Plant Syst. Evol. 239: 171-185 (2003) Plant Systematics DOI 10.1007/s00606-003-0003-4 and Evolution Printed in Austria
A phylogeny of the Munnoziinae (Asteraceae, Liabeae): circumscription of Munnozia and a new placement of M. pevfoliata
H.-G. Kim1 2, V. A. Funk2, A. Vlasak13, and E. A. Zimmer1 2
'Laboratory of Analytical Biology, Museum Support Center, Smithsonian Institution, Suitland MD ^Department of Systematic Biology, Division of Botany, National Museum of Natural History, Smithsonian Institution, Washington DC ^Current address: Department of Biology, University of Pennsylvania, Philadelphia, PA
Received March 19, 2001; accepted January 23, 2003 Published online: July 31, 2003 © Springer-Verlag 2003
Abstract. The tribe Liabeae (Compositae, Cicho- having black or dark brown anther theca and rioideae) comprises three subtribes, Liabinae, sordid or reddish pappus and re-organized. Munnoziinae, and Paranepheliinae. For one of these, the Munnoziinae, which contains the genera Key words: Asteraceae, Cichorioideae, Liabeae, Munnozia, Chrysactinium, Erato, and Philoglossa, Munnoziinae, Liabinae, Paranepheliinae, Phylogeny, the nuclear ITS (internal transcribed spacer) internal transcribed spacer (ITS), parsimony and region was sequenced to examine the monophyly maximum likelihood analyses, monophyly, para- of the subtribe and the core genus Munnozia phyly, biogeography. within it. Thirty-six samples representing four currently recognized genera of Munnoziinae and two outgroups were included in this study. Mo- One of the most successful families in the lecular phylogenetic analyses confirm the close flowering plants, the Compositae, consists of relationship of Munnozia with Chrysactinium, and approximately 20 tribes with distinctive mor- Erato with Philoglossa. However, the monophyly phologies and molecular markers (Kim and of the Munnoziinae and Munnozia is not support- Jansen 1995, Bremer 1996). Only one tribe is ed, in disagreement with the current morpholog- neotropical in its origin and distribution, the ical findings. The discrepancies were attributed to Liabeae. The Liabeae has approximately 180 the placements of Munnozia perfoliata outside the species grouped into 15 genera. They are Munnoziinae and Munnozia, and Chrysactinium divided into three subtribal groups, Munnozii- within Munnozia. The resulting tree indicates that nae, Paranepheliinae, and Liabinae, based on first, M. perfoliata needs to be moved out of palynological characters (Robinson 1983). the munnoziinae and second, Chrysactinium originated from within Munnozia. For the first Results from the studies by Robinson finding, morphological and palynological revalu- (1983), Bremer (1994), and Funk et al. (1996) ation of this species with allegedly related species demonstrated the monophyly of the first two reveals additional support in agreement with subtribes but not the Liabinae. These modern molecular data. Therefore we propose that the findings differ from older treatments as is genus Munnozia be re-delimited to the members evidenced by the controversial history of 172 H.-G. Kim et al.: A phylogeny of the Munnoziinae (Asteraceae, Liabeae) classification (Cassini 1823, 1825, 1830; Les- Munnozia, the most species-rich genus in sing 1832; De Candolle 1836; Weddell 1855- the tribe, contributes significantly to morpho- 1857; Hoffmann 1890-1984; Rydberg 1927; logical and biogeographical diversity in the Blake 1935; Cabrera 1954; Sandwith 1956; subtribe and tribe. The characteristic chaffy D'Arcy 1975; Cronquist 1955; Carlquist 1976; receptacle was originally used to define Mun- Nash and Williams 1976) and point out the nozia by Cassini and various other authors current ambiguity in the placement and rela- (1823, 1825, 1830). Later the series of works by tionships of the three sub tribes. An example of Robinson (1974, 1983) modified the diagnostic the blurred areas of tribal and subtribal characters for the genus. Consequently, Rob- relationships has been focused on the Liabi- inson transferred several Liabum species and nae. Since 1983, the circumscriptions of Mun- related to Munnozia. Munnozia is taxonomi- noziinae have not been questioned and all cally complex due not only to the large number recent studies agree on the monophyly of of species in the genus, but also to ambigu- Munnoziinae. To the contrary, our prelimi- ously demarcated generic delimitation from nary result of molecular investigation on Liabum. Nevertheless, the naturalness of the Liabeae draws attention to this subtribe, genus has not been examined in the phyloge- revealing that the currently circumscribed netic context. As the first step in a compre- Munnoziinae and Munnozia are not mono- hensive phylogenetic study of the Liabeae, we phyletic. have sequenced DNAs of the Munnoziinae Circumscribed by the synapomorphic char- using the nuclear ITS (internal transcribed acter, black or very dark brown anther thecae, spacer) region. The goals of this study are to the subtribe Munnoziinae contains four genera (1) clarify the phylogeny of the Munnoziinae, and about 60 species. The Munnoziinae is (2) examine the monophyly of Munnozia readily divided into two groups (Robinson itself, (3) assess the phylogenetic position of 1983, Bremer 1994, Funk et al. 1996): one M. perfoliata, and (4) evaluate the usefulness lineage includes Munnozia and Chrysactinium, of the ITS markers for the generic and species having spines on pollen grains regularly dis- level relationships in the tribe. posed, subquadrate raphids in the cypsela walls, and the other lineage includes Erato and Philoglossa, having stiff hairs with bulbous Materials and methods bases and 2^1 angles on the achenes (Robinson Taxon sampling. Thirty-six samples from twenty- 1983). A few traditional and cladistic works six species of the Munnoziinae and nine species have presented the intergeneric relationship of from two outgroup genera were used in this study the Munnoziinae. However, the circumscrip- (Table 1). These cover the morphological and tion and relationship to the sister group of biogeographic diversity of both the ingroup, the Munnoziinae still remains controversial due subtribe Munnoziinae, and the outgroup. Taxon not only to the lack of congruence among the names and voucher information with geographical works but also to the lack of understanding of distribution and their collecting area are listed in the core genus. According to the treatment Table 1. Most voucher specimens are housed in the (Robinson 1983), Munnozia has ca. 46 species US National herbarium. All of the samples used in and represents the morphological diversity and this study are from personal collections, identified by the primary collector. biogeographic distribution pattern of the Ingroup sampling. The ingroup contains the subtribe. Munnozia may be used to understand four genera of the subtribe Munnoziinae. For the origin of speciation and pattern of diver- Munnozia, the largest genus in the subtribe with sification of the subtribe. Therefore, under- over ca. 40 species, we have included 14 species standing of Munnozia can be pivotal for representing all of the morphologically distinctive examining the evolution and phylogeny of clades in the genus. For instance, M. campii has the Munnoziinae. white rays and M. jussieui has whitish rays H.-G. Kim et al.: A phylogeny of the Munnoziinae (Asteraceae, Liabeae) 173
d 2 a O ON NC ON r- OO CM tN •* m r- in ^o ON m NO in ^h en cd cd u o "C 'C oj OS C3 a C OH „ > 3 a TO ]> ]> 43 C3 43 43 u O u 53 T3 43 0 O o S E 'a 3 0 O 5 O CO E s 3 CO CO CO _3 s _3 T3 _o N OS a m •- U O O o 3 - o 0 3 3 3 U U s 0 U O C s o o Q S3 a s s PH pj PJ > PH PH PH U _2 ' cd ,-> C3 o u o o "u OS u a C3 0 -g 3 3 3 u 43 43 43 Pi a 5 s o and central Ame o and central Ame o o s s o c O Bolivia •0 43 3 -g a 1 -a TS •o •o %; 0) TO OJ O U E •o , Bolivia mbia , Bolivia mbia, Venezuela, a E E -3 OS - aj csj C5j 03 csj csj a a cd a & _3 csj 330 3 _o 3 60 0 3 _3 oO _0o E3 „ a C a 3 3 3 3 3 TO -H tN u 3 3 q R 2 ^% 3 R Q 60 %; .9 .9 .» § C3 • SH 5 5 § bo be X a a 5J a Q 5*2»' O O a a So R 2 3 .be JS ,0 ,0 ,0 GO S S » b« -R .a « 5 O, O, O, •? t4—1o a5 Ea < « « a a e .« « « a a e .« wi N N K| N N N 1 1 Q Q s 0 0 O O O O O O 3 a aaaRRRRRR R R R R R R R R R -c -0-0-S3RRRRRR R R R R R R R R R 3 a k" k" -C3 C3- Q- a a a js a a a a a K u a ,g .§ .§ a a ^a ^a a a H 0 bq bq hq uj < ^CjHjZs^g-g-Zgls %s «q «^ "^ "^ 174 H.-G. Kim et al.: A phylogeny of the Munnoziinae (Asteraceae, Liabeae) o Z oo o m o m rt r~ NO in a m a. a. HO _S -H CC >H o t3 to ss r: - o O 3 3 U U 3 p. u w W o o . J9 O ^ cd < 13 '3 3 3 | (D oc O -^ o -s 'x 3 O Q. o o - PO CQ S2 SUO - < 3 P 3 3 u M NO >, u 00 J3 H3 =k o < /"NX 3 < < < 0 t/2 00 < 1/2 GO £ 3 3 O o cd s a X3 J= o o o O 3 Pi Pi 3 Ep2 H 9 9 PP u OP o o 3 3 3 ° Pi Pi s s >^ _. s 3 £ 3 15 'iT O g« T3 O ci g N N 3 3 _. N o a o O 3 3 •^cd • 93 a 3 3 15 u OJ lH J5 3 u X ^ PC ^ i4 a> X ii-1/2 < cd S -3 Q Q 7g 3 .g -g K S -S5 s 5 -3 K S •C bo u CO 5, S, St 5S SS o 3 _« a o o .q .« « N N N O a s K S S S? bo 3 % K s s -2 ^§ 3 3 a u u u z u -5 s s •S -S -S Q H a ^^ CO % % co % * H.-G. Kim et al.: A phylogeny of the Munnoziinae (Asteraceae, Liabeae) 175 (sometimes turning lavender) that are rarely DNA extraction and PCR amplification. Most yellow. Munnozia nivea and M. pinnatipartita, of the leaf material used was collected in the field which belong to a small separate subgenus, have and stored in silica gel. In a few cases, either frozen the leaves pinnate or pinnatifid while all other tissue or herbarium specimens from US or TEX members of the other subgenus have simple leaves. served as source material (Table 1). Total genomic Munnozia perfoliata is a small creeping annual DNA was isolated from leaves using a modified 2X while most of the members of this genus are CTAB procedure (Doyle and Doyle 1987). Some perennials, having either a shruby or subshruby DNAs contaminated with high concentrations of habit. Eventually, two different individuals of the polysaccharides were purified using a Geneclean same collection of M. perfoliata were included IIR kit (Bio 101 Inc.). The concentration of DNA because of the unusual position within the analysis. samples was checked on 1% agarose gel/IX TBE Although they were from the same collection buffer run with Hindlll-digested lambda and numbers, they were from separate extractions of Haelll-digested DNA standards. herbarium and fresh material and the habit of the Amplifications were performed in 50 ul reac- plant made it likely that they were sampled from two tions with 1 ul of 10-40 ng genomic DNA, 1 ul of different individuals. The other Munnozia species 10 uM primers (ITS5HP and 1TS4), 5 ul of 10X Tfl sampled were from throughout the range of the polymerase buffer, 2.5 mM of each dNTP and genus from an endemic in Costa Rica, M. wilburii, to 0.5 ul of 5 units/ul Taq polymerase (Promega). a widespread variable taxon from the Andes, DNA was initially denatured at 94 °C for 5 min, M. senecionidis. Three species that have particular followed by 30 amplification cycles consisting of morphological problems are represented by two 1 min denaturation at 94 °C, 1 min annealing of collections each for a total of 17 collections sampled primer at 50 °C and 1 min 30 sec extension at from Munnozia. The other genus, Chrysactinium, 72 °C. Amplification was terminated by a final includes six species and we sampled two collections extension cycle of 72 °C for 7 min and a 4 °C of the largest and most variable species C. acaule soaking file. The PCR product was precipitated from different sites in Ecuador. The other two genera using a 20% Polyethylene Glycol solution (PEG are represented by two distinct samples of a single 8000/2.5 M NaCl). When necessary, additional variable species. The genus Philoglossa includes five purification of the amplified ITS region was species, but the widest ranging and most variable is accomplished by gel purification followed by P. mimuloides. One accession each of P. mimuloides beta-agarase (New England Biolab) treatment. was sampled from Ecuador and Peru. Two of the The two amplification primers, ITS 5HP (Suh four species of Erato were represented in the study, et al. 1993) and ITS 4 as well as two internal Erato polymnioides from both Ecuador and Peru, primers (ITS 2 and ITS 3) were used for sequencing and E. vulcanica from Costa Rica and Ecuador. (White et al. 1990). Outgroup sampling. Sinclairia and Liabum DNA sequencing and alignment. The purified were used as outgroups in the molecular analysis templates were labeled by cycle sequencing using based on their biogeographic diversity and on the FS chemistry dye terminators according to condi- results of the previous morphological analysis tions recommended by the supplier (Applied Bio- (Funk et al. 1996). The genus Liabum is represented systems). Excess dye terminators were removed by three taxa. One variable taxon, L. bourgeaui, with Sephadex (G-50) spin columns. Sequences represented by two collections, is native to Mexico were obtained on Applied Biosystems model 373 and Central America, and was collected from Costa and and 377 automated fluorescent DNA sequencers. Rica. The second species Liabum igniarium is from Data collection was carried out with Applied Colombia and Ecuador, and the final species, Biosystem Sequence analysis• 3.1, implemented L. baharonense, is an endemic from the Dominican on Macintosh G3 computers, followed by contig Republic. The second outgroup, Sinclairia, is assembly using Sequencer• from Gene Codes. represented by four species, S. angustissimum from All sequences were manually realigned using Mexico, S. liebmannii and S. vagans from Guate- Se-Al version l.Oal (Rambaut 1996). There were no mala, and S. polyantha from Costa Rica. Thus, a ambiguous regions in the alignment. The ITS region total of seven species were used to define the boundaries were defined by comparison with previ- outgroup. ously published sequences of Asteraceae (Kim and 176 H.-G. Kim et al.: A phylogeny of the Munnoziinae (Asteraceae, Liabeae) Vernonieae Ferreyranthus Oligactis — Liabum Bishopanthus Cacosmia Chionopappus Microliabum —— Liabellum rc; —- Sinclairia • Pseudonoseris L—— Paranephelius Erato Philoglossa Chrysactinium Munnozia Vernonieae Pseudonoseris Erato Philoglossa Chrysactinium Munnozia Microliabum Liabellum Sinclairia Cacosmia Figs. 1 and 2. Redrawn from Funk et al. Chionopappus (1996). The two equally parsimonious trees Bishopanthus of the Liabeae generated from 42 morpho- Ferreyranthus logical characters, Tree length (L) = 93, I Oligactis Consistency Index (CI) = 0.71, Retention ' Liabum Index (RI) = 0.44 Jansen 1994, Kim et al. 1998). All sequences are EST DESCENT on. Starting trees were con- submitted to GenBank (See Table 1 for accession structed using 100 replicates with random numbers). addition sequence. Assuming an unequal sub- stitution rate in the ITS region, the data were Phylogenetic analyses analyzed as follows: 1) ITS1 and ITS2 se- Parsimony analyses. Phylogenetic analyses quences were considered separately; 2) ITS1 were conducted using PAUP* 4.0 (Swofford and ITS2 data sets were combined; and 3) the 1998). All characters were unordered and entire ITS region as a whole was analyzed equally weighted. Gaps were coded as hyphens (ITS1, 5.8 rDNA and ITS2 (ITS region)). In (-) in the PAUP* analyses. Several ambiguous order to assess node support, bootstrap ana- sites were encountered in the ITS1 regions of lyses (Felsenstein 1985, Hillis and Bull 1993) Munnozia lyrata, M. nivea and M. gigantea. and decay analyses (Bremer 1988, Donoghue Those sites were coded using IUPAC (Inter- et al. 1992) were performed. In the bootstrap national Union of Pure and Applied Chemis- runs, PAUP was set for 100 bootstrap repli- try) ambiguity codes. cates with TBR and MULPARS options. Two To find the most parsimonious trees, heu- outgroup genera, Liabum and Sinclairia, were ristic searches were conducted using random selected based on the morphological grounds addition with TBR, MULPARS, and STEEP- cited above. To compare the length of the H.-G. Kim et al.: A phylogeny of the Munnoziinae (Asteraceae, Liabeae) 177 shortest trees, a Branch-and-Bound search was complete ITS regions (Fig. 3) are congruent performed with MULPARS ON. The tree with one of the trees randomly chosen with one length and its branching order were identical exception, the placement of Munnoziafosbergii. to the tree generated by the heuristic search. This taxon varies from being within Clade H in For this study, we present the tree obtained by the MP tree but is placed with M. gigantea and the Branch-and-Bound search. Clade E in the consensus tree for ITS 1. Despite Maximum likelihood. To compare and to the instability of the Munnoziinae (Clade D), all assess whether the short internal branch length of the strict consensus trees from the four data in several clades affects the placement of M. sets are congruent with respect to (1) recogni- perfoliata under different evolutionary criteria, tion of the currently circumscribed Munnozii- additional analyses were performed. First the nae as paraphyletic, (2) paraphyly of Munnozia, original data set of thirty-six was reduced to and (3) placement of M. perfoliata close to the fifteen taxa (Table 1). A maximum parsimony outgroup. However, the branching order within heuristic search was performed with the same Clade D (Fig. 3), which includes subclades, E, options as the ones for the original data sets, F, G, H, I, Munnozia lanceolata, M.foliosa and using PAUP 4.0* (Swofford 1998). Using the M.fosbergii, and within the Sinclairia clade are maximum parsimony tree, the program Model- incongruent among the data sets. The poor test 3.0 (Posada and Crandall 1998) was utilized resolution may be attributed to using a short to determine the model that fits best for the data and conservative region for this recently derived tested by the hierarchical likelihood ratio (LR) group. test (a = 0.01) under the nested requirement. To confirm the result of the parsimonious When the competing models were nested, the tree search, the maximum likelihood tree LR test statistic (5) is distributed as x2 distribu- search was conducted. Modeltest 3.0 version tion with degree of freedom equal to the identified TrNef + G with log likelihood score difference in number of free parameters between of 3069.2844 as the best model of the DNA two models (Huelsenbeck and Crandall 1997). evolution for the ITS data set. With the best For ITS data, the model selected is the Tamura- model TrNef + G chosen, a heuristic tree Nei model (Tamura and Nei 1993) with gamma- search was performed. Focusing on our inter- distributed site-to-site rate variation. With the est in the placement of M. perfoliata, the model determined, the maximum likelihood tree parsimonious tree search was performed with search was done using PAUP 4.0*. the same data set to compare the topologies. In both topologies (Figs. 3 and 5), M. perfoliata appeared outside the Munnoziinae clade with Results strong support (bootstrap 100%, DI > 10). Phylogenetic analyses. Of the 641 aligned ITS Proposed relationships based on ITS nucleotide positions, 210 appeared to be poten- data. The ITS trees generated by parsimony tially informative (33%), and 357 were constant and likelihood (Figs. 3 and 5) identified three (56%). Initially, seventy-four characters includ- major clades: Clade A and the Munnoziinae ing gaps, were treated as ambiguous or missing clade consisting of Clades C + D. (11 %). We performed both maximum parsimo- Clade A. Clade A, under both analytical ny (MP) and maximum likelihood (ML) an- methods, consists of two accessions of Mun- alyses using the aligned ITS data sets. With MP, nozia perfoliata. Both maximum parsimony and the combined data sets, as well as alignments for likelihood are consistent in the placement of M. each of the two separate regions of the spacer perfoliata near the base of the ITS cladogram, were analyzed. Since many trees with equal and below Sinclairia which was used as an lengths were generated from each data set, strict outgroup (Clade B, Figs. 3 and 4). Clade A is consensus trees were also utilized to compare strongly supported (bootstrap 100%, DI > 10, the branching orders (Fig. 4). The trees of the Figs. 3-5). This placement is also suggested by 178 H.-G. Kim et al.: A phylogeny of the Munnoziinae (Asteraceae, Liabeae) Munnozia pinnulosa = Munnozia pinnulosa Munnozia gigantea Munnozia hastifolia - Munnozia pinnatipartita~\ F Munnozia nivea ' Munnozia senecionidisrx Munnozia senecionidiS2 H Munnozia wilburii •Munnozia senecionidis3 Munnozia fosbergii 7j— Chrysactiniumunrysactimum acauleacauie ~i 100M cP— Chrysactinium acaule —I Munnozia lyrata Munnozia jussieui 99/ 65/+2 c Munnozia campii 3% +10 9748 c Munnozia foliosa 3 Fig. 3. One of the 22 most 13 Munnozia lanceolata I— Erato polymnioides equally parsimonious trees 8B/+4 Erato polymnioides obtained from analyses of 61A3 16 11 j— Erato vulcanica ITS sequence data matrix 98/+8 100M0I—Erato vulcanica (L = 530, CI = 0.704, 22. 16 i— Philoglossa mimuloides Rl = 0.863). The numbers 100/ iooz>io I— Philoglossa mimuloides—' above the branch indicate >10 3 i—Sinclairia poiyantha the number of characters Sinclairia vagans changed under ACC- 18, 22 \- Sinclairia moorei TRAN optimization using ICQ 00/>10 >10 -h Sinclairia angustissima PAUP 4.0. The numbers Sinclairia liebmannii below the branches indi- cate bootstrap value and 15 100M0 Munnozia perfoliata2 3' decay index, respectively. Liabum barahonense The black bar mapped on Liabum igniahum the node indicates the non- Liabum bourgeaui homoplastic informative Liabum bourgeaui gaps the sequence divergence estimate for M. perfo- Subtribe Munnoziinae. The subtribe Mun- liata relative to the majority of Munnozia noziinae clade consists of two lineages (Fig. 3): examined, ranging from 11.75%-18.75%, one (Clade C) containing Philoglossa and Erato, making it the most distant species from core and the other (Clade D) with Munnozia and Munnozia. To force M. perfoliata into the Chrysactinium. The parsimony tree supports Munnozia (Clade D) would required 22 extra their close relationships with 13 synapomor- steps over the most parsimonious arrangement. phies (bootstrap 98%, DI = 7) and their sister Thus, it appears that M. perfoliata is an relationship is also supported in the strict outgroup for Munnoziinae as well as Sinclairia. consensus tree (Fig. 4). Phylogenetic analyses Clade B. The five species of Sinclairia used of ITS sequence data have shown that as as part of the outgroup form a monophyletic currently circumscribed the subtribe is not group with strong support (bootstrap 100%, monophyletic (Figs. 3-5) because of the place- DI > 10) in the ITS tree. This clade appears to ment of M. perfoliata outside the Munoziinae be sister to the subtribe Munnoziinae. clade. In addition, Chrysactinium is nested H.-G. Kim et al.: A phylogeny of the Munnoziinae (Asteraceae, Liabeae) 179 Munnozia pinnulosa — rC Munnozia pinnulosa Munnozia hastifolia Munnozia gigantea Munnozia pinnatipartita Munnozia nivea Munnozia senecionidisi rC Munnozia senecionidis2 Munnozia wilburii Munnozia senecionidisi Munnozia fosbergii Munnoziinae Chrysactinium acaule (C+D) -C Chrysactinium acaule Munnozia lyrata Munnozia jussieui •C Munnozia campii Munnozia foliosa -c Munnozia foliosa Munnozia lanceolata Erato polymnioides HZ Erato polymniodes Erato vulcanica Erato vulcanica Philoglossa mimuloides Philoglossa mimuloides —I Sinclairia polyantha — Sinclairia vagans K5 Sinclairia moorei Outgroup Sinclairia angustissima (B) Sinclairia liebmannii — Munnozia perfoliatal — Munnoziinae Munnozia perfoliata2 — (A) Liabum barahonense — Fig. 4. The strict consensus tree of Liabum igniarium 22 most equally parsimonious Outgroup Liabum bourgeaui trees, generated using ITS se- Liabum bourgeaui quence data inside Munnozia. If both Chrysactinium and M. together as closely related taxa. As mentioned perfoliata are forced to meet the monophyly of above, the ITS tree shows Munnozia to be Munnoziinae, 26 extra steps are required rela- paraphyletic. The branch leading to Chrysac- tive to the shortest tree for the ITS sequence tinium is very long, with 57 autapomorphies, in data. fact it is the longest on the cladogram. To Clade C. The ITS phylogeny (Figs. 3 and disrupt the integrated relationship of Chrysac- 4) supports the strong sister relationships tinium within Munnozia, four extra parsimoni- between Philoglossa and Erato, and identifies ous steps are required. these two genera as a monophyletic group. The Within Munnozia (Clade D, Fig. 3), the ITS clade is stable with strong support (bootstrap tree identifies a core Munnozia group consisting 98%, DI = 8). The ITS tree is consistent with of the clades F, H, I and M.forsbergii with weak the result from morphological cladistic analy- support (bootstrap 55%, DI = 3). The rest of the ses (Robinson 1983, Funk et al. 1996). clades within Clade D are weakly resolved, Clade D: Clade D (Fig. 3) contains the forming a polytomy among M. foliosa, genera Munnozia and Chrysactinium. These M. lanceolata, M. lyrata and Clade G. With two genera have traditionally been placed respect to floral and leaf features, there are two 180 H.-G. Kim et al.: A phylogeny of the Munnoziinae (Asteraceae, Liabeae) Munnozia pinnulosa 59 Munnozia pinnatipartita 67 - Munnozia senecionidis 71 Munnozia fosbergii 100 Chrysactinium acauie 55 — Munnozia lyrata Munnoziinae 98 Erato vulcanica 100 100 — Philoglossa mimuloides — Sinclairia angustissima 98 100 •— Sinclairia polyantha Munnozia perfoliata Fig. 5. The maximum likelihood • Liabum igniarium tree of 15 ITS sequences generat- ed using the best model Liabum barahonense Trnef + G. The data set for max- Liabum bourgeaui imum likelihood reduced from the original data set of thirty-six. Liabum bourgeaui The numbers below the branches indicate bootstrap value. The scale bar corresponds to 0.05 0.05 substitutions/site substitutions per site clades of interest identified in the ITS tree variable in morphology and the most frequently (Fig. 3). One is Clade F (M. pinnatipartita and collected species is M. senecionidis. Three spe- M. nivea) and the second is Clade G {M.jussieui cies are sequenced and M. senecionidisl and 2 and M. campii). The strongly supported, Clade are placed on the ITS tree close to M. wilburii F, belongs to the subgenus Kastnera sensu and M. senecionidis! (Fig. 3), forming a poly- Robinson and Marticorena (1986) with boot- tomy. Even though these taxa are unresolved strap value of 88% and decay index of 7. Clade relative to each other, this clade is strongly G is a morphologically distinct group by whitish supported (bootstrap 100%, DI = 7). flower, and is identified in the ITS tree. How- Variability of the ITS region. The results ever, Clade G is not strongly supported (boot- of phylogenetic analyses of all four separate strap 65%, DI = 2) and its relationship to sister data sets are summarized in Table 2. The ITS group is not resolved in the ITS tree (Figs. 3 and + 5.8S region within Munnozia varied from 4). Clade I, sister to Clade F, consists of 624 to 629 bp, with M. lyrata having the M. pinnulosa, M. gigantea and M. hastifolia. longest length. Erato polymnioides was the This clade is fairly stable (bootstrap 89%, shortest (617 bp). The 5.8S region of all taxa DI = 5). Of all the species in Munnozia, the most examined was 169 base pairs long, 4-5 base H.-G. Kim et al.: A phylogeny of the Munnoziinae (Asteraceae, Liabeae) 181 Table 2. A summary of the results of phylogenetic analyses using parsimony for four separate data sets (ITS1, ITS2, ITS1 + 2, and ITS1 + 5.8S + ITS2). The asterisk (*) indicates the sequence divergence in pair- wise comparisons among all taxa examined ITS 1 ITS 2 ITS 1& 2 ITS1/2/5.8S Total length (bps) 256 216 472 641 Length variation without gaps 244-255 209-211 454^64 617-629 #s of the most parsimonious trees 14 52 52 22 Tree length 257 211 487 530 Consistency Index (CI) 0.743 0.706 0.698 0.704 Retention Index (RI) 0 896 0.865 0.866 0.863 #s informative characters (%) 117(46%) 83 (3894) 200 (42%) 210 (33%) #s variable chr, but uninformative 24(9%) 30 (14%) 54(11%) 74 (12%) % of G + C content 0.48 0.53 0.50 0.51 *(%) seq. divergence 0.3-28.6% 0.4-20.8% 0.4-27.4% 0.3-21.5% pairs longer than those of other angiosperms The results of our DNA study are consis- published (Baldwin et al. 1995, Wen and tent with Bremer's (1994) and one of the Zimmer 1996, Kami et al. 2000). Both ITS1 scenarios suggested by Funk et al. (1996; and ITS2 contributed to the spacer length Fig. 2). Considering the traditional delimita- variation, with the degree of length variation's tion of Munnoziinae sensu Robinson (1983), being similar between the two regions (Ta- the ITS tree identifies two separate groups, one ble 2). However, the ITS1 contained more including M. perfoliata and the other consist- informative characters (46%) with a lower GC ing of the remainder of the subtribe (Clade C content (53%). The entire sequence alignment and D). However, because of the close place- contained 45 gaps (7.2%). Of those, 27 were ment of Sinclairia, the outgroup, to the mem- informative (60%). The non-homoplastic in- bers of the Munnoziinae clade the final formative gaps are mapped on Fig. 3. At the conclusions on circumscription of Munnozii- interspecific level the ITS sequence divergence nae and its sister relationship need to wait until ranged from 4.42% to 18.74% within the an on-going study on the tribe is completed (in ingroup, and from 7.82% to 21.5% between preparation). In particular, the large genus ingroup and outgroup. Chrysactinium acaule is Liabum needs comprehensive study, first be- the most divergent (20.65%) among ITS cause it is known to be morphologically and sequences examined. biogeographically diverse, and second because the ambiguity of its generic delimitation from Munnozia has been documented. Discussion The Munnoziinae clade now consists of Monophyly and sister relationship. All of the two major lineages. One includes Erato and parsimony analyses agree with one another on Philoglossa, which have been recognized pre- four points. First, of the taxa examined in this viously as a group supported by several analysis Sinclairia (Clade B) is the sister group morphological synapomorphies: pollen grains to the Munnoziinae clade (Clade C and D); with regularly dispersed spines, and leaves second, the subtribe Munnoziinae as currently and stems with tomentum (Robinson 1983, circumscribed is not monophyletic; third, the Funk et al. 1996). The ITS phylogeny cor- currently circumscribed genus Munnozia may roborates the morphological study showing a not be monophyletic; and fourth, M. perfoliata strong support (bootstrap 100%, DI = 8) for (Clade A) needs new taxonomic placement this clade. The other, consisting of the Mun- within the Liabeae. nozia clade and Chrysactinium, has been 182 H.-G. Kim et al.: A phylogeny of the Munnoziinae (Asteraceae, Liabeae) considered to be a congeneric group based on correct for phylogenetic inference, we propose the reduced acaulescent habit, long scapose that the classification of Munnozia be reorga- heads, black anther theca and regularly dis- nized in the hope of clarifying the generic posed spines on the pollen wall (Robinson delimitation and interspecific relationship. 1983, 1986). Their close relationship has been More extensive sampling of both Munnozia emphasized by previous workers (Robinson and Chrysactinium will help clarify the phy- 1983, Bremer 1994, Funk 1996, Figs. 1 and logeny and relationship of this genus within 2). Our ITS tree (Figs. 3 and 4) places Munnoziinae. Chrysactinium within Munnozia with moder- Phylogenetic position of M. pevfoliata. ate support. This result supports the previous Munnozia perfoliata is an annual herbaceous morphological findings. The long branch species originally described as Liabum perfo- length of Chrysactinium may result from liatum by Blake (1927) and transferred to the rapid evolution after having diverged (Figs. currently recognized status by Robinson and 3 and 4) and may be confounding the Brettell (1974). Subsequently the palynologi- topology. Therefore, the phylogenetic rela- cal data collected by Robinson and Marti- tionship of Chrysactinium to the members of corena (1986) confirmed its placement in Munnozia is not completely certain (bootstrap Munnozia (Robinson and Marticorena 38%, DI=1, Figs. 3 and 4). 1986). The authors noted that the species is Munnozia, the most diverse genus, both in very distinctive in size of columellae cluster. terms of biogeography and morphology, has Our molecular study (Clade A of Figs. 3-5), been divided into two subgenera based on however, shows that M. perfoliata is outside morphological characters, Munnozia and Kast- the major clade of Munnozia, and lies next to nera (Robinson 1983). The palynological study the outgroup Sinclairia. The phylogenetic contradicted the morphological finding, re- trees (Figs. 3-5) show that ITS sequence vealing that the internal structure of pollen divergence of M. perfoliata from other Mun- wall represents several types in Munnozia nozia species ranges between 11.86% to (Robinson 1986). On the ITS tree, the genus 17.19%. However, the sequence divergence appears to split off into several clades: the of Munnozia to the sister group Sinclairia is subgenus Munnozia which is core Munnozia significantly less, with ranges between 8.52% with typical distinctive morphological elements and 12.78%, and thus is more closely related (Clades F, H, I), the subgenus Kastnera (Clade to Sinclairia taxa than to those in Munnozia. F), and the last four consisting of four mono- This is in disagreement with the current typic or small clades (M. foliosa, M. lyrata, M. morphological understanding. Following the lanceolata and Clade G). Despite weak support results of this molecular study, morphologi- within the Munnoziinae clade (D), the ITS tree cal and additional SEM investigations have identifies several morphological subclades rec- been conducted on M. perfoliata, another ognized by a previous worker (Robinson closely related species, M. chachapoyensis, 1983): 1) the subgenus Kastnera (Clade F) and a new species. Those detailed examina- defined by the lack of projections on the tions corroborate the result of our molecular receptacle, 2) Clade G having the only white- study with characters including the presence rayed species, 3) Clade H endemic to Costa of bullate leaf surfaces, pale yellow anther Rica, M. wilburii, and three accessions to a thecae, and irregularly dispersed spines on widespread variable taxon, M. senecionidis, the pollen. Considering their distinctiveness and 4) Clade I containing M. pinnulosa, in morphological and palynological features, M. gigantea and M. hastifolia. and the relative divergence in DNA In our study, portions of the traditional sequence, these taxa have been moved to a subgeneric classification are not supported by new genus Dillandia (Funk and Robinson the ITS study. Given that the ITS tree is 2001). H.-G. Kim et al.: A phylogeny of the Munnoziinae (Asteraceae, Liabeae) 183 Utility of ITS sequence data for developing morphology and biogeographic distribution, phylogeny of closely related genera of Astera- vary from 0.16% to 17.07%. Among all of the ceae. In many angiosperm groups that have genera and species of Munnoziinae examined, been studied over the last decade, ITS sequences the pair-wise sequence divergence ranged from have proven the most valuable for examining 0.3% to 20.65%. Considering a phylogenetic relationships within genera and among the more tree with reasonably high CI (0.704) and RI closely related genera within a tribe (Downie (0.861), ITS sequence data contained substan- et al. 2000) and family (Baldwin 1992, Baldwin tial phylogenetic signal. Several deep nodes are et al. 1995). However, across the tribes of highly supported (bootstrap > 75%) although a Asteraceae, divergence among ITS sequences few nodes of Clade D within the Munnoziinae is so high that problems with alignment and are very weakly supported (bootstrap > 50%) homoplasy are sufficient to make family-wide and collapsed in the strict consensus tree (Figs. 3 phylogenetic studies impossible with this mo- and 4). The poor resolution at the base of the lecular marker (Baldwin 1992, Kim and Jansen Munnoziinae clade may be attributed to low 1994). Our primary interest in using ITS se- sequence divergence in ITS data. Ultimately the quence data was to evaluate the possibility of combining of data from various sources can utilizing the region to infer the phylogeny of the lead to a better resolution (Donoghue and subtribe, Munnoziinae and later for the tribe Sanderson 1992, Olmstead and Sweere 1994). Liabeae. Numerous studies have shown the ITS In this study, nevertheless, ITS data provides a region to be sufficiently variable and to be useful useful measure of phylogenetic relationships at in comparisons at the generic level and below in the generic and specific level within Munnozii- Asteraceae. It has a divergence level of 0.2 to nae. 15% for species within the Madiinae (Baldwin 1992), 0 to 8.6% for Calycadenia (Baldwin We thank Fernanda Zermoglio and Matt Unwin 1993), 0.8 to 10.6% for species in Ajrzgza (Kim for assistance in collecting some of the preliminary and Jansen 1994), 0 to 4.1% for the aureoid data and Alice Tangerini for the illustration. We also thank J. Panero and M. Dillon for kindly complex of Senecio (Bain and Jansen 1995), 0.9 providing leaf material from their collections. This in Cheirolophus to 5.9% in Centaurea within work was supported by funding from the Smithso- genera, and intergeneric divergence of 3.3% to nian Institution's Office of Fellowships and Grants 20.6% within Centaureinae (Susanna et al. Postdoctoral Fellowship program (H-G. Kim and 1995), 0-1.11% in Argyranthemum and 0.2- V. Funk), the Scholarly Studies Program (V. Funk, 16.5% among its closely related genera (Fran- H. Robinson, and E. Zimmer) and the Laborato- cisco-Ortega et al. 1997), and 1-10% among ries of Analytical Biology (E. Zimmer). Anna Eupatorium and 8 to 27% within Eupatorieae Vlasek was supported by NSF grant #GER- (Schmidt and Schilling 2000). Within the family 9350106. Asteraceae, the ITS region has changed to the extent that it is possible to use ITS sequences to References infer the phylogeny even among distinct allop- atric populations in Calycadenia; the range of Bain J. F., Jansen R. K. (1995) A phylogenetic ITS sequence divergence is up to 3.7% with this analysis of aureoid Senecio (Asteraceae) complex molecular marker (Baldwin 1993). Our ITS based on ITS sequence data. Plant Syst. Evol. sequence data show a sequence divergence from 195: 209-219. Baldwin B. G. (1992) Phylogenetic utility of the 0.3% to 20.65% at both interspecific and internal transcribed spacers of nuclear ribosomal intergeneric levels. For example, congenera of DNA in plants: an example from the Compos- Sinclairia (Clade B, Fig. 3) show pair-wise itae. Mol. Phyl. Evol. 1: 3-16. sequence divergences from 0.8% to 2.8%, while Baldwin B. G. (1993) Molecular phylogenetics of those for the genus Munnozia, which is consid- Calycadenia (Composiate) based on ITS ered to be the most variable of the tribe in both sequences of nuclear ribosomal DNA: 184 H.-G. Kim et al.: A phylogeny of the Munnoziinae (Asteraceae, Liabeae) Chromosomal and morphological evolution re- Dipsacales based on rbcL sequences. Ann. Mis- examined. Amer. J. Bot. 80: 222-238. souri Bot. Card. 79: 333-345. Baldwin B. G., Sanderson M. J., Porter J. M., Donoghue M. J., Sanderson M. J. (1992) The Wojciechowski M. F., Campbell C. S., Dono- suitability of molecular and morphological ghue M. J. (1995) The ITS region of nuclear evidence in reconstructing plant phylogeny. In: ribosomal DNA: a valuable source of evidence Soltis P. S., Soltis D. E., Doyle J. J. (eds.) on angiosperm phylogeny. Ann. Missouri Bot. Molecular systematics of plants. Chapman and Card. 82: 247-277. Hall, London, pp. 340-368. Blake S. F. (1927) New South American species of Downie S. R., Katz-Downie D. S., Spalik K. (2000) Liabum. J. Wash. Acad. Sci. 17: 288-303. A phylogeny of Apiaceae tribe Scandiceae evi- Blake S. F. (1935) The genus Chionopappus of dence from nuclear ribosomal DNA internal Bentham (Asteraceae). J. Wash. Acad. Sci. 25: transcribed spacer sequences. Amer. J. Bot. 488-493. 87(1): 76-95. Bremer K. (1988) The limits of amino acid Doyle J. J., Doyle J. L. (1987) A rapid DNA sequence data in angiosperm phylogenetic re- isolation procedure for small quantities of fresh construction. Evolution 42: 795-803. leaf tissue. Phytochem. Bull. 19: 11-15. Bremer K. (1994) Asteraceae: cladistics and classi- Felsenstein J. (1985) Confidence limits on phytog- fication. Timber Press, Portland, Oregon. enies: An approach using the bootstrap. Evolu- Bremer K. (1996) Major clades of Asteraceae. In: tion 39: 783-791. Hind D. J. N., Beentje H. J. (eds.) Compositae: Francisco-Ortega J., Santos-Guerra A., Hines A., systematics. Proceedings of the International Jansen R. K. (1997) Molecular evidence for a Conference, Kew, pp. 1-7 Mediterranean Origin of the Macaronesian En- Cabrera A. L. (1954) Compuestas sudamericanas demic genus Argyranthemum (Asteraceae). nuevas o criticas, II. Notas Mus La Plata, 17(84): Amer. J. Bot. 84: 1595-1613. 71-80. Funk V., Robinson H., Dillon M. (1996) Liabeae: Carlquist S. (1976) Tribal interrelationships and taxonomy, phylogeny and biogeography. In: phylogeny of the Asteraceae. Aliso 8: 465-492. Hind D. J. N., Beentje H. (eds.) Compositae: Cassini H. (1823) Liabon. In: Cuvier G. (ed.) systematics. Proceedings of the International Dictionnaire des Sciences Naturelles, vol. 26: Compositae Conference, Kew, pp. 546-567. 203-211. Paris. [Reprinted in R. M. King & Funk V., Robinson H. (2001) A bully new genus H. W. Dawson (eds.), 1975. Cassini on Com- from the Andes (Compositae: Liabeae). Syst. positae. Oriole Editions, New York]. Bot. 26(2): 216-225. Cassini H. (1825) Oligacte. In: Cuvier G. (ed.) Hillis D., Bull J. J. (1993) An empirical test of Dictionnaire des Sciences Naturelles, vol. 36: 16- bootstrapping as a method for assessing confi- 18. Paris. [Reprinted in R. M. King & H. W. dence in phylogenetic analysis. Syst. Biol. 42: Dawson (eds.), 1975. Cassini on Compositae. 182-192. Oriole Editions, New York]. Hoffmann O. (1890-1894) Compositae. In: Engler Cassini H. (1830) Zyegee. In: Cuvier G. (ed.) A., Prantl K. (eds.) Die Naturlichen Pflanzen- Dictionnaire des Sciences Naturelles, vol. 60: familien. 4(5): 87-387. 560-619. Paris. [Reprinted in R. M. King & H. Huelsenbeck J. P., Crandall K. A. (1997) Phylog- W. Dawson (eds.), 1975. Cassini on Compositae. eny estimation and hypothesis testing using Oriole Editions, New York]. maximum likelihood. Ann. Rev. Ecol. and Syst. Cronquist A. (1955) Phylogeny and taxonomy of 28: 437-466. the Compositae. Amer. Midi. Nat. 53: 478-511. Karol G. K., Suh Y., Schatz G, Zimmer E. (2000) D'Arcy W. G. (1975) Flora of Panama. 9. Ann. Molecular evidence for the phylogenetic position Missouri Bot. Card. 62: 835-1322. of Takhtajania in the Winteraceae: Inference De Candolle A. P. (1836) Prodromus Systematis from nuclear ribosomal and chloroplast gene Naturalis Regni Vegetabilis, vol. 5. Paris Treut- spacer sequences. Ann. Missouri Bot. Gard. 87: tel & Wiirtz. 414-432. Donoghue M. J., Olmstead R. G, Smith J. F., Kim H.-G, Keeley S. C, Vroom P., Jansen R. K. Palmer J. D. (1992) Phylogenetic relationships of (1998) Molecular evidence for an African origin H.-G. Kim et al.: A phylogeny of the Munnoziinae (Asteraceae, Liabeae) 185 of the Hawaiian endemic Hesperomannia (Aster- Eupatorium) based on nuclear ITS sequence aceae). Proc. Natl. Acad. Sci. USA 95: 15440- data. Amer. J. Bot. 87: 716-726. 15445. Suh Y., Thien L. B., Reeve H. E., Zimmer E. A. Kim K.-J., Jansen R. K. (1994) Comparisons of (1993) Molecular evolution and phylogenetic phylogenetic hypotheses among different data implications of internal transcribed spacer se- sets in dwarf dandelions (Krigia, Asteraceae): quences of ribosomal DNA in Winteraceae. additional information from internal transcribed Amer. J. Bot. 80: 1042-1055. spacer sequences of nuclear ribosomal DNA. Susanna A., Garcia-Jacas N., Soltis D. E., Soltis P. Plant Syst. Evol. 190:157-185. (1995) Phylogenetic relationships in the tribe Kim K.-J., Jansen R. K. (1995) ndh¥ sequence Cardueae (Asteraceae) based on ITS sequences. evolution and the major clades in the sunflower Amer. J. Bot. 82: 1056-1068. family. Proc. Natl. Acad. Sci. USA 92:10379- Swofford D. L. (1998) PAUP*. Phylogenetic anal- 10383. ysis using parsimony and other methods. Ver- Lessing C. F. (1832) Synopsis Generum Composi- sion 4. 0. Sinauer, Sunderland, Mass. tarum. Earumque dispositionis novae tentamen Tamura K., Nei M. (1993) Estimation of the monographus multarum capensium interjectis. number of nucleotide substitutions in the Berlin. control region of mitochondrial DNA in humans Nash D. L., Williams L. O. (1976) Flora of and chimpanzees. Molec. Biol. Evol. 10: 512- Guatemala 12. Fieldiana, Bot. 24: 1-603. 526. Olmstead R. G., Sweere J. A. (1994) Combining Weddell H. A. (1855-1857) Chloris Andina, Vol- data in phylogenetic systematics: an empirical ume 1. Paris. approach using three molecular data sets in the Wen J., Zimmer E. A. (1996) Phylogeny, biogeog- Solanaceae. Syst. Biol. 43: 467-481. raphy of Panax L. (the ginseng genus Aralia- Posada D., Crandall K. A. (1998) MODELTEST ceae). Inferences from ITS sequences of nuclear 3.0: testing the model of DNA substitution. ribosomal DNA. Molec. Phylogenet. Evol. 6: Bioinformatics 14(9): 817-818. 167-177. Rambaut A. (1996) Se-Al Sequence Aligment White T. J., Birns S., Lee S., Taylor J. Editor Version 1.0 alpha 1. Department of (1990) Amplification and direct sequencing of Zoology, University of Oxford, South Parks fungal ribosomal RNA genes for phylogenetics. Road, Oxford OX1 4JD, U. K. In: Innis M., Gelfand D., Sninsky J., White T. Robinson H. (1978) Studies in the Liabeae (Aster- (eds.) PCR protocols: a guide to methods and aceae), XII: New species of Munnozia from applications. Academic Press, New York, Costa Rica. Phytologia 39(5): 331-333. pp. 315-322. Robinson Ft. (1983) Generic review of the tribe Liabeae (Asteraceae). Smith. Contr. Bot. 54: 1- 69. Robinson H., Marticorena C. (1986) A palynolog- Address of the authors: Hyi-Gyung Kim* (e-mail: ical study of the Liabeae (Asteraceae). Smith. [email protected]), Elizabeth A. Zimmer Contr. Bot. 64: 1-50. (e-mail: [email protected]), Laboratories of Robinson H., Brettell R. D. (1974) Studies in the Analytical Biology, Museum Support Center, Liabeae (Asteraceae). II: Preliminary survey of MRC 534, Smithsonian Institution, 4210 Silver Hill the genera. Phytologia 28: 44-63. Road, Suitland, MD 20746, USA. Vicki A. Funk Rydberg P. A. (1927) Cardueae, Liabeae, Neuro- (e-mail: [email protected]), Depart- laeneae, Senecioneae (pars.). North American ment of Systematic Biology, Division of Botany, Flora. 34: 289-360. National Museum of National History, MRC 166, Sandwith N. Y. (1956) Contributions to the flora of Smithsonian Institution, Washington DC, USA. tropical America. LXI: Notes on Philoglossa. ^Present address: e-mail: [email protected]. Kew Bull. 1956: 289-293. edu. Section of Integrative Biology, Department Schmidt G. J., Schilling E. E. (2000) Phylogeny and of Biological Sciences, The University of Texas, biogeography of Eupatorium (Asteraceae: Austin, 78713, USA.