Molecular phylogenetic analyses of Drakaeinae: Diurideae (Orchidaceae) based on DNA sequences of the internal transcribed spacer region

One of the most intriguing groups of Australian terrestrial orchids, with colloquial names to match, are the ant, bird, wasp, hammer, flying duck, elbow and truffle orchids (Jones 2006). Traditionally, this group of orchids has been classified as subtribe Drakaeinae Schltr. and had been interpreted as comprising Arthrochilus F.Muell., Caleana R.Br., Chiloglottis R.Br., Drakaea Lindl., Paracaleana Blaxell and Spiculaea Lindl., which is how they were treated in the most recent taxonomic treatment of the group (Jones and Clements 2001) or in separate revision of genera Paracaleana (Hopper and Brown 2006) and Drakaea (Hopper and Brown 2007). Drakaeinae are unusual because, although small in number (~66 species), they occur across mesic Australia, with outliers in New Zealand and New Guinea.

The 14 species of elbow and truffle orchids, Arthrochilus, inhabit woodlands, forests, heathlands and savanna woodlands of eastern and northern Australia, including the Torres Strait Islands, with outliers in the Kimberley in north-western Western Australia and Arnhem Land in the Northern Territory, as well as occurring in the savanna woodlands of southern New Guinea. The monospecific Caleana, flying duck orchid, is locally common and widespread throughout south-eastern Australia, where it favours sandy habitats. The 27 species of Chiloglottis, the ant, bird and wasp orchids, are also locally common and widespread throughout the forests and woodlands of south-eastern Australia, with two species also occurring in New Zealand, one as far south as the subantarctic Campbell Island. Drakaea, the hammer orchids, with 10 described species, are confined to south-western Western Australia, where they grow in sandy open-area habitats in woodlands. The 13 species of Paracaleana, also know as duck orchids, are distributed across southern and eastern Australia, with an outlier in the north island of New Zealand, where they favour sandy open forest habitats. Spiculaea, also an elbow orchid, is a monospecific genus endemic to south-western Western Australia, mainly inhabiting shallow moss-covered soils on granite tors.

Like so many other Australian taxa, the group has been in taxonomic turmoil, with no resolution because of conflicting alternative generic concepts coupled with the discovery of an earlier valid generic name for Paracaleana. The primary cause of these conflicts is differing interpretations of the level of significance placed on various morphological traits found in this unusual group of orchids. Although the initial taxonomic changes were by Blaxell (1972), when he described the segregate genus Paracaleana, the catalyst for a renewed wave of taxonomic activity was the splitting of Chiloglottis and recognition of the segregate genus, Simpliglottis (Szlachetko 2001), which was soon followed by the further division of that genus and Arthrochilus (Jones and Clements 2005; Jones et al. 2006). The resulting nomenclatural instability has inhibited conservation efforts of some of the rare taxa.

Lindley (1840a) first treated Caleana, Chiloglottis, Drakaea and Spiculaea, along with other predominantly Australian genera, such as Thelymitra, Pterostylis, Lyperanthus, Microtis and Acianthus, in tribe Arethuseae Div. 2 Euarethuseae. Bentham and Hooker (1883) transferred these to tribe Diurideae. The German morphologist, Pfitzer (1887, 1889), proposed an alternative classification based on possession of the hinged labellum, which united Caleana and Spiculaea (including Arthrochilus) into the subtribe Pterostylidinae, a position followed by Dressler and Dodson (1960).

Table 1.  Alternative taxonomies of the Drakaeinae

Subtribe Drakaeinae was created by Schlechter (1911) to encompass Arthrochilus, Caleana, Chiloglottis, Drakaea and Spiculaea, which all possess winged columns, as distinct from Pterostylidinae (now considered part of the Cranichideae). Rogers (1921) did not accept the Drakaeinae and linked the group to Caladeniinae. Dressler’s interpretation of the group changed in his various treatments of the family (Dressler 1974, 1979, 1981, 1983, 1986; Dressler and Dodson 1960), until he reinstated Schlechter’s concept, albeit with the proviso that it may prove merely a subclade of the Caladeniinae (Dressler 1993). Developmental embryological studies of the tribe Diurideae confirmed the alliance of Drakaeinae as part of tribe Diurideae, where they all shared a ‘diurid’ embryo developmental pattern (Clements 1995, 1996, 1999), although this character state is not restricted to this tribe (Kores et al. 2001). Molecular data from rbcL (Kores et al. 1997; Cameron et al. 1999), matK and trnL–F plastid genes (Kores et al. 2000, 2001), psaB (Cameron 2004) and internal transcribed spacer (ITS) nuclear rDNA (Clements et al. 2002; Górniak et al. 2010) confirmed this relationship of subtribe Drakaeinae within the tribe Diurideae, although there are still conflicting data on its position and monophyly. Other authors, such as Szlachetko (1991), have proposed alternative interpretations of these taxa, on the basis of morphological studies of the column, and created the separate subtribes Chiloglottidinae and Caleaninae. Chase et al. (2003) recognised fewer subtribes, amalgamating subtribe Drakaeinae with Thelymitrinae, within tribe Diurideae, on the basis of a combined analysis of plastid atpB, rbcL, matK, psaB, trnL–F, nuclear 26S rDNA and nad1 intron and mitochondrial (non-coding) DNA.

The systematics of the group have also been the subject of change and controversy at the generic level. Initially, confusion existed about the limits of Drakaea, Spiculaea and Arthrochilus. Blaxell (1972) eventually clarified the situation, recognising each genus in its own right, and also splitting Caleana by creating Paracaleana for the two species with tuberculate labella. The concept of Paracaleana was initially accepted by many authors of popular and scientific literature (Erickson et al. 1973; Clements 1982; Hoffman and Brown 1984; Woolcock and Woolcock 1984; Green 1985; Weber and Bates 1986; Rye 1987; Jones 1988; Dixon et al. 1989). Paracaleana was rejected by Clements (1989) on the basis that none of the characters used was sufficiently different to maintain it as separate genus from Caleana. Despite this assessment, Paracaleana was accepted by most subsequent authors (Hoffman and Brown 1992, 2011; Backhouse and Jeanes 1995; Kores et al. 2001; Jones et al. 2002; Jeanes and Backhouse 2001; Hopper and Brown 2006; Brown et al. 2008). A preliminary analysis of the ITS nrDNA (Clements et al. 2002) failed to resolve this issue because of low taxon sampling. Additional controversy also surrounds this genus, with the recent discovery of an earlier published generic name, Sullivania, that was inadvertently, although legitimately, published for the rare apomitic form with fixed, vestigial labellum Caleana sullivanii (Jones and Clements 2005; Hopper and Brown 2006; St George 2010).

Additional names were added when Szlachetko (2001) created the genus Simpliglottis for a group of species related to Chiloglottis gunnii. Following detailed study of freshly collected material representative of all genera in the subtribe, further divisions were proposed (Jones et al. 2002). The following two segregate genera were split off from Arthrochilus: Phoringopsis, to account for those species with linear lanceolate leaves, and reverse position of the labellum callus, represented by P. byrnesi and P. dockrillii; and Thynninorchis for the semi-mycrohetrotrophic species, T. hunteriana, that inhabits the southern mountain regions of south-eastern Australia isolated from the remaining species. Jones and Clements (2005) later recognised and segregated a small group of species from Chiloglottis, as Myrmechila. This group of species can be distinguished by its winter to spring flowering, suberect to erect flowers, very short sepaline osmophores and suberect to erect labella.

Most of these proposed changes of traditional established genera have, in general, not been well received and adopted throughout the botanical community, both in Australia and overseas. A major objection to these proposed changes was their destabilising effect on traditional orchid nomenclature and classification (Hopper and Brown 2006, 2007; Hopper 2009). There were also objections about methods used, as almost all were exclusively morphologically based, without any cladistic or morphometric analyses of the data to support the proposed changes. Considering the high levels of current uncertainty surrounding the classification of taxa within Drakaeiinae and the lack of a published robust phylogeny of the group, one of the main aims of the present paper was to address this situation. A secondary aim was to test the monophyly of all traditional and recently proposed genera. We hypothesise that some of the characters used to define these genera were autapomorphies for subgroups within genera, rather than being synapomorphies for the entire genus. We discuss the clades and important taxonomic characters, test alterative classifications and propose a classification based on our results.

DNA sequences for the nuclear rDNA ITS were obtained from 55 species, including representatives of all traditional or recently recognised higher taxa within Drakaeinae (Jones et al. 2002, 2006), and outgroups from Calochilus, Cryptostylis, Epiblema and Thelymitra. Successful analysis, based only on ITS data, has previously allowed establishment of viable, acceptable taxonomic outcomes, from which the results of further multigene analyses have deviated little. For example, Cox et al. (1997) on the phylogeny of the slipper orchids, and recent publications on Brassicaceae (Warwick et al. 2010), provided evidence of the utility of results from ITS sequence phylogenetic analysis. Details of source material and provenance, and GenBank accession numbers used are listed in Table 2. Reference vouchers for collections used in the study are housed either at CANB or CHR. Taxa were chosen on the basis of previous broader-level studies by Cameron et al. (1999), Clements et al. 2002), Cameron (2004), Salazar et al. (2003, 2009) and Górniak et al. (2010).

Table 2.  List of specimens examined
Reference vouchers of the Australian collections are housed in CANBR and the New Zealand collections are housed in CHR. Abbreviations: ORG, Orchid Research Group; H, herbarium specimen; Can, Canberra; Ncc, New South Wales (NSW) Central Coast; Nct, NSW Central Tablelands; Nnc, NSW northern coast; Nnt, NSW Northern Tablelands; Nsc, NSW southern coast; Nst, NSW Southern Tablelands; Qco, Queensland (Qld) Cook; Qdd, Qld Darling Downs; Qle, Qld Leichhardt; Tas, Tasmania; Vgi, Victoria (Vic) Gippsland; Vwh, Vic western highlands; Wav, Western Australia (WA) Avon; Wcv, WA Carnavon; Wda, WA Darling; Wey, WA Eyre; AU, Australia; NZ, New Zealand

Genomic DNA was extracted from 10–100 mg of fresh or silica gel-dried leaf tissue, or from herbarium material, using the DNeasy Plant Mini Kit (Qiagen, Melbourne, Vic., Australia) either individually or in the 96-well plate format. The complete ITS region of the 18S–26S nrDNA was amplified by polymerase chain reaction (PCR), following methods outlined in Clements et al. (2002). The following primer pairs from Sun et al. (1994) were used: ITS4–ITS5, 17SE–26SE or 17SE–ITS4.

Contiguous sequences were assembled and edited using Sequencher v.3.0 (Gene Codes Corp., AnnArbor, MI, USA) and manually aligned in BioEdit sequence alignment editor v.4.8.6 (Hall 1999). Sequence alignments and Nexus formatted files are available in TreeBase (accession number 15661) and all sequences are lodged in GenBank (see Table 2). Any uncertain base positions, generally located close to priming sites, and highly variable regions with equivocal sequence homology, were excluded from phylogenetic analysis. Individual base positions were coded as unordered multistates and potentially informative insertions/deletions (indels) were manually coded as additional binary characters.

Bayesian analyses were performed using MrBayes version 3.1.2. (Ronquist and Huelsenbeck 2003). The GTR+I+gamma model was applied to the ITS alignment. Indel characters were included as a separate partition and a standard (morphology) discrete state model with a gamma-shape parameter was applied to this partition. The Markov chain Monte Carlo search was run for 20 million generations with trees sampled every 50 000 generations. MrBayes performed two simultaneous analyses starting from different random trees (Nruns = 2), each with six Markov chains. In total 25% of the trees were discarded and the run ended with an average standard deviation of split frequencies of <0.005. A Bayesian consensus phylogram with posterior probability values plotted was calculated in MrBayes. Maximum parsimony analyses were performed with the heuristic search option (excluding uninformative characters) in PAUP* 4.02 (Swofford 2003). A two-step search method for multiple islands was performed with 10 000 random replicates with tree bisection reconnection (TBR) on, and saving only one of the shortest trees per replicate. The saved trees were then swapped to completion, saving all shortest-length trees. Support for internal branches was evaluated by the heuristic bootstrap method with 10 000 bootstrap replicates, each with 10 random-addition sequences with TBR and MULPARS activated (Felsenstein 1985).

The aligned ITS dataset comprised 55 DNA sequences, with a matrix length of 762 nucleotide sites. There was a total of 396 shared polymorphisms. Six indels were scored. The parsimony analyses identified 143 equal-length trees of 1320 steps (consistency index (CI) = 0.60, rescaled consistency index (RCI) = 0.50).

The Bayesian analysis of the ITS region resolved a phylogenetic tree with moderate support for the monophyly of Drakaeinae (posterior probability, PP = 0.83), and Spiculaea cilata sister to the rest of the Drakineae. The parsimony analysis (not shown) was congruent with the Bayesian analysis, except that Spiculaea was not supported as part of the Drakineae, rather it was supported in a clade containing all the outgroups as sister to the Drakineae. A clade containing Thynninorchis and Arthrochilus (with Phoringopsis) (PP = 1.0, bootstrap values, BV = 80) was sister to a clade containing Caleana, Drakaea and Chiloglottis. Arthrochilus s.lat. was well supported, as were each of the internal clades, namely, Thynninorchis, (PP = 1.0, BV = 100) and a clade (PP = 1.0, BV = 97) comprising Arthrochilus (PP = 1.0, BV = 100) and Phoringopsis (Arthrochilus dockrillii, Fig. 1). Caleana s.lat. was sister to a clade containing Drakaea and Chiloglottis s.lat.

Fig. 1.  Internal transcribed spacer (ITS) Bayesian consensus tree, with branch lengths of Drakaeninae (Orchidaceae). Numbers above branches indicate the Bayesian posterior probability, followed by bootstrap values.

Caleana s.lat. was well supported (PP = 1.0, BV = 100) and showed support for monophyletic Caleana (PP = 0.97) and Paracaleana (PP = 98, BV = 69). Drakaea (PP = 1.0, BV = 100) was strongly supported but showed very little differentiation among species. Chiloglottis s.lat. (PP = 1.0, BV = 100) was also well supported, with strong support on short branches for Simpliglottis (PP = 1.00, BV = 99) and Myrmechila (PP = 1.00, BV = 100), whereas Chiloglottis s.s. was unresolved.

Although it is recognised that the present study was based on the use of a single marker region, ITS, nevertheless it has provided an important analysis of the group. Our results, which are based on analysis of 32 species, increase the representation of all genera within Drakaeinae compared with previous studies (Fig. 1). The results are highly congruent with the previously published nuclear (Clements et al. 2002; Górniak et al. 2010) and plastid gene phylogenies (Kores et al. 2001; Cameron 2004), which were based on small samples of species in each genus. Our Bayesian results confirmed the monophyly of Drakaeinae consistent with outcomes of other broader based studies of the Diurideae (Weston et al., in press).

There are areas of incongruence between the previously published nuclear tree of Clements et al. (2002) and our current results. In particular, Arthrochilus s.lat. (including Thynninorchis) is sister to Drakaea and Chiloglottis, and the remainder of the Drakaeinae (Spiculaea, Caleana and Paracaleana), the latter forming a clade. However, there is little bootstrap support for phylogenetic relationships in this part of the backbone of the phylogeny. In the previous study, where the overall structure and phylogeny of the tribe was the focus, and no particular subtribe was investigated in detail (Clements et al. 2002), only 10 species were used as representatives of Drakaeinae. In the present study, we have expanded the range of collections, including several newly described species and additional material from disparate locations of some critical species, such as Caleana major and C. (Paracaleana) minor. The inclusion of sequence data from Caleana (Paracaleana) lyonsii, a species from south-western Western Australia, proved decisive in the assessment of that genus. C. (Paracaleana) lyonsii with a multiflowered inflorescence and erect, narrow lanceolate leaf is sister to the rest of Paracalena. C. (Paracaleana) nigrita, with its single flower and reptant, broadly ovate to ovate lanceolate leaf, is embedded within and not isolated from that clade.

Spiculaea (Fig. 2A), as originally described (Lindley 1840b), is a monospecific genus, endemic to the inland south-west of Western Australia. Phylogenetically, it is sister to the rest of the Drakaeinae, but not well supported. Its inclusion in the Drakaeinae, therefore, is equivocal. Spiculaea is characterised by a single relatively broad, glabrous basal leaf, with a thin, wiry scape narrower at the base and fleshy, resupinate, yellowish-brown, sequentially developing flowers. The fleshy scape contains sufficient nutrients and water to enable flowers and capsules of pollinated flowers to develop to maturity, even when the basal stem dries during the extreme heat of late spring and early summer. Spiculaea is also characterised by several autapomorphies, including the prominent basal labellum stalk hinged to a very short column foot, a fleshy insectiform labellum callus, with strap-like apical appendage, and prominent, sickle-like column wings that are serrated along the inner margin.

Fig. 2.  Representative of genera within Drakaeinae (Orchidaceae). A. Spiculaea ciliata, The Humps, Western Australia (WA); B. Thynninorchis hunteriana, Mount Bollard, New South Wales (NSW); C. Arthrochilus prolixus, Bulahdelah, NSW; D. A. (Phoringopsis) dockrillii, Little Forks, Queensland (Qld); E. Caleana triens, Dale State Forest, WA (G. Brockman); F. C. major, Mount Jerrabomberra, NSW (Z. Groeneveld); G. Drakaea thynniphila, Walpole, WA (C. Busby); H. D. glyptodon, Bayonet Head, WA (R. Heberle); I. Chiloglottis diphylla, Bulahdelah, NSW; J. C. (Myrmechila) formicifera, Mill Creek, NSW; K. C. (Myrmechila) platyptera, Barrington Tops, NSW; L. C. (Simpliglottis) jeanesii, Toorongo,Victoria.

Historically, Arthrochilus has been linked or confused with Drakaea and Spiculaea (Reichenbach 1871; von Mueller 1889; Schlechter 1926), and comprises three morphologically distinct groups. These groups were segregated by Jones and Clements (2005) into the following three genera, based primarily on differences in plant habit, but also on floral morphology: the dimorphic Arthrochilus (10 spp.), the monomorphic Phoringopsis (3 spp.) and mycroheterotrophic Thynninorchis (2 spp.).

Arthrochilus (Fig. 2C) predominantly inhabits the Australian tropical and temperate coastal woodlands and heathlands, extending along the eastern coast as far south as Sydney. Plants occur in colonies as rosettes or as multi-flowered racemes. Flowers are non- resupinate, greenish with inconspicuous sepals and petals and a prominent intricately insectiform, delicately hinged labellum that dangles like a fishing lure, and column with two unequal pairs of projecting column wings.

Phoringopsis (Fig. 2D) is distributed disjunctly in northern tropical Australia and south-eastern Papua New Guinea and was segregated from Arthrochilus (Jones et al. 2002) on the basis of possession of one or two elongate, distichous basal leaves, inflorescence emerging with the leaves, smooth, non-papillate column foot and sepal bases, labellum attached basally (not via a peltate stalk), and the callus fungiform and ornamented with penicillate calli.

Thynninorchis (Fig. 2B) comprises two leafless species that inhabit the montane eucalypt forests of south-eastern Australia, including Tasmania. These are the only fully mycroheterotrophic species of Drakaeinae and Jones and Clements (2005) erected the genus on account of this plant habit, but also because of possession of a hinged, insectiform labellum ornamented with long, multi-layered, barbed, caudiform cells, with the main head deeply divided into two lobes, each ending in a swollen, knob-like structure that hangs at the end of a long, narrow column foot, and column with two pairs of similar-sized, projecting, column wings that are reminiscent of those present in Spiculaea.

Sister to Thynninorchis is the clade containing Arthrochilus (10 species and the segregate genus Phoringopsis (3 species)). Phoringopsis is distinguished from Arthrochilus by possession of one or two narrow, erect, linear, lanceolate leaves and similarity of leaf morphology in flowering and sterile plants, flowers with smooth tepal bases, labellum hinge attached in the centre of the column foot, labellum lacking a peltately attached supporting stalk and the strongly swollen, fungi-like callus ornamented with long, thin, penicillate glands. Phoringopsis has a scattered disjunct distribution across northern Australia and one species, P. lavarackiana, also inhabits sites in the Melaleuca-dominated savanna woodlands of southern New Guinea.

Arthrochilus inhabits coastal, sandy heathlands and woodlands or mountain woodlands along the eastern coast of Australia from Sydney to north Cape York and some Torres Strait islands, with an isolated disjunct species, A. latipes, in Arnhem Land in the Northern Territory. Both Arthrochilus and Phoringopsis are colony-forming species, found in small clusters or in great numbers in some areas. Flowering occurs with or without leaves in Arthrochilus, depending on the species involved. Morphologically, these taxa have much in common despite these noted differences, particularly in plant habit.

The older (Blaxell 1972) and more recent treatment of Phoringopis as Arthrochilus s.lat. (Szlachetko 2003) is possibly more appropriate than the recognition of these taxa as separate genera. Thynninorchis is distinctive, supported by a long branch with the nrDNA ITS data, but the possession of several unique vegetative and floral characters makes it difficult to identify synapomorphies shared by it and Arthrochilus and Phoringopsis. Clearly, these DNA data can be interpreted to recognise alternative classifications. Features and characteristics that facilitate ready recognition of and isolation of Thynninorchis from Arthrochilus are the mycroheterotophic leafless plant habit, presence of protocorm or protocorm-like structures and absence of true tubers, labellum hinged to the end of a long narrow column foot and dangled like a fishing lure, labellum lamina simple, unlobed, with a long, peltately attached basal stalk, a callus intricately insectiform and dominating the labellum lamina, ornamented with long, multi-layered, barbed caudiform cells, with a main head that is deeply divided into two lobes, each lobe ending in a swollen knob-like structure, along with being distributed in tall wet forests throughout mountainous regions of south-eastern Australia. On this basis, we support a classification that recognises both Arthrochilus and Thynninorchis (Fig. 1).

A clade containing a monophyletic Caleana (Fig. 2F) and a monophyletic Paracaleana (Fig. 2E) is strongly supported. Caleana major (Fig. 2F), the type of that genus, is sister to representatives of Paracaleana (and Sullivania); however, both of these groups are on shorter branches than is the combined Caleana–Paracaleana clade. Caleana, the unmistakable flying duck orchid, is a south-eastern Australian species with disjunct outliers in the Mount Lofty Ranges in South Australia and Carnarvon Gorge–Blackdown Tableland region in central Queensland. P. minor often occurs sympatrically and has a distribution similar to that of C. major, although also occurring on the North island of New Zealand (St George 1999). However, the greatest number of Paracaleana species (11 spp.) inhabit areas in the south-west of Western Australia (Hopper and Brown 2006; Brown et al. 2008; Hoffman and Brown 2011). All are colony-forming species.

Caleana (Fig. 2F) is distinct from Paracaleana (Fig. 2E) on morphological grounds. This is reflective of the pollination syndrome and associated development of morphological features, such as tuberculate labella surface, first identified by Blaxell (1972) when describing the genus (Hopper and Brown 2006). Possession of a glabrous labellum or the apparent lack of glands on the labellum of C. major is attributable to the pollination syndrome, where flowers are visited by males of the long-tailed sawfly, Lophyrotoma leachii (Kirby), that land cross-wise on the smooth labellum surface (Cady 1965; Bates 1989; Bower 2001). All other species are apparently pollinated by male thynnine wasps (Erione spp.: Tiphiidae) that are attracted to the flower by allomones and psuedocopulate with the insectiform labellum (Hopper and Brown 2006). Possession of a hinged labellum sensitive to touch is characteristic for all species in this major clade and this is the only group in Drakaeinae with this character. Column wings are also highly developed, extending the whole length of that organ, forming a cup-like structure that temporarily holds the insect in place during pollination. From the results of our study, we conclude that Paracaleana (and Sullivania) are synonymous with Caleana, as originally proposed by Clements (1989).

The three sampled species representative of the genus Drakaea (Fig. 2G, H) form a strongly supported clade. These unique hammer orchids of south-western Western Australia possess several synapomorphies and the genus is now recognised as containing at least 10 species (Hopper and Brown 2006). Their separation from all other members of Drakaeinae belies past confusion surrounding their recognition as a distinct genus. Recent research (Phillips et al. 2011) has suggested that most species inhabit highly specialised microhabitats in open areas of sandy ground and that each species is associated with a single specific mycorrhizal fungus. All species are reported as being pollinated by thynnine wasps (Zaspilothynnus spp.) during attempted pseudocopulation with the highly specialised insectiform labellums (Hopper 2009). Column wings are more or less vestigial in all species and the anther is very prominent. In addition, Drakaea is characterised by possession of a single, short, reinform, spongy, glabrous leaf, a thin, wiry, glabrous scape that gradually widens towards the apex, and a one-flowered inflorescence, with a labellum stalk basally hinged and able to pivot on the hinge, returning to its original position.

Chiloglottis s.lat. comprises the final clade. Our results suggested that although species attributed to both Myrmechila (represented by C. truncata, C. platyptera, C. trapeziformis and C. formicifera) (Fig. 2J, K) and Simpliglottis (represented, for example, by C. cornuta, C. valida, C. chlorantha) (Fig. 2L) are monophyletic (PP = 1.00, BV = 100), on short branches, the rest of Chiloglottis (Chiloglottis s.s.) (Fig. 2I) is unresolved within Chiloglottis s.lat. Analysing a combined matrix of both ITS and trnL–trnF cpDNA intergenic spacer sequences, Mant et al. (2002) found strong support for the monophyly of two groups, where Chiloglottis s.str. was sister to Myrmechila, and these combined where sister to Simpliglottis, contrary to what was found here. These conflicting results are likely to be the result of ITS being a faster-evolving locus than trnL–F in these taxa. The arrangement of taxa where the Simpliglottis clade is sister to the remainer of Chiloglottis is also reflected in possession of a simpler, less elaborate floral morphology found in species of Simpliglottis. The nature of these conflicting results and the minimal nature of the morphological differences among taxa, in particular Myrmechila and Chiloglottis s.s., and the difficulty in applying this taxonomic segregation in practice, suggest that the original decision by Szlachetko (2001) and follow-up decisions by Jones and Clements (2005) to split Chiloglottis s.lat. were unnecessary. We, therefore, argue that Robert Brown’s original concept of a single Chiloglottis, supported by most subsequent authors, is the best taxonomic interpretation of these data.

Chiloglottis (Fig. 2I–L) is distributed throughout south-eastern Australia, including Tasmania, as well as occurring in New Zealand where it reaches as far south as the Auckland Island, where C. cornuta was first discovered and described by Hooker (1844–1847). Chiloglottis is characterised by possession of the following: two subequal, opposed leaves; a one (rarely two) flowered, fleshy, ephemeral inflorescence arising from the centre of the two leaves; flowers suberect to horizontal; sepals with cylindrical, apical osmophores; a prominent rhomboid-trapeziform, caudate, stiffly, hinged labellum with a wide lamina and prominent columnar; and stalked calli, weakly or strongly insectiform; and column wings extending the length of the column.

Although only based on analysis of ITS nuclear rDNA sequences, our results provided sufficient basis for a reinterpretation of generic cicumscription within subtribe Drakaeinae. These results are supported by those generated using both nuclear and chloroplast genes (Weston et al., in press), where the position and status of Drakaeinae, relative to the remainder of Diurideae, is the subject of further consideration. Given the very high levels of support and degree of genetic divergence, coupled with possession of readily definable morphological synapomorphies for each major clade, the Drakaeinae require significant taxonomic changes to conform to a monophyletic interpretation of the tribe and its genera. A reclassification of all the species involved, including some reallocation of taxa to appropriate genera, is provided (Appendix 1).