ISSN 1346-7565 Acta Phytotax. Geobot. 70 (2): 103–118 (2019) doi: 10.18942/apg.201822

Taxonomical Review of ser. Verticillatae () in Taiwan

1,* 2 3 Kaori Murayama , Richard H. Ree , Kuo-Fang Chung , 4 5 Chih-Chieh Yu , Noriyuki Fujii

1Graduate School of Science and Technology, Kumamoto University, 2-39-1 Kurokami, Chuo-ku, Kumamoto 860-8555, Japan. * [email protected] (author for correspondence); 2Integrative Research Center, Field Museum, 1400 S Lake Shore Drive, Chicago, IL 60605, USA; 3Biodiversity Research Center, Academia Sinica, Nangang, Taipei 115, Taiwan; 4School of Forestry and Resource Conservation, National Taiwan University, Taipei 23652, Taiwan; 5Faculty of Advanced Science and Technology, Kumamoto University, 2-39-1 Kurokami, Chuo-ku, Kumamoto 860-8555, Japan

To classify the species of Pedicularis ser. Verticillatae Maxim. (Orobanchaceae) in Taiwan, phylogeo- graphic analyses using internal transcribed spacer (ITS) regions of nuclear ribosomal DNA (nrDNA) and trnK (partial) regions of chloroplast DNA (cpDNA) were performed. It was determined that the Taiwan- ese populations formed a single strongly-supported clade with P. refracta (Maxim.) Maxim., an endem- ic of Kyushu, Japan. Little genetic variation between the Taiwanese populations and P. refracta was de- tected. Morphological analyses showed that the Taiwanese populations and P. refracta differed in three key characteristics: corolla length, distance between the rachis and the sinuses of the pinnae, and height. It was concluded that the Taiwanese populations should be treated as a variety of P. refracta (Ky- ushu populations); P. refracta var. transmorrisonensis (Hayata) Hurus. (lectotypified here). In addition, the phylogeographic analyses suggested that the disjunct distribution pattern between Taiwan and Ky- ushu was due to Quaternary vicariance events rather than to long distance dispersal of .

Key words: disjunct distribution, eastern Asia, Kyushu, morphology, Pedicularis refracta, Pedicularis verticillata, phylogeny, Taiwan, taxonomical review

In recent decades, as a result of improvements nomic classifications. The related findings are in molecular techniques, it has become possible important for understanding the biogeographic to accurately estimate interpopulation and inter- origins and evolutionary histories of local floras. species phylogenetic relationships of various or- Eastern Asia has a relatively rich flora and ganisms (Avise 2004). Methods of inferring phy- many endemic species because of its complicated logenetic trees have also developed rapidly (Soltis geographical history, temperate climate and high et al. 1998, Salemi & Vandamme 2003, Felsen- precipitation (Qiu et al. 2011). The islands in this stein 2004). Consequently, taxonomic research region lie on the boundary of the continent and and reexamination of various plant groups has in- have been connected to and separated from the creased accordingly (Dowie & Palmer 1992, mainland several times (Liew & Chung 2001). Johnson & Soltis 1994, Nakamura et al. 2012, Because of this complicated geologic history, Ja- Fior et al. 2013, Fujii et al. 2014). Analyses using pan and Taiwan along the eastern boundary of molecular makers have revealed the existence of Asia, although small in area, exhibit relatively cryptic species that are not identifiable by mor- high species diversity and many endemic . phological analysis (Liu et al. 2011). Delimitation The flora of these islands has been well investi- of taxonomic boundaries is important because it gated and documented in the literature (Japan; can prompt the reexamination of previous taxo- Ohwi 1965, Iwatsuki et al. 1993, 1995a, 1995b, 104 Acta Phytotax. Geobot. Vol. 70

found misidentification in populations of Pe- dicularis in Hakusan in central Honshu, Japan, through molecular phylogenetic and morpho- logical analysis. The plants were not P. verticil- lata L. but P. spicata Pall. This revealed the disjunct distribution of P. spicata of ser. Verti- cillatae, between the Asian continent and Ja- pan. The findings suggest that examination of the species of Pedicularis ser. Verticillatae has not been sufficient. In Taiwan, a single species of Pedicularis ser. Verticillatae, for which three scientific names have been proposed, has been identified (referred to here as the Taiwanese populations, Fig. 1). It was first described as P. transmorri- sonensis Hayata (Hayata 1915, Limpricht 1924, Tsoong 1963, Yang et al. 1998), a species en- demic to Taiwan. Hurusawa (1948) recognized it as a variety of an endemic Japanese species, P. refracta (Maxim.) Maxim. var. transmorri- sonensis (Hayata) Hurus. It was then identified as P. verticillata by Taiwanese botanists (Li 1978, Liu 1998), a species widely distributed in the arctic-alpine regions of the northern hemi- sphere. Thus, a consensus regarding the taxo- Fig. 1. Populations of Pedicularis refracta var. transmorriso- nensis in their natural habitat on Nanhu Shan, Taiwan (A), nomic identity of the Taiwanese plants has not and a close up of a (B). yet been reached. This paper aims to clarify the identity of the Taiwanese populations based on molecular 1999, 2001, 2006, and Taiwan; Huang 1993, 1994, phylogenetic analyses of Pedicularis ser. Verti- 1996, 1998, 2000, 2003). However, in recent years, cillatae in eastern Asia and morphological many more taxa and geographic occurrences have comparisons among the Taiwanese popula- been reported in Japan and Taiwan (Nakamura et tions, P. refracta and P. verticillata. We also al. 2012, Tanaka 2013, Yu & Chung 2014, Shiga & present a phylogeographic discussion of the Kadono 2015), indicating a need for further taxo- disjunct distribution pattern among the islands nomic study in the region. An accurate understand- on the eastern edge of Asia. ing of the regional flora is therefore important for exploring its establishment. Pedicularis ser. Verticillatae Maxim. (Oro- Materials and Methods banchaceae) includes 37 species that are widely dis- tributed throughout Europe, Asia, including Japan Materials and Taiwan, and northwestern North America. The We analyzed the nuclear ribosomal DNA areas with the greatest diversity are in eastern Asia. (nrDNA) of the internal transcribed spacer China alone is home to 30 species (Yang et al. (ITS) region of 76 individuals from 33 popula- 1998). Although various species are reported for tions, including several Taiwanese populations this region, reexamination of the of the from different areas (Hehuan Shan, Nanhu species is scarce. For example, Fujii et al. (2014) Shan, and Xue Shan) and six species (Pedicu- June 2019 Murayama & al. – Taxonomy of Pedicularis in Taiwan 105

Kokonoe Minami-Oguni Korean Peninsula

Myougahara Hijyudai Takamori Gokase-cho Nishi-Otsuka Yadake Kyushu Japan (P. refracta) China Hehuan Shan Nanhu Shan

Ryukyu Islands Xue Shan

200 km Taiwan (Taiwanese populations)

Fig. 2. Map of eastern Asia and adjacent area, including Taiwan and Japan. Black dots indicate Taiwanese population and Pedicularis refracta sampling sites. laris verticillata, Pedicularis refracta, P. spicataFIG. 2 dicularis by Ree (2005) and Tkach et al. (2014). Pall., P. szetschuanica Maxim., P. metaszetsch- To compare the morphological characteristics uanica Tsoong, and P. lineata Franch. ex Max- of the Taiwanese populations, we integrated the im.) considered to be related to them based on morphological data obtained during our field col- previous phylogenetic studies by Ree (2005), lection of the Taiwanese populations and P. re- Yang & Wang (2007), Eaton et al. (2012), Fujii et fracta and from herbarium specimens of P. verti- al. (2014), and Tkach et al. (2014) (Table 1). The cillata. During the field research, we collected nucleotide sequences of 26 individuals from 20 of and leaves from each individual of each these populations were taken from the DDBJ/ population. In total, we sampled 131 individuals EMBL/GenBank databases. The remaining sam- from the three Taiwanese populations (Hehuan ples were newly collected and analyzed in the Shan, Nanhu Shan, and Xue Shan) and 142 indi- present study. For the phylogenetic analysis of the viduals from the four P. refracta populations chloroplast DNA (cpDNA) using the trnK (par- [Aso (Minami–Oguni), Aso (Takamori), Hitoyo- tial) region (includes part of trnK 5’ intron and shi, and Miyazaki]. The Kokonoe and Hijyudai part of coding region of matK), we used 37 indi- populations were excluded from the morphologi- viduals from 20 populations, including the three cal analysis because the number of individuals in Taiwanese populations and the six species listed those populations was small. Voucher specimens above. The nucleotide sequences of 13 individu- were deposited in the herbarium at the Faculty of als from 12 of these populations were taken from Science, Kumamoto University (KUMA). For the studies by Ree (2005), Eaton et al. (2012), Fujii et herbarium survey, we examined 127 specimens al. (2014), and Tkach et al. (2014). The remaining of P. verticillata, collected mainly around eastern samples were newly collected and analyzed in the Asia, at Kanazawa University (KANA) and the present study. For phylogenetic analyses using University of Tokyo (TI). the ITS and trnK (partial) regions, we selected three outgroup species (P. amoena Adams ex. Methods Steven, P. pycnantha Boiss. and P. chamissonis Leaves dried using silica gel were powdered Steven), based on the phylogenetic studies of Pe- in a mixer (Multi-Beads Shocker MB455U, Yasui 106 Acta Phytotax. Geobot. Vol. 70

Table 1. Materials and their sources, analyzed for the cpDNA and nrDNA of Pedicularis ser. Verticillata. N1 Accession Nos. Species or population and locality Collecter, voucher or reference ITS trnK(partial) ITS trnK(partial) Taiwanese population Hehuan Shan, Yilan County, Taiwan N. Fujii & K. Murayama F03373 (KUMA) 9 6 LC377615 LC377628 Nanhu Shan, Yilan County, Taiwan N. Fujii & K. Murayama F03366 (KUMA) 15 4 LC377616 LC377629 Xue Shan, Miaoli County, Taiwan N. Fujii F03394 (KUMA) 6 5 LC377617 LC377630 Pedicularis refracta (Maxim.) Maxim. Takamori, Aso, Kumamoto Pref., Kyushu, M. Teramoto & N. Fujii T2 (KUMA), Fujii 1 1 AB754823 AB754828 Japan et al. (2014) Minami-Oguni, Aso, Kumamoto Pref., N. Fujii & K. Murayama F03406 (KUMA) 4 2 LC377618 LC377631 Kyushu, Japan Myougahara, Aso, Kumamoto Pref., N. Fujii May 13, 2007 (KUMA) 1 - LC377619 - Kyushu, Japan Yadake, Hitoyoshi, Kumamoto Pref., N. Fujii & K. Murayama F03416 (KUMA) 6 3 LC377620 LC377632 Kyushu, Japan Nishi-Otsuka, Hitoyoshi, Kumamoto N. Fujii & K. Murayama F03413 (KUMA), 2 2 AB754824 AB754829 Pref., Kyushu, Japan Fujii et al. (2014) Gokase-cho, Nishiusuki, Miyazaki Pref., N. Fujii & K. Murayama F03429 (KUMA) 2 2 LC377621 LC377633 Kyushu, Japan Kokonoe, Kusu, Oita Pref., Kyushu, Japan M. Teramoto & N. Fujii F02551 (KUMA), 1 1 AB754823 AB754828 [=T018, Fujii et al. (2014)] Hijyudai, Kusu, Oita Pref., Kyushu, Japan M. Teramoto & N. Fujii F02557(KUMA) 1 - LC377622 - Pedicularis verticillata L. Taisetsusan, Hokkaido, Japan N. Fujii July 22, 1995 (KUMA)[=NF151, 1 1 AY949698 AY949762 Ree (2005)] Shibutsusan, Gunmma Pref., Honshu, H. Yoshii F00092 (KUMA), Fujii et al. 1 - AB754820 - Japan (2014) Kitadake, Nagano Pref., Honshu, Japan N. Shirai F02523 (KUMA) [=N. Fujii 1 - AB754819 - 198987, Fujii et al. (2014)] Yatsugadake, Nagano Pref., Honshu, Japan N. Fujii F00655 (KUMA), Fujii et al. (2014) 1 1 AB754819 AB754826 Tolbachik, Kamchatka Peninsula, Russia S. Okitsu 195143 (KUMA), Fujii et al. 1 - AB754821 - (2014) Jiuzhi, Qinghai, China Boufford et al. 39306 (A), Fujii et al. (2014) 1 1 AB754822 AB754827 Garze, W. Sichuan, China YANG0121, Yang & Wang (2007) 1 - AY155291 - Ivano-Frankivsk region, Nadvirna O. Peregrym (no voucher data) 1 - LC377623 - district, Ukraine Switzerland RN, 15 July 1999 (A), Ree (2005) 2 1 AY949706 AY949769 Seward Peninsula, Alaska, USA G. Mansion & A. Guggisberg alask03-2 1 - LC377624 - (SZU) Nome, Alaska, USA K. Marr et al. 12-0077 (V), 12-0118 (V) 2 - LC377625 - 1N = number of analyzed samples

Kikai Corp, Osaka, Japan) and total genomic in a total volume of 50 µL. The PCR program ran DNA was extracted from 0.01 g of the dried pow- for 5 min at 94 oC for initial denaturalization, fol- der using a slightly modified version of the cetyl- lowed by 30 cycles of denaturation at 94 oC for 1 trimethyl ammonium bromide (CTAB) method min, primer annealing at 50 oC for 1 min and ex- described by Doyle & Doyle (1987). Polymerase tension at 72 oC for 2 min. The reactions were chain reaction (PCR) was performed for the ITS then extended by 7 min at 72 oC. After DNA am- and trnK (partial) regions using the following plification was confirmed, we used the Illustra primers: ITS; ITS4 and ITS5 (White et al. 1990), ExoProStar (GE Healthcare, Tokyo) to exclude trnK (partial); trnK11 and matK510R (Young et the extra primer and nucleotide. For sequencing al. 1999). The PCR reaction mixtures contained analyses, the same primers were used (final con- 50–100 ng template DNA, 5 µL 10×PCR buffer, 4 centration 9.6 pM). We employed Eurofins Ge- µL 2.5 mM of each deoxyribonucleotide, 2.5 µL nomics Corp. (Tokyo, Japan) analyze the DNA 0.5 µM of each of the primer pairs, and 0.25 µL sequences. The following additional primers ExTaq DNA polymerase (Takara Bio Inc., Tokyo) were used to read entire regions: ITS; ITS2 and June 2019 Murayama & al. – Taxonomy of Pedicularis in Taiwan 107

Table 1. Continued. N1 Accession Nos. Species or population and locality Collecter, voucher or reference ITS trnK(partial) ITS trnK(partial) Pedicularis spicata Pall. Shakasindo, Hakusan, Ishikawa Pref., N. Shirai Sept. 20, 2009 (KANA), 4 1 AB754817 AB754825 Honshu, Japan [=F12115, Fujii et al. (2014)] Sannomine, Hakusan, Ono-gun, Fukui T. Shimizu 176148 (KANA), Fujii et al. 1 - AB754816 - Pref., Honshu, Japan (2014) Mentougao, W. Beijing, China YANG0401, Yang & Wang (2007) 1 - AY881148 - Primorskij-region, Russia K. Ueda & C. Suyama 1575a, 1575b 2 - AB754818 - (KANA), Fujii et al. (2014) Pedicularis szetschuanica Maxim. Rangtang (Zamtang), Sichuan, China Boufford et al. 39167 (A), Tkach et al. 1 - HG424161 HG423980 (2014) Xiaojin, Sichuan, China Boufford et al. 38479 (A), Tkach et al. 1 1 JN252802 JN252960 (2014) Xiaojin, Sichuan, China Boufford et al. 38407 (A) 1 1 LC377626 LC377634 Pedicularis metaszetschuanica Tsoong Maerkang (Barkam), Sichuan, China Boufford et al. 39853 (A), Eaton et al. 1 1 JN252773 JN252932 (2012) Hongyan, Sichuan, China Boufford et al. 39973 (A) 1 1 LC377627 LC377635 Maerkang (Barkam), Sichuan, China Boufford et al. 27898 (A), Ree (2005) 1 1 AY949665 AY949735 Pedicularis lineata Franch. ex Maxim. Maerkang (Barkam), Sichuan, China Boufford et al. 27942 (A), Ree (2005) 1 1 AY949664 AY949734

Outgroups Pedicularis amoena Adams ex. Steven Altayskiy kray, Russia A. Tribsch & F. Essl 10090 (SZU), Tkach et 1 1 HG424070 HG423888 al. (2014) Pedicularis pycnantha Boiss. Kopet-Dag, Turkmenistan I. A. Gubanov s.n. (LE), Tkach et al. (2014) 1 1 HG424188 HG424007 Pedicularis chamissonis Steven Kitadake, Nagano Pref., Honshu, Japan N. Fujii, Aug. 22, 1995 (KANA) [=NF919, 1 1 AY949631 AY949709 Ree (2005)] 1N = number of analyzed samples

ITS3 (White et al. 1990), trnK (partial); matK- tions/deletions (indels) and missing data were 53F (Young et al. 1999), trnK600R and mat- treated as complete deletions. Clade support was K840R (Fujii et al. 2014). The nucleotide se- estimated using bootstrap analysis (Felsenstein quences obtained were edited using ChromasPro 1985) based on 10,000 replicates. In the MP anal- v.1.7.6 (www.technelysium.com.au/ChromasPro. ysis, trees were constructed solely based on the html) and aligned using MEGA v.6.06 (Tamura et substitution data using the branch and bound al. 2013). When double peaks were observed in search option with MulTrees on. The additional the nucleotide sequences of the ITS region we ed- sequence option was set to farthest. Clade sup- ited the uncertain sequence to “N” and treated it port was estimated using bootstrap analysis based as missing data. Phylogenetic analyses were con- on 1,000 replicates in PAUP* using the fast step- ducted with the neighbor-joining (NJ) method us- wise-addition option as the type of bootstrap ing MEGA v.6.06 and the maximum parsimony search. (MP) method using PAUP* v.4.0b10 (Swofford In the morphological analysis, we measured 2003). For NJ analysis, best-fit substitution mod- and compared nine characters to identify differ- els were determined using a model test in MEGA ences among the Taiwanese populations, Pedicu- with Akaike’s information criterion (Akaike laris verticillata and P. refracta: (a) galea apex 1974). The best-fit model for the present ITS data angle (°) (upper lip of the corolla), (b) corolla was the Kimura two-parameter model (Kimura length (mm) (excluding the calyx tube), (c) leaf 1980), and that for the cpDNA data was the Tamu- blade length (mm), (d) distance between the ra- ra three-parameter model (Tamura 1992). Inser- chis and the longest pinna (mm), (e) distance be- 108 Acta Phytotax. Geobot. Vol. 70

Fig. 3. Floral and leaf blade characters measured for determination of variance. a: galea apex angle, b: corolla length (excluding calyx tube), c: leaf blade length, d: distance between the rachis and the longest pinna, e: distance between the rachis and the sinuses of the pinna. In addition, we calculated the ratios of charac- teristics c/d (f) and e/d (g) and measured the number of pinnae (h) and aboveground plant height (i). Flower and leaf in this figure are from plants from Nanhu Shan, Taiwan (Table 1).

Pedicularis refracta (Aso, Hitoyoshi, Miyazaki, Oita), Taiwanese populations (Hehuan Shan, Nanhu Shan, Xue Shan) 91 / 83 P. spicata Beijing, China 99 / 99 P. spicata Primorskij, Russia

88 / 78 P. spicata Hakusan, Japan 69 / 75 P. spicata Hakusan, Japan 55 / - 60 / - P. verticillata Kitadake, Yatsugadake, Japan 75 / 87 P. verticillata Shibutsusan, Japan

87 / 92 P. verticillata Alaska, Switzerland, Ukraine P. verticillata Taisetsusan, Japan 86 / - 99 / 98 67 / 65 P. verticillata Tolbachik, Russia

99 / 100 P. verticillata Garze, China P. verticillata Jiuzhi, China P. metaszetscuanica Hongyan, China 100 / 95 P. lineata Maerkang, China P. metaszetscuanica Maerkang, China 58 / - 16 / - P. szetscuanica Rangtang, China 7 / - P. metaszetscuanica Maerkang China

36 / - P. szetscuanica Xiaojin, China P. szetscuanica Xiaojin, China P. pycnantha Turkmenistan 47 / 53 0.01 P. amoena Russia P. chamissonis Japan

Fig. 4. Neighbor-joining (NJ) tree produced using the internal transcribed spacer (ITS) regions of nuclear ribosomal DNA from Pedicularis refracta vars. refracta and transmorrisonensis (Taiwanese populations), P. verticillata and other related species. Genetic distances were calculated basedFIG. on Kimura 4 ITS two-parameter tree model. Maximum parsimony (MP) tree shows fundamentally same topology as NJ tree. Numbers along branches indicate bootstrap values (%); the former are based on NJ analysis, the latter are based on MP analysis. June 2019 Murayama & al. – Taxonomy of Pedicularis in Taiwan 109 tween the rachis and the sinuses of the pinna the measurements (a–e, h, i) and produced a scat- (mm), (f) ratio of leaf blade length to distance be- ter diagram showing component 1 on the x-axis tween the rachis and the longest pinna (c/d; leaf and component 2 on the y-axis. Eigenvalues > 0.5 shape), (g) ratio of distance between the rachis were used as the criteria for extracting compo- and the longest pinna to distance between the ra- nents. chis and the sinuses of the pinna (e/d; the degree of dissection of the leaf pinnae), (h) number of pinnae, and (i) aboveground plant height (cm) Results (Fig. 3). The measurements, excluding leaf blade length (c) and aboveground plant height (i), were DNA sequence variation taken using a portable digital scope (Handy The aligned ITS dataset was 559 bp in length Scope 130s, ScienceEye, Saitama) with DinoCap- (including outgroups); 106 bp of variable sites ture 2.0 v.1.3.5 software (AnMo Electronics (including 96 bp of nucleotide substitutions and Corp., Taipei). A vernier caliper was used to mea- 10 bp of 5 indels) were observed in all accessions sure leaf blade length (c) and a ruler was used to and 75 bp (including 66 bp of nucleotide substitu- measure the aboveground plant height (i). JMP tions and 9 bp of 4 indels) were observed in in- v.11.0.0 software (SAS Institute Inc., Cary, NC, group accessions. There were no sequence differ- USA) was used for one-way analysis of variance ences between the Taiwanese populations and (ANOVA) and Tukey-Kramer HSD tests were Pedicularis refracta. The calculated value of used to assess differences between populations pairwise uncorrected p-distance (Nei & Kumar and/or species. Furthermore, we performed prin- 2000) between the accessions of the Taiwanese cipal component analysis (PCA) using some of populations and P. refracta was 0.0%. The uncor-

95 / 93 Taiwanese populations (Hehuan Shan, Nanhu Shan, Xue Shan) Pedicularis refracta (Aso, Hitoyoshi, Miyazaki, Oita) P. metaszetschuanica Hongyan, China

65 / 64 P. metaszetschuanica Maerkang, China P. metaszetschuanica Maerkang, China 30/ 72 60 / - P. lineata Maerkang, China P. verticillata Yatsugatake, Japan 99 / 100 64 / 66 P. verticillata Switzerland P. verticillata Taisetsusan, Japan

39 / - P. verticillata Jiuzhi, China P. szetschuanica Xiaojin, China 59 / 62 P. szetschuanica Xiaojin, China 53 / - P. szetschuanica Rangtang, China P. spicata Hakusan, Japan P. chamissonis Japan 80 / - 99 / 98 P. pycnantha Turkmenistan 0.002 P. amoena Russia

Fig. 5. Neighbor-joining (NJ) tree produced using trnK (partial) regions of chloroplast DNA from Pedicularis refracta vars. refracta and transmorrisonensis (Taiwanese populations), P. verticillata and related species. Genetic distances were cal- culated based on Tamura three-parameter model. Maximum parsimony (MP) tree had fundamentally the same topology as the NJ tree. Numbers along branches indicate bootstrap values (%); the former are based on NJ analysis, while the lat- ter are based on MP analysis. FIG.5 trnK(partial) tree 110 Acta Phytotax. Geobot. Vol. 70

Table 2. Analysis of variance of morphological characters between the Taiwanese populations, Pedicularis verticillata, and P. refracta. Taiwanese populations P. refracta P. verticil- Aso Characters Hehuan 1 Aso Nanhu Shan Xue Shan lata (Minami- Hitoyoshi Miyazaki Shan (Takamori) Oguni) a. galea apex 63.0 ± D3 67.7 ± CD 70.3 ± BC 83.9 ± A 73.8 ± B 70.0 ± BC 68.3 ± BCD 67.4 ± CD angle (°) 5.352 5.89 7.04 11.3 6.48 7.68 5.74 3.96 (N = 31) (N = 41) (N = 60) (N = 95) (N = 30) (N = 43) (N = 38) (N = 31) b. corolla length 8.7 ± C 9.2 ± C 8.7 ± C 11.0 ± B 13.7 ± A 13.4 ± A 14.1 ± A 13.0 ± A (mm) 0.92 0.97 0.94 1.32 1.04 2.43 2.96 0.70 (N = 31) (N = 41) (N = 60) (N = 92) (N = 30) (N = 43) (N = 38) (N = 31) c. leaf blade 32.0 ± B 31.8 ± B 26.9 ± B 20.0 ± C 32.2 ± B 29.0 ± B 43.1 ± A 30.8 ± B length (mm) 2.29 1.59 7.01 7.71 6.51 6.34 13.25 8.54 (N = 31) (N = 41) (N = 60) (N = 66) (N = 30) (N = 37) (N = 38) (N = 31) d. distance 7.1 ± 7.0 ± 5.7 ± 4.4 ± 7.9 ± 6.9 ± 10.4 ± 7.1 ± B B C D B BC A B between the 1.97 2.13 1.31 1.79 1.79 1.51 3.27 1.57 rachis and the (N = 31) (N = 41) (N = 60) (N = 66) (N = 30) (N = 37) (N = 38) (N = 31) longest pinna (mm) e. distance 1.0 ± CD 0.9 ± CD 0.9 ± CD 0.81 ± D 2.0 ± A 1.4 ± B 1.9 ± A 1.1 ± BC between the 0.34 0.36 0.48 0.42 0.54 0.44 0.73 0.28 rachis and the (N = 31) (N = 41) (N = 60) (N = 66) (N = 30) (N = 37) (N = 38) (N = 31) sinuses of the pinnae (mm) f. (c) / (d)4 4.4 ± A 4.6 ± A 4.8 ± A 4.7 ± A 4.2 ± A 4.3 ± A 4.4 ± A 4.4 ± A 0.81 0.87 0.83 1.28 0.83 0.93 0.99 0.96 (N = 31) (N = 41) (N = 60) (N = 66) (N = 30) (N = 37) (N = 38) (N = 31) g. (e) / (d)5 0.141 ± C 0.132 ± C 0.162 ± BC 0.186 ± B 0.261 ± A 0.198 ± B 0.192 ± B 0.166 ± BC 0.040 0.042 0.080 0.059 0.051 0.049 0.069 0.042 (N = 31) (N = 41) (N = 60) (N = 72) (N = 30) (N = 37) (N = 38) (N = 31) h. number of 15.8 ± BC 17.5 ± AB 17.8 ± A 14.2 ± C 15.7 ± BC 14.9 ± C 19.2 ± A 18.1 ± A pinnae 1.86 2.34 3.10 4.42 1.27 2.02 2.28 1.63 (N = 31) (N = 41) (N = 60) (N =126) (N = 30) (N = 37) (N = 38) (N = 31) i. aboveground 20.2 ± CD 20.6 ± C 14.0 ± E 12.1 ± E 28.3 ± B 15.3 ± DE 37.3 ± A 28.7 ± B plant height (cm) 9.13 8.65 4.85 5.44 7.38 2.70 8.93 8.18 (N = 31) (N = 41) (N = 60) (N = 72) (N = 30) (N = 30) (N = 38) (N = 31) 1We examined 127 specimens of P. verticillata, collected mainly around East Asia, at Kanaza University (KANA) and the University of Tokyo (TI). 2Average ± standared error (N = number of samples). 3Alphabetical characters, A to E, indicate the result of the multiple comparison test (Tukey-Kramer HSD test) (p < 0.05). 4Ratio of leaf blade length to ditance between the rachis and the longest pinna. 5Ratio of distance between the rachis and the longest pinna to distance between the rachis and the sinuses of the pinna. rected p-distance values ranged from 0.4 to 1.7% the P. verticillata accessions (Table 4). in the accessions of Pedicularis verticillata and 0.2–1.3% in the accessions of P. spicata (Table 4). Phylogenetic analysis The aligned trnK (partial) region of the cpD- In the NJ tree of the ITS region, the acces- NA dataset was 1,055 bp in length (including out- sions of the Taiwanese populations were not in groups). There were 103 bp of variable sites (in- the same clade as those of Pedicularis verticillata cluding 72 bp of nucleotide substitutions and 31 (Fig. 4). As mentioned above, the nucleotide se- bp of 9 indels) in all accessions and 78 bp (includ- quences of the accessions of P. refracta and the ing 48 bp of nucleotide substitutions and 30 bp of Taiwanese populations were identical. Both se- 5 indels) were observed in ingroup accessions. quences were therefore treated as a single acces- The Taiwanese populations and P. refracta were sion. Moreover, the clade of P. refracta and the distinguished by a single nucleotide substitution Taiwanese population was sister to the P. spicata (sequence position 315). The pairwise uncorrect- clade with 91% bootstrap probability. The acces- ed p-distance value between the accessions of the sions of P. verticillata were in a monophyletic Taiwanese populations and P. refracta was 0.1% group with a high bootstrap value (99%). This (Table 4). The values ranged from 0.1 to 0.3% in clade was sister to the clade of P. refracta, the June 2019 Murayama & al. – Taxonomy of Pedicularis in Taiwan 111

Table 3. Factor loading, eigenvalue, contribution, and accu- tween P. szetschuanica, P. metaszetschuanica, mulated contribution of principal components. and P. lineata was unclear. In the MP analysis of 1 Components Characters PC1 PC2 the ITS region, 8,505 MP trees, requiring 153 a. galea apex angle (°) -0.05860 -0.40830 steps, were obtained [consistency Index includ- b. corolla length (mm) 0.66280 -0.08398 ing uninformative characters (CI) = 0.8235; Re- c. leaf blade length (mm) 0.40845 0.87120 d. distance between the rachis and tention Index (RI) = 0.8933]. The topology of the 0.66487 0.57231 the longest pinnae (mm) strict consensus tree of the 8,505 MP trees was e. distance between the rachis and 0.77437 0.22962 the sinuses of the pinnae (mm) almost the same as for the NJ tree, excluding the h. number of pinnae -0.01263 0.55155 clades of P. verticillata and the three Chinese en- i. aboveground plant height (cm) 0.59894 0.54267 demic species (P. szetschuanica, P. metaszetsch- Eigen value 3.4932 1.3703 contribution (%) 49.903 19.576 uanica, and P. lineata) with 56% bootstrap prob- Accumlated contribution (%) 49.903 69.479 ability (data not shown). 1The ratio characters (f and g) were excluded in this analysis In the NJ tree of trnK (partial) region, the ac- (see, Table 2). cessions of the Taiwanese populations were not in Taiwanese populations, and Pedicularis spicata same clade as those of P. verticillata, but were in with weak (55%) bootstrap value. Pedicularis the same clade as the accessions of P. refracta szetschuanica, P. metaszetschuanica and P. lin- with a bootstrap value of 95% (Fig. 5). The acces- eata, which are endemic to southwest China, sions of P. verticillata were in a clade with those were in a single clade with 100% bootstrap sup- of P. metaszetschuanica and P. lineata with 99% port. Together with the clade of the Taiwanese bootstrap probability. The accessions of P. populations, P. refracta, P. verticillata and P. spi- szetschuanica were in a single clade with 59% cata, they constituted a monophyletic group with bootstrap probability. The relationships between a bootstrap value of 99%. The relationships be- the above three clades and P. spicata were un-

4

S 3 S A G G 3 3 A B Taiwanese populations 3 3 2 3 GG (Xue Shan) 3 A G A A G G 3 AS A3 J Taiwanese populations 3 3 A A A A 3 S3SAS G GG 3 A A (Hehuan Shan) 333 33 AASASA AAAA 1 3 3 SS H Taiwanese populations 3 3 AB A A GE G E S A A A G 19.6 %) 3 S 3G A (Nanhu Shan) S S EG SSE ( 3 3 B E G 3S HSS EGGEE G G GG E P. refracta (Miyazaki) 3 BE S E A E G G 0 3B B E G EE E HG B B G 3JBBB J B EEEE EBG HB B SJ BH JS E E S P. refracta (Takamori) HES J E GGEB SH JJHH BJBJH B H HE H H B H BJ EJG GH G G B 3B3 BHB H JH JHE H G -1 B HHBBB B BJ J J G J G A Component 2 B B BBB HHBHBB H H3 J  P. refracta (Minami-Oguni) J BJ B B H 3 HJ HJ E B BBHH JB J GJ H BHJBHB B HJ B 3 J G P. refracta (Hitoyoshi) B B H -2 B B B H B B B J H 3 P. verticillata

-3 B

-4 -4 -2 0 2 4 6 8 Component 1(49.9%)

Fig. 6. Scatter diagram of principal components 1 and 2 from 291 samples of Pedicularis refracta vars. refracta and trans- morrisonensis (Taiwanese populations) and P. verticillata.

FIG. 6 112 Acta Phytotax. Geobot. Vol. 70

Table 4. The values of the pairwise uncorrected p-distance between two populations of ITS and trnK (partial). Tai- trnK wanese P. re- P. verticillata P. spicata (partial) pops. fracta

1 All 3 2 Taiset- Shubut- Kita- Yatsug- Tolba- Swit- Shaka Sanno- Men- Primor- ITS All pops. pops. susan susan dake adake chik Jiuzhi Garze zerland Alaska sindo mine tougao sikij All Taiwanese popula- 0.001 0.015 – – 0.012 – 0.014 – 0.013 – 0.011 – – – pops. 1 tions All P. refracta popula- 0.000 0.016 – – 0.013 – 0.015 – 0.014 – 0.012 – – – tions2 Taisetsusan 0.051 0.051 – – 0.001 – 0.002 – 0.003 – 0.019 – – – Shubutsu- 0.051 0.051 0.011 – – – – – – – – – – – san Kitadake 0.049 0.049 0.010 0.002 – – – – – – – – – – Yatsuga- – – – – – – 0.001 – 0.002 – 0.016 – – – P. dake verticillata Tolbachik 0.045 0.045 0.006 0.006 0.004 0.011 – – – – – – – – Jiuzhi 0.049 0.049 0.017 0.013 0.011 0.006 0.011 – 0.003 – 0.018 – – – Garze 0.047 0.047 0.015 0.011 0.009 0.004 0.009 0.009 – – – – – – Switzerland – – – – – – – – – – 0.017 – – – Alaska3 0.051 0.051 0.011 0.004 0.002 – 0.006 0.013 0.011 – – – – – Shakasindo 0.035 0.035 0.057 0.053 0.055 – 0.055 0.055 0.051 – 0.057 – – – Sannomine 0.033 0.033 0.055 0.055 0.053 – 0.053 0.053 0.049 – 0.055 0.002 – – P. spicata Mentougao 0.037 0.037 0.055 0.055 0.053 – 0.053 0.053 0.051 – 0.055 0.013 0.011 – Primorsikij 0.031 0.031 0.053 0.053 0.051 – 0.051 0.051 0.049 – 0.053 0.004 0.002 0.009 1Taiwanese populations include the three all populations, see Table 1. 2P. refracta include the eight all populations, see Table 1. 3Alaska include the two populations, see Table 1. clear. In the MP analysis of the trnK (partial) re- 9.2 mm) than in both P. verticillata (11.0 mm) and gion, 2,430 MP trees, requiring 79 steps, were P. refracta (13.0–14.1 mm). There were signifi- obtained; CI = 0.9367; RI = 0.9412. The topology cant differences in mean leaf blade length (c) be- of the strict consensus tree of these 2,430 MP tween the Taiwanese populations (26.9–32.0 mm) trees was fundamentally the same as for the NJ and P. verticillata (20.0 mm), but not between the tree. In the MP tree, four major clades were iden- Taiwanese populations and P. refracta (26.9–32.0 tified as follows: the Taiwanese populations and mm, excluding the Hitoyoshi population). The Pedicularis refracta (93%); P. verticillata, P. mean distance between the rachis and the longest metaszetschuanica, and P. lineata (100%); P. spi- pinna (d) of P. verticillata was the smallest (4.4 cata; and P. szetschuanica (62%) (Fig. 5). mm) among the three groups. The distance was similar between Taiwanese populations and P. re- Morphological analysis fracta (7.0–7.9 mm), excepting the Xue Shan and We compared nine morphological characters Hitoyoshi populations. There were no significant (a–i) in the Taiwanese populations, P. verticillata differences in the mean distance between the ra- and P. refracta (Table 2). The mean angle of the chis and the sinuses of the pinnae (e) between Tai- galea apex (a) was significantly smaller in the Tai- wanese populations (0.9–1.0 mm) and P. verticil- wanese populations (63.0°–70.3°) than in P. verti- lata (0.81 mm), however, it was significantly cillata (83.9°), but not significantly different from greater (1.4–2.0 mm, excluding the Miyazaki the Taiwanese populations and P. refracta (67.4°– population) in P. refracta. The mean ratio of leaf 73.8°). The mean corolla length (b) was signifi- blade length to distance between the rachis and cantly shorter in the Taiwanese populations (8.7– the longest pinna [f (c/d)], was not significantly June 2019 Murayama & al. – Taxonomy of Pedicularis in Taiwan 113 different between Taiwanese populations (4.4– populations 4.8), Pedicularis verticillata (4.7), and P. refracta In previous taxonomic studies of Pedicularis, (4.2–4.4). The mean ratio of distance between the the Taiwanese populations have been identified rachis and the longest pinna to the width of the as P. verticillata (Li 1978, Liu 1998), P. trans- rachis and the sinuses of the pinna [g (e/d)] was morrisonensis (Hayata 1915, Limpricht 1924, smaller in Taiwanese populations (0.132–0.162) Tsoong 1963, Yang et al. 1998), and P. refracta than in P. refracta (0.166–0.261) and P. verticil- var. transmorrisonensis (Hurusawa 1948). In our lata (0.186), but not significantly different among molecular phylogenetic analysis using ITS re- the populations/species. The mean number of gions of nrDNA, the Taiwanese populations pinnae (h) was almost the same in the Taiwanese formed a strongly supported clade with P. refrac- populations (15.8–17.8) as in P. refracta (14.9– ta. Pedicularis verticillata was positioned in an- 19.2) and less in P. verticillata (14.2). Mean other clade (Fig. 4), as were P. spicata, P. aboveground plant height (i) was greater in the metaszetschuanica, P. lineata, and P. szetschua- Taiwanese populations (14.0–20.6 cm) than in P. nica. In the analysis of trnK (partial) regions of verticillata (12.1 cm), but the Taiwanese popula- cpDNA, the Taiwanese populations and P. re- tions were shorter than P. refracta (15.3–37.3 fracta were in a single clade with high bootstrap cm). There were significant differences in mean probability, and P. verticillata was in another plant height among the Taiwanese populations, P. clade (Fig. 5), as were P. metaszetschuanica, P. verticillata and P. refracta, excluding the Xue lineata, P. szetschuanica, and P. spicata. The re- Shan (14.0 cm) and Takamori populations (15.3 sults suggest that the Taiwanese populations are cm) of P. refracta. obviously not P. verticillata and should be con- To determine the relationship between the sidered to be P. refracta. Taiwanese populations, P. verticillata and P. re- In the morphological analysis using ANOVA, fracta, we performed PCA using characteristics there were seven characteristics (a–d, g–i) that from seven of the measured characters (a–e, h, i) differed between the Taiwanese populations and from 291 specimens. A scatter diagram of princi- P. verticillata (Table 2). Among the seven were pal components 1 and 2 is shown in Fig. 6. The clear significant differences in four: (a) galea contribution of component 1 was 49.9% and that apex angle, (b) corolla length, (c) leaf blade of component 2 was 19.6%. The accumulated length, and (d) distance between the rachis and contribution of components 1 and 2 represented the longest pinna. In comparison, only two sig- 69.5% of the total variation (Table 3). Component nificant differences were detected between the 1 primarily contributed to the distance between Taiwanese populations and P. refracta: (b) corol- the rachis and the longest pinnae (+ 0.66487) and la length and (i) aboveground plant height. Plants the distance between the rachis and the sinuses of in the Takamori population of P. refracta were the pinnae (+ 0.77437). Component 2 primarily smaller than in the other three populations, contributed to leaf blade length (+ 0.87120) and whereas they were the same as those of the Tai- the distance between the rachis and the longest wanese populations and P. verticillata. However, pinnae (+ 0.57231). Most specimens of P. verticil- the collections from the Takamori population lata, P. refracta, and the Taiwanese populations were made somewhat earlier than those from oth- clustered within the range for each species/popu- er populations. Controlled burning and grass cut- lation, although all clusters were continuously ting is carried out annually at the site of the Taka- distributed (Fig. 6) mori population. The results of PCA showed that the Taiwanese populations, P. refracta, and P. verticillata each form a morphological cluster Discussion (Fig. 6), although the distribution of the clusters was continuous. The findings show that each pop- Taxonomic attributes of Taiwanese Pedicularis ulation is discriminable and that the Taiwanese 114 Acta Phytotax. Geobot. Vol. 70 populations are morphologically closer to Pedic- sis in the Flora of China (Yang et al. 1998). In the ularis refracta than to P. verticillata. Flora of Taiwan, 1st and 2nd editions the Taiwan- We therefore evaluated the taxonomic status ese plants are treated as P. verticillata (Li 1978, of the Taiwanese populations based on the fol- Li 1998), without P. refracta var. transmorriso- lowing three points. First, there were small ge- nensis being listed in synonymy. We therefore netic differences between the accessions of the propose reviving the name P. refracta var. trans- Taiwanese populations and those of P. refracta. morrisonensis for plants in the populations in The values of intraspecific pairwise p-distance Taiwan. among the accessions of P. verticillata and P. spi- cata were larger than between the accessions of Phylogeography of the Pedicularis refracta vars. the Taiwanese populations and P. refracta (Table refracta and transmorrisonensis 4). The degree of genetic differentiation between In the following discussion, we refer to the the Taiwanese populations and P. refracta is plants of Taiwan as var. transmorrisonensis and therefore smaller. Second, the Taiwanese popula- those from Kyushu as var. refracta. In the phylo- tions and P. refracta differ in corolla length and genetic analyses using the trnK (partial) region, plant height (Table 2). Plants in the Taiwanese the bootstrap probabilities of most clades in the populations had relatively shorter corollas and tree were low (Fig. 5), but the tree based on the were shorter overall than plants in the popula- ITS regions indicated relatively clear phylogenet- tions of P. refracta. Third, the habitat of the Tai- ic relationships among the species/populations wanese populations differs from the habitat of P. (Fig. 4). Therefore, the following phylogeograph- refracta. The Taiwanese populations are in the al- ic discussion is based on the ITS tree. pine zone of mountains at approximately 2,500– The monophyly of P. spicata, P. verticillata, 3,500 m in Taiwan, while P. refracta occurs in P. metaszetschuanica, P. lineata, and P. szetsch- semi-natural grasslands in the mountainous areas uanica was strongly supported in a previous of Kyushu, Japan at approximately 400–800 m. study (P. verticillata lineage; Tkach et al. 2014). In short, plants in the Taiwanese populations and In the present analyses, vars. refracta and trans- those of P. refracta exhibit morphological and morrisonensis were in a monophyletic group with ecological differences, although we found little P. spicata (Fig. 4). Therefore, we consider that genetic variation between them. We conclude they should be included in the P. verticillata lin- that the plants in the Taiwanese populations eage. Although only P. verticillata is widely dis- should be treated at an infraspecific rank underP. tributed in the arctic and alpine regions of the refracta. northern hemisphere (Hultén 1968, Yamazaki Hayata (1915) first described plants in the 1993, Fujii & Senni 2006, Fujii et al. 2014). Other populations from Taiwan as the endemic P. trans- species of the P. verticillata lineage are endemic morrisonensis. In the literature, P. transmorriso- to eastern Asia. Pedicularis spicata is distributed nensis is distinguished from P. verticillata by a from northern Japan to northern Korea, north and slightly longer, narrower corolla tube and leaves northeastern China, Mongolia, and eastern Sibe- arranged remotely along the length of the stem, ria (Ivanina 1991, Yamazaki 1993, Yang et al. not clustered under the . Hurusawa 1998, Grubov 2001, Fujii et al. 2014). Pedicularis (1948), however, noted morphological similari- metaszetschuanica, P. lineata, and P. szetschua- ties between P. transmorrisonensis and the Ky- nica are only in southeastern China: P. ushu populations of P. refracta and treated the metaszetschuanica in Sichuan Province, P. linea- Taiwanese populations as P. refracta var. trans- ta in Gansu, Shaanxi, Sichuan, and Yunnan prov- morrisonensis. P. refracta was described earlier inces, and P. szetschuanica in Gansu, Qinghai, than P. transmorrisonensis (Maximowicz 1881). Sichuan, and Xizang provinces (Yang et al. 1998). The name P. refracta var. transmorrisonensis This suggests that P. refracta originated from was treated as a synonym of P. transmorrisonen- part of the P. verticillata lineage and differenti- June 2019 Murayama & al. – Taxonomy of Pedicularis in Taiwan 115

Fig. 7. Lectotype of Pedicularis refracta var. transmorrisonensis (Hayata) Hurus (= P. transmorrisonensis Hayata) (A), the flower and leaf collections in paper capsules of the specimen, the anatomical illustration of corolla in Hayata (1915) was resemble to these collections (B). 116 Acta Phytotax. Geobot. Vol. 70 ated at the eastern boundary of Asia. 2000). It is considered that most plants in the As mentioned above, vars. refracta and trans- Ryukyu Islands originated in China and arrived morrisonensis are genetically close to each other via Taiwan (Shimabuku 1997, Hsieh 2002). Al- and have been treated as conspecific. However, though, Pedicularis ser. Verticillatae is not in the Taiwan is geographically separated from Kyushu Ryukyu Islands, we considered that the vegeta- by approximately 1,200 km (Fig. 2), and the pop- tion and climate of Taiwan and the Ryukyus was ulations are highly disjunct. We propose two pos- similar and that both attained elevations of 1,000– sible hypotheses to explain the establishment of 1,500 m between 2 and 1.2 Mya (Kuroda et al. this distribution pattern. The first is that it oc- 2002, Fujiki & Ozawa 2008). Furthermore, fossil curred through vicariance events. During the pollen of Abies, Tsuga, Pinus subgen. Haploxy- Pleistocene, the ancestors of Pedicularis refracta lon, Keteleeria, and Cryptomeria, currently ab- were widely and continuously distributed across sent from the Ryukyu Islands, has been found eastern Asia over land bridges that connected there (Fujiki & Ozawa 2008). It is therefore pos- lands that are now islands. They were then sepa- sible that the ancestors of the Taiwanese and Ky- rated and isolated by rising sea levels during the ushu populations expanded their range through Quaternary climate oscillations. They now sur- the Ryukyus during the late Pliocene to early vive as relict, discontinuous populations on Tai- Pleistocene. To further evaluate this hypothesis wan and on Kyushu. The second hypothesis is and to estimate the time of divergence of P. re- that the disjunct distribution pattern is the result fracta vars. refracta and transmorrisonensis, an of recent long distance dispersal between analysis using high resolution markers is re- Taiwan and Kyushu. Typically, it is believed that quired. long distance seed dispersal in plants is accom- plished by birds or wind (birds; Kleyheeg et al., 2015, Viana et al., 2016, wind; Nathan et al., Taxonomy 2002, Tackenberg 2003). It is thought that the seeds of Pedicularis are mainly dispersed close Pedicularis refracta Maxim. var. transmorri- to their origin by gravity (Juan et al. 2000, Liu et sonensis (Hayata) Hurus. in J. Jap. Bot. 22: 74 al. 2013) because they are relatively large and (1948). — Pedicularis transmorrisonensis Haya- lack appendages for wind dispersal. Further- ta in Icon. Pl. Formosan. 5: 126 (1915). more, there is little possibility for long distance Lectotypus. TAIWAN: in Morrison, U. Mori s.n. (TI!, dispersal by birds, because they have neither hic designatus) —Fig. 7. fleshy fruits that birds would eat nor clinging structures to attach to birds (Howe & Smallwood Distribution. Taiwan, ca. 2,500–3,500 m alt. 1982). Thus, the recent long distance dispersal hypothesis is unlikely to explain the disjunct dis- tribution pattern between the populations on Tai- We thank the students in the laboratory of Kuo-Fang wan and Kyushu. Therefore, we conclude that the Chung and Takuro Ito for their cooperation in collecting samples in the alpine regions of Taiwan. We also thank, first Quaternary vicariance hypothesis is a more Jin Murata and Tetsuo Toma (TI) and Kunihiko Ueda likely explanation for the disjunct distribution (KANA) for granting permission to examine specimens. pattern in P. refracta on Taiwan and Kyushu. Thanks also to Hidetoshi Nagamasu for his help to ob- According to Chiang & Schaal (2006) the serve the type specimen of Pedicularis refracta var. Ryukyu Islands played an important role in the transmorrisonensis (= P. transmorrisonensis). Finally, we also thank David E. Boufford, Minoru N. Tamura and evolutionary history of populations of plants on Koji Takayama for their help in checking this manuscript. Taiwan and Kyushu. The Ryukyu Islands are This study was supported by JSPS KAKENHI Grant geographically located between Taiwan and Ky- Numbers JP25440215. ushu (Fig. 2) and separated from the Asian main- land in the late Pliocene (Ota 1998, Motokawa June 2019 Murayama & al. – Taxonomy of Pedicularis in Taiwan 117

Editorial Committee of the Flora of Taiwan, Depart- References ment of Botany. National Taiwan University, Taipei. Huang, T.-C. (ed.). 1996. Flora of Taiwan 2nd ed., vol. Ⅱ. Akaike, H. 1974. A new look at the statistical model iden- Editorial Committee of the Flora of Taiwan, Depart- tification. IEEE transactions on automatic control 19: ment of Botany. National Taiwan University, Taipei. 716–723. Huang, T.-C. (ed.). 1998. Flora of Taiwan 2nd ed., vol. Ⅳ. Avise, J. C. 2004. Molecular markers, natural history and Editorial Committee of the Flora of Taiwan, Depart- evolution. 2nd ed. Sinauer, Sunderland, Massachu- ment of Botany. National Taiwan University, Taipei. setts. Huang, T.-C. (ed.). 2000. Flora of Taiwan 2nd ed., vol. Ⅴ. Chiang, T. Y. & B. A. Schaal. 2006. Phylogeography of Editorial Committee of the Flora of Taiwan, Depart- plants in Taiwan and the Ryukyu Archipelago. Taxon ment of Botany. National Taiwan University, Taipei. 55: 31–41. Huang, T.-C. (ed.). 2003. Flora of Taiwan 2nd ed., vol. Ⅵ. Dowie, S. R. & J. D. Palmer. 1992. Use of chloroplast Editorial Committee of the Flora of Taiwan, Depart- DNA rearrangements in reconstructing plant phylog- ment of Botany. National Taiwan University, Taipei. eny. In: Soltis, D. E., P. S. Soltis & J. J. Doyle. (eds.), Hultén, E. 1968. Flora of Alaska and neighboring territo- Molecular Systematics of Plants. pp. 14–35. Chapman ries: a manual of the vascular plants. Stanford Univ. & Hall, New York. Press., Stanford. Doyle, J. J. & J. L. Doyle. 1987. A rapid DNA isolation Hurusawa, I. 1948. Pedicularis refracta (Maxim.) Maxim. var. procedure for small quantities of fresh leaf tissue. transmorrisonensis (Hayata) Hurus. J. Jap. Bot. 22: 74. Phytochem. Bull. 19: 11–15. Ivanina, L. I. 1991. Pedicularis L. In: Charkevicz, S. S. (ed.), Eaton, D. A. R., C. B. Fenster, J. Hereford, S. Huang & R. Plantae Vasculares Orientis Extremi Sovietici, vol. 5. pp. H. Ree. 2012. Floral diversity and community struc- 334–359. Nauka, Saint Petersburg (in Russian). ture in Pedicularis (Orobanchaceae). Ecology 93: Iwatsuki, K., T. Yamazaki, D. E. Boufford & H. Ohba (eds.). 182–194. 1993. Flora of Japan, vol. Ⅲa. Kodansha, Tokyo. Felsenstein, J. 1985. Confidence limits on phylogenies: an Iwatsuki, K., T. Yamazaki, D. E. Boufford & H. Ohba (eds.). approach using the bootstrap. Evolution 39: 783–791. 1995a. Flora of Japan, vol. Ⅲb. Kodansha, Tokyo. Felsenstein, J. 2004. Inferring phylogenies. Sinauer, Sun- Iwatsuki, K., T. Yamazaki, D. E. Boufford & H. Ohba (eds.). derland, Massachusetts. 1995b. Flora of Japan, vol. Ⅱ. Kodansha, Tokyo. Fior, S., M. Li, B. Oxelman, R. Viola, S. A. Hodges, L. Iwatsuki, K., D. E. Boufford & H. Ohba (eds.). 1999. Flora of Ometto & C. Varotto. 2013. Spatiotemporal recon- Japan, vol. Ⅱc. Kodansha, Tokyo. struction of the Aquilegia rapid radiation through Iwatsuki, K., D. E. Boufford & H. Ohba (eds.). 2001. Flora of next-generation sequencing of rapidly evolving cpD- Japan, vol. Ⅱb. Kodansha, Tokyo. NA regions. New Phytol. 198: 579–592. Iwatsuki, K., D. E. Boufford & H. Ohba (eds.). 2006. Flora Fujii, N. & K. Senni. 2006. Phylogeography of Japanese of Japan, vol. Ⅱa. Kodansha, Tokyo. alpine plants: biogeographic importance of alpine re- Johnson, L. A & D. E. Soltis. 1994. matK DNA sequences gion of central Honshu in Japan. Taxon 55: 43–52. and phylogenetic reconstruction in Saxifragaceae s. Fujii, N., M. Teramoto, R. H. Ree, N. Shirai, C. Suyama, str. Syst. Bot. 19: 143–156. K. Ueda & H. Takahashi. 2014. Relict distribution in Juan, R., J. Pastor & I. Fernández. 2000. SEM and light Pedicularis spicata Pall. (Orobanchaceae): a new lo- microscope observations on fruit and seeds in Scroph- cality in central Honshu, Japan. Acta Phytotax. Geo- ulariaceae from Southwest Spain and their systematic bot. 65: 75–87. significance. Ann. Bot. 86: 323–338. Fujiki, T. & T. Ozawa. 2008. Vegetation change in the Kimura, M. 1980. A simple method for estimating evolu- main island of Okinawa, southern Japan from late tionary rates of base substitutions through compara- Pliocene to early Pleistocene. Quant. Int. 184: 75–83. tive studies of nucleotide sequences. J. Molec. Evol. Grubov, V. I. 2001. Key to the vascular plants of Mongolia 16: 111–120. (with an atlas) vol. 2. Science Publishers, Enfield. Kleyheeg, E., C. H. A. van Leeuwen, M. A. Morison, B. A. Hayata, B. 1915. Pedicularis transmorrisonensis Hayata Nolet & M. B. Soons. 2015. Bird-mediated seed dis- sp. nov. Icon. Pl. Formosan. 5: 126–128. persal: Reduced digestive efficiency in active birds Howe, H. F. & J. Smallwood. 1982. Ecology of seed dis- modulates the dispersal capacity of plant seeds. Oikos persal. Annual Rev. Ecol. Syst. 13: 201–228. 124: 899–907. Hsieh, C.-F. 2002. Composition, endemism and phytogeo- Kuroda, T., T. Ozawa & H. Furukawa. 2002. Paleoenvi- graphical affinities of the Taiwan flora. Taiwania. 47: ronment of the Ryukyu arc. From paleontology. In: 298–310. Kimura, M. (ed) The formation of the Ryukyu arc. Huang, T.-C. (ed.). 1993. Flora of Taiwan 2nd ed., vol. Ⅲ. and migration of biota to the arc. Okinawa Times, Editorial Committee of the Flora of Taiwan, Depart- Naha. pp. 85–102 (in Japanese). ment of Botany. National Taiwan University, Taipei. Li, H. L. 1978. Pedicularis L. In: Li, H.-L., Liu, T.-S., Huang, T.-C. (ed.). 1994. Flora of Taiwan 2nd ed., vol. Ⅰ. Huang, T.-C., Koyama, T. & E. C. Devol. (eds.), Flora 118 Acta Phytotax. Geobot. Vol. 70

of Taiwan 1st ed., vol. 4. pp. 583. Republic of China, Soltis, D. E., P. S. Soltis & J. J. Doyle. 1998. Molecular Taipei. Systematics of Plants II. Springer, Boston. Liew, P.-N. & N.-J. Chung. 2001. Vertical migration of for- Swofford, D. L. 2003. PAUP*. Phylogenetic analysis using ests during the last glacial period in subtropical. West. parsimony (*and other methods). Version 4.0b10, Pacific Earth Sci. 1: 405–414. Sinauer, Sunderland, Massachusetts. Limpricht, W. 1924. Pedicularis transmorrisonensis Tackenberg, O. 2003. Modeling long distance dispersal of Hayata. In: Rep. Specierum Nov. Reg. Veg. 20: 204. plant diaspores by wind. Ecol. Monogr. 73: 78–85. Liu, H. Y. 1998. Pedicularis L. In: Huang, T.-C. (ed.), Flo- Tamura, K. 1992. Estimation of the number of nucleotide ra of Taiwan 2nd ed., vol. 4. pp. 619. Department of substitutions when there are strong transition-trans- Botany. National Taiwan University, Taipei. version and G+C-content biases. Molec. Biol. Evol. 9: Liu, J., M. Möller, L. M. Gao, D. Q. Zhang & D. Z. Li. 678–687. 2011. DNA barcoding for the discrimination of Eur- Tamura, K., G. Stecher, D. Peterson, A. Filipski & S. Ku- asian yews (Taxus L., Taxaceae) and the discovery of mar. 2013. MEGA6: Molecular evolutionary genetics cryptic species. Molec. Ecol. Resour. 11: 89–100. analysis version 6.0. Molec. Biol. Evol. 30: 2725–2729. Liu, M. L., W. Bin Yu, D. Z. Li, R. R. Mill & H. Wang. Tanaka, N. 2013. A new species of Chionographis (Mel- 2013. Seed morphological diversity of Pedicularis anthiaceae) from Japan. J. Jap. Bot. 88: 30–35. (Orobanchaceae) and its taxonomic significance. Plant Tkach, N., R. H. Ree, P. Kuss, M. Röser & M. H. Hoff- Syst. Evol. 299: 1645–1657. mann. 2014. High mountain origin, phylogenetics, Maximowicz, C. J. 1881. Pedicularis refracta (Maxim.) evolution, and niche conservatism of arctic lineages in Maxim. Mel. Biol. 11: 247. the hemiparasitic genus Pedicularis (Orobanchaceae). Motokawa, M. 2000. Biogeography of living mammals in Molec. Phylogenet. Evol. 76: 75–92. the Ryukyu islands. Tropics 10: 63–71. Tsoong, P. C. 1963. Pedicularis transmorrisonensis Haya- Nakamura, K., K. F. Chung, C. J. Huang, Y. Kono, G. ta. In: Chien, S. S. & W. Y. Chun. (eds.) Fl. Reipubl. Kokubugata & C. I. Peng. 2012. Extreme habitats that Popularis Sin. vol. 68: 174. Science Press, Beijing (in emerged in the Pleistocene triggered divergence of Chinese). weedy Youngia (Asteraceae) in Taiwan. Molec. Phylo- Viana, D. S., L. Santamaría & J. Figuerola. 2016. Migra- genet. Evol. 63: 486–499. tory birds as global dispersal vectors. Trends Ecol. Nathan, R., G. G. Katul, H. S. Horn, S. M. Thomas, R. Evol. 31: 763–775. Oren, R. Avissar, S. W. Pacala & S. A. Levin. 2002. White, T. J., T. Bruns, S. Lee & J. Taylor. 1990. Amplifica- Mechanisms of long distance dispersal of seeds by tion and direct sequencing of fungal ribosomal RNA wind. Nature 418: 409–413. genes for phylogenetics. In: Innis, M., D. Gelfand, J. Nei, M. & S. Kumar. 2000. Molecular Evolution and Phy- Sninsky & T. White. (eds.), PCR protocols: a guide to logenetics. Oxford university press. New York. methods and applications, pp. 315–322. Academic Ohwi, J. 1965. Flora of Japan (English ed.). Smithsonian Press, San Diego. Inst., Washington DC. Yamazaki, T. 1993. Pedicularis. In: Iwatsuki, K., T. Ota, H. 1998. Geographic patterns of endemism and specia- Yamazaki, D. E. Boufford & H. Ohba. (eds.), Flora of tion in amphibians and reptiles of the Ryukyu archi- Japan, vol. IIIa. pp. 364–371. Kodansha, Tokyo. pelago, Japan, with special reference to their paleogeo- Yang, F. S. & X. Q. Wang. 2007. Extensive length varia- graphical implications. Res. Popul. Ecol. 40: 189–204. tion in the cpDNA trnT-trnF region of hemiparasitic Qiu, Y. X., C. X. Fu & H. P. Comes. 2011. Plant molecular Pedicularis and its phylogenetic implications. Plant phylogeography in China and adjacent regions: Trac- Syst. Evol. 264: 251–264. ing the genetic imprints of Quaternary climate and en- Yang, H., N. H. Holmgren & R. R. Mill. 1998. Pedicularis. vironmental change in the world’s most diverse tem- In: Wu, Z. Y. & P. H. Raven. (eds.), Flora of China, perate flora. Molec. Phylogenet. Evol. 59: 225–244. vol. 18, pp. 97–209. Science Press, Beijing & Missouri Ree, R. H. 2005. Phylogeny and the evolution of floral di- Botanical Garden Press, St. Louis. versity in Pedicularis (Orobanchaceae). Int. J. Plant Young, N. D., K. E. Steiner & C. W. DePamphilis. 1999. Sci. 166: 595–613. The evolution of parasitism in Scrophulariaceae/Oro- Salemi, M. & A. M. Vandamme. 2003. Handbook of phy- banchaceae: plastid gene sequences refute an evolu- logenetic methods. Cambridge Univ. Press, Cam- tionary transition series. Ann. Missouri Bot. Gard. 86: bridge. 876–893. Shiga, T. & Y. Kadono. 2015. Nuphar saikokuensis (Nym- Yu, C. C. & K. F. Chung 2014. Systematics of Berberis phaeaceae), a new species from central to western Ja- sect. Wallichianae (Berberidaceae) of Taiwan and Lu- pan. J. Jap. Bot. 90: 22–28. zon with description of three new species, B. schaali- Shimabuku, K. 1997. Check list vascular flora of the Ryukyu ae, B. ravenii, and B. pengii. Phytotaxa 184: 61–99. Islands. Revis Edn. Kyushu Univ. Press, Fukuoka.

Received September 29, 2018; accepted November 7, 2018