American Journal of 92(4): 625±633. 2005.

MOLECULAR PHYLOGENY OF () BASED ON ITS AND TRNL-F SEQUENCES1

SHAOTIAN CHEN,2 KAIYUN GUAN,2 ZHEKUN ZHOU,2,3 RICHARD OLMSTEAD,4 AND QUENTIN CRONK5

2Kunming Institute of Botany, Chinese Academy of Sciences, Heilongtan, Kunming, Yunnan, 650204, P. R. China; 4Department of Biology, University of Washington, Hitchcock Hall Rm 423, Seattle, Washington 98195 USA; 5UBC Botanical Garden & Centre for Research, Faculty of Agricultural Sciences, University of British Columbia, 6804 Southwest Marine Drive, Vancouver, British Columbia V6T 1Z4 Canada

Incarvillea is a herbaceous and temperate member of Bignoniaceae, previously divided into four subgenera, Niedzwedzkia, Amphi- come, Incarvillea, and Pteroscleris. Niedzwedzkia and Amphicome have in the past been treated as independent genera. Different relationships have been proposed for the four subgenera. Here, maximum parsimony analysis using ITS and trnL-F sequences resulted in similar trees and showed that the is monophyletic. Analysis of the combined data resulted in a single tree with ®ve major clades highly supported and well resolved. The relationships of the ®ve major clades are (subgenus Niedzwedzkia (Incarvillea olgae (subgenus Amphicome (subgenus Incarvillea, subgenus Pteroscleris)))). All four subgenera are well supported for monophyly, with the exception of subgenus Incarvillea, represented here by I. sinensis and I. olgae. Incarvillea olgae is not closely related to I. sinensis, a conclusion supported by morphology. The two basal monotypic subgenera are found in Central Asia. The most -rich subgenus, Pteroscleris, has 10 species in the Himalaya-Hengduan Mountains and may have dispersed early from central Asia to eastern Asia. Short branch lengths on the molecular trees within Pteroscleris suggest a recent and rapid radiation of this rosette-forming subgenus, perhaps connected with the uplift of the Himalaya-Hengduan massif.

Key words: Bignoniaceae; classi®cation; Himalaya-Hengduan Mountains; Incarvillea; ITS; phylogenetics; radiation; trnL-F.

Incarvillea Juss. is notable for being a temperate and her- was put into Bignoniaceae as an independent genus (Vassil- baceous genus of the primarily tropical and woody family Big- czenko, 1957, 1958). noniaceae (Fig. 1). It is composed of 16 species (Grierson, The geographical distribution of the genus is interesting. 1961; Zhao, 1988; Wang et al., 1991; Grey-Wilson, 1998), and Most of the species occur in the Himalayas and S. W. China, the modern classi®cation of Incarvillea was established by where the diversity of the mainly herbaceous species appears Grierson (1961) who divided the genus into four subgenera, to be related to the uplift of the Himalaya-Hengduan Moun- Amphicome (Royle) R. Br. apud Royle, Incarvillea, Pteros- tains (Chen et al., 2003). However, a number of species (all cleris Baillon, and Niedzwedzkia (B. Fedtsch.) Grierson. Al- with basal woodiness) are distributed through Central Asia and though the subgenera differ considerably in many characters into Mongolia (Fig. 3). Subgenera Amphicome and Incarvillea (including habit, calyx, stamens, capsule texture, and seeds), are found in eastern and central Asia, whereas subgenus Pter- morphological classi®cation is complicated due to extensive oscleris is endemic to the Himalaya-Hengduan region (E. apparent character convergence. Therefore, Grierson (1961) Asia). Subgenus Niedzwedzkia and the distinctive I. olgae Re- drew only tentative conclusions about evolutionary relation- gel are endemic to central Asia. ships within Incarvillea (Fig. 2A). Three subgenera, Incarvil- In this paper, we used ITS and trnL-F sequences to produce lea, Amphicome, and Pteroscleris are found in China, and in a phylogenetic framework on which to assess previous work Flora Reipublicae Sinicae (Wang et al., 1990), a different re- on the classi®cation of Incarvillea, particularly the number and lationship between them (Fig. 2B) was postulated from that of relationships of natural groups in the genus. Grierson. Furthermore, subgenus Amphicome was once placed MATERIALS AND METHODS variously in or Bignoniaceae as a genus (Don, 1837; Bureau, 1864; Burtt and Grierson, 1953). Based on the Samples and DNA extractionÐWe sampled 13 species of Incarvillea rep- karyomorphology and morphology, Chen et al. (2004) sug- resenting all four subgenera (Grierson, 1961). The three species not sampled gested that the three subgenera, Amphicome, Incarvillea, and are two that may be extinct (I. altissima G. Forrest and I. forrestii Fletcher, Pteroscleris, should all be treated as genera. The subgenus both in subgenus Pteroscleris) and I. potaninii Batalin from Mongolia, which Niedzwedzkia was originally described in Pedaliaceae and then was not obtainable for this study. Considering the historically unstable posi- tions of subgenera Amphicome and Niedzwedzkia in Bignoniaceae, Gesneri- aceae, and Pedaliaceae in the history of the genus, we selected one species 1 Manuscript received 21 January 2004; revision accepted 16 December from each of these families in order to be certain of the placement of Incar- 2004. villea. Tecomaria capensis (Thunb.) Spach, Sesamum indicum L., and Ne- The authors thank Dr. Mark Chase (Royal Botanic Gardens, Kew) for gift- matanthus strigillosus (Mart.) H. E. Moore were thus added into the analysis ing the DNA sample of Incarvillea emodi. We are grateful to the National Key Basic Research Program (973) (2003CB415102), National Natural Sci- as outgroups, and the tree was rooted on Scrophularia ningpoensis Hemsl. ence Foundation (40332021), and the Chinese Academy of Science Innovation In addition, we selected six further outgroups in Bignoniaceae for the trnL- Project (KSCX2±1±09) for support. F analyses to test the monophyly of the genus. These are Campsidium val- 3 Author for correspondence (e-mail: [email protected]) phone: 86± diviana (Phil.) Bull, Campsis radicans Seem., Pandorea jasminoides Schum., 871±5219932, fax: 86±871±5150227 Podranea ricasoliana Sprague, Tecoma stans Juss. and Tecomanthe volubilis 625 626 AMERICAN JOURNAL OF BOTANY [Vol. 92

Fig. 1. Growth habits of Incarvillea arguta, I. zhongdianensis, and I. mairei var. mairei. Incarvillea arguta has the much branched habit characteristic of subgenus Amphicome and contrasts with I. zhongdianensis and I. mairei var. mairei with the herbaceous habit characteristic of subgenus Pteroscleris. (A, B) I. arguta. (C) I. zhongdianensis. (D) I. mairei var. mairei.

L.S. Gibbs subsp. volubilis, all related Incarvillea, following the results of a gel-dried leaves using a DNA extraction kit (Watson Biotechnologies, Shang- study of the family based on three genes by Olmstead et al. (2002). hai, P.R. China). The Incarvillea semiretschenskia Grierson sample was collected from a herbarium specimen (E), and DNA of I. emodi (Lindl.) Chatterjee was pro- Ampli®cation, sequencing and sequence alignmentÐThe whole internal vided by Dr. Mark Chase. Sequences of Sesamum indicum and Nematanthus transcribed spacer region was ampli®ed using primers ITS4 and ITS5 (White strigillosus were obtained from GenBank (Beardsley and Olmstead, 2002; et al., 1990) for all materials except I. semiretschenskia, for which four prim- Bremer et al., 2002; Zimmer, 2002). Other materials were collected from the ers (ITS2, ITS3, ITS4, and ITS5) were used. The universal primers trnc and ®eld and Kunming Botanic Garden (Appendix, see Supplemental Data accom- trnf of Taberlet et al. (1991) were used for the trnL-F ampli®cation, according panying the online version of this article). DNA was extracted from silica- to the following protocol: 94ЊC for 3 min; 35 cycles of 94ЊC for 30 s, 53ЊC

Fig. 2. Comparison between three kinds of phylogenetic arrangements of groups in Incarvillea. (A) based on the revision of Grierson. (B) based on Flora Reipublicae Sinicae. (C) based on combined analysis of ITS and trnL-F sequences. April 2005] CHEN ET AL.ÐMOLECULAR PHYLOGENY OF INCARVILLEA 627

Fig. 3. Distribution of the genus Incarvillea, showing the E Asian distribution of subgenus Pteroscleris (solid line) and (dotted line). Other species are scattered through Central Asia and Mongolia with the exception of I. arguta (subgenus Amphicome) in E Asia. for 30 s, 72ЊC for 1 min; 72ЊC for 7 min. PCR products were subsequently imum parsimony (MP) approach. Characters were weighted equally and were visualized on a 1% low-melting temperature agarose gel, then were cut from unordered. Gaps were treated as missing data rather than as a ®fth character. the gel and puri®ed using a gel extraction kit (Watson Biotechnologies). A heuristic research was conducted using 1000 random taxon addition repli- Puri®ed products were sequenced in both directions, using the same con- cates, and 10 trees were held in each step during stepwise addition. Other tree ditions as the PCR. ITS sequences of all species were obtained using primer search options included tree-bisection-reconnection (TBR) branch swapping, ITS4 and ITS5, except for I. semiretschenskia for which ITS2 and ITS3 were steepest descent off, MulTrees on. To estimate the support for individual used. Four primers, trnc, trnd, trne and trnf were used for trnL-F sequencing clades, heuristic bootstrapping (1000 resampling replicates) was performed. of all material. Sequencing was carried out on an ABI 3100 automated se- Decay indices (Bremer, 1988) were calculated manually. quencer (Applied Biosystems, Foster City, California, USA) at Kunming In- The ITS sequences of six other outgroups in trnL-F data set were not stitute of Zoology. Sequences were aligned using Clustal W (Higgins, 1994) obtained in this study, so we combined ITS and trnL-F sequences of all taxa with minor manual adjustments. except these six relatives as a combined data set. We assessed the degree of The trnL-F sequences of six further outgroups were obtained using the phylogenetic incongruence between ITS and trnL-F partitions using the in- following procedures. The protocol for PCR and sequencing was 35 cycles congruence-length difference (ILD) test (Mickevich and Farris, 1981; Farris of 94ЊC for 1 min, 50ЊC for 1 min, and 72ЊC for 1 min. PCR products were et al., 1995) in PAUP* 4.0b10, with 1000 replications using a heuristic tree puri®ed using Qiaquick spin-columns (Qiagen Inc., California, USA) accord- search with 10 addition sequence replicates. To ensure the accuracy of the P ing to the manufacturer's protocol. Sequences of both strands of the PCR value in the ILD test, only parsimony informative positions were used. product were generated on an ABI 377 (Applied Biosystems). The resulting sequences were compiled and compared using the program Sequencher ver- RESULTS sion 3.0 (Gene Codes Corp., Ann Arbor, Michigan, USA). Sequences of all primers used for PCR ampli®cation and sequencing are Sequence characteristicsÐITSÐThe length of ITS (ITS1, listed in Table 1. ITS2, including 5.8S) sequences in Incarvillea ranged from 622 to 628 bp, giving an aligned matrix of 667 characters, 152 Phylogenetic analysisÐPhylogenetic analyses were performed using (22.79%) of which were phylogenetically informative. ITS1 PAUP*: phylogenetic analysis using parsimony (* and other methods), version varied from 236 to 240 bp in length with an aligned length of 4.0 Beta10 (Swofford, 2001) on a Macintosh (eMac) computer using a max- 258. Of these, 67 sites were informative. ITS2 ranged from 227 to 231 bp, with an aligned length of 251 characters, 81 of which were informative. The length of 5.8S varied from TABLE 1. Sequences of all primers used for PCR ampli®cation and 150 to 152 bp (aligned 152 with three informative sites). The sequencing. remaining six sites in the aligned matrix are from 26S.

Primer names Sequences (5Ј to 3Ј) trnL-FÐA poly-G region of about 18 guanines at approxi- ITS2 GCTCGTTCTTCATCGATGC Ј ITS3 GCATCGATGAAGAACGTAGC mately 120 bp from the 3 end in Incarvillea proved dif®cult ITS4 TCCTCCGCTTATTGATATGC to sequence accurately, so the sequences were excluded at this ITS5 GGAAGTAAAAGTCGTAACAAGG point. The trnL-F sequences of Incarvillea used in this study trnc CGAAATCGGTAGACGCTACG varied in length from 715 to 753 bp. The length of the trnL trnd GGGGATAGAGGGACTTGAAC intron region varied from 434 to 446 bp. The trnL 3Ј exon trne GGTTCAAGTCCCTCTA was 50 bp. The sequenced part of the trnL-F spacer region trnf ATTTGAACTGGTGACACGAG varied from 236 to 257 bp in length. The alignment resulted 628 AMERICAN JOURNAL OF BOTANY [Vol. 92

TABLE 2. Sequence characteristics of species studied.

Parameter ITS trnL-F Length range (bp) 588±628 715±778 Length range (ingroup) (bp) 622±628 715±753 Length range (outgroup) (bp) 588±628 751±778 Aligned length (bp) 667 826 G ϩ C content range (%) 55.15±66.78 33.59±37.40 G ϩ C content mean (%) 58.59 35.88 Sequence divergence (ingroup) (%) 0±8.85 0±8.45 Sequence divergence (in/outgroup) (%) 13.63±22.38 3.79±14.87 Size of indels (ingroup) 1±2 1±11 Size of indels (total) 1±21 1±11 Number of indels (ingroup) 13 20 Number of indels (total) 42 44 Number of variable sites (%) 254 (38.08) 195 (23.61) Number of constant sites (%) 413 (61.92) 631 (76.39) Number of informative sites (%) 152 (22.79) 115 (13.92) Number of autapomorphic sites (%) 102 (15.29) 80 (9.69) Transitions 200 37 Transversions 111 57 Transitions/transversions 1.8 0.65 in a data set with 826 characters, including 115 phylogenetic The combined analysis had higher bootstrap values on the informative sites (13.92%) and 69 variable but uninformative ingroup nodes than in the separate analyses of the two se- sites. quences. The genus Incarvillea is strongly supported as mono- Sequence characteristics are given in Table 2. phyletic with a bootstrap value of 100%, and the same ®ve major clades are generated within the genus as in the separate Phylogenetic analysisÐIndividual analysesÐParsimony analyses. The relationships between the ®ve major clades all analysis of the ITS data matrix resulted in only one most par- have bootstrap support, and a further ®ve clades are also sup- simonious tree of 482 steps (Fig. 4, CI ϭ 0.720, RI ϭ 0.667). ported. The genus Incarvillea is a monophyletic group with 100% bootstrap support and is sister to Tecomaria (bootstrap value DISCUSSION ϭ 90%). There are ®ve major clades in the genus and in all eight clades have bootstrap support of 50% or higher. Four of the ®ve major clades are equivalent to the four subgenera in The genus Incarvillea is a natural groupÐA persistent the system of Grierson except that I. olgae, which was put into question in Incarvillea concerns the rank and po- subgenus Incarvillea by Grierson, does not group with I. si- sition of Amphicome and Niedzwedzkia (Don, 1837; Bureau, nensis Lam. The relationship between the ®ve major clades is 1864; Burtt and Grierson, 1953; Vassilczenko, 1957, 1958; fully resolved but without good bootstrap support. Two sub- Grierson, 1961). In this study, Incarvillea as delimited by genera, Amphicome and Pteroscleris have bootstrap support of Grierson (including the genera Amphicome and Niedzwedzkia) greater than 50% (53% and 100%, respectively). is monophyletic with respect to Tecomaria with very strong MP analysis of the trnL-F data set generated 16 equally support (100% bootstrap) in analysis of ITS and combined parsimonious trees of 260 steps with consistency index (CI) data sets. Monophyly is also con®rmed by the analysis of the ϭ 0.892 and retention index (RI) ϭ 0.948. In the strict con- trnL-F data set with more Bignoniaceae taxa (100% boot- sensus tree (Fig. 5), seven clades in the genus Incarvillea have strap). None of the three data sets (ITS, trnL-F, and combined) bootstrap support of 50% or higher. The genus Incarvillea is con¯ict with the placement of the genera Amphicome and monophyletic with strong support (100%), and nested in Big- Niedzwedzkia in the genus Incarvillea sensu lato (s.l.), al- noniaceae (bootstrap value ϭ 100%), but the relationship with though the sister group relationship of Niedzwedzkia with the other Bignoniaceae taxa is not resolved. Five major clades rest of Incarvillea does not rule out separate generic treatment (identical to those suggested by ITS) were recovered but the for this taxon. The sole species of Niedzwedzkia, now called relationships between them were poorly resolved with the ex- I. semiretschenskia, has several unique characteristics within ception of I. sinensis being sister to the subgenus Pteroscleris the genus, such as a winged capsule, septifragal dehiscence, clade (bootstrap value ϭ 73%). and calyx tubular only at base (Table 3). The similarity of the species in karyomorphology and pollen Combined analysesÐIn the homogeneity test, the combined characters supports the genus as a natural group. The chro- data provided a P value of 0.597: the null hypothesis of con- mosome number of all species reported is 2n ϭ 22 (Sugiura, gruence is not rejected; therefore, we combined the ITS and 1936; Bowden, 1940; Fedorov, 1969; Xiao et al., 2002; Chen trnL-F data sets into a single matrix for phylogenetic analyses. et al., 2003), and their interphase nuclei and prophase chro- Phylogenetic analysis of the combined data set yielded one mosomes are of the simple chromocenter and interstitial type, most parsimonious tree of 733 steps with CI ϭ 0.795 and RI respectively, although there is some variation in karyotype fea- ϭ 0.803 (Fig. 6). This tree has the same topology in the genus tures (Xiao et al., 2002; Chen et al., 2003). The species also Incarvillea as the ITS tree, differing only in the relative ar- have very similar pollen morphology (Wang et al., 1997; Wei rangement of Sesamum indicum and Nematanthus strigillosus. et al., 2001; Chen et al., 2003). April 2005] CHEN ET AL.ÐMOLECULAR PHYLOGENY OF INCARVILLEA 629

Fig. 4. The single most parsimonious tree based on ITS sequences. Length ϭ 482, CI ϭ 0.720, RI ϭ 0.667, RC ϭ 0.481. Numbers below branches indicate bootstrap values (%), and branch length and decay value (in parentheses) are above branches. Names of subgenera are shown on right. O ϭ Incarvillea olgae. N ϭ Subgenus Niedzwedzkia.

Subdivision of the genusÐThe separate and combined analyses. This re¯ects morphology, as I. olgae is different analyses all support the existence of ®ve major clades: these from other species of subgenus Incarvillea in morphological are (1) subgenus Niedzwedzkia (I. semiretschenskia), (2) I. ol- characters such as the paniculate in¯orescence, opposite gae, (3) subgenus Amphicome, (4) Incarvillea, and (5) sub- leaves, calyx teeth base not swollen, and no calyx tube groove genus Pteroscleris (Fig. 2C). Two subgenera in the system of between calyx teeth (Table 3). These characters are peculiar Grierson (besides the monotypic subgenus Niedzwedzkia), Am- to I. olgae. We therefore suggest that I. olgae should be sep- phicome and Pteroscleris, are monophyletic groups. In all arated from subgenus Incarvillea as a new monotypic subge- analyses, the Amphicome clade has bootstrap support of more nus. I. potaninii, on the other hand, closely resembles I. si- than 50%, and the Pteroscleris clade is very strongly sup- nensis in all morphological features. ported (bootstrap value ϭ 100%). All groups suggested here are clearly distinctive on the basis In the system of Grierson, subgenus Incarvillea is composed of morphology (Table 3), allowing the genus to be readily of three species, Incarvillea olgae, I. sinensis, and the Mon- divided into ®ve formal taxonomic units. golian I. potaninii. Material of I. potaninii was not available Relationship between clades in the genusÐGrierson to us, but I. olgae was not related to I. sinensis in any of the (1961) drew tentative conclusions about the evolutionary re- 630 AMERICAN JOURNAL OF BOTANY [Vol. 92

Fig. 5. The strict consensus tree of four equally parsimonious trees based on trnL-F. Length ϭ 260, CI ϭ 0.892, RI ϭ 0.948, RC ϭ 0.846. Numbers below branches indicate bootstrap values (%), and branch length and decay value (in parentheses) are above branches. Names of subgenera are shown on right. N ϭ Subgenus Niedzwedzkia.Oϭ Incarvillea olgae. lationships between his four subgenera based on the distribu- tween this treatment and Grierson's is in the phylogenetic ar- tion of characters in the Bignoniaceae generally: he considered rangement of subgenus Amphicome and subgenus Incarvillea. subgenus Amphicome and subgenus Incarvillea relatively Our study suggests relationships between subgenera as follow: primitive, although he regarded the Incarvillea as more ad- (subgenus Niedzwedzkia (Incarvillea olgae (subgenus Amphi- vanced in view of the presence of an annual species. Subgenus come (subgenus Incarvillea, subgenus Pteroscleris)))). This Pteroscleris he regarded as advanced and more closely related supports Grierson's treatment, with subgenus Niedzwedzkia to subgenus Incarvillea (Fig. 2A). There are three subgenera basal, but with a separate position for I. olgae (Fig. 2). In this in China, and Fig. 2B shows the classi®cation used in Flora context, taking the subgeneric rank of the other four clades Reipublicae Sinicae (Wang et al., 1991). The difference be- into account, we think I. olgae should be accommodated in a April 2005] CHEN ET AL.ÐMOLECULAR PHYLOGENY OF INCARVILLEA 631

Fig. 6. The single most parsimonious tree based on the combined data set of ITS and trnL-F sequences. Length ϭ 733, CI ϭ 0.795, RI ϭ 0.803, RC ϭ 0.639. Numbers below branches indicate bootstrap values (%). Branch length and decay value (in parentheses) are above branches. Names of subgenera are shown near the tree. Five major branches and their distributions are shown on the right. O ϭ Incarvillea olgae.Nϭ Subgenus Niedzwedzkia.Cϭ Central Asia. CA ϭ Central Asia and Afghanistan. AH ϭ Afghanistan and Himalaya-Hengduan Mountains. H ϭ Himalaya-Hengduan Mountains. HM ϭ Area from Himalaya-Hengduan Mountains to Mongolia. new monotypic subgenus, which will be described in a future to dehiscent fruits with the septum parallel to the valves. Fruits paper. dehiscing perpendicular to valves may be a plesiomorphy Gentry (1980) maintained that it is very dif®cult to under- (Spangler and Olmstead, 1999). Incarvillea semiretschenskia stand the phylogeny within Bignoniaceae because of ``rampant maintains this possibly ancestral state and occupies the basal parallelisms and convergence in nearly every taxonomically position in the genus. Subgenus Pteroscleris is herbaceous and important characteristics.'' The same applies to the genus In- other groups are suffruticose, supporting the notion of Pter- carvillea. On the basis of the molecular data, however, it is oscleris as a group with several advanced characters. Some evident that some morphological characters are useful for de- populations of I. sinensis (subgenus Incarvillea) are annual, ®ning the subgenera (Table 3). which is also probably a derived condition. As regards pollen Incarvillea semiretschenskia is the only species with septif- morphology, in features such pollen grain size and the number ragal capsules. In the family, the phylogenetic arrangement of and shape of colpi, subgenus Pteroscleris has also been con- Gentry (1980) implied that dehiscent fruits with the septum sidered to possess advanced characters and Amphicome prim- perpendicular to the valves gave rise to indehiscent fruits and itive ones, while subgenus Incarvillea is transitional between 632 AMERICAN JOURNAL OF BOTANY [Vol. 92

TABLE 3. Comparison of morphology between ®ve major clades.

Subgen. Subgen. Niedzwedzkia Subgen. Amphicome Incarvillea Subgen. Pteroscleris Incarvillea olgae Habit Suffruticose Suffruticose Suffruticose Herbaceous Suffruticose Leaf Alternate Alternate Alternate Alternate or radical Opposite Calyx teeth 5 5 10 5 5 Calyx Tube at base Tube Tube Tube Tube Calyx teeth base Not swollen Not swollen Swollen Not swollen Not swollen In¯orescence Racemose Racemose Racemose Racemose Paniculate Anther Glabrous Pilose Glabrous Glabrous Glabrous Capsule shape Ovate Elongate, linear cylindri- Cylindrical Cylindrical Cylindrical cal Capsule texture Subligneous Fibrous Coriaceous Subligneous Coriaceous Capsule dehiscence Septifragal Loculicidal Loculicidal Loculicidal Loculicidal Capsule wing Six wings, longitudinal No No No No Seed Opaque wing Coma at ends Hyaline wing Opaque wing Hyaline wing

the two (Chen et al., 2003). In karyomorphology too, features LITERATURE CITED such as the size of chromosomes and the asymmetry of the karyotype, suggest that subgenus Amphicome has primitive BEARDSLEY, P. M., AND R. G. OLMSTEAD. 2002. Rede®ning Phrymaceae: the characters and subgenus Pteroscleris advanced ones, while placement of Mimulus, tribe Mimuleae, and Phryma. American Journal subgenus Incarvillea is intermediate between the two subgen- of Botany 89: 1093±1102. era (Chen et al., 2004). These characteristics are all consistent BOWDEN, W. M. 1940. Diploidy, polyploidy and winter hardiness relation- with Grierson's views of the evolution of these three subgenera ships in the ¯owering . American Journal of Botany 27: 357±371. BREMER, K. 1988. The limits of amino acid sequence data in angiosperm (1961) (Fig. 2A), with the exception of I. olgae. phylogenetic reconstruction. Evolution 42: 795±803. In subgenus Pteroscleris, two sections have been proposed, BREMER, B., K. BREMER,N.HEIDARI,P.ERIXON,R.G.OLMSTEAD,A.A. caulescent and acaulous (Zhao, 1985), but this division is not ANDERBERG, AND M. KAÈ LLERSJOÈ . 2002. Phylogenetics of based supported by our studies. Of the species sampled from sub- on 3 coding and 3 non-coding chloroplast DNA markers and the utility genus Pteroscleris, two species, I. beresowskii Batalin and I. of non-coding DNA at higher taxonomic levels. Molecular Phylogenetics lutea Bur. et Fanch., are caulescent. Incarvillea compacta and Evolution 24: 273±300. BUREAU, E. 1864. Monographie des BignoniaceÂes: recherches sur les carac- Maxim. has a short stem, and the others, I. mairei (LeÂvl.) teÁres, la composition, les limites de cet ordre naturel de plantes. L. Mar- Grierson, I. dissectifoliola Q. S. Zhao, I. zhongdianensis Grey- tinet, Paris, France. Wilson, I. younghusbandii Sprague, and I. delavayi Bur. et BURTT, B. L., AND A. J. C. GRIERSON. 1953. The name Incarvillea arguta. Fanch., are acaulous. Two caulescent species are scattered Notes from the Royal Botanic Garden Edinburgh 19: 155±156. among the acaulous species in all three analyses (Figs. 4±6). CHEN, S. T., K. Y. GUAN,Z.K.ZHOU, AND T. F UJIKI. 2003. Pollen mor- phology of the genus Incarvillea (Bignoniaceae). Acta Botanica Yunnan- ica 25: 457±467. Biogeographical considerationsÐTwo basal subgenera, CHEN, S. T., Z. K. ZHOU,K.Y.GUAN, AND M. NAKATA. 2004. Karyomor- phological study of Incarvillea Juss. (Bignoniaceae) and its implications Niedzwedzkia and ``Incarvillea olgae,'' are endemic to central in distribution and taxonomy. Botanical Journal of the Linnean Society Asia, while the apparently most derived group, subgenus Pter- 144: 113±121. oscleris, is distributed in the eastern Himalaya and the Heng- DON, G. 1838. A general history of the dichlamydous plants. J. G. and F. duan Mountains. The short branch lengths on the phylogenetic Rivington, London, UK. trees within subgenus Pteroscleris may indicate that these spe- FARRIS, J. S., M. KAÈ LLERSJOÈ ,A.G.KLUGE, AND C. BULT. 1994. Testing cies have undergone a recent radiation, perhaps related to the signi®cance of incongruence. Cladistics 10: 315±319. FEDOROV, A. N. 1969. Chromosome numbers of ¯owering plants. Akademiaà uplift of the Himalaya-Hengduan massifs. The Himalaya and Nauk, Leningrad, USSR. Hengduan Mountains are currently the center of diversity for FELSENSTEIN, J. 1985. Con®dence limits on phylogenies: an approach using the genus with three subgenera and 14 species (Fig. 3). This the bootstrap. Evolution 39: 783±791. suggests the dispersal of Incarvillea may have been from cen- GENTRY,A.G.,AND A. S. TOMB. 1979. Taxonomic implications of Bignon- tral Asia to eastern Asia. If so, this dispersal must have hap- iaceae palynology. Annals of the Missouri Botanical Garden 66: 756± pened at least before the Miocene, after which the Himalaya 777. GENTRY, A. H. 1980. Bignoniaceae. Flora Neotropica Monograph 25: 1± Mountains would have been an effective barrier to migration. 130. Subgenus Amphicome is interesting because it has one species GOLDBLATT,P.,AND A. H. GENTRY. 1979. Cytology of Bignoniaceae. Bo- shared between central Asia and eastern Asia, perhaps indi- tanica Notiser 132: 475±482. cating ancient dispersal of this relatively basal group. Pteros- GREY-WILSON, C. 1998. A new look at the subgenus Pteroscleris. The New cleris is adapted to the high altitude environment of the Him- Plantsman 1998: 76±98. alaya-Hengduan Mountains and has there undergone a consid- GRIERSON, A. J. C. 1961. A revision of the genus Incarvillea. Notes from erable radiation. The continuous distribution of subgenus In- the Royal Botanic Garden Edinburgh 23: 303±354. HIGGINS, D. G. 1994. CLUSTAL W: multiple alignment of DNA and protein carvillea from Himalaya-Henduan Mountains to the Russian sequences. In A. M. Grif®n and H. G. Grif®n [eds.], Computer analysis Far East, on the other hand, could be at least partly the result of sequence data, part II, 307±318. Humana Press, Totowa, New Jersey, of Quaternary expansion as part of its range was glaciated. USA. April 2005] CHEN ET AL.ÐMOLECULAR PHYLOGENY OF INCARVILLEA 633

MICKEVICH,M.F.,AND J. S. FARRIS. 1981. The implications of congruence WANG, F. S., N. F. CHIEN, AND Y. L. ZHANG. 1997. Pollen ¯ora of China, in Menidia. Systematic Zoology 30: 351±370. 2nd ed., 97±101. Science Press, Beijing, China. OLMSTEAD, R. G., M. ZJHRA,S.GROSE,A.ECKERT, AND R. E. SPANGLER. WANG, W. T., K. Y. PAN,Z.Y.ZHANG, AND Z. Y. LI. 1990. Incarvillea Juss. 2002. Molecular phylogeny, evolution, and classi®cation of the Bignon- Flora Reipublicae Popularis Sinicae 69: 34±49. iaceae. [online proceedings] http://www.2002.botanyconference.org/ WEI, Z. X., H. WANG, AND Q. J. JIN. 2001. Pollen morphology of Bignon- sauthor.shtml (Abstract). iaceae in China. Acta Botanica Yunnanica 23: 225±230. SPANGLER,R.E.,AND R. G. OLMSTEAD. 1999. Phylogenetic analysis of WHITE, T. J., T. BRUNS,S.LEE, AND J. TAYLOR. 1990. Ampli®cation and Bignoniaceae based on the cpDNA gene sequences rbcL and ndhF. An- direct sequencing of fungal ribosomal RNA genes for phylogenetics. In nals of the Missouri Botanical Garden 86: 33±46. M. Innis, D. Gelfand, J. Sninsky, and I. White [eds.], PCR protocols: a SUGIURA, T. 1936. Studies on the chromosome numbers in higher plants, with guide to methods and application, 315±322. Academic Press, San Diego, special reference to cytokinesis. Cytologia 7: 544±595. California, USA. SWOFFORD, D. L. 2001. PAUP*: phylogenetic analysis using parsimony (* XIAO, H., Q. X. ZHOU,Z.J.GU, AND K. Y. GUAN. 2002. Karyomorphology and other methods), version 4.0 Beta10. Sinauer, Sunderland, Massachu- of six Incarvillea species. Acta Botanica Yunnanica 24: 87±93. setts, USA. ZHAO, Q. S. 1985. A study on the genus Incarvillea Juss. of the Transverse TABERLET, P., L. GIELLY,G.PAUTON, AND J. BOUVET. 1991. Universal prim- Mountain Range. Exploration of Nature 12: 170±174. ers for ampli®cation of three non-coding regions of choloroplast DNA. ZHAO, Q. S. 1988. A new species of Incarvillea (Bignoniaceae) from Sichuan. Plant Molecular Biology 17: 1105±1109. Acta Phytotaxonomica Sinica 26: 78±79. VASSILCZENKO. 1957. Niedzwedzkia. Notulae Systematicae Unionis Rerum- ZIMMER, E. A., E. H. ROALSON,L.E.SKOG,J.K.BOGGAN, AND A. IDNURM. publicarum Socialisticarum Sovieticarum 18: 219±224. 2002. Phylogenetic relationships in the Gesnerioideae (Gesneriaceae) VASSILCZENKO. 1958. Niedzwedzkia. Flora Unionis Rerumpublicarum So- based on nrDNA ITS and cpDNA trnL-F and trnE-T spacer region se- cialisticarum Sovieticarum 23: 5. quences. American Journal of Botany 89: 296±311.