Erschienen in: Botanical Journal of the Linnean Society ; 179 (2015), 3. - S. 511-525 https://dx.doi.org/10.1111/boj.12336

Pollinator rarity as a threat to a with a specialized pollination system

RYAN D. PHILLIPS1,2,3*, ROD PEAKALL1, BRYONY A. RETTER2,3, KIRKE MONTGOMERY2, MYLES H. M. MENZ4, BELINDA J. DAVIS2,3, CHRISTINE HAYES1, GRAHAM R. BROWN5,6, NIGEL D. SWARTS7,8 and KINGSLEY W. DIXON2,9

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 Plant Biology, The University of , Crawley, WA 6009, Australia 4Institute of Ecology and Evolution, University of Bern, Baltzerstrasse 6, 3012 Bern, Switzerland 5Museum and Art Gallery of Northern Territory, GPO Box 4646, Darwin, NT 0801, Australia 6Research institute for Environment and Livelihoods, Charles Darwin University, Darwin, NT 0909, Australia 7Royal Tasmanian Botanic Gardens, Queens Domain, Hobart, TAS 7000, Australia 8Tasmanian Institute of Agriculture, University of Tasmania, Private Bag 98, Hobart, TAS 7000, Australia 9Department of Agriculture and Environment, Curtin University, WA 6102, Australia

An increasing diversity of highly specialized pollination systems are being discovered, many of which are likely to be vulnerable to anthropogenic landscape modification. Here, we investigate if a specialized pollination system limits the persistence of Caladenia huegelii (), an endangered species pollinated by sexual deception of thynnine wasps. Once locally common in part of its geographical range, C. huegelii is now largely restricted to small habitat remnants in urban areas. Pollinator surveys coupled with DNA barcoding detected a single pollinator taxon, a small form of Macrothynnus insignis. Phylogenetic analysis revealed that small M. insignis from within the range of C. huegelii are strongly divergent from other wasp populations, suggesting that some reproductive isolation may exist. Although common in intact landscapes outside the range of C. huegelli, small M. insignis individuals were recorded at only 4% of sites in suitable C. huegelii habitat. Accordingly, reproductive success in C. huegelii was low compared with related Caladenia spp., with 33–60% of populations failing to set fruit in any given year. As such, populations are likely to now persist primarily through individual plant longevity rather than reproduction. Due to the low reproductive success of C. huegelii, ongoing human intervention will almost certainly be needed to sustain the species. Future research will need to focus on optimizing hand pollination to maintain reproduction and high seed fitness. © 2015 The Linnean Society of London, Botanical Journal of the Linnean Society, 2015, 179, 511–525.

ADDITIONAL KEYWORDS: habitat fragmentation – orchid – sexual deception.

INTRODUCTION pollinated are specialized on a subset of the available pollinator community (Fenster et al., 2004; Although pollination by multiple animal species pro- Rosas-Guerrero et al., 2014). Further, there is vides greater reproductive assurance, most animal- increasing recognition that there is a wide range of specialized pollination systems, with plants being *Corresponding author. E-mail: [email protected] reliant on just one or a few species of pollinators

511

Konstanzer Online-Publikations-System (KOPS) URL: http://nbn-resolving.de/urn:nbn:de:bsz:352-2-ctuy3kdcag0q3 512

(Johnson, 2010). In these systems, specialization can analysis in south-western Australia (Phillips et al., be mediated through a diversity of processes includ- 2011). Further, a detailed investigation of correlates ing exaggerated morphology for which only one polli- of rarity in the sexually deceptive genus Drakaea nator species can reach the nectar (Nilsson et al., Lindl. suggested that rarity was most likely caused by 1987), unusual rewards of interest to a small portion the absence of pollinators from otherwise suitable of the pollinator community (Houston et al., 1993; orchid habitat, rather than low levels of reproductive Pauw & Bond, 2011) or chemical attractants that success within existing populations (Phillips et al., exploit sensory biases of a particular pollinator 2014a). However, meta-analyses have shown that in (Schiestl et al., 2003; Brodmann et al., 2008; Bohman other genera sexually deceptive species tend to have et al., 2014). As a consequence of relying on just a few lower levels of fruit set than related orchids with pollinator species for sexual reproduction, low density more generalized pollination strategies (Phillips or narrow geographical range of the pollinator can et al., 2009b; Gaskett, 2011), suggesting low fruit set contribute to rarity of the plant (Phillips et al., could cause rarity in some sexually deceptive species. 2014a). Further, should a pollinator species undergo a Caladenia R.Br. is a diverse genus of terrestrial decline following anthropogenic landscape modifica- orchids (>375 taxa) with centres of diversity in south- tion, plant populations are also likely to decline western and south-eastern Australia (Phillips et al., (Anderson et al., 2011; Pauw & Hawkins, 2011; 2009a), two regions with extensive habitat removal. Geerts & Pauw, 2012). In extreme scenarios, local Although many Caladenia taxa are pollinated by extinctions of pollinators can leave plant populations nectar-foraging insects (Faast et al., 2009; Phillips entirely reliant on clonal reproduction. This situation et al., 2009b), and a small portion are self-pollinating is exemplified by the oil-producing orchids pollinated (Jones, 2006), > 100 species representing three subgen- by the bee Rediviva peringueyi, with local extinction era are believed to be pollinated by sexual deception of of the bees leading to local extinctions of the orchids, thynnine wasps (Stoutamire, 1983; Bower, 2001; and the less clonal species suffering extinction more Phillips et al., 2009b). Caladenia contains > 70 taxa rapidly (Pauw & Hawkins, 2011). that are currently federally listed as threatened or Orchidaceae are characterized by a high incidence endangered (Department of the Environment, 2015), of specialized pollination strategies, with many over half of which are believed to be pollinated via orchid species reliant on only one or few pollinator sexual deception. This trend raises the question of species (Tremblay, 1992; Schiestl & Schluter, 2009). whether pollination by sexual deception is contribut- Of these strategies, one of the most diverse and ing to rarity either through intrinsic rarity of some geographically widespread is pollination by sexual wasp species or vulnerability of specialized pollination deception (Gaskett, 2011; Phillips et al., 2014b). In systems to landscape modification. Given the diversity this pollination strategy, long-range attraction of pol- of the genus, understanding causes of rarity in sexu- linators occurs through mimicry of the sex phero- ally deceptive Caladenia may yield useful information mone of female insects, with pollination achieved by for the management of a large number of threatened the male during attempted courtship or copulation taxa. with the flower (Schiestl et al., 1999, 2003; Bohman Caladenia huegelii Rchb.f. is a rare species endemic et al., 2014). As a by-product of mimicking the spe- to south-western Australia (Hoffman & Brown, 2011) cific sex pheromones of insects, sexually deceptive (Fig. 1). It is currently known only from the Swan orchids often have only a single known pollinator Coastal Plain between Perth and Busselton (Fig. 2), species (Paulus & Gack, 1990; Phillips et al., 2009b; preferring areas of woodland on the Bassen- Peakall et al., 2010; Gaskett, 2011). However, studies dean dune system (Hoffman & Brown, 2011). Once investigating populations of sexually deceptive locally common in at least part of its range (Hopper & orchids over a broad geographical range are increas- Brown, 2001), C. huegelii has undergone an extensive ingly revealing geographical variation in pollinator decline associated with the large scale habitat species (Breitkopf et al., 2013; Menz et al., 2015b; destruction for agriculture, housing and light Phillips et al., 2015). Furthermore, DNA barcoding industrial areas, which has continued in recent has demonstrated that some groups of sexually decades (Brown, Thomson-Dans & Marchant, 1998; deceived pollinators contain morphologically cryptic Department of Environment and Conservation, 2008). taxa (Griffiths et al., 2011; Menz et al., 2015b), poten- Of the 48 known populations, no plants were recorded tially obscuring the exact number of species actually at 21 of them during the last major surveys involved in pollination. (Department of Environment and Conservation, Due to their high pollinator specificity, lineages of 2008). Further, 17 of the surviving populations con- sexually deceptive orchids may show higher levels of tained less than five flowering plants (Department of intrinsic rarity than other pollination strategies, a Environment and Conservation, 2008). Due to the prediction that has been supported by a regional small and isolated nature of the remaining popula- 513

Figure 1. Macrothynnus insignis s.l. (A), the pollinator of the sexually deceptive orchid Caladenia huegelii (B). Panel (A) illustrates specimens of the large and small forms of M. insignis s.l. Only the small form is responsible for pollination of C. huegelii. Photographs (A) by Belinda Davis, and (B) by Ryan Phillips.

C. huegelii C. thinicola C. huegelii fruit set Place names Perth

Kwinana

Busselton Yallingup

Northcliffe

Broke Inlet Albany

−350 50 −34 100 km −33 −32 −31

114 115 116 117 118 119 120

Figure 2. Geographical range of Caladenia huegelii and C. thinicola based on records in the Western Australian Herbarium, the C. huegelii recovery plan and personal observations. Marked in red are sites where fruit set for C. huegelii was recorded. tions, C. huegelii has been listed as endangered under populations may be limited through interspecific the Australian federal EPBC Act. interactions, particularly in a now heavily frag- Although habitat clearance has been the major mented landscape. Investigations into the mycorrhi- driver of species decline in C. huegelii, surviving zal ecology of C. huegelii have shown that, although it 514 relies on a single species of mycorrhizal fungus for MATERIALS AND METHODS germination and annual growth (Swarts et al., 2010), POLLINATOR SURVEYS AND OBSERVATIONS field germination rates are comparable to co-occurring common Caladenia spp. (Swarts et al., 2010) and the Observations of pollinator behaviour were under- fungus is geographically widespread (Davis et al., taken using the pollinator baiting method, in which 2015). Alternatively, relatively little is known about picked orchid flowers are moved to a new part the pollination biology of the species. Like most of the landscape to renew the pollinator response members of Caladenia subgenus Calonema (Benth.) (Stoutamire, 1974; Peakall, 1990). Bait flowers were Hopper & A.P.Br., pollinator exclusion experiments picked at the base of the stem and kept fresh in vials have shown that C. huegelii is reliant on a vector for of water for up to 2 weeks in a portable refrigerator pollination and is self-compatible (Retter, 2009). at c. 4 °C. Although observations in other Caladenia However, a single season survey by See (2006) indi- spp. suggest there is no detectable decrease in polli- cated that the availability of pollinators could be nator attraction over this period, most bait flowers limiting reproductive success of this species. Prelimi- were used in experiments within 4 days of being nary observations revealed that on rare occasions picked. In all baiting trials, two flowers of C. huegelii C. huegelii attracts a large form of the thynnine were used. For C. huegelii, flowers were sourced from wasp Macrothynnus insignis with behaviour indica- populations represented by the voucher specimens tive of sexual deception, in which the wasp PERTH07501080 and PERTH07439938 in the approaches the flower with the sustained zig-zag Western Australian Herbarium (locations withheld as flight characteristic of its behaviour when tracking a a requirement of permits to study rare flora). Polli- sex pheromone plume (e.g. Peakall, 1990; Phillips nator observations were undertaken in sunny et al., 2013). However, this wasp is too large to be a weather when ambient temperatures exceeded 18 °C. pollinator and, even when locally common, only rarely We recorded if the responding pollinator exhibited exhibits attraction to the flower. A small form of the zig-zag flight associated with tracking a sex M. insignis pollinates the putative sister species, pheromone plume, if they attempted copulation with Caladenia thinicola Hopper & A.P.Br. (Phillips et al., the flower and if they came into contact with 2009b), raising the possibility that this wasp or a the column (see Phillips et al., 2013). Where possible, close relative could also be involved in the pollination before leaving the flower, floral visitors were of C. huegelii. The morphological study of Brown netted for identification and subsequent genetic (1995) did not reveal any morphological difference analysis. between the small and large forms of M. insignis Due to the rapid attraction of pollinators to experi- s.l. other than a disjunction in size (Fig. 1; total mentally presented bait flowers, this method can also length of small form: mean = 19.6 ± 0.5 mm (s.e.); be used as an effective tool to survey the abundance range = 15.2–23.9 mm (n = 17); total length of large and distribution of sexually deceived pollinators form: mean = 31.2 ± 0.5 mm; range = 27.2–34.3 mm (Phillips et al., 2014a). Pollinator surveys were under- (n = 15); R.D. Phillips, unpublished data), but the taken at 92 sites across the full geographical range of apparent differences in response to the orchids sug- C. huegelii. Sites were chosen based on soil type gests that they may represent different taxa. and broad vegetation community matching that of The overall objective of this study was to determine C. huegelii populations (Fig. 3; see Department of if a specialized pollination system could limit persis- Environment and Conservation, 2008 for more precise tence of C. huegelii. Specifically, we sought to answer details of habitat). Due to the extensive habitat clear- the following questions: ing in this region, our survey was able to cover most large bushland remnants on publically accessible (1) Which taxon is responsible for the pollination of land. Surveys were undertaken progressively through C. huegelii? September and October of 2008, 2011 and 2012. (2) What is the current abundance of the pollinator? Baiting was conducted in a 20m×40m rectangle (3) What is the geographical range of the pollinator with bait positions at each corner of the rectangle and relative to the orchid? mid-way along its two long sides, giving a total of six (4) Is C. huegelii characterized by low reproductive baiting positions at each site. Each position was success compared to related species? baited for 2 min. It has previously been shown for other orchids pollinated by sexual deception of thyn- To help resolve the of the putative polli- nine wasps that this survey design yields 90% con- nators, we take advantage of DNA barcoding, which gruence in presence/absence of the pollinator across has proved to be highly informative for detecting years (Phillips et al., 2014a). Nonetheless, this design morphologically cryptic species of thynnine wasps was repeated across 2 years at 22 sites and three (Griffiths et al., 2011; Menz et al., 2015b). years at 17 sites. 515

Baiting with C. huegelii Baiting with other species M. insignis present

−350 50 −34 100 km −33 −32 −31

114 115 116 117 118 119 120

Figure 3. Sites where the small form of Macrothynnus insignis has been recorded and sites where surveys for it were undertaken using orchids as bait flowers.

DISTRIBUTION OF THE POLLINATOR – MUSEUM PERTH08604347; PERTH08604460, RDP0276, RECORDS AND BAITING WITH OTHER ORCHIDS RDP0299). At multiple sites, representatives of the In light of evidence in 2011 that the small form of visiting wasps were collected for identification and M. insignis is a pollinator of C. huegelii (see results), DNA barcoding. we took two complementary approaches to resolve the distribution of this taxon beyond the range of C. hue- gelli. One of us (GRB) searched the collections of the DNA BARCODING – IDENTIFICATION AND Western Australian Museum, the Western Australian DISTRIBUTION OF THE POLLINATOR Department of Agriculture and the Australian Due to recent discoveries of morphologically cryptic National Insect Collection for records of this wasp. species in some genera of thynnine wasps (Griffiths Secondly, following the same survey design as outlined et al., 2011; Menz et al., 2015b), genetic analysis was above, baiting was undertaken with orchids known to undertaken to help resolve species boundaries within attract M. insignis, either as a pollinator or a ‘minor M. insignis s.l. Specimens of M. insignis collected responder’, defined as wasps that are sexually from all orchid species used in the baiting studies attracted to the flower, but exhibit inappropriate size were used for genetic analysis (Supplementary or behaviour to achieve pollination (sensu Bower, Table S1). These specimens extended along the coast 1996). The small form of M. insignis pollinates from Kwinana in the north to Albany in the south- C. thinicola (Phillips et al., 2009b), but it is also occa- east, a distance of c. 520 km (Fig. 4). In addition, sionally attracted to C. attingens Hopper & A.P.Br. specimens were included that were collected by oppor- subsp. attingens and C. granitora Hopper & A.P.Br., tunistic sweep-netting of patrolling males or catching but has not been observed to actually alight on the mating pairs while feeding on nectar plants. In total, flower in either of these species (Phillips et al., 2009b) 20 and 19 individuals, respectively, of the small and (RD Phillips et al., unpublished data). We undertook large forms of M. insignis s.l. were sequenced. baiting with C. thinicola (32 sites), C. attingens subsp. DNA barcoding using the mitochondrial DNA CO1 attingens (60 sites) and C. granitora (five sites) within locus was used to investigate species boundaries in the natural distribution of these orchids (Fig. 3; M. insignis s.l. This locus has previously been useful voucher populations: C. attingens subsp. attingens: for resolving species-level differences in thynnine PERTH08604347, RDP0229, RDP0281; RDP0286; wasps (Griffiths et al., 2011; Menz et al., 2015b). DNA C. granitora: PERTH005536294; C. thinicola: extraction followed the methods of Griffiths et al. 516

C. huegelii M. insignisscp clade M. insignissouth clade

−350 50 −34 100 km −33 −32 −31

114 115 116 117 118 119 120

Figure 4. Records of the two clades of the small form of Macrothynnus insignis relative to the geographical range of Caladenia huegelii. Distributional data for C. huegelii come from the Western Australian Herbarium and the C. huegelii recovery plan.

(2011). Initially, the primer set of Folmer et al. (1994), were analysed using maximum likelihood and Bayes- Hebert et al. (2004) and Hajibabaei et al. (2006) was ian analyses. The maximum likelihood analysis was used for DNA amplification. However, this set of run in PHYML (Guindon & Gascuel, 2003) via a primers failed to readily amplify DNA of M. insignis plug-in in Geneious v7.1.4 (Kearse et al., 2012), with s.l. Following failed attempts to develop primers for 1000 bootstrap replicates using the GTR + G model of this locus, we instead used the Simon et al. (1994) nucleotide substitution. The Bayesian analysis was primer set. These primers amplify a region of CO1 run in Mr Bayes 3.2.2 (Huelsenbeck & Ronquist, immediately adjacent to the region amplified by the 2001) through the CIPRES portal (Miller, Pfeiffer & Folmer primers, in the 3′ direction. This set of Schwartz, 2011). Two parallel runs of the analysis primers successfully amplified DNA for all individuals were run for 2 million generations (first 5% discarded of M. insignis, except those that responded to C. hue- as burn in, Markov chain sampled every 1000 gen- gelii. For these individuals we developed a final set of erations with four chains), using the GTR + G model. primers (MthynCO1B.1F GATTYTTTGGWCAYCCA- However, the analysis ceased after c. 860 000 genera- GAAG; MthynCO1B.1R AATTTTTTAAAGTAT- tions as the standard deviation of the split frequen- GATGGT), which successfully amplified these cies reached <0.01. We quantified the maximum level specimens for the Folmer et al. (1994) region. To of sequence divergence within putative species using confirm that the CO1 region used by Simon et al. Geneious. To confirm that the Simon et al. (1994) (1994) yields comparable results to the Folmer et al. region of CO1 yielded comparable results to the (1994) region, we sequenced between four and 28 Folmer et al. (1994) CO1 region previously used for individuals of several Zaspilothynnus spp., using the barcoding studies of thynnine wasps, we calculated same individuals as Menz et al. (2015b). Zaspilothyn- the correlation between the level of sequence diver- nus was chosen as the genus for comparison as phy- gence within species for both regions of the CO1 locus logenetic evidence provides strong support that using JMP v9.0.0 (SAS Institute Inc., 2010). Macrothynnus is nested within Zaspilothynnus (RD Phillips et al., unpublished data). REPRODUCTIVE SUCCESS A multiple sequence alignment was performed in Hand pollination experiments have shown that fruit- Geneious v7.1.4 (Kearse et al., 2012). Sequence data ing of C. huegelii is pollen- rather than resource- 517 limited, with no difference in the proportion of seeds DISTRIBUTION OF THE POLLINATOR BASED ON containing embryos between self-pollination and out- MUSEUM RECORDS crossing treatments (Retter, 2009). Although the Searches of museum collections revealed that prior to number of seeds produced by C. huegelii has not been this study the small form of M. insignis was known quantified, the closely related C. arenicola Hopper & from only three specimens, collected from Kwinana, A.P.Br. produces c. 30 000 seeds per (Batty, Yallingup and Northcliffe (Fig. 3). Of these locations, 2001). Fruit set was assessed at a total of ten popula- Kwinana is broadly within the natural range of tions in 2008, 2011 and 2012. Data collection focused C. huegelii and the other locations are within the on urban remnants in the south-eastern suburbs of range of C. thinicola. In addition to wasps collected Perth, Western Australia, the core range of C. huegelii within the core range of C. huegelii and C. thinicola, (Fig. 2). Fruit set was recorded in these populations in baiting with orchids yielded specimens of the small 2008 and 2012. In 2011, fruit set was recorded in some form of M. insignis from Albany (C. granitora), North- of these suburban sites plus three sites from the cliffe (C. attingens subsp. attingens) and Broke Inlet southern part of the geographical range of C. huegelii. (C. thinicola) (Fig. 4). The ability to collect data was limited in some sites by low flowering in some years and permission to access study sites. Percentage fruit set was calculated as the DNA BARCODING – IDENTIFICATION AND number of fully formed fruits divided by the number of DISTRIBUTION OF THE POLLINATOR flowers in the population, though C. huegelii is primar- Comparison with the data of Menz et al. (2015b) dem- ily single-flowered (Hopper & Brown, 2001). For each onstrated that sequence divergence within species is site, the percentage fruit set was averaged across years strongly correlated for the two regions of the CO1 to give a mean for that site. The average percentage sequence locus (R = 0.97), thereby confirming the fruit set across populations of C. huegelii was com- utility of the Simon et al. (1994) region for thynnine pared with data available in the literature for other wasps. Phylogenetic analysis resolved three strongly members of Caladenia subgenus Calonema (Peakall & supported clades within M. insignis s.l. (Fig. 5). The Beattie, 1996; Phillips et al., 2009b, 2013). As a large form of M. insignis was sister to the two clades measure of the actual reproductive output per popula- of the small form of M. insignis. One of these clades tion, the number of fruits formed in each year was used contained only wasps from the to calculate a mean for each population and an overall (SCP clade), including all three sequenced pollinators mean across populations. of C. huegelii. The other clade contained samples ranging from Yallingup to Albany (South clade), RESULTS including wasps attracted to C. thinicola, C. granitora and C. attingens subsp. attingens. Maximum POLLINATOR SURVEYS AND OBSERVATIONS sequence divergence within each of these clades was Pollinator observations across the entire geographical low (0% in the SCP clade; 1.2% in the South clade), range of C. huegelii revealed that only the small form but comparable to that seen in some other species of M. insignis was both attracted to C. huegelii and of thynnine wasp (Table 1). The minimum level of exhibited the necessary behaviour for pollen removal sequence divergence between the large form of and deposition. During the present study, there were M. insignis and the other two clades was c. 10%, no instances of the large form of M. insignis being which is comparable with levels of interspecific attracted to C. huegelii. Despite an extensive multi- variation in other genera of thynnine wasps. The year survey, only six M. insignis individuals were minimum sequence divergence between the SCP attracted to C. huegelii, none of which carried pollinia. clade and the South clade was 4.7%, which is less All of the responding wasps displayed the character- than that seen within the widespread Z. nigripes, but istic zig-zag flight associated with tracking a sex greater than in other thynnine species. pheromone plume. Three out of the six responding wasps crawled into the position needed for pollination, with at least two attempting copulation with the REPRODUCTIVE SUCCESS flower. Pollinators were observed at only four of the 92 Averaged across all populations, fruit set for C. hue- sites surveyed (4.3%), with only one of these sites gelii was 3.8 ± 1.1 (S.E.)% (Table 2), which is lower supporting a known population of C. huegelii. At each than that observed in other sexually deceptive of these sites they were only detected in 1 of the 2 years members of Caladenia subgenus Calonema (range for in which they were surveyed. By comparison, when other species: 9.9–20.5%; Table 3). Out of the ten baiting with C. thinicola outside the range of C. hue- populations, fruit set was observed in seven, with gelii, the small form of M. insignis was recorded at 13 33–60% of populations failing to set fruit in any given of the 32 sites surveyed (40.6%). year. Of the seven populations in which fruit set data 518

Figure 5. Phylogenetic analysis of the mtDNA CO1 locus (‘Simon’ region) for Macrothynnus insignis s.l. and related species in the genus Zaspilothynnus. One of the clades of M. insignis is the pollinator of Caladenia huegelii. SCP clade: wasps collected from the Swan Coastal Plain; South clade: wasps collected from the southern part of the range of M. insignis. Numbers at nodes above the line are bootstrap values from a maximum likelihood analysis, numbers below are posterior probabilities from a Bayesian analysis. Numbers in parentheses indicate the number of identical sequences. was recorded in multiple years, five populations expe- deceived pollinator, the small form of M. insignis, rienced at least one year with no fruit set. Overall, despite the presence of diverse thynnine wasp com- there was on average 1.0 ± 0.3 fruits produced per munities in these habitats (Menz et al., 2015a). This population per year. apparently high specificity matches other sexually deceptive orchids, which are typically pollinated by one or few insect species (Phillips et al., 2009b; DISCUSSION Gaskett, 2011). Although males of the large form of POLLINATOR SPECIES OF C. huegelii M. insignis were observed patrolling for females Baiting at over 90 sites across the geographical range at >15 survey sites, none of them was attracted to the of C. huegelii yielded a single taxon of sexually bait flowers, confirming that this taxon is not a pol- 519

Table 1. Percentage sequence divergence in clades of Macrothynnus insignis s.l. and closely related members of the genus Zaspilothynnus. The Folmer et al. (1994) region is that typically used for DNA barcoding studies. The Simon et al. (1994) region is immediately adjacent to the Folmer et al. (1994) region but in the 3′ direction. ‘Small’ and ‘large’ refers to two morphologically distinct forms of M. insignis. X: data could not be obtained due to consistent failure to amplify DNA. N = number of wasp individuals sequenced

Taxon % divergence ‘Folmer’ CO1 % divergence ‘Simon’ CO1

Macrothynnus insignis – large X 0.5 (N = 19) Macrothynnus insignis – small – SCP clade X 0 (N =4) Macrothynnus insignis – small – South clade X 1.2 (N = 16) Macrothynnus insignis – small – both clades X 4.7 (N = 20) Zaspilothynnus dilatatus 0.6 (N = 4) 0.5 (N =5) Zaspilothynnus gilesi (red-legged) 2.3 (N = 7) 1.5 (N =7) Zaspilothynnus gilesi (black-legged) 0.7 (N = 12) 0.6 (N = 10) Zaspilothynnus nigripes 5.8 (N = 28) 5.4 (N = 28) Zaspilothynnus sp. A 1.5 (N = 6) 0.5 (N =4) Zaspilothynnus sp. B 2.6 (N = 10) 1.5 (N = 10) Zaspilothynnus rugicollis 0(N =7) 0(N =5) Zaspilothynnus seductor 1.1 (N = 11) 0.7 (N =9)

Table 2. Percentage fruit set in populations of Caladenia huegelii. Numbers in parentheses are the number of flowering plants in that year. IF = insufficient flowering for reproductive success to be scored. (–) population was not surveyed

2008 2011 2012 Average

Wandi 0 (7) IF IF 0 Fraser S 0 (12) 0 (9) 0 (14) 0 Dennis B 14.28 (7) IF 0 (5) 7.14 Ken Hurst L 0 (18) 7.1 (28) 2.94 (34) 3.35 Ken Hurst S 0 (10) – 7.69 (13) 3.85 Caladenia 9.09 (11) – 3.37 (27) 6.4 Fraser L 11.76 (33) – 4.28 (70) 8.02 Shirley Balla – 0 (8) – 0 Kooljenerrup – 9.1 (22) – 9.10 Ruabon NR – 0 (4) IF 0

linator of C. huegelii. Furthermore, the degree of CO1 related members of Zaspilothynnus. Although levels divergence between the small and large forms of of divergence within the clades are low across mod- M. insignis (10%) falls well outside typical intraspe- erate distances (SCP = 0% over 90 km; South 1.2% cific variation (most species <2%; Table 1), leading us over 300 km), the 4.7% sequence divergence occurs to conclude that these sympatric forms represent dif- over a distance of only 50 km between clades. This ferent species. higher than expected genetic divergence raises the Baiting for pollinators with other sexually decep- possibility that a level of reproductive isolation may tive Caladenia spp. revealed that the small form of exist between these two clades and that M. insignis M. insignis occurs over a wide area beyond the range s.l. could represent a complex of up to three species. of C. huegelii. Indeed, in some of the southern parts of Biogeographic evidence supports the plausibility its distribution it appears to be locally common in of this situation, as the SCP is characterized by areas with large amounts of intact habitat. However, high levels of endemism in both plants and animals populations from the SCP clade and southwards along (Hopper & Gioia, 2004; Cogger, 2014), often with the coast from Yallingup (South clade) form two diver- considerable changes in community composition with gent clades. While the level of genetic divergence adjoining regions. between these clades (4.7%) is comparable with that Further resolution of the species boundaries within seen in the widespread Zaspilothynnus nigripes,it M. insignis s.l. could be important for understand- exceeds the level of divergence seen in other closely ing distributional limits in C. huegelii. Critically, 520

Table 3. Percentage fruit set in members of Caladenia subgenus Calonema, listed by pollination strategy. Species have only been included if data is available for three or more populations. Numbers are the percentage fruit set ± standard error with the number of populations measured in parentheses

Taxon Fruit set (%) Reference

Sexual deception Caladenia arenicola 9.9 ± 3.1 (4) (Phillips et al., 2009b) Caladenia attingens subsp. attingens 19.9 ± 7.3 (12) (Phillips et al., 2009b) Caladenia ferruguinea 14.4 ± 9.9 (3) (Phillips et al., 2009b) Caladenia huegelii 3.8 ± 1.1 (10) Present study Caladenia pectinata 12.3 ± 5.6 (8) (Phillips et al., 2013) Caladenia tentaculata 14.7 ± 8.6 (9) (Peakall & Beattie, 1996) Caladenia thinicola 20.5 ± 3.1 (5) (Phillips et al., 2009b) Food deception Caladenia longicauda subsp. eminens 42.6 ± 9.2 (4) (Phillips et al., 2009b) Caladenia serotina 50.2 ± 15.7 (3) (Phillips et al., 2009b) Caladenia speciosa 40.0 ± 6.9 (3) (Phillips et al., 2009b) Nectar producing Caladenia rigida 33.7 ± 3.1 (8) (Bickerton, 1997; Phillips et al., 2009b) Unresolved strategy Caladenia behrii 14.9 ± 1.7 (3) (Petit & Dickson, 2005) experimental opportunities provided by sexually the geographical range of C. huegelii, with only one deceptive pollination systems may enable tests of containing a known population of C. huegelii. Due to whether the SCP and South clades are likely to the low number of thynnine wasp specimens collected exhibit reproductive isolation via differences in sex from the SCP in previous decades, it is impossible to pheromones. Experimental presentation of calling judge whether the small form of M. insignis is natu- females is one plausible method for conducting choice rally rare in this region or has declined through tests with male thynnines (Alcock, 1981). Similarly, habitat clearance and fragmentation of the remaining choice tests, in which both pollinator clades are vegetation. However, at least some decline seems exposed to C. huegelii and C. thinicola (as per Bower, likely to be based on the extensive declines, and in 1996), could provide evidence of differences in sex some cases local extinction, documented for many pheromones should the two orchid species differ vertebrate species on the SCP following extensive in floral odour. Testing pollinator responses to the habitat clearance (Storr & Johnstone, 1988; How & floral odour of C. huegelii and C. thinicola with Dell, 2000; Davis, Gole & Roberts, 2013). gas chromatography-electroantennographic detection The current scarcity of the small form of M. insignis (GC-EAD) would provide a test of whether the two on the SCP suggests that an absence of pollinators clades of pollinator detect the same components of the from areas of suitable habitat for C. huegelii could be floral odour. Although this technique has been used as contributing to population decline in this orchid. This a starting point to determine sex pheromone systems prediction is supported by the low levels of reproduc- in other thynnine wasps, synthesis of compounds and tive success in C. huegelii compared to related sexu- experimental validation of sexual attraction in the ally deceptive orchids (Table 3), with 33–60% of field are needed to confirm that compounds detected populations failing to set fruit in any given year. in GC-EAD function in attraction (Schiestl et al., Comparison with the closely related C. thinicola pro- 2003; Peakall et al., 2010; Bohman et al., 2014). vides further support for this argument. Caladenia However, due to the large number of experimental thinicola is pollinated exclusively by the small form of animals needed, larger populations of M. insignis M. insignis, but is allopatric to C. huegelii. Within the from the SCP clade will need to be found before these range of C. thinicola, the small form of M. insignis approaches will become feasible in this system. was recorded at 40.6% of sites (versus 4% in C. hue- gelii), with much higher fruit set in C. thinicola than in C. huegelii (20.5% for C. thinicola vs. 3.8% for POLLINATOR RARITY AND LOW REPRODUCTIVE C. huegelii). SUCCESS IN C. huegelii Low levels of fruit set in C. huegelii, despite the Despite intensive survey effort, the small form of apparent absence of a thynnine wasp pollinator the M. insignis was only recorded from four sites within study populations, raises the question of how these 521 pollination events are achieved. One possibility is of the remaining populations, future management of that the small form of M. insignis does occur in these C. huegelii may require direct intervention through remnants at low abundance, but our surveys did not hand pollination, supplemental planting or the estab- detect them. Alternatively, pollination may occur in lishment of new populations. Although supplemen- the absence of a sexually deceived pollinator by occa- tary hand pollination has been shown to increase sional food deception events. Although to the human population sizes in Caladenia hastata (Nicholls) eye, C. huegelii has the dull red labellum character- Rupp (N. Reiter, pers. comm.), there are limited data istic of many sexually deceptive Caladenia (Phillips available on the costs to the individual of repeated et al., 2009b), it also has large, cream coloured peri- fruiting for this group of orchids (but see Coates & anth parts that may occasionally attract nectar- Duncan, 2009). Given the small and isolated nature of foraging insects. This possibility is supported by an many populations of C. huegelii (Department of observation of pollination of the sexually deceptive Environment and Conservation, 2008), which may Caladenia behrii Schltdl. by a food-foraging fly already exhibit some inbreeding depression, we would (Dickson & Petit, 2006). On rare occasions, thynnine strongly recommend consideration of the possible wasp individuals belonging to species that typically benefits of ‘genetic rescue’ by hand pollination do not alight on or attempt copulation with orchids between populations (Frankham, 2015). Species can achieve pollination (Peakall et al., 2010). such as C. huegelii, which are not autogamous, have Although such events may contribute to pollination of consistent habitat preferences (Department of C. huegelii, observations in other sexually deceptive Environment and Conservation, 2008) and have orchids suggest that they are likely to be rare and populations that were recently connected (Swarts therefore difficult to observe (Phillips et al., 2009b; et al., 2009), are highly likely to show increased Peakall et al., 2010; Phillips, 2010). Should C. hue- fitness following pollen transfer between populations, gelii have the capacity to attract pollinators by with minimal risk of outbreeding depression either of these mechanisms, it may confer some resil- (Frankham et al., 2011; Frankham, 2015). ience in the absence of its primary sexually deceived Bolstering or establishing populations by the plant- pollinator. ing of seedlings is also likely to be feasible for this Given the recent fragmentation of the landscape species. Techniques for the cultivation and planting of and the evidence for long life spans in at least Calad- Caladenia are well refined, with multiple successful enia subgenus Calonema (25+ years, Swarts et al., reintroductions being conducted in south-eastern Aus- 2009), several of these populations of C. huegelii may tralia (Wright et al., 2009). Critically, pollinator be below the threshold for persistence but have not surveys provide an important method for establishing yet suffered local extinction, a situation that has been candidate sites for the establishment of new popula- termed an ‘extinction debt’ (Tilman et al., 1994). This tions. Alternatively, although reintroduction of polli- phenomenon has been shown for both plant and pol- nating insects has been undertaken on other linator communities, but the response to fragmenta- continents (e.g. http://www.bumblebeereintroduction tion may be more rapid for pollinators (Sang et al., .org/), reintroductions of thynnine wasps are likely to 2010; Pauw & Bond, 2011; Pauw & Hawkins, 2011) be challenging. Thynnine wasps have a complex life due to their short generation times (Kuussaari et al., cycle in which females parasitize underground scarab 2009). Because of the consistently low fruit set and beetle larvae (Ridsdill-Smith, 1970) and adults are the rarity of the pollinating thynnine wasp, detailed reliant on either nectar or sugary secretions from empirical data on population growth rates are needed other insects as a food source (Brown & Phillips, to assess the potential for long-term population per- 2014). Given that the specific requirements are sistence and the predicted time until extinction (e.g. known for few, if any, thynnine species, it is unlikely Kuussaari et al., 2009). Although some groups of that pollinator reintroductions or management of Caladenia show extensive clonal reproduction via the habitat specifically for pollinators will be achieved in proximal production of daughter tubers (Dixon & the foreseeable future without a concerted research Tremblay, 2009), clonality in wild plants is believed to effort to understand the biology of these species be rare or absent in C. huegelii and related species, better. suggesting that persistence of populations is more likely to be related to longevity than clonality. IMPLICATIONS FOR FLORA THAT REQUIRE POLLINATION MANAGEMENT PROSPECTS FOR THE MANAGEMENT OF C. huegelii To manage cases successfully in which the pollinator Given the decline of the species as a whole (Brown is virtually absent, some critical information will et al., 1998; Department of Environment and be needed before implementing hand pollination. Conservation, 2008) and the low reproductive success For example, as some plant species show a cost of 522 reproduction (Primack & Hall, 1990; Obeso, 2002), Bower CC. 2001. Determination of the pollinators of sexually there will be an optimum interval between artificial deceptive orchids in the subtribes Drakaeinae and Calad- reproductive events. Similarly, there may be an eniinae, Technical Report, 1997–2000. Melbourne: Austral- optimal distance range of pollen transfer (within ian Orchid Foundation. populations, between nearby populations, between Breitkopf H, Schluter PM, Xu S, Schiestl FP, Cozzolino distant regions) that maximizes fitness or adaptive S, Scopece G. 2013. Pollinator shifts between Ophrys sphe- potential. Ideally, the success of crosses should be godes populations: might adaptation to different pollinators evaluated with germination experiments in the field, drive population divergence? Journal of Evolutionary as the effects of inbreeding may increase at later life Biology 26: 2197–2208. Brodmann J, Twele R, Francke W, Holzler G, Zhang QH, history stages (e.g. Smithson, 2005). Further, when Ayasse M. 2008. Orchids mimic green-leaf volatiles to undertaking crosses between plants it will be neces- attract prey-hunting wasps for pollination. Current Biology sary to avoid artificial selection (e.g. favouring larger 18: 740–744. flowered individuals). We are unaware of any cases Brown AP, Thomson-Dans C, Marchant N 1998. Western involving management to sustain populations in the Australia’s threatened flora. Perth: Department of Conser- absence of pollinators. However, given the declines vation and Land Management. recorded in many groups of pollinators (Potts et al., Brown GR. 1995. Revision of the Australian wasp genus 2010), we predict that these are going to become Macrothynnus Turner (Hymenoptera: Tiphiidae: Thynni- increasingly important topics for the conservation of nae). Records of the Western Australian Museum 17: 267– rare plant species (Menz et al., 2011). 275. Brown GR, Phillips RD. 2014. A review of the diet of flower wasps (Hymenoptera: Thynnidae; Thynninae). Northern ACKNOWLEDGEMENTS Territory Naturalist 25: 50–63. Funding was provided by Jandakot Airport Holdings Coates F, Duncan M. 2009. Demographic variation between and an Australian Research Council Linkage grant populations of Caladenia orientalis – a fire-managed threat- (LP110100408) to RP and KWD, a grant from the ened orchid. Australian Journal of 57: 326–339. Australian Orchid Foundation to RDP, and grants Cogger HG. 2014. Reptiles and amphibians of Australia. from the Holsworth Wildlife Research Endowment to Melbourne: CSIRO Publishing. Davis BJ, Phillips RD, Wright M, Linde CC, Dixon KW. RDP and MHMM. During parts of this project RDP 2015. Continent-wide distribution in mycorrhizal fungi: and MHMM were supported by Australian Postgradu- implications for the biogeography of specialised orchids. ate Awards. Thank you to Keith Smith, Alyssa Wein- Annals of Botany 116: 413–421. stein and Jeff Hardwick for assistance with fieldwork. Davis RA, Gole C, Roberts JD. 2013. Impacts of urbanisa- tion on the native avifauna of Perth, Western Australia. REFERENCES Urban Ecosystems 16: 427–452. Department of Environment and Conservation. 2008. Alcock J. 1981. Notes on the reproductive behaviour of some Grand spider orchid (Caladenia huegelii) draft interim Australian thynnine wasps (Hymenoptera, Tiphiidae). recovery plan 2008–2013. Interim recovery plan 272. Perth: Journal of the Kansas Entomological Society 54: 681– Department of Environment and Conservation. 693. Department of the Environment. 2015. EPBC Act list Anderson SH, Kelly D, Ladley JJ, Molloy S, Terry J. of threatened flora. Available at: http://www.environment 2011. Cascading effects of bird functional extinction reduce .gov.au/cgi-bin/sprat/public/publicthreatenedlist.pl?wanted pollination and plant density. Science 331: 1068–1071. =flora Batty AL. 2001. The role of symbiotic seed germination in the Dickson CR, Petit S. 2006. Effect of individual height and conservation of selected Western Australian terrestrial labellum colour on the pollination of Caladenia (syn. Arach- orchids. PhD Thesis, University of Western Australia. norchis) behrii (Orchidaceae) in the northern Adelaide Bickerton D. 1997. Pollinators, herbivores or loss of habitat: region, South Australia. Plant Systematics and Evolution why are there so few Caladenia rigida? Adelaide: University 262: 65–74. of Adelaide. Dixon K, Tremblay RL. 2009. Biology and natural history of Bohman B, Phillips RD, Menz MHM, Berntsson BW, Caladenia. Australian Journal of Botany 57: 247–258. Flematti GR, Barrow R, Dixon KW, Peakall R. 2014. Faast R, Farrington L, Facelli JM, Austin AD. 2009. Bees The discovery of alkyl- and hydroxyl pyrazines as insect sex and white spiders: unravelling the pollination syndrome of pheromones and orchid semiochemicals. New Phytologist Caladenia rigida (Orchidaceae). Australian Journal of 203: 939–952. Botany 57: 315–325. Bower CC. 1996. Demonstration of pollinator-mediated Fenster CB, Armbruster WS, Wilson P, Dudash MR, reproductive isolation in sexually deceptive species of Chilo- Thomson JD. 2004. Pollination syndromes and floral spe- glottis (Orchidaceae: Caladeniinae). Australian Journal of cialization. Annual Review of Ecology Evolution and Sys- Botany 44: 15–33. tematics 35: 375–403. 523

Folmer O, Black M, Hoeh W, Lutz R, Vrijenhoek R. 1994. Jones DL. 2006. A complete guide to the native orchids of DNA primers for amplification of mitochondrial cytochrome Australia. Sydney: Reed New Holland. c oxidase subunit 1 from diverse metazoan invertebrates. Kearse M, Moir R, Wilson A, Stones-Havas S, Cheung M, Molecular Marine Biology and Biotechnology 3: 294–299. Sturrock S, Buxton S, Cooper A, Markowitz S, Duran Frankham R. 2015. Genetic rescue of small inbred popula- C, Thierer T, Ashton B, Mentjies P, Drummond A. tions: meta-analysis reveals large and consistent benefits of 2012. Geneious basic: an integrated and extendable desktop gene flow. Molecular Ecology 24: 2610–2618. software platform for the organization and analysis of Frankham R, Ballou JD, Eldridge MDB, Lacy RC, Ralls sequence data. Bioinformatics (Oxford, England) 28: 1647– K, Dudash MR, Fenster CB. 2011. Predicting the prob- 1649. ability of outbreeding depression. Conservation Biology 25: Kuussaari M, Bommarco R, Heikkinen RK, Helm A, 465–475. Krauss J, Lindborg R, Ockinger E, Partel M, Pino J, Gaskett AC. 2011. Orchid pollination by sexual deception: Roda F, Stefanescu C, Teder T, Zobel M, pollinator perspectives. Biological Reviews 86: 33–75. Steffan-Dewenter I. 2009. Extinction debt: a challenge for Geerts S, Pauw A. 2012. The cost of being specialized: biodiversity conservation. Trends in Ecology & Evolution 24: pollinator limitation in the endangered geophyte Brunsvigia 564–571. litoralis (Amaryllidaceae) in the Cape Floristic Region of Menz MHM, Brown GR, Dixon KW, Phillips RD. 2015a. South Africa. South African Journal of Botany 78: 159–164. Absence of nectar resource partitioning in a community of Griffiths KE, Trueman JWH, Brown GR, Peakall R. 2011. parasitoid wasps. Journal of Insect Conservation. doi: Molecular genetic analysis and ecological evidence reveals 10.1007/s10841-015-9793-2 multiple cryptic species among thynnine wasp pollinators of Menz MHM, Phillips RD, Anthony JM, Bohman B, Dixon sexually deceptive orchids. Molecular Phylogenetics and KW, Peakall R. 2015b. Ecological and genetic evidence for Evolution 59: 195–205. cryptic ecotypes in a rare sexually deceptive orchid, Guindon S, Gascuel O. 2003. A simple, fast, and accurate Drakaea elastica. Botanical Journal of the Linnean Society algorithm to estimate large phylogenies by maximum like- 177: 124–140. lihood. Systematic Biology 52: 696–704. Menz MHM, Phillips RD, Winfree R, Kremen C, Aizen Hajibabaei M, Janzen DH, Burns JM, Hallwachs W, MA, Johnson SD, Dixon KW. 2011. Reconnecting plants Hebert PDN. 2006. DNA barcodes distinguish species of and pollinators: challenges in restoration of pollinator ser- tropical Lepidoptera. Proceedings of the National Academy vices. Trends in Plant Science 16: 4–12. of Sciences of the United States of America 103: 968–971. Miller MA, Pfeiffer W, Schwartz T. 2011. The CIPRES Hebert PDN, Penton EH, Burns JM, Janzen DH, science gateway: a community resource for phylogenetic Hallwachs W. 2004. Ten species in one: DNA barcoding analyses, Proceedings of the 2011 TeraGrid conference: reveals cryptic species in the Neotropical skipper butterfly extreme digital discovery. Salt Lake City: ACM. Astraptes fulgerator. Proceedings of the National Academy of Nilsson LA, Jonsson L, Ralison L, Randrianjohany E. Sciences of the United States of America 101: 14812–14817. 1987. Angrecoid orchids and hawkmoths in central Mada- Hoffman N, Brown AP. 2011. Orchids of south-west Aus- gascar – specialized pollination systems and generalist for- tralia. Perth: Scott Print. agers. Biotropica 19: 310–318. Hopper SD, Brown AP. 2001. Contributions to Western Obeso JR. 2002. The costs of reproduction in plants. New Australian orchidology: 2. New taxa and circumscriptions in Phytologist 155: 321–348. Caladenia. Nuytsia 14: 27–307. Paulus HF, Gack C. 1990. Pollination of Ophrys (Orchi- Hopper SD, Gioia P. 2004. The Southwest Australian Flo- daceae) in Cyprus. Plant Systematics and Evolution 169: ristic Region: evolution and conservation of a global hot spot 177–207. of biodiversity. Annual Review of Ecology Evolution and Pauw A, Bond WJ. 2011. Mutualisms matter: pollination Systematics 35: 623–650. rate limits the distribution of oil-secreting orchids. Oikos Houston TF, Lamont BB, Radford S, Errington SG. 1993. 120: 1531–1538. Apparent mutualism between Verticordia nitens and V. Pauw A, Hawkins JA. 2011. Reconstruction of historical aurea (Myrtaceae) and their oil-ingesting bee pollinators pollination rates reveals linked declines of pollinators and (Hymenoptera, Colletidae). Australian Journal of Botany plants. Oikos 120: 344–349. 41: 369–380. Peakall R. 1990. Responses of male Zaspilothynnus triloba- How RA, Dell J. 2000. Ground vertebrate fauna of Perth’s tus Turner wasps to females and the sexually deceptive vegetation remnants: the impacts of 170 years of urbaniza- orchid it pollinates. Functional Ecology 4: 159–167. tion. Pacific Conservation Biology 6: 198–217. Peakall R, Beattie AJ. 1996. Ecological and genetic conse- Huelsenbeck JP, Ronquist F. 2001. MRBAYES: Bayesian quences of sexual deception in the orchid Caladenia ten- inference of phylogenetic trees. Bioinformatics (Oxford, taculata. Evolution 50: 2207–2220. England) 17: 754–755. Peakall R, Ebert D, Poldy J, Barrow RA, Francke W, Johnson SD. 2010. The pollination niche and its role in the Bower CC, Schiestl FP. 2010. Pollinator specificity, floral diversification and maintenance of the southern African odour chemistry and the phylogeny of Australian sexually flora. Philosophical Transactions of the Royal Society B 365: deceptive Chiloglottis orchid: implications for pollinator- 499–516. driven speciation. New Phytologist 188: 437–450. 524

Petit S, Dickson CR. 2005. Grass-tree (Xanthorrhoea semi- syndromes: do floral traits predict effective pollinators? plana, Liliaceae) facilitation of the endangered pink-lipped Ecology Letters 17: 388–400. spider orchid (Caladenia syn. Arachnorchis behrii, Orchi- Sang A, Teder T, Helm A, Partel M. 2010. Indirect evidence daceae) varies in South Australia. Australian Journal of for an extinction debt of grassland butterflies half century Botany 53: 455–464. after habitat loss. Biological Conservation 143: 1405–1413. Phillips RD. 2010. Landscape, pollinator and mycorrhizal SAS Institute Inc. 2010. JMP v9.0.0. Cary: SAS Institute specificity and their contribution to rarity in Drakaea Inc. (hammer orchids). Perth: University of Western Australia. Schiestl FP, Ayasse M, Paulus HF, Lofstedt C, Hansson Phillips RD, Backhouse G, Brown AP, Hopper SD. BS, Ibarra F, Francke W. 1999. Orchid pollination by 2009a. Biogeography of Caladenia, with special reference to sexual swindle. Nature 399: 421–422. the Southwest Australian Floristic Region. Australian Schiestl FP, Peakall R, Mant JG, Ibarra F, Schulz C, Journal of Botany 57: 259–275. Franke S, Francke W. 2003. The chemistry of sexual Phillips RD, Bohman B, Anthony JM, Krauss SL, Dixon deception in an orchid-wasp pollination system. Science 302: KW, Peakall R. 2015. Mismatch in the distribution of floral 437–438. ecotypes and pollinators: insights into the evolution of sexu- Schiestl FP, Schluter PM. 2009. Floral isolation, specialized ally deceptive orchids. Journal of Evolutionary Biology 28: pollination, and pollinator behavior in orchids. Annual 601–612. Review of Entomology 54: 425–446. Phillips RD, Brown AP, Dixon KW, Hopper SD. 2011. See J. 2006. Investigations into the pollination of Caladenia Orchid biogeography and the factors associated with rarity huegelii and its sister species Caladenia arenicola by in a biodiversity hotspot: the Southwest Australian Floristic thynnid wasps (Tiphiidae: Thynninae). Perth: The Univer- Region. Journal of Biogeography 38: 487–501. sity of Western Australia. Phillips RD, Faast R, Bower CC, Brown GR, Peakall R. Simon C, Frati F, Beckenbach A, Crespi B, Liu H, Flook 2009b. Implications of pollination by food and sexual decep- P. 1994. Evolution, weighting, and phylogenetic utility of tion on pollinator specificity, fruit set, population genetics mitochondrial gene sequences and a compilation of con- and conservation of Caladenia. Australian Journal of served polymerase chain reaction primers. Annals of the Botany 57: 287–306. Entomological Society of America 87: 651–701. Phillips RD, Peakall R, Hutchinson HF, Linde CC, Xu T, Smithson A. 2005. Pollinator limitation and inbreeding Dixon KW, Hopper SD. 2014a. Specialized ecological depression in orchid species with and without nectar interactions and plant species rarity: the role of pollinators rewards. New Phytologist 169: 419–430. and mycorrhizal fungi across multiple spatial scales. Bio- Storr GM, Johnstone RE. 1988. Birds of the Swan Coastal logical Conservation 169: 285–295. Plain and adjacent seas and islands. Perth: Western Aus- Phillips RD, Scaccabarozzi D, Retter BA, Hayes C, tralian Museum. Brown GR, Dixon K, Peakall R. 2014b. Caught in the Stoutamire WP. 1974. Australian terrestrial orchids, act: pollination of sexually deceptive trap-flowers by fungus thynnid wasps, and pseudocopulation. American Orchid gnats in Pterostylis (Orchidaceae). Annals of Botany 113: Society Bulletin 43: 13–18. 629–641. Stoutamire WP. 1983. Wasp-pollinated species of Caladenia Phillips RD, Xu T, Hutchinson MF, Dixon KW, Peakall (Orchidaceae) in southwestern Australia. Australian R. 2013. Convergent specialisation – the sharing of polli- Journal of Botany 31: 383–394. nators by sympatric genera of sexually deceptive orchids. Swarts ND, Sinclair EA, Francis A, Dixon KW. 2010. Journal of Ecology 101: 826–835. Ecological specialization in mycorrhizal symbiosis leads to Potts SG, Biesmeijer JC, Kremen C, Neumann P, rarity in an endangered orchid. Molecular Ecology 19: 3226– Schweiger O, Kunin WE. 2010. Global pollinator declines: 3242. trends, impacts and drivers. Trends in Ecology & Evolution Swarts ND, Sinclair EA, Krauss SL, Dixon KW. 25: 345–353. 2009. Genetic diversity in fragmented populations of the Primack RB, Hall P. 1990. Costs of reproduction in the pink critically endangered spider orchid Caladenia huegelii: ladys slipper orchid–a4year experimental study. The implications for conservation. Conservation Genetics 10: American Naturalist 136: 638–656. 1199–1208. Retter BA. 2009. Factors affecting pollination success and Tilman D, May RM, Lehman CL, Nowak MA. 1994. offspring viability of orchids in a fragmented landscape. Habitat destruction and the extinction debt. Nature 371: Perth: The University of Western Australia. 65–66. Ridsdill-Smith TJ. 1970. The biology of Hemithynnus hya- Tremblay RL. 1992. Trends in the pollination ecology of the linatus (Hymenoptera: Tiphiidae), a parasite on scarabaeid Orchidaceae – evolution and systematics. Canadian Journal larvae. Journal of the Australian Entomological Society 9: of Botany 70: 642–650. 183–195. Wright M, Cross R, Dixon K, Huynh T, Lawrie A, Nesbitt Rosas-Guerrero V, Aguilar R, Marten-Rodriguez S, L, Pritchard A, Swarts N, Thomson R. 2009. Propaga- Ashworth L, Lopezaraiza-Mikel M, Bastida JM, tion and reintroduction of Caladenia. Australian Journal of Quesada M. 2014. A quantitative review of pollination Botany 57: 373–387. 525

SUPPORTING INFORMATION

Table S1. Voucher details for thynnine wasps used in DNA barcoding study of Macrothynnus insignis. See Figure 5 for designation of clades.