Introduction

The flora of Korea, which belongs to the eastern Asiatic floristic region, is exceptionally rich and diverse due to the topographic and climatic complexities of the Korean peninsula (Takhtajan 1986; Park 2005). Approximately 3,954 species and infraspecific taxa of vascular plants in 207 families and 1,048 genera are currently distributed in Korea. Of nearly 4,000 vascular plants, only six genera (all monotypic) are endemic to the Korean peninsula: Mankyua Sun et al. (Ophioglossaceae), Megaleranthis Ohwi (Ranunculaceae), Pentactina Nakai (Rosaceae), Echinosophora Nakai (Fabaceae), Abeliophyllum Nakai (Oleaceae), and Hanabusaya Nakai (Campanulaceae) (Park 2005; Sun et al. 2001). Recent molecular phylogenetic studies shed light on the origin of these six endemic genera in Korea. For example, Sun et al. (2010) demonstrated that Makyua is a well defined genus based on morphology and is sister to the Botrychioid lineage in the molecular phylogenetic study. Kim et al. (2009) reported that Megaleranthis is deeply embedded within Trollius based on ITS and cpDNA matK sequences, questioning its endemic genus status. In addition, Lee et al. (2004) demonstrated that Echinosophora is embedded in the highly polyphyletic genus Sophora and treated it as a species of Sophora (i.e., S. koreensis). In the case of Abeliophyllum, Kim et al. (2000) showed that this endemic genus shares its most recent common ancestor with Forsythia and supported its endemic genus status. Wallander and Albert (2000) also showed a sister relationship between these two genera in the phylogeny of Oleaceae. Finally, Kim et al. (1999) demonstrated that Hanabusaya is sister to monophyletic Adenophora. However, they did not draw any conclusions about the endemic status of Hanabusaya due to very limited sampling of Adenophora (i.e., suspecting paraphyly of Adenophora). In a recent and more comprehensive phylogeny of Campanulaceae, Roquet et al. (2008) demonstrated that Hanabusaya is embedded within Adenophora, making Adenophora paraphyletic. Despite all these efforts to determine the phylogenetic position of the six endemic genera and their generic status in Korea, little is known about Pentactina, and a molecular phylogenetic study to determine its phylogenetic position has never been conducted. This is due, in part, to difficulty in obtaining the species for molecular phylogenetic study.

Pentactina Nakai is a poorly known monotypic endemic genus composed of P. rupicola Nakai (Korean vernacular names: Geum-gang-in-ga-mok or geum-gang-guk-su-na-mu) from the Korean peninsula. It is a highly isolated taxon, restricted to a limited geographical area of Mt. Geumgang (Biro-bong—near the Biro summit: 128°6′18.25″E 38°39′13.25″N; near the Boduck Cave: 128°4′55.43″E 38°38′ 8.08″N) and is currently located in the Gangwon Province of North Korea, where it is restricted to mountain habitats (400–800 m) on rocky cliffs or crevices. It was originally collected and described from the rocky cliffs of Haekeum River in 1917 by Nakai but has never been collected anywhere else other than Geumgang Mountain in North Korea. According to a recent report, this taxon has also been found in two additional nearby regions (e.g., Changdo County, Cujen-Ri; Hoeyang County, Pochon-Ri, both in Gangwon Province) of North Korea (http://nm.nktech.net/index.jsp). Because of its restricted geographic distribution and rarity, P. rupicola has been ranked as a near threatened (NT) and rare species (Anonymous 2008). At present, this taxon is legally protected as National Monument no. 232 in North Korea (http://nm.nktech.net/index.jsp).

Pentactina rupicola is a perennial, leaf-shedding dwarf shrub that reaches a height of up to 70 cm. The stem is slender with pendulous branches. The inflorescences are racemes with white flowers. The flowers of P. rupicola are unique within the tribe Spiraeeae and are characterized by long linear petals (Nakai 1917; Lee 2003, 2007). The generic rank for Pentactina has been controversial since it was established by Nakai (1917). Schulze-Menz (1964) and Takhtajan (1997, 2009) agreed with Nakai’s taxonomic concept and recognized Pentactina as the distinct genus. On the other hand, Hutchinson (1964) and Kalkman (2004) considered this monotypic Korean genus to be a synonym of the genus Spiraea L.

The systematic position of Pentactina within either subfamily Spiraeoideae Agardh or tribe Spiraeeae DC. has only been disputed by one palynological study (Lee et al. 1993). Its monotypic status is supported by pollen morphology. However, these authors suggest that Pentactina is derived from within Spiraea or from a common ancestor of Spiraea and Pentactina.

Its correct placement within the new emerging tribal classification of Spiraeeae within Rosaceae remains untested by recent molecular data (Potter et al. 2007a, b). Potter et al. (2007a, b) were not able to include Pentactina in their studies due to the difficulty of getting material, but they strongly suggested that including P. rupicola in future studies is necessary to establish with certainty the number of genera that should be recognized in Spiraeeae (Potter et al. 2007b). Thus, including Pentactina is important for reconstructing the phylogeny of the tribe Spiraeeae sensu Potter et al. as well as for ascertaining the possible geographical origin of Pentactina.

Recently, one of present authors (S.-P. Hong) with his colleague (H.S. Roh) studied the morphology and anatomy of the tribe Spiraeeae (Roh 2010; Roh and Hong, unpublished data). During the course of study he was able to obtain a sample of P. rupicola from the Royal Botanic Garden, Edinburgh (RBGE) that was originally collected by Ernest H. Wilson (possibly collected in 13–14 October, 1917, acc. to Kim et al. 2010, and also see the details in “Materials and methods”).

In this paper, we conducted a phylogenetic study of the tribe Spiraeeae including monotypic Pentactina to determine its phylogenetic position within the tribe and to evaluate its endemic genus status. We sequenced both the nuclear ITS region and the chloroplast trnL-trnF intergenic spacer of Pentactina and analyzed it along with other members of Spiraeeae (Potter et al. 2007b).

Materials and methods

Plant material

The distribution of Pentactina rupicola Nakai is highly restricted to rocky cliffs or crevices of Biro-bong (near the Biro summit) and near the Boduck Cave of Geumgang Mountain, at Kangwon Province in North Korea. Ernest H. Wilson collected Pentactina in North Korea during the early 1900s, and the only living collection outside of North Korea currently available is at the Rock Garden (R36) in the Royal Botanic Garden Edinburgh (RBGE) (accession number of 19240109; EH Wilson collection number 10247). Some original collections by Wilson from 1917 and 1918 are deposited primarily at the Harvard University Herbarium (GH) (Barcodes 288830–5) and Arnold Arboretum Cultivated Herbarium (A) (Barcodes 00185582–4; these specimens were based on Wilson 10247). It appears that one living collection (EH Wilson collection number 10247) of Pentactina was cultivated at the Arnold Arboretum (but now deceased), and this same collection has been cultivated at the RBGE since 1924. Subsequently, the RBGE holds three specimens (Barcodes E00228832, E00228269, and E00228874) based on the single living material and one duplicate specimen (E00314767) was obtained from Arnold Arboretum Cultivated Herbarium. We were only able to obtain a leaf tissue from one living sample of Pentactina rupicola from RBGE in 2006 to determine its phylogenetic position within Spiraeeae.

DNA isolation, amplification, and sequencing

Total genomic DNA from dried leaf tissue was extracted using the DNeasy Plant Mini Kit (Qiagen, Valencia, CA, USA). The nuclear rDNA ITS region was amplified using primers ITS 6 and ITS 9 developed by Sang-Hun Oh (Potter et al. 2007b). The chloroplast trnL-trnF region was amplified using primers trnc and trnf of Taberlet et al. (1991). We followed the same conditions for PCR amplification and subsequent DNA sequencing as described by Bortiri et al. (2001).

Sequence editing and alignment

Sequence fragments were assembled and edited using Sequencher version 4.2.2 (Gen Codes, Ann Arbor, MI, USA). For cpDNA trnL-trnF intergenic spacer (Genbank accession number, EU595410), the corrected consensus sequence was aligned manually with other members of Spiraeeae using MacClade version 4.06 (Maddison and Maddison 2000) due to its low sequence divergence. For the ITS of nrDNA (Genbank accession number, EU595411), we used ClustalX (Thompson et al. 1997) to align the edited consensus sequence of Pentactina with other members of Spiraeeae due to relatively high sequence divergence. We used several different alignment parameters for gap opening and gap extension penalties for multiple alignment: default (gap opening 15 and gap extension 6.66), gap opening 25 and gap extension 6.66, gap opening 5 and gap extension 6.66, and gap opening 5 and gap extension 3. Subsequent phylogenetic analyses based on these different alignment parameters without further manual adjustment did not change the phylogenetic position of Pentactina, and thus we present the results based on default alignment parameters. For the trnL-trnF intergenic spacer, seven indels were coded as binary characters and polarized based on the outgroup. There were no complex indels in the trnL-trnF intergenic spacer, and thus they were simply coded as presence or absence of indels.

Phylogenetic analysis

In the phylogenetic study of tribe Spiraeeae using the nrDNA ITS sequences and trnL-trnF intergenic spacer sequences (Potter et al. 2007b), two outgroup genera Adenostoma and Gillenia were used. In this study, we included several other outgroups including Chamaebatiaria, Sorbaria, Neillia, and Exochorda (Potter et al. 2002). For the nrDNA ITS region, a total of 39 species (6 outgroup and 33 ingroup species) were analyzed. For the cpDNA trnL-trnF spacer, we included a total of 39 species (7 outgroup and 32 ingroup species). For the combined data set analysis, we included a total 41 species (7 outgroup and 34 ingroup species). Although four species (i.e., Sibiraea croatica, Xerospiraea hartwegiana, Gillenia trifoliata, and Kelseya uniflora) have only either ITS or trnL-trnF sequences (not both), we included them in the combined data set analysis.

Three data sets were analyzed using an equally weighted, unordered maximum parsimony (MP) approach (Fitch 1971) implemented in PAUP* version 4.0 (Swofford 2002). The MP analyses for each data set included a default heuristic search for the most parsimonious trees: starting trees were obtained via stepwise addition. Sequences were added via simple addition with one tree held at each step. Branch swapping was performed via tree-bisection-recombination (TBR), and steepest descent and MulTrees options were in effect. Branches were collapsed if maximum branch length was zero, and topological constraints were not enforced. Support for groups was calculated by bootstrap analysis with 1,000 bootstrap replicates (Felsenstein 1985) with the same heuristic options. Congruence between ITS and trnL-trnF data sets was tested using the incongruence length difference (ILD) test (Farris et al. 1995) as implemented by the partition homogeneity test in PAUP* for 100 replicates (heuristic search, simple addition, TBR branch swapping), each saving a maximum of 1,000 most parsimonious trees per replicate.

Each data set was also analyzed using the maximum likelihood approach to determine the stability of the parsimony results with an explicit model-based approach (Felsenstein 1981). Optimal models of molecular evolution were chosen, using the likelihood ratio test (Goldman 1993) implemented in the program ModelTest, version 3.7 (Posada and Crandall 1998). Model parameters were then imported into PAUP*, and heuristic searches were executed.

Results

ITS of nrDNA data set

The aligned data matrix of ITS consisted of 760 total characters for the default alignment: 367 constant (48.3%), 118 variable but parsimony uninformative (15.5%), and 275 parsimony informative (36.2%) characters. MP analysis found 72 equally most parsimonious trees with a tree length (TL) of 1,008, consistency index (CI) of 0.6319 (0.5745 excluding uninformative characters), retention index (RI) of 0.7455. The strict consensus tree (not shown) was well resolved and showed that Pentactina rupicola is sister to the western North American genus Petrophyton. To find the most appropriate DNA substitution model, the Akaike Information Criterion (AIC) and the Hierarchical Likelihood Ratio Tests (hLRTs) were calculated using ModelTest 3.7 (Posada and Crandall 1998). Basically the ModelTest can give us two analyses: one based on AIC and the other on hLRTs. We selected the AIC, and the model selected based on hLRTs also gave us the same tree topology. Like the MP strict consensus tree, it shows that Pentactina is sister to Petrophyton (Fig. 1). However, this relationship is only weakly supported with a bootstrap value of 55%. This clade (Pentactina-Petrophyton) is in turn sister to the strongly supported (96%) monophyletic genus Spiraea.

Fig. 1
figure 1

ITS maximum likelihood tree of the tribe Spiraeeae including Pentactina. Bootstrap values above 50% are shown

Full size image

trnL-trnF of cpDNA data set

The aligned data matrix of trnL-trnF consisted of 1,116 characters: 896 constant (80.2%), 100 variable but parsimony uninformative (8.9%), and 120 potentially parsimony informative (10.8%). The MP analysis excluding indels recovered 20 equally parsimonious trees with TL of 301, CI of 0.8571 (0.7795 excluding uninformative characters), and RI of 0.9071. The strict consensus tree (not shown) recovers an unresolved grade including Pentactina, Kelseya, Sibriaea, and Petrophyton, that is sister to Spiraeeae. The MP analysis including indels and nucleotide characters also found 20 equally most parsimonious trees with TL of 309, CI of 0.8576 (0.7822 excluding uninformative characters), and RI of 0.9106. The strict consensus tree (not shown) is the same as the one recovered by analyses that exclude indels. The AIC of ModelTest selected “K81uf+G” as the best fit model (−ln L = 3312.1506) for this data set, and subsequent ML analysis recovered one tree (Fig. 2). The ML tree suggests that Pentactina shares the most recent common ancestor with the clade containing Kelseya, Sibiraea, and Petrophyton. This relationship, however, is supported by low bootstrap support of 56%. Like the nrDNA ITS ML tree, the clade containing the four genera (Pentactina-Sibiraea-Petrophyton-Kelseya) is sister to the strongly supported (98%) monophyletic genus Spiraea.

Fig. 2
figure 2

Chloroplast DNA trnL-trnF spacer sequence maximum likelihood tree of the tribe Spiraeeae including Pentactina. Bootstrap values above 50% are shown

Full size image

Combined data set

The partition homogeneity test for the ITS and trnL-trnF indicated that there was no significant incongruence between these two regions (P = 0.63). The aligned data matrix consisted of 1,876 characters: 1,263 constant (67.3%), 218 variable but parsimony uninformative (11.6%), and 395 potentially parsimony informative (21%). The MP analysis found 72 equally most parsimonious trees with TL of 1,319, CI of 0.6785 (0.6063 excluding uninformative), and RI of 0.7793. The strict consensus tree (not shown) shows that Pentactina is sister to Kelseya. The AIC of ModelTest selected “GTR+I+G” as the best fit model (−ln L = 9561.8750) and ML analysis found one tree (Fig. 3). The ML tree suggests that Pentactina is sister to Petrophyton, which occurs exclusively in western North America. This relationship, however, is weakly supported (<50% bootstrap value). This clade is in turn sister to the clade containing Kelseya and Spiraea.

Fig. 3
figure 3

Maximum likelihood tree based on the combined nrDNA ITS and cpDNA trnL-trnF data set. Bootstrap values above 50% are shown

Full size image

Discussion

The present study represents the first attempt to place Pentactina within the phylogeny of Spiraeeae. This was accomplished by sampling the living plant material, which, outside its limited native range, is only found in the collections at RBGE (see “Materials and methods”). The current study suggested that Pentactina is closely related to other small genera, such as Sibiraea, Kelseya, and Petrophyton from the Northern Hemisphere. The tree topology inferred in the present study based on combined nrDNA ITS and cpDNA trnL-trnF spacer sequences (Fig. 3) is identical to previous studies (see Fig. 1 in Potter et al. 2007b) with the exception that we also include Pentactina. Thus, we feel that the phylogenetic position of Pentactina within the Spiraeeae is stable and meaningful.

Kelseya Rydb. is a monospecific genus [i.e., K. uniflora (Wats.) Rydb] of cushion-forming shrublets with solitary pink flowers and simple entire leaves. This species is found on limestone in restricted areas of the mountains of Montana, Wyoming, and Idaho in the United States. Petrophyton includes four prostrate shrubby species that occur exclusively in western North America (WNA). Petrophyton species have simple serrate leaves; flowers are small, white to light pink with five petals, and they have long stamens that are arranged in dense clusters on spike-like racemes. Sibiraea comprises five erect shrub species with simple entire leaves and dense spike-like white flowers in panicles. Members of this genus are distributed in southeastern Europe (Croatia and Bosnia), Russia (Siberia), Kazakhstan, and western Asia. Pentactina, an erect shrub with simple serrate leaves and with raceme inflorescences can be distinguished from Kelseya by the latter possessing biternate leaves and solitary flowers. Pentactina shares a similar habit (erect shrub) and leaves (simple entire) with Sibiraea, but the inflorescence (raceme) of Pentactina is more similar to Petrophyton. The nuclear phylogeny (Fig. 1) shows that Pentactina shares its most recent common ancestor with Petrophyton, whereas the cpDNA phylogeny (Fig. 2) places Pentactina as sister to a clade comprising Kelseya, Sibiraea, and Petrophyton. Our analyses of the combined nrDNA-cpDNA data set analysis (Fig. 3) resolve a clade including Pentactina and Petrophyton, but with less than 50% bootstrap support. Therefore, based on the present molecular sequence data, we suggest that Pentactina possibly has a close relationship with the small WNA genus Petrophyton, although receiving a weak support (BS < 50%). However, the gene regions used did not fully resolve the possible insights into the relationships within the tribe Spiraeeae. In addition, our current study suggests that Pentactina is not part of the large genus Spiraea as previously hypothesized by Hutchinson. Further studies utilizing gene regions with higher rates of nucleotide substitutions as well as various nonmolecular data are needed to infer with more precision the evolutionary relationships of this rare Korean endemic taxon, Pentactina in the tribe Spiraeeae.

Potter et al. (2007b) suggested, based on parsimony-based character reconstructions, that the common ancestor of Spiraeeae occurred in WNA, with independent migrations to the Old World by Aruncus, Sibiraea, and Spiraea. Based on our current study, it is plausible that Pentactina also independently migrated to Korea from the common ancestor with Petrophyton in WNA. The origin of Spiraea can be dated to at least the Eocene (56.5–35.4 mya) based on the fossil records and molecular dating (Zhang et al. 2006). Since the four genera, Sibiraea, Pentactina, Petrophyton, and Kelseya diverged earlier than the origin of Spiraea (i.e., the branch lengths between these genera and Spiraea are much longer), it is plausible that the origin of Pentactina is much older than the Eocene, likely during the early Tertiary (the Paleocene) or even earlier (Cretaceous). Given the current narrow, limited distribution of Pentactina, Petrophyton, Kelseya, and Sibiraea, we cannot rule out the possibility that the common ancestors of those genera were widely distributed in the northern hemisphere during the Late Cretaceous and early Tertiary and became extinct after the origin of those reduced growth form genera within the Spiraeeae.

A recent biogeographic investigation of seven diverse genera (e.g., Cornus, Boykinia, Tiarella, Trautvetteria, Aralia, Calycanthus, and Adiantum) by Xiang et al. (1998) attempted to identify a general model of relationships for several disjunct genera with distributions in East Asia (EA) and both eastern and western North America (ENA + WNA). Of the seven taxa studied by Xiang et al. (1998), Boykinia and Calycanthus are most similar to the Pentactina/Petrophyton disjunction pattern. The other case of EA-WNA disjunction is in Ribes (Schultheis and Donoghue 2004) and Thuja (Li and Xiang 2005). As summarized by Donoghue and Smith (2004), EA-WNA disjunction is much rarer than a EA-ENA disjunction. Therefore, the present case for Pentactina/Petrophyton disjunction pattern provides an additional opportunity to understand better the EA-WNA disjunction.

Finally, our present study provides one significant implication with regard to the number of endemic genera in Korea. Currently, we recognize six monotypic endemic genera (Sun et al. 2001; Park 2005). Based on the earlier and current molecular phylogenetic studies, it appears that only three genera (Abeliophyllum, Pentactina, and Mankyua) should be recognized as distinct endemic genera in Korea. Both Megaleranthis and Echinosophora were deeply embedded within Trollius and Sophora, respectively, and thus, endemic genus status of these two genera is questionable. The validity of Hanabusaya as a distinct endemic genus is at least questionable based on the large scale phylogeny of Campanulaceae (Roquet et al. 2008). Future detailed molecular phylogenetic studies of the Korean endemic genera will provide more information about the number of endemic genera in the Korean peninsula.