To assess taxonomic position of P. rungsuriyanum within the genus Paphiopedilum, we compared the cytological, molecular and morphological data obtained from the representative species of each subgenus in Paphiopedilum according to the report by Gorniak et al. (2014). Furthermore, we investigated the distribution patterns of rDNA signals in P. rungsuriyanum and P. canhii for cytotaxonomic reference. The major significant characters among the subgroups are summarized in Table 2.
Cytological and rDNA FISH analysis
Previous cytological studies on Paphiopedilum species have provided valuable data for cytotaxonomy (Karasawa and Saito 1982). In Paphiopedilum, the diploid chromosome number ranges from 2n = 26 (all metacentric chromosomes) to 2n = 42 [with the conserved arm number (n.f.) = 52]. Species of subgenera Parvisepalum and Brachypetalum, and the three sections of the subgenus Paphiopedilum, i.e. Paphiopedilum (except P. druryi and P. spicerianum with 2n = 30) Coryopedilum and Pardalopetalum possess 2n = 26, while two other sections of the subgenus Paphiopedilum, i.e. Barbata and Cochlopetalum, have the chromosome complement of 2n = 28–42 (n.f. = 52–54) and 2n = 30–37 (n.f. = 50), respectively. In section Cochlopetalum, their common ancestor might lose either two telocentric chromosomes or a single metacentric chromosome before divergence of extant species (Cox et al. 1998). Phylogenetic analyses have indicated that plesiomorphic karyotype for Paphiopedilum possessed 26 metacentric chromosomes with increases in chromosome number accomplished by centric fission (Cox et al. 1997, 1998). In our karyotype analysis, P. rungsuriyanum has the chromosome complement of 2n = 26 (Fig. 3), belonging to the groups with plesiomorphic karyotype. Therefore, we may exclude P. rungsuriyanum as a member of sections Barbata and Cochlopetalum.
In Paphiopedilum, the numbers and distribution patterns of rDNA loci exhibit a considerable diversity that correlates well with phylogenetic lineages and provide important markers for cytotaxonomy (Lan and Albert 2011). The most parsimonious ancestral number of 25S rDNA sites in Paphiopedilum is two, and duplication of 25S rDNA loci could be detected in subgenus Parvisepalum and in sections Coryopedilum and Pardalopetalum of subgenus Paphiopedilum. Massive duplication event of 5S rDNA loci occurred in all five sections of subgenus Paphiopedilum, while the early diverging subgenera, i.e. Parvisepalum and Brachypetalum retained two 5S rDNA sites (Table 2). In this study, both P. rungsuriyanum and P. canhii possess only two 45S rDNA sites and significant duplication of 5S rDNA sites (Fig. 4a, b). In P. rungsuriyanum but not P. canhii, one of the major 5S rDNA signals are closely linked with the 45S array that is similar to the pattern of rDNA signals in section Paphiopedilum. From the rDNA FISH data, we may exclude P. rungsuriyanum as a member of subgenera Parvisepalum and Brachypetalum and suggest a closer relationship to subgenus Paphiopedilum.
Comparative analysis of molecular and morphological data
Paphiopedilum rungsuriyanum and P. canhii are found in the limestone areas in Laos. Although both of them have the miniature plants with tessellated leaves and the chromosome number of 26, their flowers are clearly different and distinct from species in the other subgenera/sections (see Additional file 9: Table S4). A new subgenus Megastaminodium (Braem and Gruss 2011) or a new section Pygmaea (Averyanov et al. 2011) has been proposed to accommodate P. canhii, but it now looks difficult to place P. rungsuriyanum and P. canhii into the same group. The phylogenetic analyses using multiple genes would be helpful in the treatment of systematic position at subgenus/section levels. For the study on the taxonomic position of P. rungsuriyanum, the present phylogenetic analyses are primarily conducted based on the molecular dataset published by Guo et al. (2015). The results are consistent with the previous molecular studies (Chochai et al. 2012; Gorniak et al. 2014), indicating that the well-supported division of the genus Paphiopedilum into three subgenera Parvisepalum, Brachypetalum and Paphiopedilum.
In this study and the previous report by Gorniak et al. (2014), the positions of P. rungsuriyanum and P. canhii are discordant between plastid and nuclear gene trees. On the ITS-based tree (Additional file 4: Figure S1), P. rungsuriyanum is sister to species of the section Paphiopedilum (PP = 0.91), while P. canhii is sister to a clade comprising species of the subgenus Brachypetalum and section Barbata, but without bootstrap support. In the present phylogenetic analyses, P. rungsuriyanum is grouped with P. canhii in the same lineage (PP = 1.00) based on the ACO tree (Additional file 5: Figure S2). On the DEF4-based tree (Additional file 6: Figure S3), P. rungsuriyanum and P. canhii are embedded in the section Paphiopedilum (BP = 84; PP = 0.99). On the RAD51-based tree, P. canhii is embedded in the section Paphiopedilum, while P. rungsuriyanum is sister to the clade comprising species of sections Paphiopedilum, Barbata, Coryopedilum and Pardalopetalum (Additional file 7: Figure S4). Based on the plastid tree (Additional file 8: Figure S5), P. rungsuriyanum is embedded in the clade comprising sections Barbata and Paphiopedilum, while P. canhii is sister to the subgenus Paphiopedilum. According to the analysis from combined data (Fig. 5), both P. rungsuriyanum and P. canhii are sister to the section Paphiopedilum and embedded in the subgenus Paphiopedilum. The incongruence between plastid and nuclear gene trees may be caused by horizontal gene transfer, hybridization, and/or incomplete lineage sorting (Nishimoto et al. 2003; Maddison and Knowles 2006; Kim and Donoghue 2008; Petit and Excoffier 2009; Yu et al. 2013). In Paphiopedilum, based on the multiple low-copy nuclear genes and the network analyses, subgenus Paphiopedilum (particularly sections Barbata, Cochlopetalum and Paphiopedilum) had a higher species diversification rate than the other subgenera of Paphiopedilum, suggesting that hybridization plays an important role in speciation (Guo et al. 2015). Due to the lack of strong interspecific reproductive barriers in Paphiopedilum species, it is proposed that as the geographic and ecological changes (e.g. sea-level fluctuations) disrupted the species boundaries, the interspecific hybridization may lead to the genome introgression across species barriers and contribute to the reticulate evolution in Paphiopedilum (Guo et al. 2015).
In P. canhii and P. rungsuriyanum, their miniature plants with marbled leaves can be easily allied to the taxonomic position associated with species of the sections Parvisepalum (subgenus Parvisepalum) and Barbata (subgenus Paphiopedilum) as suggested by Averyanov et al. (2010). However, from molecular analyses, it is hard to connect P. rungsuriyanum to any species of subgenus Parvisepalum. Species of subgenus Parvisepalum have markedly different floral morphology from those of P. rungsuriyanum, such as the staminode without any umbo, the mammillated stigmatic surface and the granular pollinia (Additional file 9: Table S4). In Paphiopedilum, the staminode morphology provides taxonomically important information for species delimitation (Braem 1988; Cribb 1998). Morphologically, the staminode of P. rungsuriyanum looks intermediate between those of sections Barbata and Paphiopedilum, being half-moon shaped with three lobes and a slight umbo in the middle (Fig. 1c). The staminode of section Barbata is characterized by semi-lunate shape and more or less tri-lobed or tri-dentate, without any umbo. P. rungsuriyanum and species in section Barbata are alike in the staminode. Besides, as compared with other morphological characteristics, such as marbled leaves, single-flowered inflorescence and petal/sepal ratio, we may possibly suggest a close relation of this species with the section Barbata (Additional file 9: Table S4). Nevertheless, the close affinity to section Barbata (forming a clade with both sections Barbata and Paphiopedilum) is only revealed by the plastid analysis with weak support values (Additional file 8: Figure S5). In the analysis of combined data, P. rungsuriyanum is clustered with section Paphiopedilum species with high support values (Fig. 5). Although the floral morphology of P. rungsuriyanum does not resemble those of section Paphiopedilum species, it is noteworthy that section Paphiopedilum is characterized by staminode with a prominent umbo, and P. rungsuriyanum has a slight umbo in the middle of staminode as well. Guo et al. (2015) indicated that the sympatric distribution and the weak interspecific reproductive isolation may have facilitated the interspecific hybridization and led to higher diversification rate in subgenus Paphiopedilum. In Paphiopedilum, thousands of artificial interspecific hybrids have been made between species from different subgenera/sections and registered in the Royal Horticultural Society (http://apps.rhs.org.uk/horticulturaldatabase/orchidregister/orchidregister.asp), and we can observe various staminode morphologies in these artificial interspecific hybrids. Since Indochina is the hotspot of species in sections Barbata and Paphiopedilum, the intermediate staminode morphology of P. rungsuriyanum might be the results from introgression between sections of subgenus Paphiopedilum in the process of hybrid speciation.