Specklinia dunstervillei, a New Species Long Confused with Specklinia endotrachys (Orchidaceae: Pleurothallidinae)

Specklinia dunstervillei is described as a new species in recognition of the distinctness of a Venezuelan species related to and confused with Specklinia endotrachys. It was illustrated for the first time by G. C. K. Dunsterville in 1963 from a plant collected in Trujillo on the Cordillera de Merida. The newly named species can be easily recognized by its small habit, short leaves and small reddish-orange flowers, the non-ascending dorsal sepal and the obtuse petals that are shortly apiculate. Specklinia dunstervillei is formally described and illustrated once again and compared morphologically and genetically with its closest relatives.


Introduction
In the systematic study of the Specklinia endotrachys species complex, Pupulin et al. [1] found that, aside from the latter, Specklinia pfavii (Rchb.f.) Pupulin & Karremans, S. remotiflora Pupulin & Karremans and S. spectabilis (Ames) Pupulin & Karremans could all be recognized as distinct species in the complex. Specklinia endotrachys (Rchb.f.) Pridgeon & M.W.Chase had traditionally been considered a widely distributed and highly variable albeit morphologically quite unique species, nevertheless, the four proposed taxa could be easily distinguished morphologically and ecologically from each other. At the time, the authors recognized that not all the available material labeled S. endotrachys could be adequately placed into one of those four species concepts, and suspected more taxa would be recognized in the complex in the future [1,2,3].
The fifth species to be added to the complex was the Guatemalan Specklinia juddii (Archila) Pupulin & Karremans, unknown at the time of the first publication and later placed amongst its relatives [2,3]. It had been published under the generic name Empusella (Luer) Luer, a monotypic genus typified by S. endotrachys and here regarded as a synonym of Specklinia Lindl. [4,5,6]. Similarly, the Venezuelan material labeled S. endotrachys and illustrated for the first time in Dunsterville and Garay [7], morphologically did not resemble the other five species in the complex and had remained a mystery to the authors [1,2,3]. Now with additional material at hand we are able to compare the Venezuelan material with the other species of this complex, both morphologically (for all species currently recognized) and genetically (for all except S. juddii).

Materials and Methods
This study was conducted at Jardín Botánico Lankester (JBL) of the Universidad de Costa Rica and Naturalis Biodiversity Center-Leiden University, from 2012 to 2014. Plants were collected under the scientific permits handed by the Costa Rican Ministry of Environment (MINAE) to researchers at JBL. Individual plants were photographed, illustrated and preserved as DNA samples, herbarium specimens and spirit specimens in formaldehyde: acetic acid: ethanol [FAA (53% ethanol, 37% water, 5% formaldehyde and 5% glycerol)] (only including flowers) for future reference, deposited at JBL-spirit and L-spirit and the DNA bank of Naturalis Biodiversity Center. Taxon names mostly follow Pridgeon [8].

Photography
The Lankester Composite Digital Plate (LCDP) and color illustrations of the flowers were made using a Nikon D5100 digital camera with a AF-S VR Micro-NIKKOR 105mm f/2.8G IF-ED lens and an Epson Perfection Photo Scanner V600, and a Leica MZ9.5 stereomicroscope fitted with a Leica DFC295 digital microscope color camera with Leica FireCam 3.4.1 software.

Phylogenetic analysis
The data matrix included DNA sequences of 50 individuals (Table 1), 27 of which were produced in this study. The remaining data were obtained from GenBank [4,5,9]. Fresh leaf and flower cuttings of approximately 1 cm 2 were dried with silica gel. Samples (20 mg) were pulverized and extraction performed following the DNEasy procedure (Qiagen). The nuclear ribosomal internal transcribed spacer (nrITS) region was amplified using the methods and primers for sequencing and amplification described by Sun et al. [10], and Sanger sequencing was done commercially by Macrogen on a 96-capillary 3730xl DNA Analyzer automated sequencer (Applied Biosystems, Inc.) using standard dye-terminator chemistry (Macrogen, Inc.).
The Staden et al. [11] package was used for editing of the sequences. Contigs were exported as.fas files and opened in Mesquite v2.72 (Maddison & Maddison [12]), where they were checked for base calling errors, the matrix was aligned manually (S1 File: Sequence Matrix). The ends of each data set were trimmed to eliminate possible erroneous data, and gaps were regarded as missing data (filled with Ns). Phloeophila peperomioides was used as the outgroup, as it was found to be one of the most distantly related of all included species in this phylogenetic analysis (Pridgeon et al. [4]). The trees were produced with an analysis of the nrITS dataset of 43 sequences using BEAST v1.6.0. (Drummond & Rambaut [13]). Parameters were set to preset, except for substitution model GTR with 10 categories, clock model uncorrelated lognormal, tree prior Yule process, and number of generations 20,000,000. The resulting trees were combined using TreeAnnotator v1.6.0., where the first 3000 trees were used as burn-in. FigTree v1.3.1. (Rambaut [14]) was used to edit the resulting tree. Posterior probabilities are given for each node in decimal form.
A pairwise comparison of the ITS sequence of the accessions of S. dunstervillei, S. endotrachys, S. montezumae (as an outgroup), S. pfavii, S. spectabilis and S. remotiflora is presented in Table 2. Each different base was counted as an individual change, even when concurrent; insertions and deletions were counted as a single change regardless of length. All the mentioned accession of each species in Table 1 were combined and used for the comparison except for AF262859, a sequence labeled S. endotrachys by Pridgeon et al. [4] but which we suspect (based on DNA data) should be S. remotiflora or a closely related species.

Nomenclature
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, and hence the new names contained in the electronic publication of a PLOS ONE article are effectively published under that Code from the electronic edition alone, so there is no longer any need to provide printed copies.

Results Photography
The color illustrations of species of the Specklinia endotrachys complex (Fig 1) shows a morphologically distinct entity, Specklinia dunstervillei (Fig 1A and 1B), recognized amongst others by the smaller flowers and shortly apiculate petals.

Phylogenetic analysis
The consensus gene tree (Fig 2) was obtained from a BEAST analysis of a matrix of 45 ITS sequences (Table 1) (Table 2) shows that the sequences of Specklinia endotrachys and S. spectabilis do not differ from each other, while S. dunstervillei differs in 2 bases from those species. Specklinia pfavii and S. remotiflora differ in 4 and 10 bases respectively, from the three before mentioned species. Specklinia montezumae differs in 9 to 11 bases from each of the members of the S. endotrachys species complex.  The species is similar to Specklinia endotrachys but can be distinguished by the small habit, shorter leaves, smaller flowers, the flat, non-ascending dorsal sepal (vs. ascending), and the obtuse, shortly apiculate (vs. emarginate and long mucronate) petals.
Note: Only the specimens from Venezuela were used for the description (Dunsterville 757 and Karremans 5966).
Etymology. The name honors G. C. K. Dunsterville, who prepared the first known illustration of the species. Other records. Colombia. Without collecting data, illustration of voucher in Ortiz and Uribe [15].
Distribution. The material collected by Dunsterville in Venezuela comes from the road between Boconó and Guaramacal in Trujillo, on the Cordillera de Merida, a branch of the Colombian Andes, at an elevation of about 2400 m. A few specimens that have been found in private collections in Costa Rica and Colombia lack precise location data.
Key to the species of the empusellous species of Specklinia.  Ecology. The typical form of Specklinia dunstervillei is that found in Colombia and Venezuela (Figs 1a, 3 and 6). The material illustrated by Dunsterville, which lacks an herbarium voucher (Gustavo Romero pers. comm.), was collected on the Cordillera de Merida at about 2400 m in elevation. That makes the new species the one with the most Easternly distribution in the complex and, together with S. remotiflora, the only one to grow in cloud forests at high elevations above 1800 m. A photograph of this species was published in Ortiz and Uribe [15] without precise locality, nonetheless it was definitively taken from a Colombian plant (Carlos Uribe V. pers. comm.). Considering that the Cordillera de Merida extends into Colombia, it is indeed not surprising that the species is also found there. Finally, we have chosen to regard the material found in Costa Rican private collections as S. dunstervillei as they are morphologically most similar to that species concept, nevertheless they do show quite some variation (Figs 1b, 4 and 5). The studied Costa Rican specimens lack specific collection data.
Like the other species of the S. endotrachys complex, S. dunstervillei has nectar secreting stomata placed on the apex of the warts that cover the surface of the sepals [1,2,3,6]. The released nectar gives the sepals a wet-glossy appearance, as noted by Dunsterville and Garay [7], and is given special attention by visiting fruit flies. At Lankester Botanical Garden the flowers were frequently visited by up to 6 individuals of Drosophila spp. The flies wander about on the sepals, sucking on the papillae rich surfaces, occasionally removing the pollinia, after stepping on the movable lip and being adpressed against the column (Fig 8).
Morphological recognition. Specklinia dunstervillei is morphologically similar to other members of the S. endotrachys complex. Within Specklinia they are recognized by the relatively large habit and flowers, the long, successive, flat, ancipitous inflorescence, the bright yellow to orange-red flowers, the verrucose sepals, the relatively minuscule petals and the highly sensitive strongly concave lip. Amongst its closest relatives, S. dunstervillei is recognized by the tiny habit, caespitose plants, short leaves (up to about 10 cm), the small flowers (dorsal sepal 14-16 × 5-6 mm), the flattening, non-ascending dorsal sepal, the lateral sepals that spread, slightly twisting downwards and the obtuse petals, which are shortly apiculate.
Genetic recognition. The nrITS sequence of the accession of Specklinia dunstervillei differs in two base pairs (out of just under 800bp) from the accessions of both S. endotrachys and S. spectabilis, which are identical to each other ( Table 2). Considering that the species of the S. endotrachys complex differ in only 9 to 11 bases (between 1.1 and 1.4%) from an outgroup as morphologically distinct as S. montezumae, it becomes highly plausible that species with significant morphological and ecological differences within the complex itself, such as S. endotrachys and S. pfavii, can differ in just about 0.5% bases. Therefore it is also not unusual that more closely related species such as S. endotrachys and S. dunstervillei differ in only 2 bases, or that the sisters S. endotrachys and S. spectabilis do not differ at all in this particular DNA barcode.
The internal transcribed spacer (ITS) of nuclear ribosomal DNA has consistently shown high levels of discrimination among many species of plants [16] including Pleurothallidinae species, and are well suited for a broad range of phylogenetic studies [4,10,17,18]. However, in current literature DNA data are mostly used to support so called "cryptic" novel taxa, when the accession of a said taxon does not cluster with other accessions of the a priori believed same taxon [19,20,21,22]. When different accessions of a species complex cluster together authors seem to believe that there is evidence that they are a single species, however that is not only an illogical conclusion but also frequently untrue. Be it accessions of the same species, or sister species, or species of a single genus or the same family, any two accessions will cluster together with high support if they are relatively much more similar to each other than to the other sequences analyzed. The only measurable difference between the mentioned sister pairs would be the branch length (as a measure of number of base pair changes). However, as there is no established similarity threshold above which one should consider two sequences to belong to two different species, or below which they should be considered a single species, it is at this time also not possible to infer from branch lengths alone if we are dealing with a single or more than one species. Meyer and Paulay [23] found that this especially counts for taxonomically understudied groups that are not yet thoroughly sampled.
In Pleurothallidinae, DNA data have been seldomly used to support novel species descriptions. Ramos-Castro et al. [24] used an nrITS based phylogeny to prove that their novel taxon Stelis zootrophionoides Castañeda-Zárate & Ramos-Castro belonged in that genus, however they did not venture into using DNA data to set their novelty apart from other taxa in the genus. In the phylogeny presented, the single accession of S. zootrophionoides clusters, as would be expected, with that of the morphologically similar S. nigriflora (L.O.Williams) Pridgeon & M.W.Chase. Therefore, their phylogeny by itself could not prove that the accessions belonged to two different species, and the authors relied exclusively on morphological differences to establish their novelty. A species labeled Specklinia sp. in Chiron et al. [9], of which the nrITS sequence clustered with that of Specklinia marginata (Lindl.) Pridgeon & M.W. Chase, was later published as a distinct species using mainly morphological characters. In both cases the authors could also have argued that the nrITS sequence of the accession of their novelty differed in about 1% from the sequence of the accession of its closest relative with which it clustered. A different approach was that of Meyer et al. [25], who used the presence of insertions/deletions in the trnH-psbA and ycf1 sequences of accessions of Dracula radiosa (Rchb.f.) Luer and close relatives to support the establishment of their novel species Dracula trigonopetala Gary Mey. & L. Baquero. In that case their novelty differed in less than 0.3% from its closest relatives (as an insertion/deletion counts for a single change), however, the 26 bp unique insertion in the trnH-psbA sequence of their novelty stands out in an otherwise quite conserved region and therefore easily sets it apart from its close relatives.

Conclusions
A common misconception amongst modern authors is that DNA data will in itself resolve taxonomic issues. In fact, that two sequences are identical in a particular region does not guarantee that they belong to the same species, nor does the fact that they differ guarantee that they belong to different species. DNA data albeit less subjective than morphological characterization, are nonetheless subject to many of the same pitfalls, which can basically be summarized as: 1) interpretation of the type specimens and correct application of names; 2) reduction or elimination of data reading mistakes; 3) adequate interpretation of the observed variation.
Considering all the evidence at hand we conclude that the specimen illustrated in Dunsterville and Garay [7] from Venezuela and labeled Specklinia endotrachys actually represents an until now unnamed species. With the inclusion of the here proposed novelty, Specklinia dunstervillei, the S. endotrachys species complex now consists of six distinctly named species, instead of a single variable species. Its highest diversity is found in Costa Rica and Panama, but this might change as additional material becomes available from less sampled areas; we therefore do not reject the possibility of discovering other novelties in this group in the future.
Supporting Information S1 File. Alignment matrix of 43 nrITS sequences used to infer the molecular phylogeny presented in Fig 2. The ends were trimmed and gaps were regarded as missing data. (NEX)