Sequence characteristics

The general features of the DNA regions that were used are presented in Table 1. The attempt to amplify psaB from the new myco-heterotrophic orchid (Yunorchis pingbianensis) failed. It is possible that this region is lacking. A total of 1047 bp of rbcL and 1458 bp of matK sequences (GenBank accession KM526763 and KM526774) were assessed, and none obvious deletion/addition loci were observed among the sequences.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Parsimony statistics from phylogenetic analyses of the various datasets.

https://doi.org/10.1371/journal.pone.0123382.t001

In the subfamily-wide matrix, 104 DNA sequences of rbcL and matK, 87 of psaB, and 70 of Xdh were obtained. The most variable dataset was Xdh with 47.25% potentially informative sites (excluding uninformative characters). Parsimony analyses resulted in a parsimonious tree of 2150 steps (TL) with a consistency index (CI) = 0.4647 and a retention index (RI) = 0.5843. The second variable dataset was matK (31.92% potentially informative sites), which had the lowest CI = 0.3560 and RI = 0.5182. The third most variable dataset was rbcL with 16.05% potentially informative sites. The least variable dataset was psaB with 11.75% potentially informative sites, which had the highest CI = 0.5835 and RI = 0.6291. In the combined three chloroplast markers (rbcL, matK, psaB) matrix, which comprised 4646 aligned nucleotides with 931 potentially informative sites (20.04%), TL = 5140, CI = 0.4811 and RI = 0.5309. In the combined chloroplast and nuclear Xdh markers (963 bp) matrix, which comprised 5609 aligned nucleotides with 1386 potentially informative sites (24.71%), TL = 7333, CI = 0.4739 and RI = 0.5438.

In the tribe-wide matrix, 29 DNA sequences of rbcL, 33 of matK, and 32 of ITS were obtained. The most variable dataset was ITS, with 30.46% potentially informative sites. Parsimony analyses resulted in a parsimonious tree of 641 steps with the lowest CI = 0.7285 and RI = 0.7573. The second most variable dataset was matK with 12.27% potentially informative sites, which had the highest CI = 0.8390 and RI = 0.9378. The least variable dataset was rbcL with 8.72% potentially informative sites. In the combined two chloroplast markers (rbcL, matK) matrix, which comprised 3000 aligned nucleotides with 322 potentially informative sites (10.73%), TL = 913, CI = 0.8160 and RI = 0.8918. In the combined two chloroplast and nuclear ITS markers (673 bp) matrix, which comprised 3673 aligned nucleotides with 527 potentially informative sites (14.35%), TL = 1593, CI = 0.7659 and RI = 0.8362 (Table 1).

Phylogeny of the subfamily Epidendroideae

The results from the single and combined analyses were presented by the Bayesian trees. The values of the Bayesian Posterior probabilities (PP) and Bootstrap percentage (BS_ML and BS_MP) are indicated at each node. The branches that collapse in the strict consensus tree are indicated by “-”. In the nuclear Xdh analysis (S1 Fig), the results with added new molecular data are not completely in accordance with the results of Górniak et al. [11], especially the topology. The tribe Calypsoeae, including the new molecular taxa (Dactylostalix and Danxiaorchis), is supported as monophyletic with weak support (PP = 0.36, BS_ML = 30, and BS_MP = 60) and sister to Epidendreae (excluding Coelia).

No strongly supported, incongruent patterns of relationship were detected in the individual chloroplast markers analyses, so a combined analysis of all of the chloroplast data (rbcL, matK, and psaB) was performed. The trees of this analysis are shown in S2 Fig. The paraphyletic Calypsoeae (two supported groups) were observed, and the results were consistent with previous molecular studies. The nuclear and chloroplast trees that were generated by the Bayesian inference analyses (BI) were congruent with those that were retrieved by the ML and MP analyses except for the poorly supported nodes (PP < 0.90).

Except for Calypsoeae and/or Wullschlaegelia, there were no strongly supported incongruent results in the topology of the Xdh trees and the combined chloroplast trees. Although the Calypsoeae is paraphyletic in chloroplast trees, the weak support and the presence of winter leaves in some members (except for the Wullschlaegelia) of the two supported groups indicate a close relationship. Therefore, the chloroplast and nuclear DNA were combined into a single dataset for the phylogenetic analyses (Fig 1). Our findings are consistent with the overall topology of the trees that are produced by the BI, ML, and MP methods, except for a few collapsed nodes (such as Dendrobieae and Malaxideae). The Posterior Probability values (PP) were often higher than the Bootstrap values (BS). The tree that was recovered in the combined analysis is more similar to that based on the Xdh than the combined chloroplast.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Phylogenetic relationships of the subfamily Epidendroideae based on the chloroplast DNA rbcL, matK, and psaB and the nuclear DNA Xdh.

The numbers near the nodes are Bayesian posterior probabilities and bootstrap percentages (PP left, BS_ML middle, and BS_MP right). “*” indicates that the node is 100% supported. “-” indicates that the node is incongruent between the topology of the MP/ML trees and the Bayesian tree.

https://doi.org/10.1371/journal.pone.0123382.g001

In the Lower or “basal” Epidendroideae, highly supported clades include a monophyletic clade (PP = 1.00 and BS = 100) consisting of Neottieae sister to all of the remaining Epidendroideae, followed by Sobralieae (PP = 1.00 and BS > 50) and the other four tribes with a relatively weak bootstrap and an unstable topology. Higher Epidendroideae taxa include most of the clades that are successively (PP > 0.90 and BS > 50) sister to the grade of clades representing previously recognized tribes. Arethuseae are monophyletic as sister to all of the other Higher Epidendroideae with strong support (PP = 1.00 and BS > 90), followed by Cymbidieae (PP = 0.90 and BS = 65), and then Malaxideae plus Dendrobieae with a relatively weak bootstrap and an unstable topology. The next most supported clade (PP = 0.99) is divided into three tribes: Collabieae, Podochileae and Vandeae (excluding Agrostophyllinae) and strongly support (PP > 0.90 and BS >75) were observed between them. The last clade is divided into four groups, comprising the first group Epidendreae (excluding Coelia), sister to the second group Calypsoeae (excluding Wullschlaegelia) with a stable topology, the third group Coelia (PP = 0.95) with an unstable topology, and the fourth group Agrostophyllinae with strong support (PP = 1.00 and BS >75). The new holomycotrophic orchid, as a sister to Calypso, is a member of the tribe Calypsoeae.

Phylogeny of Calypsoeae

Based on the ITS sequence data, the relationships in the strict consensus tree for most of the genera are resolved and well supported (PP > 0.95 and BS > 75) with only one genus (Oreorchis) receiving a paraphyly (S3 Fig). The basal group comprises Calypso, Changnienia, Dactylostalix, and Yunorchis. A relatively weak bootstrap in Bayesian posterior probabilities (PP = 0.61) and stable topologies in bootstrap analyses (BS = 100) between this group and others were observed. The new orchid Yunorchis pingbianensis is a sister to Dactylostalix ringens with a PP of 0.96, BS_ML of 64 and BS_MP of 85.

The 12 genera can be distinguished easily from the phylograms that are based on the chloroplast sequence data analysis. The phylogenetic topologies that are generated from BI are consistent with those from the ML and MP analyses (S4 Fig). The basal group comprises Calypso, Changnienia, Dactylostalix, Tipularia, and Yunorchis. Yunorchis pingbianensis is a sister to Calypso bulbosa with a PP of 0.96, BS_ML of 85 and BS_MP of 83. The chloroplast sequence data analysis showed a relatively stronger bootstrap between this group and others, with PP = 1.00 and BS > 75.

In this study, ITS, matK and rbcL were combined into a single dataset. The results showed that the currently defined Calypsoeae is subdivided into four clades (Fig 2). Clade I, comprising the genera Calypso, Changnienia, Dactylostalix, Tipularia and Yunorchis, is strongly supported (PP = 1.00 and BS = 100) and is followed by the others. Yunorchis pingbianensis, as a sister to Dactylostalix ringens, has been recognized as a natural genus within this tribe. Clade II includes only the genus Govenia, which is strongly supported (PP = 1.00, BS_ML = 72, and BS_MP = 99) and is followed by a complex group (clade III), which includes Aplectrum, Cremastra, Danxiaorchis, Oreorchis, and Yoania. The topologies of this group were different, with a PP of 1.00 and a relatively weak bootstrap and an unstable topology of ML and MP phylograms (no show), and were subdivided into four subclades (Fig 2). The last clade contains a single genus, Corallorhiza, which comprises 13 taxa, although the phylograms showed that the data are not intragenetically well supported.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Phylogenetic relationships of the tribe Calypsoeae based on the combined nrDNA ITS and chloroplast DNA rbcL, matK.

The numbers that are near the nodes are Bayesian posterior probabilities and bootstrap percentages (PP left, BS_ML middle, and BS_MP right). “*” indicates that the node is 100% supported. “-” indicates that the node is incongruent between the topology of the MP/ML trees and the Bayesian tree.

https://doi.org/10.1371/journal.pone.0123382.g002

Subfamily-wide analysis of Epidendroideae

The use of multiple genes is helpful in the accuracy of phylogenetic reconstruction [19,20]. Xiang et al. [15] combined the rbcL, matK, and psaB sequences into a single dataset for the analysis of Epidendroideae, but the sampling and data support for Calypsoeae was weak. In this study, the data for the combined chloroplast genes (rbcL, matK, and psaB) indicate that the tribe Calypsoeae is divided into two supported groups (PP < 0.50, BS < 50). The results of these chloroplast analyses are consistent with those that were obtained in previous studies [11,13,14]. Górniak et al. [11] indicated that 1) the low-copy nuclear protein-coding gene of Xdh is appropriate for resolving the intrafamilial relationships, 2) the phylogenetic relationships of Xdh within Orchidaceae are consistent with those of previous analyses using other DNA data, 3) Xdh is more variable than the plastid genes matK and ycf1 [21] and also can be used at the subtribal and generic levels, and 4) the “difficult” Calypsoeae taxa have been resolved better using Xdh. In this study, the results of the Xdh analyses are consistent with those of Górniak et al. [11] and those of previous studies [4,5,21], but the topologies and higher supports were not entirely consistent with the results of previous studies. The chloroplast genes in many myco-heterotrophs are primarily lacking or are highly divergent, which causes biases in inferred phylogenies [2,16]. The Calypsoeae primarily comprise holomycotrophs or heteromycotrophs that obtain carbon from a symbiotic fungus; this arrangement suggests that the chloroplast genes are lacking, deleted or highly divergent [2,22]. These results indicate that the phylogenetic systematic position of the results for Calypsoeae, which lack some chloroplast genes, would be more reasonable based on the combination of rbcL, matK, psaB, and Xdh into a single dataset for the phylogenetic analysis of Epidendroideae. The sister relationship of Calypsoeae and Epidendreae was also recovered in the result of Nuebig et al. [21] based on the plastid gene ycf1 using parsimony analysis.

The relationships of Coelia (Coeliinae) to other genera have been problematic for more than a century. Based on multiple DNA (rbcL, matK, trnL-F, and ITS) analyses, van den Berg et al. [23] described a circumscription of Epidendreae that includes only five New World subtribes, namely, Bletiinae, Chysinae, Laeliinae, Pleurothallidinae, and Ponerinae, and concluded that Coelia is a sister to Calypsoeae. However, Pridgeon et al. [4] placed Coelia as a subtribe within Epidendreae based on their unpublished analyses. In our analyses (Fig 1 and S1 Fig), Coelia is a sister to Epidendreae plus Calypsoeae. Additionally, van den Berg et al. [23] indicated that the subtribe Agrostophyllinae of Vandeae is a sister to the “core Vandeae” (Aeridinae, Angraecinae, and Polystrachyinae) plus the tribe Cymbidieae, but the node was weakly supported. In Bayesian analysis based on the Xdh gene, Górniak et al. [11] recovered similar relationships; Earina (Agrostophyllinae) was far from the core Vandeae, being a sister to Epidendreae plus Calypsoeae. In our chloroplast analysis (combining rbcL, matK, and psaB), Agrostophyllinae was a sister to one of the groups of Calypsoeae (S2 Fig). In the nuclear Xdh (S1 Fig) and combined rbcL, matK, psaB and Xdh results (Fig 1), the Agrostophyllinae was a sister (PP > 0.90 and BS > 75) to the group that was formed by Coeliinae being a sister to Calypsoeae/Epidendreae. Therefore, the inclusion of Coeliinae (only Coelia) and Calypsoeae in Epidendreae and the system position of Agrostophyllinae warrant further study.

The systematic position of neotropical Wullschlaegelia has long been in trouble. Szlachetko [24] described the genus as a monotypic subtribe within Gastrodieae based on the similar seed morphology, sectile pollinia, cellular viscidium, and the lack of chlorophyll. However, these morphological data are not supported by chloroplast DNA data, suggesting that this genus nests in Calypsoeae. Based on the sequences of the chloroplast genes rbcL and matK, Pridgeon et al. [4] and Zhai et al. [5] placed this genus within Calypsoeae. Based on the sequence of nuclear Xdh, Górniak et al. [11] indicated that this genus is a sister to Xerorchis within the “lower” Epidendroideae. Szlachetko [24] also indicated differences in the floral morphology and the position of the anther with Gastrodieae. Because Wullschlaegelia is a holomycotrophic plant [25], the nuclear evidence appears to be more credible [2]. Moreover, Wullschlaegelia has walls in a striate-rugulate pattern with free-standing baculae between them or coarsely verrucate-striate or gemmate [4]. The pollinia of Orchidaceae have been considered informative sources of both taxonomic and phylogenetic information and are thought to be less susceptible to parallelism than other floral features [26]. Indeed, the number of pollinia in Wullschlaegelia is two, and the members of Calypsoeae are four. These data suggest that this genus and Calypsoeae are not closely related. Therefore, we conducted analyses with Wullschlaegelia, which is a sister to Xerorchis, based on our results (Fig 1 and S1 and S2 Figs) and the conclusion of Górniak et al. [11].

Phylogenetic relationships within Calypsoeae

The present phylogenetic study was primarily conducted based on the molecular dataset that was described by Zhai et al. [5]. In this study, we performed an additional DNA sequence analysis using the excluded highly homoplasious morphological characters. Although the phylogenetic relationships were fully assessed and discussed in the previous study using the combined analysis of molecular and morphological characters, most of the support was weak. Based on the new molecular systematic evidence with greater support and in the interest of nomenclatural stability, we divided this tribe into four clades, and we redefined the four subtribes, namely, Aplectrinae, Cremastrinae, Wullschalaegeliinae (the genus Oreorchis in this study), and Yoaniinae, which were circumscribed by Zhai et al. [5], as four subclades of clade II.

In the phylogenetic analyses (Fig 2 and S3 and S4 Figs), the new holomycotrophic orchid were not nested into the same group with the other three (Corallorhiza, Danxiaorchis and Yoania) holomycotrophic genera of Calyposeae. The Corallorhiza comprising 11 species that are confined to the northern hemisphere of the New World except for one circumboreal species (C. trifida), is one of the largest leafless genera in Orchidaceae [9], and the species are characterized by the coralloid rhizome and unlobed labellum narrowing to a claw at base with two free lamellae [4]. Danxiaorchis is founded in Guangdong, China, and the species possesses a labellum with a large Y-shaped callus and fleshy seeds [5]. Yoania is sister to Danxiaorchis in the phylogeny of combined chloroplast rbcL and matK (Fig 2 and S4 Fig). The same relationships were observed in Zhai et al. studies [5] based on the combined molecular markers and/or morphological character matrix. These indicated that Yoania is most closely related to Danxiaorchis. Additionally, the inflorescences of Yoania stand straight up, the labellum bears a prominent spur and the pollinarium have four superposed pollinia [4]. However, the subumbel inflorescences of the new orchid grow underground, the sac-shaped labellum has hook and two fragments at the base of the hook, the pollinarium are in two pairs and the two pollinia of each pair are unequal in size. These features make it easy to distinguish the new orchid from the other holomycotrophic genera of Calyposeae.

The new orchid, Yunorchis, which is a sister to Dactylostalix (Fig 2 and S1 Fig) and/or Calypso (S2 Fig), Changnienia, and Tipularia, form an independent root clade (clade I) of Calypsoeae. The genera Calypso, Dactylostalix, and Yunorchis have a trilobed labellum. The lip lobe of Calypso occupies the first half of the sac border, the lobe base has tufted setae, and the labellum tip has two calcariform objects. This green leaf plant has an erect raceme and is primarily distributed in temperate regions. While the lip lobe of Yunorchis is small and is located in the middle of the labellum sac, the sac border before the lobe closes upward and forms the unciform tip. The entire plant, including the inflorescence, grows underneath the ground, and the flower stretches toward the earth’s surface when blooming; therefore, only a few flowers are observed scattering on the ground, and this plant is completely saprophytic and is distributed in tropical regions. Both genera have obvious plant morphology and habitat differences and are different from Changnienia and Tipularia, which have a distinct calcar in the labellum, green leaves, a scape growing along the ground, and inflorescences with one or many tiny flowers.

Pridgeon et al. [4] indicated that the Dactylostalix and Ephippianthus have viscidia with only the rudiments of a stipe and are unusual in having only slightly thickened rhizomes and no corms. Moreover, both genera show leaves in some seasons. Unfortunately, the molecular materials of the other genus Ephippianthus could not be obtained. Zhai et al. [5] indicated that the genus formed a sister with Dactylostalix based on morphological characteristics. The inflorescence of Dactylostalix, which has one or two sheathing bracts, solitary flowers, pollinarium with four superposed pollinia and a viscidium with a rudimentary stipe [4], and the few-flowered plants of Ephippianthus, which have membranous bracts, an entire labellum, and pollinarium with four superposed pollinia [4], are significantly different from the new holomycotrophic orchid and the other orchids of the Calypso Clade.

Etymology

The generic name alludes to the type locality, Yunnan, and incorporates the Greek name for orchid; Yunorchis refers to an orchid genus in Yunnan province, China. The specific epithet pingbianensis refers to the type of species of Yunorchis that grows in Pingbian county of Yunnan.

Diagnosis

This new remarkable genus is distinct from all known orchid genera; it comprises an entire plant with a subumbel inflorescence growing underground, flowers that each have a long pedicel that ruptures the ground to open and that has a sac-shaped labellum, a sac mouth that is connate on the front half edges, forming a hook and two fragments at the base of the hook, and four waxy pollinia in two pairs, each containing two pollinia that are unequal in size, and there are no conspicuous caudicles that are attached directly to a common viscidium.

Description

Holomycotrophic plants 6–9 cm tall, rhizome tuberous, cylindrical, 4–6 cm long, 5–7 mm thick, with many short branches. Entire plant growing underground, scape arising from the terminal of rhizome, 1 or 2 inflorescences, terete, pale red-brow, 4–6-sheathed; sheaths, cylindrical, clasping scape stem, membranous, 1.0–1.5 cm long; subumbel inflorescence, 0.5–0.8 cm long, 2–6-flowered; flowers rupturing the ground when open, floral bracts broadly ovate or ovate-elliptic, 1–1.6 cm long, obtuse at apex; pedicel and ovary 3.8–5.0 cm long, glabrous, twisted; sepals pale purple-red; petals white; lip white, spotted with purple-red at apex and inside; dorsal sepals oblong, 2–2.3 cm long, 6.0–7.5 mm wide, obtuse at apex; lateral sepals ovate-elliptic, concave, with incurved margins, 2.0–2.3 cm long, 0.8–1.0 cm wide, obtuse at apex; petals narrowly oblong, 1.9–2.1 cm long, 1.6–1.8 cm wide, obtuse at apex; lip semiglobose-saccate, 1.7–1.9 cm long; sac mouth connate on front half edges forming a hook, with the lateral lobes reduced at the base of hook; disk with a purple prominent central papillate-pubescent callus extending from base to apex, column 7–8 mm long, 5.0–5.5 mm wide, with a 1.7–1.9 mm long fingerlike process projecting on either side of the anther at the apex, with a 2–3 mm long column foot at the base; anther terminal, anther cap ellipsoid, with a horn at the apex; four pollinia, in two pairs, narrowly elliptic, waxy, each pair containing two pollinia that are unequal in size, lacking conspicuous caudicles, attached directly to a common viscidium; and stigma concave, broad and large; rostellum conspicuous; capsule fusiform. Fl. May–June.

Ethics statement

The locations of the field studies are neither private lands nor protected areas but are controlled by the State Forestry Administration of China, with which our institution is affiliated. The State Forestry Administration authorized us to conduct scientific observations or tests in the regions that it controls. A valid permit was obtained for testing the genes of Yunorchis.

Taxon and gene sampling

To reconstruct a robust phylogeny for determining the circumscription and systematic positions of Calypsoeae, we sampled as many taxa as possible from all of the tribes recognized within Epidendroideae by Chase et al. [27] and Pridgeon et al. [4,28,29]. A total of 104 genera were sampled. The outgroup included two genera from the subfamily Cypripedioideae and one genus from the subfamily Apostasioideae. The sampled strategy is not only important for examining relationships, but the inclusion of DNA sequence data from four nuclear and chloroplast regions was expected to permit possible reinterpretation of previous hypotheses that included only few of these genomes at a time [11,12,13]. A composite of slowly evolving genes was used to infer the phylogenetic relationships at the tribal/subtribal levels. All of the terminal taxa represent a single genus and include at least three of the four DNA markers (rbcL, matK, psaB and Xdh), except for the new orchid and Agrostophyllum, which include two markers (rbcL and matK). The Dactylostalix is sampled for the first time in a broad molecular phylogenetic study. The voucher information and GenBank accession numbers are listed in S1 Table.

A second series of analyses focused on the tribe Calypsoeae. A composite of slowly (rbcL and matK) and quickly (ITS) evolving genes and DNA regions was used to infer the phylogenetic relationships. The analysis is primarily based on the molecular dataset that was published by Zhai et al. [5] based on the combined analysis of molecular and morphological characters. In this study, we performed additional DNA extraction (six DNA sequences of rbcL, four of matK, and five of the ITS were newly obtained) excluding the highly homoplasious morphological characters for molecular phylogenetic analyses. This tribal-wide matrix included 34 in-group species and two outgroups (Agrostophyllum majus and Earina autumnalis). The voucher information and GenBank accession numbers are listed in S2 Table. DNA extraction, PCR amplification, sequencing, sequence editing and assembly were performed according to Zhang et al. [30]. The primers are showed in S3 Table.

Phylogenetic analyses

Phylogenetic analyses were performed under Bayesian inference (BI) and the MP and ML methods. We performed the BI analysis using MrBayes 3.1.2 [31]. For the ML analysis, we used RAxML version 7.2.8 with 100 bootstrap replicates and settings as described in Stamatakis et al. [32]. The homogeneities of the subfamily Epidendroideae between nuclear Xdh and combined chloroplast DNA and the tribe Calypsoeae between nuclear ITS data and the combined chloroplast DNA dataset were tested using the incongruence length difference (ILD) test [33], as implemented in PAUP* version 4.0b10 [34]. The ILD test was conducted with 1000 replicates, each with 10 random addition sequence replicates and TBR branch swapping, maintaining no more than 100 trees per random addition replicate. Following Cunningham [35], a significance level of P = 0.01 was used for this test. The MP analyses were performed using PAUP* version 4.0b10 [34]. The test settings included 1000 replications of randomly added sequences and a heuristic search with tree bisection-reconnection branch swapping. Table 1 lists the tree length (TL), consistency index (CI), and retention index (RI).

Nomenclature acts

The electronic version of this article in Portable Document Format (PDF) in a work with an ISSN or ISBN will represent a published work according to the International Code of Nomenclature for algae, fungi, and plants; therefore, the new names that are contained in the electronic publication of a PLoS ONE article are effectively published under that Code from the electronic edition alone. There is no longer any need to provide printed copies.

In addition, the new names that are contained in this work have been submitted to IPNI, from where they will be made and are available to the Global Names Index. The IPNI LSID 77145331–1 can be resolved and the associated information viewed through any standard web browser by appending the LSID contained in this publication to the prefix http://ipni.org/. The online version of this work is archived and available from the following digital repositories: PubMed Central, LOCKSS.