman wing fanning displays followed by anntenation with et al., 2012, 2014). Due to the high specificity of females, with subsequent copulation if the male is mating signals, sexually deceptive systems accepted (Fisher, 1959; Juillet, 1959; Kugler & Woll- typically involve specialised plant–pollinator rela- berg, 1967; Spradbery, 1969; Cole, 1970; Morey, tionships, in which each orchid species has one or 1971; Vinson, 1972; Slobodchikoff, 1973; Dowell & few pollinator species (Paulus & Gack, 1990; Phil- Horn, 1975; Barrows, 1976; Gordh & Hendrickson, lips et al., 2009, 2014a; Peakall et al., 2010; Gas- 1976; Crankshaw & Matthews, 1981; Rotheray, kett, 2011). The reproductive success and offspring 1981; van Veen, 1982; Dyer & Landis, 1997). If fitness of a sexually deceptive orchid will therefore rejected by the female, males often continue to exhi- be strongly dependent upon its efficacy at eliciting bit courtship behaviour (Spradbery, 1969; Gordh & pollinator sexual attraction, and upon the mate Hendrickson, 1976; Crankshaw & Matthews, 1981; search behaviour and home range of its specific pol- Rotheray, 1981; van Veen, 1982). There is strong linator species. Surprisingly, some elements of the experimental evidence that male ichneumonids are behaviour of sexually deceived pollinators remain attracted via a sex pheromone (Heatwole, Davis & poorly studied, despite their importance for plant Wenner, 1964; Cole, 1970; Vinson, 1972; Rotheray, fitness. For example, detailed studies of the extent 1981; Jewett & Carpenter, 1999). However, while and temporal period of pollinator revisitation are antennal glands have been suggested to play a role mostly lacking (though see Whitehead & Peakall, in courtship via transfer of a contact pheromone (Bin 2013). Further, most detailed studies of the beha- et al., 1999; Kolarov & Gurb€ uz,€ 2007; Klopfstein, viour of sexually deceived pollinators have focused Quicke & Kropf, 2010; Steiner et al., 2010), the sex on thynnine wasps (e.g. Peakall (1990), Peakall & pheromone has yet to be resolved for any ichneu- Beattie (1996), Alcock (2000), Menz et al. (2013), monid. There have been few studies investigating Whitehead & Peakall (2013), De Jager & Peakall ichneumonid mate searching behaviour, with move- (2015) but see Peakall (1989), Peakall & Schiestl ment distances reported for just five species (Juillet, (2004), De Jager & Ellis (2014)). 1959; Heatwole & Davis, 1965), ranging from a maxi- A diversity of families of solitary wasps, bees and mum of 37 m for Eulimneria rufifemur (Juillet, flies is known to be involved in pollination by sexual 1959), to a maximum of 107 m for Ephialtes ruficollis deception (Paulus & Gack, 1990; Ellis & Johnson, (Juillet, 1959). 2010; Gaskett, 2011; Phillips et al., 2014b). The con- The present study aimed to investigate the beha- trasting life histories of these groups may have dif- viour and spatial movement patterns of Lissopim- ferent consequences for pollen movement and pla excelsa, and their implications for the reproductive success. In an unusual case of pollina- pollination of Cryptostylis ovata R.Br. The beha- tor sharing, the ichneumonid wasp Lissopimpla viour of L. excelsa was investigated at multiple excelsa Costa pollinates all five Australian species of spatial scales relevant to pollen movement: on a C. Cryptostylis, and is the only member of the Ichneu- ovata inflorescence, within C. ovata patches, and monidae known to visit sexually deceptive flowers between C. ovata patches. A combination of obser- (Coleman, 1927a, b, 1929, 1930a, b, 1938; Nicholls, vations of artificially presented C. ovata patches, 1938). The Cryptostylis–L. excelsa system is further and mark-recapture techniques were implemented. unusual in that it is one of only two documented Further, we aimed to investigate the consequences cases in which the sexual attraction of deceived polli- of ichneumonid pollination by comparing pollinator nators to flowers can lead to pollinator ejaculation movement distances, pollinator sexual attraction to (Coleman, 1931; Erickson, 1951; Blanco & Barboza, a flower, and pollination rate between the Cryp- 2005; Gaskett, Winnick & Herberstein, 2008). While tostylis system and other sexually deceptive sys- it is likely that floral odour plays a role in pollinator tems in the literature. 285

METHODS ichneumonid courtship behaviours of wing and anten- STUDY SYSTEM nal movement have not been observed in L. excelsa individuals on flowers. In some species of Cryptostylis, The genus Cryptostylis is comprised of 25 species, including C. ovata, male L. excelsa have been found through south-east Asia, Australia, and New observed to ejaculate onto the flowers (Coleman, 1928; Guinea, and extending to New Zealand (Pridgeon, Erickson, 1951; Gaskett et al., 2008). Artificial self- Cribb & Chase, 2001). Five species are present in pollination of C. ovata usually results in seed set (A. Australia, with Cryptostylis ovata R.Br. occurring Weinstein & R. Phillips, unpubl. data). exclusively in southwestern Australia (Brown et al., 2008), and C. erecta, C. hunteriana, C. leptochila, and C. subulata occurring in eastern Australia, sometimes FLORAL PRESENTATIONS in sympatry (Gaskett, 2012). Cryptostylis ovata flow- Lissopimpla excelsa males were attracted using the ers between November and March, and relies on baiting method developed by Stoutamire (1983) and reserves of moisture in its tubers during this period Peakall (1990), in which the experimental presenta- (Pate & Dixon, 1982). This long flowering period is tion of picked flowers at a new position within the due to the sequential opening of flowers along the landscape leads to the rapid attraction of their polli- raceme (Pridgeon et al., 2001). Racemes typically nators. Similar to the behaviour observed for thyn- have 10–15 flowers (maximum observed = 24), one to nine species (Peakall, 1990; Peakall & Beattie, 1996; three of which are usually open at any given time. Alcock, 2000; Peakall et al., 2010; Whitehead & Pea- Plants typically form clonal patches (Dodd et al., kall, 2013), the greatest number of responses of L. 1984) that in some cases can have over 100 leaves excelsa to presented flowers occurs within the first and support more than 10 racemes, although they few minutes of the presentation (Tomlinson & Phil- also occur as isolated individuals (single leaf). lips, 2012). Flowers used for baiting were collected Cryptostylis ovata is pollinated solely by the ichneu- 0 from populations in Capel (33°35 29.69″S, monid wasp Lissopimpla excelsa Costa (Coleman, 0 0 115°32 31.77″E) and Margaret River (33°58 2.21″S, 1929, 1930b). Visiting wasps land on the resupinate 0 115°0 58.37″E), southwest Western Australia. All labellum and reverse into the flower until their abdo- experiments were undertaken in a site approxi- men comes into contact with the column (Fig. 1). This mately 400 9 700 m within Kings Park and Botanic reversing behaviour is essential for pollination – 0 0 Garden (31°57 48.92″S, 115°50 18.21″E), an urban removal of pollinia is effected as the dorsal side of the bushland remnant in Perth, Western Australia. Lis- apex of the abdomen contacts the viscidium (Van der sopimpla excelsa occurs in abundance at this site, Pijl & Dodson, 1966). Once positioned, the wasp forms while C. ovata is absent. Experiments were con- a falcate curve with its abdomen and begins copula- ducted during December 2011–January 2012, and tory probing (Coleman, 1927b), although this beha- December 2013–February 2014. All experiments viour is not required to effect pollination. The typical were conducted between 06:00 and 12:00 h to coin- cide with the period of maximum wasp activity (Tom- linson & Phillips, 2012).

POLLINATOR BEHAVIOUR: ON AN AND WITHIN A PATCH To quantify pollinator behaviour on an inflorescence, and pollinator movement between inflorescences within a patch, an artificial patch was created using inflorescences with a minimum of two open flowers. Following the method of Peakall & Beattie (1996), an inflorescence was placed on each corner of a 1 9 1m quadrant, and pollinator behaviour within this artifi- cial patch was recorded by two observers. Observa- tions were conducted for 10-min periods, after which the patch was relocated a minimum distance of 20 m to achieve a renewed pollinator response (Bower, Figure 1. A male Lissopimpla excelsa on a flower of 1996; Tomlinson & Phillips, 2012). For each wasp Cryptostylis ovata. The insect has reversed into the that approached flowers in the patch it was recorded flower, bringing pollinia from another flower into contact whenever the wasp: (1) landed on an inflorescence, with the stigma. (2) contacted the column, or (3) attempted copulation 286 with a flower. Due to the large number of responding LONG-DISTANCE POLLINATOR MOVEMENT wasps, it was not possible to examine flowers for To approximate pollinator movement between patches ejaculate as confirmation of attempted copulation, of C. ovata, a larger scale mark-recapture experiment because we did not have sufficient flowers to intro- was implemented. Male L. excelsa wasps were duce a fresh flower after every wasp. Behaviour was attracted by baiting with fresh C. ovata flowers recorded from initial approach through multiple flo- between the 9 and the 16 January 2012. A 20 9 20 m ral visitations until the individual left the patch, not- grid was overlain onto the study site (600 9 200 m) ing which inflorescence(s) the individual landed on. and baiting locations determined in a random, strati- To compare the proportion of wasps alighting, con- fied fashion, so that baiting was only undertaken tacting the column, and copulating with the flowers within patches of vegetation. Baiting was undertaken over multiple visitations, G-tests were conducted in at 112 locations within the study site for 10-min peri- GenAlEx v6.5 (Peakall & Smouse, 2006, 2012). ods, with a minimum of 40 m distance between con- secutive locations. Orchid flowers were transported in a sealed container between baiting locations. Wasps POLLINATOR REVISITATION: WITHIN A DAY AND were marked on the thorax using coloured paint pens WITHIN A 7-DAY PERIOD (Uniball Posca) and nail varnish. A unique colour To determine the rate of pollinator revisitation to an combination was used for each location. Recaptured inflorescence within a day and within a week, two wasps were marked on the femur of the right hind leg mark-recapture experiments, that each comprised of with a colour unique to the recapture location, and 4 days of captures, were conducted during the week were given an additional mark upon first recapture. – – 27 December 2013 2 January 2014, and 14 20 Jan- Wasps were released at the point of capture. Potential uary 2014. Every second day during the week, an movement distances were calculated from UTM points inflorescence was presented at the exact same loca- using GenAlEx v6.5 (Peakall & Smouse, 2006, 2012). tion from 06:00 until 10:30 h. All wasps observed Observed mark-recapture distances are presented as contacting the column were caught and marked with raw frequencies, and as frequencies adjusted for the an individual colour code using nail varnish (Peakall probability of a recapture according to their distance & Schiestl, 2004), which was applied to the thorax class (calculated from the distribution of possible and hind leg to allow easy identification following movement distances). recapture. The colour ID of all recaptured wasps was recorded along with the date and time of their recap- ture. Two people were involved in wasp capture and COMPARISON WITH OTHER SEXUALLY DECEPTIVE identification during all experiments, which ensured SYSTEMS that all wasps contacting the column were captured. To investigate the consequences of ichneumonid pol- Wasps that alighted but did not contact the column lination relative to other sexually deceptive pollina- were not interfered with because: (1) our aims relate tion systems, we compared data from the present to the consequences of the behaviour of L. excelsa for study with the literature. Data were gathered for pollen movement, for which column contact is the orchid pollination rate, frequency of pollinator col- only relevant behaviour; (2) pollinator behaviour umn contact, frequency of pollinator pseudocopula- within a patch, including the full sequence of beha- tion with a flower, and pollinator movement vioural responses, was quantified in a previous distances. Species were only included if data were experiment; and (3) it is implausible to capture available for two out of the three variables of pollina- wasps without altering their subsequent behaviour tion rate, rate of pseudocopulation, and rate of col- (for example, after alighting only). Recapture rates umn contact. When behavioural data were available were calculated for both within a day and within a from multiple studies, data were selected from the week using pooled data from the two mark-recapture study providing both pseudocopulation and column = studies (data were pooled after returning P 0.5631 contact. Fruit set data were included for Pterostylis = (within day) and P 0.3775 (between days) in a sanguinea, as this species is known to be primarily – Fligner Killeen test for homogeneity of variance of pollen limited (Phillips et al., 2014b). When pollina- the number of recaptures per wasp). These recapture tion rate data were available from multiple years, rate data were collected and analysed separately to the rate was calculated by taking the mean of all the within-patch behavioural experiments. None of yearly values. When multiple sampling strategies the wasps caught in the first mark-recapture experi- (e.g. sweep netting and baiting) had been imple- ment was recaptured in the second (repeat) experi- mented in wasp movement studies, data from the ment, demonstrating that the behaviour of baiting method only were compared, to be consistent pollinators was not affected by encounters with orch- with our own sampling strategy. ids in the previous experiment. 287

For C. ovata, data for fruit set rather than pollina- Table 1. Number of Lissopimpla excelsa alighting on, tion rate were gathered. In C. ovata fruit set is pri- contacting the column of, and copulating with Cryp- marily pollen limited (A. Weinstein & R. Phillips, tostylis ovata flowers in artificially presented patches over unpubl. data), and given the long 3-month flowering multiple visitations period and sequential opening of the flowers, it was not possible to monitor populations regularly to First Second Third G-test obtain reliable pollination rate data. Fruit set for C. visitation visitation visitation P-value ovata was recorded from 16 populations (for locations N 240 59 18 see also Supporting Information, Tables S1) at the Alight (%) 89.2 91.5 100 0.128 end of the flowering season. For a total of 242 flower- Column 50.5 35.2 16.7 0.004* ing stems it was recorded whether or not capsules contact (%) developed. Fruit set was calculated as the percentage Copulation (%) 33.6 22.2 11.11 0.036* of flowers that developed capsules per inflorescence, from which population means were calculated. The Values for column contact and copulation are given as a final value for C. ovata was calculated by averaging percentage of that alighted, while those alighting the fruit set across the 16 populations. were calculated as a percentage of total responses. We aimed to understand the effect of pollinator *P < 0.05. behaviour at a flower on reproductive success. Firstly, we tested for a correlation between the fre- (7.5%) in a third, and 3 (1.3%) in a fourth. In the quency of pseudocopulation and column contact. We second visitation, 49.1% of wasps moved to a new then tested the expectation that a high frequency of inflorescence, 17.5% moved to a different flower on column contact would result in a high pollination the same inflorescence, and 33.3% returned to the rate. Correlations were tested for using R v 3.1.3 (R same flower. In the third visitation, 44.4% of wasps Development Core Team, 2015). To test if the size of moved to a new inflorescence, while the remaining the study site affects estimates of movement dis- 55.6% returned to the same flower as in their first tances, we tested for a correlation between the maxi- or second visitation. During the course of our experi- mum observed movement distances and the size of ment, 13 wasps (5.4%) contacted the column on at the study sites. Further, we present the maximum least two visitations to flowers within a single movement distances both as a raw distances and as encounter with the patch. A significant difference in percentages of the largest movement distances possi- pollinator behaviour was observed between the ini- ble within the study sites. tial and subsequent visitations, with a significant decrease being observed for both the percentage of wasps contacting the column (P = 0.004, G = 11.20, = = RESULTS d.f. 2) and copulating with flowers (P 0.036, G = 6.643, d.f. = 2) as the number of visitations POLLINATOR BEHAVIOUR: ON AN INFLORESCENCE AND increased (Table 1). WITHIN A PATCH In total, 240 wasps were observed visiting C. ovata flowers in artificially presented patches. Varying POLLINATOR REVISITATION RATE: WITHIN A DAY degrees of sexual attraction were displayed by indi- In total, 94 observations of marked wasps visiting vidual wasps – in their first visitation to a flower 214 and contacting the column were made. Of these (89.2%) wasps alighted on the flower, 108 (50.5% of wasps, 24 (25.5%) returned and contacted the column alighting wasps) contacted the column, and 72 again during the same 5½ h presentation period (33.6% of alighting wasps) attempted copulation (Fig. 2). Of the returning wasps, 18 (19.1%) returned (Table 1). Over the course of the study, both pollinia and contacted the column once, and six (6.4%) removal and deposition were observed during the returned and contacted the column two to four times reversing-in behaviour, both with and without subse- (Fig. 2). quent pseudocopulation. Approximately 10% of copu- lating wasps did not initially assume the reversed in position, and instead moved over the surface of the POLLINATOR REVISITATION RATE: WITHIN A 7-DAY labellum. Typically, these wasps eventually reversed PERIOD in to the flower, with the delay in reaching this posi- Within the two 7-day presentation periods, 31 tion often being caused by the presence of another (50.0%) of the 62 marked wasps observed contacting male already occupying the flower. the column of artificially presented flowers returned Of the wasps visiting the experimental patch, 59 and contacted the column of the presented flower (24.6%) engaged in a second floral visitation, 18 again (Fig. 3). Sixteen wasps (26.2%) returned and 288

cent (31) of wasps did not revisit the infloresence (Fig. 4).

LONG-DISTANCE POLLINATOR MOVEMENT In total, 224 male L. excelsa wasps were marked at 78 locations (Fig. 5). Wasps were not detected at 33 locations. Thirty-two individuals were recaptured (14.3% of those captured), with 30 of these being recaptured once, and four recaptured twice, total- ling 40 recapture events. The median time between capture and recapture was 1 day (range = 0– 4 days). Recapture distances ranged from 16.1 to 625.0 m, with a mean of 99.2 21.4 m (median = 49.0 m, N = 40; Fig. 6). Observed and adjusted pol- linator movement distances approximated a lep- tokurtic distribution (Fig. 6). These distributions were more positively skewed than the distribution Figure 2. Total number of Lissopimpla excelsa returning of possible movement distances, and as such a ½ to a Cryptostylis ovata flower within the daily 5 h obser- greater number of movement distances < 50 m were vation period across the eight sampling periods. observed than would be expected in a random movement pattern. contacted the column once, eight wasps (13.11%) twice, and seven wasps (11.45%) returned and con- tacted the column three to eight times (Fig. 3). COMPARISON WITH OTHER SEXUALLY DECEPTIVE In total, 64 revisitation events were observed SYSTEMS within the two 7-day trials. Fifty per cent of these Across the 16 populations of C. ovata surveyed (a revisitation events were contributed by 11.3% (7) of total of 242 plants), fruit set averaged 27.8 1.3%. the individual wasps, which revisited flowers on The mean number of flowering stems per population between three to eight occasions (Fig. 4). A further was 13.7 5.2 (range = 1–79). The mean number of 38.7% (24) of individuals revisited once or twice and flowers per raceme was 11.21 0.28 (range = 1–24). contributed the remaining revisitations. Fifty per Across the other species previously investigated and

Figure 3. Total number of Lissopimpla excelsa that returned and reversed in to a Cryptostylis ovata flower over the two 7-day observation periods (totalled from two periods). 289

Figure 4. Percentage contribution of each individual Lissompimpla excelsa wasp to total revisitations on artificially presented Cryptostylis ovata flowers across two 7-day experimental periods.

reported in the literature, pollination rate was lowest in Caladenia species (mean = 16.9%; minimum 5.9%, Caladenia ferruguinea), and highest in Drakaea spe- cies (mean = 59.4; maximum 63.2%, D. glyptodon; Table 2). The rate of pseudocopulation was lowest for Caladenia tentaculata (7.5%), and greatest in Caladenia attingens (63.1%). Column contact ranged from 1.2% in Caladenia sp. Moora to 50.5% in C. ovata. There was no significant correlation between the frequency of pseudocopulation and the frequency of column contact (R = 0.278, P > 0.05, N = 11), or between the frequency of column contact and pollina- tion rate (R = 0.270, P > 0.05, N = 9). Across the mark-recapture studies, both the mean (99.2 m) and maximum (625 m) recapture distances were greater for L. excelsa than the other species in the literature (Table 3). The shortest mean recapture distance was for Colletes cunicularius (5 m) (Colleti- dae), and the shortest maximum recapture distance for Thynnoides pugionatus (40 m) (Tiphiidae). A sig- nificant positive correlation between the maximum observed pollinator movement distance and the max- imum possible recapture distance in the study site was observed (R = 0.924, P < 0.001). However, this result is likely to arise through sites being selected to match the greater predicted pollinator movement Figure 5. Map of the study site in Kings Park and Bota- distances of larger species. Critically, Thynnoides sp. nic Garden, showing movements of male Lissopimpla A was the only species for which the maximum excelsa wasps attracted to Cryptostylis ovata bait flowers. observed recapture distance was within 10% of the Red dots are points where baiting for pollinators was con- maximum possible distance, with the observed move- ducted. ment distances for all species ranging from 40.0% to 290

Figure 6. Frequency distribution of movement distances of Lissopimpla excelsa males (movement distance categories: observed, adjusted for the probability of a recapture according to distance class, and all possible).

98.7% of the maximum possible movement distances in a laboratory bioassay after, but not before, they (Table 3). had been exposed to female extracts presented in an identical manner to the blanks (Robacker et al., 1976). Further, female ichneumonids can learn to associate artificial host shelters with the presence of DISCUSSION hosts (Wardle & Borden, 1985), and more specifically BEHAVIOUR OF LISSOPIMPLA EXCELSA WITHIN A VISIT to associate colour (Arthur, 1966; Wardle, 1990; Sch- TO A PATCH midt, Carde & Vet, 1993), size (Arthur, 1967), and Within a visit to a patch, pollinators that interacted odour cues (Arthur, 1971; Iizuka & Takasu, 1998) with multiple flowers exhibited a decrease in sexual with the presence of hosts. attraction to the flower with subsequent floral visits. While learning is likely to explain the observed As a result, while 25% of pollinators engaged in mul- decline in sexual response over time, pollinators tiple floral visitations within a single visit to the revisiting Cryptostylis flowers within a patch may patch, only a small portion (5.4%) contacted the col- experience a post-mating refractory period, poten- umn twice or more. A decrease in the degree of sex- tially caused by sperm depletion (Gordh & Hendrick- ual attraction to stimuli over time has been observed son, 1976). Although the presence or absence of in plant-pollinator interactions and laboratory exper- ejaculate on visited C. ovata flowers was not iments (Robacker, Weaver & Hendry, 1976; Ayasse assessed, this phenomenon is known in the genus et al., 2000; Gaskett et al., 2008; Whitehead & Pea- (Coleman, 1931; Erickson, 1951; Gaskett et al., kall, 2013; De Jager & Ellis, 2014). Data from the 2008). There have been only a few studies of mating present study suggest that pollinator learning is behaviour within the Ichneumonidae, making it responsible for the decrease in sexual response impossible to generalise to the family as a whole and observed in L. excelsa, as with each encounter the determine if this refractory period is related to a attraction of individual wasps to the flower shortage of sperm. Within the Ichneumonidae both decreased. Studies of ichneumonids in other contexts the presence (Gordh & Hendrickson, 1976) and have demonstrated a capacity for learning. Itoplectis absence (Vinson, 1972) of post-mating refractory conquisitor (which, like L. excelsa is in subfamily periods have been recorded. Male Bathyplectes anu- Pimplinae) males were observed to respond to blanks rus (Ichneumonidae) typically ignore females for 1 h Table 2. Rates of pollination, column contact, and pseudocopulation for sexually deceptive orchids and their pollinators

Column Pseudocopulation N Pollination rate N Orchid Pollinator contact (%) (%) (visitors) (%) (sites) Reference

Caladenia attingens Thynnoides sp. C (Tiphiidae) 21.4 63.1 130 13.4 4 R. Phillips, unpubl. data Caladenia crebra Camplyothynnus flavipictus (Tiphiidae) 12.8 60 113 ––R. Phillips, unpubl. data Caladenia ferruguinea Thynnid new genus 3 sp. B (Tiphiidae) 2.8 52.1 142 5.9 7 R. Phillips, unpubl. data Caladenia pectinata Zaspilothynnus nigripes (Tiphiidae) 4 10 233 12.3 8 Phillips et al. (2013) Caladenia sp. Moora Thynnoides sp. J (Tiphiidae) 1.2 30.6 114 ––R. Phillips, unpubl. data Caladenia tentaculata Thynnoides pugionatus (Tiphiidae) 7.5 7.5 287 36 3 Peakall & Beattie (1996) Chiloglottis Neozeleboria cryptoides (Tiphiidae) – 40 200 24.7* 2 De Jager & Peakall trapeziformis (2015), M. de Jager, unpubl. data* Chiloglottis trilabra Neozeleboria proxima (Tiphiidae) 25 25 2411 41 1 Peakall & Handel (1993) † Chiloglottis valida Neozeleboria monticola (Tiphiidae) – 28.6 48 21.7* 1 Whitehead & Peakall (2014), Whitehead et al. (2015)* ‡ Cryptostylis ovata Lissopimpla excelsa (Ichneumonidae) 50.5 30 240 28 16 Present study Drakaea glyptodon Zaspilothynnus trilobatus (Tiphiidae) 16 5.9 618 63.2* 51 Peakall (1990), Phillips et al. (2014a)* Drakaea livida Zaspilothynnus nigripes (Tiphiidae) 43 35 313 55.5 19 Phillips et al. (2013) Leporella fimbriata Myrmecia urens (Formicidae) 58 58 57 27.18 7 Peakall (1989) ‡ Pterostylis sanguinea Mycomya sp. (Mycetophilidae) – 57.8 135 23.7 19 Phillips et al. (2014b)

The rates of column contact and pseudocopulation are given as a percentage of the animals that alighted. Letters (e.g. sp. A) refer to species recognised in previ- ous papers such as Phillips et al. (2014a) and an unpublished dataset on the pollinators of Caladenia. *The origin of data where data for the same species has been sourced from two different studies. Data marked †, while being part of the referenced studies, were not explicitly stated and were obtained in personal communication with the authors. ‡ Data is for fruit set in place of pollination rate, as fruit set is in these systems known to be primarily pollen limited and collection of pollen deposition data was not feasible. 291 292

Table 3. Mean and maximum recapture distances for pollinator species of sexually deceptive orchids for which data were available

Maximum recapture distance Mean Maximum Maximum as a % of recapture recapture distance in maximum distance distance study site possible Pollinator species Family Orchid species pollinated N (m) (m) (m) distance Reference

Colletes cunicularius Colletidae Ophrys fusca, O. integra, 577 5 50 120* 41.7 Peakall & Schiestl (2004) O. exaltata, O. archipelagi Lissopimpla excelsa Ichneumonidae Cryptostylis ovata, C. erecta, 224 99.2 625 775 80.1 Present study C. subulata, C. leptochila, C. hunteriana Macrothynnus sp. A Tiphiidae Caladenia thinicola 120 56.3 470 707 66.5 M. H. M. Menz, unpubl. data Neozeleboria cryptoides Tiphiidae Chiloglottis trapeziformis 505 15.5 161 280* 57.5 Whitehead & Peakall (2013) Thynnoides pugionatus Tiphiidae Caladenia tentaculata 20 17 40 100 40.0 Peakall & Beattie (1996) Thynnoides sp. A Tiphiidae Drakaea gracilis 148 51.4 293 297 98.7 M. H. M. Menz, unpubl. data Zaspilothynnus gilesi Tiphiidae Drakaea elastica 147 55.2* 150 457 32.8 Menz et al. (2013) Zaspilothynnus nigripes Tiphiidae Drakaea livida 506 46.5* 300 457 65.6 Menz et al. (2013) † † Zaspilothynnus trilobatus Tiphiidae Drakaea glyptodon 21 32 132 160 82.5 Peakall (1990, 1987)

Letters (e.g. sp. A) refer to species recognised in previous papers such as Phillips et al. (2014a) and an unpublished dataset on the pollinators of Caladenia. Data marked *, while being part of the referenced studies, were not explicitly stated and were obtained by personal communication with the authors. † The origin of data where data for the same species has been sourced from two different studies. 293 following copulation (Gordh & Hendrickson, 1976). Chung, 2005; Mant et al., 2005), pollen transfer In contrast, male Campoletis sonorensis (Ichneu- within a patch of C. ovata inflorescences is likely monidae) have been observed to copulate a second to contribute to inbreeding. The analysis of pollina- time within minutes of the first copulation (Vinson, tor behaviour over multiple time scales revealed 1972). We predict that in L. excelsa a degree of learn- that any occurrence of pollen transfer within a ing is likely to be occurring within a visit to a patch, patch of C. ovata flowers is most likely to be due possibly in concert with the effect of sperm depletion to pollinator revisitation across several days, and in some individuals. not within a day or within a single encounter with a patch. Despite pollinators alighting on flowers multiple times within a single visit to a patch, only REVISITATION BEHAVIOUR OF LISSOPIMPLA EXCELSA 5.4% of pollinators exhibited behaviour that could During a mark-recapture study of the wasps that lead to the transfer of pollen within the patch (con- contacted the column of a single flowering plant tacting the column two or more times). While 25% over a 7-day period, 50% of marked L. excelsa were of pollinators were observed to revisit within a recaptured at least once – twice the recapture rate day, the highest rate of revisitation, and thereby observed during a single day. Similarly, in previous the greatest potential for pollen transfer within a studies of sexually deceptive orchids, the thynnine patch, occurred over a 7-day period (50% revisita- wasps Neozeleboria cryptoides and Zaspilothynnus tion rate). It must be noted that these estimates of trilobatus were observed not to revisit a site within pollen transfer within a patch do not take into the course of a day, yet to revisit the same site after account that it is unlikely that all pollinator visits a 24 h period had elapsed (Peakall, 1990; Whitehead will involve both pollinia removal and deposition, & Peakall, 2013). Given the evidence for learning and that pollinators may visit additional patches within a single visitation to a patch in L. excelsa,it between revisits to the monitored patch. is plausible that learning could also be functioning Lissopimpla excelsa had a mean recapture distance over a 24-h time period, potentially explaining the of 99 m and a capacity to travel distances of at least greater revisitation rate observed over several days. 625 m, which far exceeds the size of a C. ovata patch As this avoidance occurs for a 24-h period only, it is (typically < 5 m in diameter). Despite these large unlikely to be a strategy to avoid deceptive orchids, movement distances, 50% of pollinators were which flower for several weeks. Instead, such learn- observed to revisit patches multiple times, suggest- ing could have evolved via selection for males to ing that L. excelsa males are patrolling a home avoid sites where they had recently encountered a range. A similar degree of site fidelity has also been female, thereby optimising the discovery of newly reported for other ichneumonid species, along with emerged females. In Alabagrus texanus (Bra- the potential to move distances in excess of 100 m conidae), following initial scramble competition, (Heatwole & Davis, 1965). Considering the observed males avoid the site where a female has eclosed for movement distances of L. excelsa, and that C. ovata 1 h, after which they will again be attracted to patches are usually located within a few hundred females at the site (Goh & Morse, 2010). Only rarely metres of one another, it is likely that frequent pol- does more than one female eclose within a five 5 m len dispersal between patches or subpopulations of area on the same day, making avoiding the eclosure C. ovata is occurring. site advantageous for increasing the efficacy of mate It is of interest that despite the large number of searching (Goh & Morse, 2010). Importantly, this wasps visiting the C. ovata flowers, proportionally explanation is likely to apply regardless of whether few were responsible for the high rate of revisitation. or not the encounter with a female resulted in a This result may indicate that pollen movement successful mating. It remains unknown whether a within patches of C. ovata could be caused by only a similar phenomenon may be operating in L. excelsa, small portion of the population of visiting wasps. A as patterns of female eclosure for both L. excelsa skew in the rate of revisitations could possibly be and ichneumonid wasps in general have not been related to a variation in male fitness or in the stimu- described. lus threshold required to respond to flowers. A low stimulus threshold could be attributed to either time since emergence, or a lack of sexual experience. IMPLICATIONS FOR THE POLLINATION OF However, the one study to address the role of sexual CRYPTOSTYLIS OVATA experience in ichneumonid mating behaviour (Vin- Given the clonal nature of C. ovata (Dodd et al., son, 1972) reported no difference in the response 1984), and the prevalence of fine-scale genetic rates of mated and virgin Campoletis sonorensis (Ich- structure in orchids (Peakall & Beattie, 1996; Trap- neumonidae) males to females. Alternately, the nell, Hamrick & Nason, 2004; Chung, Nason & wasps making repeat visits may have a different 294 mate search strategy or home range than the 14 study species, suggesting that this will be an remainder of the population. important future area of study. The lack of a correlation between pseudocopulation and column contact, and between column contact BEHAVIOUR OF L. EXCELSA IN THE CONTEXT OF and pollination rate is in contrast with the finding of OTHER SEXUALLY DECEPTIVE SYSTEMS Gaskett et al. (2008) that orchids provoking more intense pollinator behaviour have higher pollination Pollinator behaviour at a flower and its impact on success. As a measure of pollinator behaviour, Gas- fruit set kett et al. (2008) assigned orchids to discrete cate- Despite being one of only two genera in which the gories of observed pollinator sexual behaviour pollinator has been observed to ejaculate onto the (ejaculation, copulation, gripping a hinged , and flower (Coleman, 1931; Erickson, 1951; Blanco & entrapment). It was found that more intense pollina- Barboza, 2005; Gaskett et al., 2008), the rate of polli- tor behaviour was associated with a higher pollina- nator pseudocopulation was low relative to other sex- tion rate. However, this result was largely driven by ually deceptive species included in the analyses. the high frequency of pollinated plants reported for However, the efficacy of C. ovata at converting polli- Cryptostylis, the only genus in the highest category nator attraction into column contact was high when of pollinator sexual response. Since the analysis of compared with other orchids. Gaskett et al. (2008), a much larger dataset has While it could be generally predicted that more become available allowing our more detailed investi- frequent pseudocopulation would increase contact gation, in which pollination rate is assessed on a per with the column, this relationship was not observed flower basis. While our comparison across genera in our analysis across sexually deceptive systems. does not support the hypothesis of Gaskett et al. Such a relationship would be dependent on pseudo- (2008) that greater sexual attraction leads to a copulation being with the part of the flower where higher pollination rate, a comparison of closely pseudocopulatory movements bring the pollinator related species with similar floral traits may support into contact with the column. While in most sexu- this prediction. ally deceptive orchids the sexual attractant is released from the labellum (Schiestl et al., 2000; Pollinator movement distances Phillips et al., 2013, 2014b; Bohman et al., 2014), Here we compare the movement distances of an ich- in Caladenia pseudocopulation frequently occurs neumonid (Lissopimpla excelsa) to those of other sex- with the glandular tips, leading to a compar- ually deceived pollinators. The sole bee species in atively inefficient pollination mechanism (Phillips this comparison, Colletes cunicularius, has a consid- et al., 2013). As demonstrated in C. ovata, the con- erably shorter mean recapture distance than the verse may also apply, where column contact can other species, which is due to the different mating occur without pseudocopulatory behaviour. There- behaviours of solitary bees and thynnine wasps fore, the frequency of pseudocopulation and column (which represent the majority of species in the com- contact may not be correlated across systems due parison). Male solitary bees patrol specific ren- to taxonomic variation in floral traits and pollinator dezvous sites (Alcock et al., 1978; Ayasse, Paxton & behaviour. Tengo,€ 2001), whereas thynnine males patrol com- It is often presumed that a high efficacy of convert- paratively large home ranges (Alcock, 1981). ing pollinator attraction into contact with the repro- Amongst the thynnine species there was a high ductive structures would be correlated with a high degree of variability in maximum movement dis- pollination rate. However, this trend was not tance, ranging from T. pugionatus (40 m) to M. observed in our analyses. While a high efficacy of insignis (470 m). However, L. excelsa has even converting attraction into column contact contributes greater mean and maximum recapture distances, towards pollination rate, the abundance and fidelity with its mean recapture distance being almost dou- of the pollinator species, and the spatial distribution ble that of all the other compared species (Table 3). of the plants, also have an effect. For example, a The difference in movement distances between L. small number of effective pollinator visits could lead excelsa and the other studied sexually deceived polli- to a similar pollination rate as could many less effec- nators may be attributable to differences in the mate tive visits. For a direct comparison of the efficiency searching or foraging behaviour between insect fami- of pollinators between systems, it may be informative lies. Previous studies of ichneumonid wasps, how- to focus on the ratio of pollen deposition relative to ever, report movement distances of no more than removal, as this measure is independent of the num- 107 m (Juillet, 1959; Heatwole & Davis, 1965). While ber of pollinator visitations. Unfortunately, these these studies were not consistent in methodology, data are currently only available for two out of the these results suggest that the home range of L. 295 excelsa may also exceed those of some other ichneu- bees and wasps. Zoological Journal of the Linnean Society monids. 64: 293–326. Arthur AP. 1966. Associative learning in Itoplectis conquisi- tor (Say) (Hymenoptera: Ichneumonidae). The Canadian Entomologist 98: 213–223. CONCLUSIONS Arthur AP. 1967. Influence of position and size of host Combining the observation of experimental C. ovata shelter on host-searching by Itoplectis conquisitor (Hyme- patches with mark-recapture studies demonstrated noptera: Ichneumonidae). The Canadian Entomologist 99: – that pollen transfer within a patch of orchids is 877 886. likely to arise most frequently through pollinator Arthur AP. 1971. Associative learning by Nemeritis canes- cens (Hymenoptera: Ichneumonidae). The Canadian Ento- revisitation over a period of several days, rather mologist 103: 1137–1141. than within a day or within a single visit. However, Ayasse M, Schiestl FP, Paulus HF, Lofstedt€ C, Hans- L. excelsa are also capable of moving distances in son B, Ibarra F, Francke W. 2000. Evolution of repro- excess of half a kilometre, demonstrating their abil- ductive strategies in the sexually deceptive orchid Ophrys ity to transfer pollen between subpopulations or far- sphegodes: how does flower-specific variation of odor sig- ther. These movement distances are the largest nals influence reproductive success? Evolution 54: 1995– reported to date for sexually deceived pollinators, 2006. and the first reported for a sexually deceived ichneu- Ayasse M, Paxton R, Tengo€ J. 2001. Mating behavior and monid, highlighting the unexplored potential for chemical communication in the order Hymenoptera. Annual variation in pollinator behaviour between taxonomic Review of Entomology 46: 31–78. groups. As such, to comprehensively understand the Barrows EM. 1976. Sexual behavior in Hyposoter fugitivus consequences of pollination by sexual deception, (Hymenoptera: Ichneumonidae). Entomological News 87: detailed investigations of the natural history of other 101–102. pollinator groups and their interaction with sexually Bin F, Wackers€ F, Romani R, Isidoro N. 1999. Tyloids in deceptive flowers are required. Such studies should Pimpla turionellae (L.) are release structures of male encompass aspects of pollinator behaviour responsi- antennal glands involved in courtship behaviour (Hymenop- ble for pollen transfer, and how these behaviours are tera: Ichneumonidae). International Journal of Insect Mor- influenced by floral traits. phology and Embryology 28: 61–68. Blanco MA, Barboza G. 2005. Pseudocopulatory pollination in Lepanthes (Orchidaceae: Pleurothallidinae) by fungus gnats. Annals of Botany 95: 763–772. ACKNOWLEDGEMENTS Bohman B, Jeffares L, Flematti GR, Phillips RD, Dixon KW, Peakall R, Barrow RA. 2012. The discovery of 2- During the course of this study, A.M. Weinstein was hydroxymethyl-3-(3-methylbutyl)-5-methylpyrazine: a supported by a Kings Park Summer Scholarship semiochemical in orchid pollination. Organic Letters 14: from the Botanic Parks and Gardens Authority of 2576–2578. Western Australia. Amber Stacey and Sarah Water- Bohman B, Phillips RD, Menz MHM, Berntsson BW, field are thanked for their assistance with fieldwork Flematti GR, Barrow RA, Dixon KW, Peakall R. 2014. as part of the Primary Industry Centre for Science Discovery of pyrazines as pollinator sex pheromones and Education Industry Placement Programme. Rod Pea- orchid semiochemicals: implications for the evolution of sex- kall and two anonymous reviewers are thanked for ual deception. New Phytologist 203: 939–952. their insightful comments on the manuscript, as is Bower CC. 1996. Demonstration of pollinator-mediated M. de Jager for the use of his unpublished data. reproductive isolation in sexually deceptive species of Chiloglottis (Orchidaceae: Caladeniinae). Australian Jour- nal of Botany 44: 15–33. REFERENCES Brown A, Dundas P, Dixon KW, Hopper SD. 2008. Orch- ids of Western Australia. Perth: University of Western Aus- Alcock J. 1981. Notes on the reproductive behavior of some tralia Press. Australian thynnine wasps (Hymenoptera: Tiphiidae). Jour- Chung M, Nason J, Chung M. 2005. Spatial genetic struc- nal of the Kansas Entomological Society 54: 681–693. ture in populations of the terrestrial orchid Orchis cyclo- Alcock J. 2000. Interactions between the sexually deceptive chila (Orchidaceae). Plant Systematics and Evolution 254: orchid Spiculaea ciliata and its wasp pollinator Thynno- 209–219. turneria sp. (Hymenoptera: Thynninae). Journal of Natural Cole LR. 1970. Observations on the finding of mates by male History 34: 629–636. Phaeogenes invisor and Apanteles medicaginis (Hymenop- Alcock J, Barrows EM, Gordh G, Hubbard L, Kirk- tera: Ichneumonoidea). Behaviour 18: 184–189. endall L, Pyle DW, Ponder TL, Zalom F. 1978. The ecol- Coleman E. 1927a. Pollination of Cryptostylis leptochila. ogy and evolution of male reproductive behaviour in the The Victorian Naturalist 44: 333–340. 296

Coleman E. 1927b. Pollination of the orchid Cryptostylis lep- Gaskett AC. 2012. Floral shape mimicry and variation in tochila. The Victorian Naturalist 44: 20–23. sexually deceptive orchids with a shared pollinator. Biologi- Coleman E. 1928. Pollination of an australian orchid by cal Journal of the Linnean Society 106: 469–481. the male ichneumonid Lissopimpla semipunctata, Kirby. Gaskett AC, Herberstein ME. 2010. Colour mimicry and Transactions of the Entomological Society of London 76: sexual deception by Tongue orchids (Cryptostylis). Natur- 533–539. wissenschaften 97: 97–102. Coleman E. 1929. Pollination of Cryptostylis subulata Gaskett AC, Winnick CG, Herberstein ME. 2008. Orchid (Labill.) Reichb. The Victorian Naturalist 46: 62–66. sexual deceit provokes ejaculation. The American Natural- Coleman E. 1930a. Pollination of Cryptostylis erecta R. Br. ist 171: E206–E212. The Victorian Naturalist 46: 236–238. Goh MZ, Morse DH. 2010. Male mate search for female Coleman E. 1930b. Pollination of some West Australian emergence sites by a parasitic wasp. Animal Behaviour 80: orchids. The Victorian Naturalist 46: 203–206. 391–398. Coleman E. 1931. Mrs. Edith Coleman’s further observa- Gordh G, Hendrickson R. 1976. Courtship behavior in tions on the fertilisation of Australian orchids by the male Bathyplectes anurus (Thomson) (Hymenoptera: Ichneu- ichneumonid, Lissopimpla semipunctata. Proceedings of the monidae). Entomological News 87: 271–274. Entomological Society of London 6: 22–24. Heatwole H, Davis DM. 1965. Ecology of three sympatric Coleman E. 1938. Further observation on the pseudocopula- species of parasitic insects of the genus Megarhyssa (Hyme- tion of the male Lissopimpla semipunctata Kirby (Hyme- noptera: Ichneumonidae). Ecology 46: 140–150. noptera parasitica) with the Australian orchid Cryptostylis Heatwole H, Davis DM, Wenner AM. 1964. Detection of leptochila F. v. M. Proceedings of the Royal Entomological mates and hosts by parasitic insects of the genus Mega- Society of London 13: 82–83. rhyssa (Hymenoptera: lchneumonidae). American Midland Crankshaw OS, Matthews RW. 1981. Sexual behavior Naturalist 71: 374–381. among parasitic Megarhyssa wasps (Hymenoptera: Ichneu- Iizuka T, Takasu K. 1998. Olfactory associative learning of monidae). Behavioral Ecology and Sociobiology 9: 1–7. the pupal parasitoid Pimpla luctuosa Smith (Hymenoptera: De Jager ML, Ellis AG. 2014. Costs of deception and Ichneumonidae). Journal of Insect Behavior 11: 743–760. learned resistance in deceptive interactions. Proceedings of Jersakov a J, Johnson SD, Kindlmann P. 2006. Mecha- the Royal Society of London B: Biological Sciences 281: nisms and evolution of deceptive pollination in orchids. Bio- 20132861. logical Reviews 81: 219–235. De Jager ML, Peakall R. 2015. Does morphology matter? Jewett DK, Carpenter JE. 1999. Chemically-mediated An explicit assessment of floral morphology in sexual decep- attraction of Ichneumon (= Pterocormus) promissorius tion. Functional Ecology 30: 537–546. (Hymenoptera: Ichneumonidae) males by females. Environ- Dodd J, Heddle E, Pate J, Dixon K. 1984. Rooting pat- mental Entomology 28: 551–556. terns of sandplain plants and their functional significance. Johnson SD, Nilsson LA. 1999. Pollen carryover, geitono- In: Pate J, Bear J, eds. Kwongan: plant life of the sand- gamy, and the evolution of deceptive pollination systems in plain: biology of a south-west Australian shrubland ecosys- orchids. Ecology 80: 2607–2619. tem. Nedlands, WA: University of Western Australia Press, Juillet JA. 1959. Immature stages, life histories, and beha- 146–177. viour of three hymenopterous parasites of the European Dowell R, Horn D. 1975. Mating behavior of Bathyplectes pine shoot moth, Rhyacionia buoliana (Schiff.) (Lepi- curculionis [Hym.: Ichneumonidae] a parasitoid of the doptera: Olethreutidae). The Canadian Entomologist 92: alfalfa weevil, Hypera postica [Col.: Curculionidae]. Ento- 342–346. mophaga 20: 271–273. Klopfstein S, Quicke DLJ, Kropf C. 2010. The evolution Dressler RL. 1981. The orchids: natural history and classifi- of antennal courtship in diplazontine parasitoid wasps cation. Cambridge, MA: Harvard University Press. (Hymenoptera, Ichneumonidae, Diplazontinae). BMC Evo- Dyer LE, Landis DA. 1997. Diurnal behavior of Eriborus lutionary Biology 10: 218. terebrans (Hymenoptera: Ichneumonidae). Environmental Kolarov J, Gurb€ uz€ MF. 2007. New data on the structure of Entomology 26: 1385–1392. the flagellum in males of Perithous divinator (Rossi) Ellis AG, Johnson SD. 2010. Floral mimicry enhances pol- (Hymenoptera, Ichneumonidae, Pimplinae). Acta Entomo- len export: the evolution of pollination by sexual deceit out- logica Serbica 12: 59–66. side of the Orchidaceae. The American Naturalist 176: Kugler J, Wollberg Z. 1967. Biology of Agrothereutes tune- E143–E151. tanus Haber.[Hym. Ichneumonidae] an ectoparasite of Erickson R. 1951. Orchids of the West. Perth: Paterson Bro- Orgyia dubia Tausch.[Lep. Lymantriidae]. Entomophaga kensha. 12: 363–379. Fisher RC. 1959. Life history and ecology of Horogenes Kullenberg B. 1961. Studies in Ophrys pollination. Zoolo- chrysostictos Gmelin (Hymenoptera, Ichneumonidae), a par- giska Bidrag fran^ Uppsala 34: 1–340. asite of Ephestia sericarium Scott (Lepidoptera, Phycitidae). Mant J, Bower C, Weston P, Peakall R. 2005. Phylogeog- Canadian Journal of Zoology 37: 429–446. raphy of pollinator-specific sexually deceptive Chiloglottis Gaskett AC. 2011. Orchid pollination by sexual deception: taxa (Orchidaceae): evidence for sympatric divergence? pollinator perspectives. Biological Reviews 86: 33–75. Molecular Ecology 14: 3067–3076. 297

Menz MHM, Phillips RD, Dixon KW, Peakall R, Didham Phillips RD, Peakall R, Hutchinson MF, Linde CC, Xu RK. 2013. Mate-searching behaviour of common and rare T, Dixon KW, Hopper SD. 2014a. Specialized ecological wasps and the implications for pollen movement of the sexu- interactions and plant species rarity: the role of pollinators ally deceptive orchids they pollinate. PLoS ONE 8: e59111. and mycorrhizal fungi across multiple spatial scales. Biolog- Morey CS. 1971. Biologıa de Campoletis grioti (Blanchard) ical Conservation 169: 285–295. (Hymen.: Ichneumonidae) parasito de la ‘lagarta cogollera Phillips RD, Scaccabarozzi D, Retter BA, Hayes C, del maız’ Spodoptera frugiperda (J. E. Smith). Revista Peru- Brown GR, Dixon KW, Peakall R. 2014b. Caught in the ana de Entomologia 14: 263–271. act: pollination of sexually deceptive trap-flowers by fungus Nicholls WH. 1938. A new species of the genus Cryptostylis gnats in Pterostylis (Orchidaceae). Annals of Botany 113: R. Br. The Victorian Naturalist 54: 182–183. 629–641. Pate JS, Dixon KW. 1982. Tuberous, cormous and bulbous Pridgeon AM, Cribb PJ, Chase MW. 2001. Genera Orchi- plants. Biology of an adaptive strategy in Western Aus- dacearum: Vol. 2. . New York: Oxford Univer- tralia. Nedlands, WA: University of Western Australia sity Press. Press. R Development Core Team. 2015. R: a language and envi- Paulus HF, Gack C. 1990. Pollination of Ophrys (Orchi- ronment for statistical computing. Vienna, Austria: R Foun- daceae) in Cyprus. Plant Systematics and Evolution 169: dation for Statistical Computing. 177–207. Robacker DC, Weaver KM, Hendry LB. 1976. Sexual Peakall RJ. 1987. Genetic systems of Australian terrestrial communication and associative learning in the parasitic orchids. PhD thesis, University of Western Australia. wasp Itoplectis conquisitor (Say). Journal of Chemical Ecol- Peakall R. 1989. The unique pollination of Leporella fimbri- ogy 2: 39–48. ata (Orchidaceae): pollination by pseudocopulating male Rotheray GE. 1981. Courtship, male swarms and a sex ants (Myrmecia urens, Formicidae). Plant Systematics and pheromone of Diplazon pectoratorius (Thunberg) (Hymenop- Evolution 167: 137–148. tera: Ichneumonidae. Entomologist’s Gazette 32: 193–196. Peakall R. 1990. Responses of male Zaspilothynnus triloba- Schiestl FP, Ayasse M, Paulus HF, Lofstedt€ C, Hansson tus Turner wasps to females and the sexually deceptive BS, Ibarra F, Francke W. 1999. Orchid pollination by orchid it pollinates. Functional Ecology 4: 159–167. sexual swindle. Nature 399: 421. Peakall R, Beattie AJ. 1996. Ecological and genetic conse- Schiestl FP, Ayasse M, Paulus HF, Lofstedt€ C, Hansson quences of pollination by sexual deception in the orchid BS, Ibarra F, Francke W. 2000. Sex pheromone mimicry Caladenia tentaculata. Evolution 50: 2207–2220. in the early spider orchid (Ophrys sphegodes): patterns of Peakall R, Handel SN. 1993. Pollinators discriminate hydrocarbons as the key mechanism for pollination by sex- among floral heights of a sexually deceptive orchid: implica- ual deception. Journal of Comparative Physiology A 186: tions for selection. Evolution 47: 1681–1687. 567–574. Peakall R, James SH. 1989. Outcrossing in an ant polli- Schiestl FP, Peakall R, Mant JG, Ibarra F, Schulz C, nated clonal orchid. Heredity 62: 161–167. Franke S, Francke W. 2003. The chemistry of sexual Peakall R, Schiestl FP. 2004. A mark-recapture study of deception in an orchid-wasp pollination system. Science male Colletes cunicularius bees: implications for pollination 302: 437–438. by sexual deception. Behavioral Ecology and Sociobiology Schiestl FP, Peakall R, Mant JG. 2004. Chemical com- 56: 579–584. munication in the sexually deceptive orchid genus Cryp- Peakall R, Smouse PE. 2006. GENALEX 6: genetic analy- tostylis. Botanical Journal of the Linnean Society 144: sis in Excel. Population genetic software for teaching and 199–205. research. Molecular Ecology Notes 6: 288–295. Schmidt JM, Carde RT, Vet LEM. 1993. Host recognition Peakall R, Smouse PE. 2012. GenAlEx 6.5: genetic analysis by Pimpla instigator F. (Hymenoptera: Ichneumonidae): in Excel. Population genetic software for teaching and preferences and learned responses. Journal of Insect Behav- research – an update. Bioinformatics 28: 2537–2539. ior 6: 1–11. Peakall R, Ebert D, Poldy J, Barrow RA, Francke W, Slobodchikoff CN. 1973. Behavioral studies of three mor- Bower CC, Schiestl FP. 2010. Pollinator specificity, floral photypes of circumflexum (Hymenoptera: Ichneu- odour chemistry and the phylogeny of Australian sexually monidae). The Pan-Pacific Entomologist 49: 197–206. deceptive Chiloglottis orchids: implications for pollinator- Spradbery JP. 1969. The biology of Pseudorhyssa sternata driven speciation. New Phytologist 188: 437–450. Merrill (Hym., Ichneumonidae), a cleptoparasite of Siricid Phillips RD, Faast R, Bower CC, Brown GR, Peakall R. woodwasps. Bulletin of Entomological Research 59: 291–297. 2009. Implications of pollination by food and sexual decep- Steiner SM, Kropf C, Graber W, Nentwig W, Klopfstein tion for pollinator specificity, fruit set, population genetics S. 2010. Antennal courtship and functional morphology of and conservation of Caladenia (Orchidaceae). Australian tyloids in the parasitoid wasp Syrphoctonus tarsatorius Journal of Botany 57: 287–306. (Hymenoptera: Ichneumonidae: Diplazontinae). Phillips RD, Xu T, Hutchinson MF, Dixon KW, Peakall Structure & Development 39: 33–40. R. 2013. Convergent specialization–the sharing of pollina- Stoutamire WP. 1974. Australian terrestrial orchids, thyn- tors by sympatric genera of sexually deceptive orchids. nid wasps, and pseudocopulation. American Orchid Society Journal of Ecology 101: 826–835. Bulletin 43: 13–18. 298

Stoutamire WP. 1983. Wasp-pollinated species of Caladenia (Hymenoptera: Ichneumonidae). Environmental Entomology (Orchidaceae) in South-western Australia. Australian Jour- 1: 409–414. nal of Botany 31: 383–394. Wardle AR. 1990. Learning of host microhabitat colour by Tomlinson S, Phillips RD. 2012. Metabolic rate, evapora- Exeristes roborator (F.) (Hymenoptera: Ichneumonidae). tive water loss and field activity in response to temperature Animal Behaviour 39: 914–923. in an Ichneumonid wasp. Journal of Zoology 287: 81–90. Wardle AR, Borden JH. 1985. Age-dependent associative Trapnell DW, Hamrick JL, Nason JD. 2004. Three- learning by Exeristes roborator (F.) (Hymenoptera: Ichneu- dimensional fine-scale genetic structure of the neotropical monidae). The Canadian Entomologist 117: 605–616. epiphytic orchid, Laelia rubescens. Molecular Ecology 13: Whitehead MR, Peakall R. 2013. Short-term but not long- 1111–1118. term patch avoidance in an orchid-pollinating solitary Van der Pijl L, Dodson CH. 1966. Orchid flowers: their wasp. Behavioral Ecology 24: 162–168. pollination and evolution. Coral Gables, FL: University of Whitehead MR, Peakall R. 2014. Pollinator specificity Miami Press. drives strong prepollination reproductive isolation in van Veen JC. 1982. Notes on the biology of Banchus femora- sympatric sexually deceptive orchids. Evolution 68: lis Thomson (Hym., Ichneumonidae) an endoparasitoid of 1561–1575. Panolis flammea (D. & S.) (Lep., ). Zeitschrift fur€ Whitehead MR, Linde CC, Peakall R. 2015. Pollination by Angewandte Entomologie 94: 300–311. sexual deception promotes outcrossing and mate diversity Vinson SB. 1972. Courtship behavior and evidence for a in self-compatible clonal orchids. Journal of Evolutionary sex pheromone in the parasitoid Campoletis sonorensis Biology 28: 1526–1541.

SUPPORTING INFORMATION

Table S1. Locations of populations of Cryptostylis ovata from which fruit set data were collected.