Monophyly or Paraphyly– The Taxonomy of Holcoglossum (Aeridinae: Orchidaceae)

Recently, there have been a lot of intense debates about the acceptance/rejection of paraphyletic groups in biological classification. On the one hand, evolutionary classification states that similarity and common descent are two criteria for biological classification and paraphyletic groups are natural units of biological classification. On the other hand, cladistic classification considers that common descent is the only criterion in biological classification and monophyly should be strictly adhered to. Holcoglossum is used herein as a case to illustrate this problem. Although Holcoglossum is a small orchid genus of less than 20 species, there is little consensus about its generic circumscription since it was established, which leads to confusion in taxonomic treatments in the Aerides-Vanda group. Based on the analyses of molecular and morphological evidence, our results suggest that the clade comprising Holcoglossum s.s., Ascolabium, Penkimia and Ascocentrum himalaicum is strongly supported as a monophyly, and that the three taxa are nested within different subclades of Holcoglossum s.s. Thus, it is reasonable to recognize a monophyletic circumscription of Holcoglossum, which is also well supported by some vegetative and floral characters. The Holcoglossum s.l. would facilitate a better understanding of pollinator-driven floral divergence and vegetative stasis than a paraphyletic and narrowly defined genus.


Introduction
In the era of integrative taxonomy, there is more major congruence in the biological classification between cladistic classification and evolutionary classification except with regard to the acceptance/rejection of paraphyletic groups. Phylogenetic (cladistic) methods and DNA sequences are routinely used in systematics and taxonomy, and both schools of classification have recognized that the principle of common decent plays a major role in biological classification. However, the fundamental question about the acceptance/rejection of paraphyletic groups remains unsolved, and recently there have been many intense debates about this issue [1][2][3][4][5][6][7][8][9][10][11]. Evolutionary classification recognizing paraphyletic groups argues the following: (1) there are two criteria for biological classification, i.e., similarity and common descent [12]; (2) many species are paraphyletic [1,[12][13][14]; (3) paraphyletic groups are natural transitional stages in the evolution of taxa and are natural units of biological classification [4]; (4) cladistic classification is incompatible with the Linnean hierarchy system [1,4,7], such as diachronous groups and (5) classification based only on common descent often fails to reflect divergence and natural selection. In contrast, cladistic classification states the following: (1) only monophyletic groups in their strictest sense (holophyly) that are evidenced by synapomorphous characters are recognized in biological classification; (2) only species and clades are objective, and supraspecific taxa are terminals in cladistic classification [15,16]; (3) there are no objective criteria to circumscribe paraphyletic groups, and paraphyletic groups are artificial classes created by taxonomists to emphasize some particular characters or divergence [5,[16][17][18] and (4) cladistic classification can be accommodated within the Linnaean system except for monotypic higher taxa and historic groups [19], with the former being neither paraphyletic nor monophyletic and the latter being dismissed [16,19]. However, as there are many theorical and lengthy discussions anywhere [1][2][3][4][5][6][7][8][9][10][11], a discussion of the merits and fallacies of each school of classification is not the major aim of present paper. Instead, we used the taxonomy of Holcoglossum (Aeridinae, Orchidaceae) as a case to illustrate this problem here.
The orchid subtribe Aeridinae is a large and well-defined horticulturally important group of approximately 1200 species in 120 genera [20,21]. However, the taxonomy of Aeridinae, particularly the generic delimitations, is difficult and has been considered as ''the black pit'' [20][21][22][23][24][25][26]. Seidenfaden [27] (page 8) even stated, ''A recurrent dilemma in the study of the Aeridinae is the allotment of a species or a group of species to a genus. 2 I have several times met with this problem, e.g., in Aerides, Holcoglossum and Ascocentrum.'' Holcoglossum is a small Asian genus, consisting of less than 20 species and mainly distributed in southwestern China and neighboring regions. Nonetheless, Holcoglossum occupies an important systematic position in the informal taxonomic group, the Aerides-Vanda alliance, which includes Aerides, Ascocentrum, Ascolabium, Holcoglossum, Neofinetia, Papilionanthe, Penkimia, Seidenfadenia and Vanda. Holcoglossum has been redefined several times since it was established by Schlechter [27], but it appears that there is little consensus about its generic circumscription, with many species being transferred among genera within the Aerides-Vanda alliance [22][23][24][28][29][30][31][32][33]. Based on plastid matK and trnL-F and nuclear ITS sequences, two most recent molecular phylogenetic studies of Holcoglossum [32,33] have helped to clarify this problem; however, their results of each are substantially different, thus further confusing the taxonomy of Vanda-Aeridies alliance [32,33].
In the present study, phylogenetic relationships were inferred using four DNA markers (plastid matK, trnH-psbA and trnL-F, and nuclear ITS sequences) and combined with morphological analyses with sampling across Aeridinae to (1) illustrate the generic circumscription problem of Holcoglossum, (2) understand the difficulty in accepting/rejecting paraphyletic groups in Holcoglossum, and (3) investigate the evolution of morphological characters within Holcoglossum.
Within Aeridinae, the ITS dataset for 138 taxa consisted of 854 characters of which 321 were parsimony informative. The Bayesian trees of each dataset were congruent with the MP trees, except for some weakly supported nodes. The interrelationships among most of the genera in Aeridinae were unresolved. Holcoglossum s.s. was not monophyletic: Ascolabium, Penkimia and Ascocentrum himalaicum were nested within different clades of Holcoglossum ( Figure S1). Four out of the five sampled Ascocentrum species (except for A. himalaicum) were nested in the clade that included Vanda, Neofinetia and Christensonia ( Figure S1).

Phylogenetic Analyses within a Reduced Holcoglossum Dataset
The ITS dataset displayed 749 characters, 118 of which were parsimony informative. The combined matrix of plastid matK, trnH-psbA and trnL-F sequences indicates 4287 sites of which 258 are parsimony informative (Table S1). The monophyly of the clade including Holcoglossum s.s., Ascolabium, Penkimia and Ascocentrum himalaicum is moderately or weakly supported based on the ITS and the combined plastid datasets, respectively (results not shown).
The combined matrix of the nuclear and plastid datasets has 5036 sites, 376 of which are parsimony informative (Table S1). The Bayesian analysis yielded trees having topologies that are consistent with those obtained using MP analysis. The clade including Holcoglossum s.s., Ascolabium, Penkimia and Ascocentrum himalaicum is a monophyly with moderate support (BS = 74, PP = 0.99). Moreover, similar to the results of Fan et al. [32], this clade subdivided into three subclades, i.e., the tropical (TC), the alpine (AC) and the intermediate subclades (HC). Penkimia and Ascocentrum himalaicum are nested in TC (BS = 93, PP = 1.00), and Ascolabium is sister to H. quasipinifolium in HC (BS = 96, PP = 1.00) ( Figure 1A).
In total, forty-five gross morphological characters are examined (Table S2). The cladograms from the morphological data based on the MP analysis were poorly resolved ( Figure S2). Based on the combination of the morphological and molecular evidence, the clade comprising Holcoglossum s.s., Ascolabium, Penkimia and Ascocentrum himalaicum was strongly supported (BS = 92, PP = 1.00) ( Figure 1B).
The cladistic and patristic distances are shown in Table S3. The patrocladistic analysis revealed that Ascolabium, Penkimia and Ascocentrum himalaicum remain embedded within different subclades of Holcoglossum s.s. (Figure 2).

Monophyly Versus Paraphyly 2 Definition of Holcoglossum
Based on the morphological and molecular evidence, our results indicated that Holcoglossum Schltr. (Holcoglossum s.s.) is paraphyletic; however, the clade including Holcoglossum s.s., Ascolabium, Penkimia and Ascocentrum himalaicum is strongly supported as a monophyly ( Figure 1). The red to yellow tiny flowers misplaced members, namely, Penkimia, the neglected Ascolabium [34], and Ascocentrum himalaicum, have independently evolved at least twice in Holcoglossum s.s. (Figure S3). We are in a dilemma to circumscribe Holcoglossum. It appears that there are three proposals for the circumscription of Holcoglossum. (1) The first is to narrow Holcoglossum, as Liu et al. [33] did: based on this, at least six genera, Holcoglossum, Ascolabium, Penkimia, Paraholcoglossum, Tsiorchis, and another new genus containing Ascocentrum himalaicum, would be recognized, yet Holcoglossum would still be paraphyletic. (2) The second is to circumscribe Holcoglossum according to Jin [30] and Jin & Wood [35]: under this scenario, two monotypic or oligotypic genera, Penkimia (with Ascocentrum himalaicum) and Ascolabium, have to be recognized in addition to the paraphyletic Holcoglossum s.s. (3) The third is to define Holcoglossum sense lato: if Holcoglossum is redefined as monophyletic, then the other four genera, i.e., Penkimia, Ascolabium, Paraholcoglossum, and Tsiorchis, will be included, and Ascocentrum himalaicum will be transferred into it. Regarding option 1, there are no distinct morphological characters to distinguish Tsiorchis and Paraholcoglossum from Holcoglossum ( Figure S3), and the narrowly redefined Holcoglossum is still paraphyletic. Furthermore, the subtribe Aeridinae is abundant in many aberrant species and plagued by the many monotypic genera; therefore, this proposal would set a precedent to separate many aberrant species as new genera. Accordingly, it seems that the recognition of many similar monotypic genera while keeping Holcoglossum as paraphyletic provides no practicability or similarity and does not reflect the maximum evolutionary information.
With regard to option 2, all the members of the paraphyletic Holcoglossum s.s. are consistent in both the vegetative characters [22,27,30,36] and floral characters, which epitomize bee-pollination syndrome [36], making it very good in practicability, high in similarity and information content; however, two small genera have to recognized, as their evolutionary information cannot be related to Holcoglossum.
For option 3, Holcoglossum s.l., including Holcoglossum s.s., two monotypic genera (Ascolabium and Penkimia ), Ascocentrum himalaicum and two newly described genera, Paraholcoglossum and Tsiorchis, is strongly supported as a monophyly by the molecular evidence and patrocladistic analyses (Figure 1 and 2). Indeed, the genetic distances revealed that these species show closer relationships to Holcoglossum s.s. than to other genera within Aeridinae (Table 1). In fact, Ascocentrum himalaicum was included in Holcoglossum by Tsi [28]  and Averyanov [37] based on morphological characters. However, Holcoglossum s.l. is greatly heterogeneous in floral traits. The floral autapomorphies of Ascolabium, Penkimia and Ascocentrum himalaicum, such as red or purple, small-sized flowers, short and densely flowered inflorescences, and cylindrical spur, evolved independently at least twice in Holcoglossum and most likely were driven by the pollinator-mediated selection (Figure 4 and 5).
Based on these data, we tentatively propose to recognize a monophyletic circumscription of Holcoglossum that comprises approximately 17 species, including Holcoglossum s.s., Ascolabium, Penkimia, and Ascocentrum himalaicum, whereas Paraholcoglossum and Tsiorchis are placed within Holcoglossum s.s. for the purposes of practicability, similarity and providing the maximum evolutionary information.

Morphological Character Evolution within Holcoglossum s.l
In comparison to the conserved vegetative characters, Holcoglossum s.l. is greatly divergent with regard to its floral characters (Table S2). Recent studies of pollination in Holcoglossum indicated that the pollination systems are more divergent than previously expected. To date, four pollination systems, autogamy [38], beetle pollination [39], bee pollination [40], and ant pollination [41], have been recorded in Holcoglossum, whereas bird pollination in H. himalaicum and H. pumilum remains to be confirmed [42]. Generally, pink flowers are considered bird pollination syndrome, and white flowers with a colored lip are considered bee pollination syndrome [43]. Two pink-flower species, H. himalaicum and H. pumilum, had previously been placed in Ascocentrum due to their floral similarities with A. ampullaceum, but our results indicated that pink flowers have independently evolved at least twice from white flowers in Holcoglossum s.l. (Figure 5, right). It appears that this shift is the result of pollinator-mediated selection ( Figure 5, left). Therefore, Holcoglossum s.l. would be better to understand pollinator-driven floral divergence and vegetative stasis than a paraphyletic and narrowly defined genus.

Conclusions
Based on the analyses of molecular and morphological characters, Holcoglossum s.s. is paraphyletic; the clade consisting of elements from five other genera, Ascocentrum, Ascolabium, Paraholcoglossum, Penkimia and Tsiorchis, is strongly supported as a monophyly. Hence, it would be better to retain Holcoglossum as monophyletic by the inclusiveness of three red to yellow species for the sake of information content, practicality and similarity. Therefore, we tentatively propose to recognize the monophyletic

Description
Epiphytic, small to moderate-size plants. Roots arising from the base of stem, white and fleshy, tip reddish. Stem short, usually clustered, enclosed by persistent leaf sheaths. Leaves articulate at base, condensed along stem, fleshy, subterete to terete, channeled adaxially. Inflorescences lateral. Flowers white to purple; pedicel and ovary long; dorsal sepal usually erect; lip 3-lobed, spurred or saccate; lateral lobes erect; middle lobe arising from spur; column winged; pollinia two, porate, usually with tapering stipe; anther cap beaked.

Ethics Statement
The species collected here are not included in the checklist of Chinese Protected Species. The fieldwork was conducted under the permission of the authority of each natural reserve, specifically Gaoligongshan National Nature Reserve (Yunnan, China), Dali Cangshan-Erhai National Natural Reserve (Yunnan, China), Jianfengling National Nature Reserve (Hainan, China) and Wuzhishan National Nature Reserve (Hainan, China). No specific permits were required for the described field studies.

Taxon Sampling
To determine the systematic positions of Holcoglossum and related genera, we sampled 78 genera and 138 samples, representing all of the major clades in the subtribe Aeridinae, based on previous molecular work (Table S4) [26,44]. Secondly, we sampled a reduced matrix containing 36 species to analyze the intraspecific relationships within Holcoglossum (Table S5).
To facilitate consistency and convenience, Paraholcoglossum and Tsiorchis were referred to Holcoglossum throughout the Results.

DNA Extraction, PCR and Sequencing
The total DNA was extracted from silica gel-dried materials using the modified CTAB method [45]. The ITS, matK and trnH-psbA primers used for the amplification and sequencing are listed in Xiang et al. [46], and the trnL-F primers are from Taberlet et al. [47]. The selected DNA regions were amplified using a standard polymerase chain reaction (PCR). The sequencing reactions were performed using the ABI Prism Bigdye Terminator Cycle Sequencing Kit (Applied Biosystems, ABI).

Genetic Distance Analyses
To estimate the variation of the ITS, matK and trnL-F sequences across Holcoglossum and related genera, we calculated the pairwise genetic p-distance for each region using MEGA v. 4 [48]. These distances were initially used to evaluate the interspecific divergence with the Kimura 2-Parameter model (K2P). TrnH-psbA was excluded under this analysis because all of the sequences were from Holcoglossum and there was a paucity of related genera.

Phylogenetic Analyses
Clustal X 1.83 [49] was used to obtain an initial alignment of the DNA sequences, followed by manual adjustment using BioEdit [50].
For the MP analyses, heuristic searches were conducted with 1,000 replicates of random addition, with one tree held at each step during the stepwise addition, tree-bisection-reconnection (TBR) branch swapping, MulTrees in effect, and the steepest descent off. All of the characters were unordered and equally weighted, and the gaps were coded as missing data. To access the node support, bootstrap analyses [53] were performed using 1,000 replicates, with 10 random taxon additions and heuristic search options.
Prior to the Bayesian analysis, the Akaike Information Criterion (AIC) implemented in ModelTest version 3.7 [54,55] was used to select the best-fit model of molecular evolution for each dataset. For the BI analyses, four chains of the Markov Chain Monte Carlo (MCMC) were run, sampling one tree every 1,000 generations for 5,000,000, starting with a random tree. Majority rule (.50%) consensus trees were constructed after removing the ''burn-in period'' samples (the first 20% of the sampled trees).

Gross Morphology and Micromorphology
To understand the morphology of Holcoglossum and relative genera, we performed herbarium examination of specimens (Table  S6) and fieldwork observations for gross morphology, and scanning electric microscope (using KYKY-1000B) for the pollinium micromorphology (Table S7). However, the morphological characters of following not Chinese native five species, Aerides krabiensis, A. thibautiana, Holcoglossum calcicola, Jumellea sagittata, Microterangis hariotiana, are from literature.

Character Mapping
To identify the synapomorphies that are congruent with each of the major clades of Holcoglossum retrieved in the molecular tree and to assess the value of the characters used in the classifications, 45 morphological characters were selected. We used the exemplar method, scoring the morphological characters in the morphological matrix for the same species as used in the molecular analyses. The complete morphological matrix, coding 45 characters for the 35 taxa, is available in Table S2. We also prepared a combined morphological and molecular matrix that included the 32 taxa for character mapping.
Character evolution was reconstructed onto a 50% major consensus tree generated in PAUP using the parsimony ancestral state reconstruction in Mesquite v.2.75 [56]. All of the morphological characters were considered unordered and unweighted.

Patrocladistic Analysis
We displayed the patrocladistic analysis according to Stussey and König [57]. This cladogram with high support based on Bayesian inference served as a structure for the patristic distance. The cladistic distances were calculated from the selected cladogram, and these values were placed in a new cladistic matrix. The patristic distance was defined as the number of apomorphic step changes separating two taxa on the cladogram. The patristic distance was then added to the cladistic distance to form the combined patrocladistic data matrix (Table S3). These combined distances are used as input into UPGMA (packages in MEGA) to construct the patrocladogram. Figure S1 Bayesian inference tree of subtribe Aeridinae based on ITS. The bootstrap percentages and posterior probability of .50% are shown above each branch. ''2'' = no value. ''*''represents data from Fan et al. [32], and ''#''represents data from Liu et al. [33]. (JPG) Figure S2 The strict consensus maximum parsimony tree of Holcoglossum s.l. based on the morphological data. The bootstrap percentages of .50% are shown above each branch. (JPG) Figure S3 Character mapping of Holcoglossum s.l. and related genera. See Table S2 for the character numbers and states. (JPG)  Table S5 Taxa and GenBank accession numbers for the ITS, matK, trnL-F and trnH-psbA sequences in phylogenetic analysis of Holcoglossum alliance. A dash indicates missing data; *represent the sequences obtained in this study, and the remaining sequences are from GenBank. (DOC)