55 (1) • February 2006: 125–137 Andersson • Phylogeny of Calceolaria On the phylogeny of the genus Calceolaria (Calceolariaceae) as inferred from ITS and plastid matK sequences Stephan Andersson Botanical Institute, Department of Systematic Botany, Göteborg University, Box 461, 405 30 Göteborg, Sweden. stephan.andersson@botany.gu.se The chloroplast gene matK and the nuclear internal transcribed spacer region (ITS) were used to explore taxonomic placement of and phylogeny within Calceolaria and related genera of Calceolarieae (= Calceolariaceae), a group traditionally referred to the polyphyletic Scrophulariaceae. Calceolarieae appear as a strongly supported monophyletic group and receive moderate support as the sister group of the family Gesneriaceae. The monotypic genus Porodittia is deeply nested inside Calceolaria and is reduced to synonymy under the latter genus. The phylogenetic trees retrieved for Calceolaria reveal that many presently recognised sections are polyphyletic. They also suggest that temperate species have retained the most ancestral character states in the genus. The phylogeny of Calceolaria features a number of well supported clades, which are difficult to characterize morphologically. A renewed interpretation of morphological characters diagnosing natural groups within Calceolaria is called for. The role of the Huancabamba Deflection as an important factor in the biogeography of Calceolaria is strongly alluded by the inferred phylogeny. KEYWORDS: Andes, Calceolaria, Calceolariaceae, Gesneriaceae, Huancabamba Deflection, Jovellana, ITS, Lamiales, matK, phylogeny, Porodittia. c INTRODUCTION Calceolaria L. is perhaps one of the most easily recognised plant genera in the Andean region. It is often found along roadsides and in seepages, and is easily identified by its remarkable, often strikingly yellow flowers with saccate lower lip. This genus consists of shrubs, lianas and herbs, and the geographic range extends from Patagonia to central Mexico. The main range is limited to the Andes but some species occur in nearby coastal regions of Ecuador, Peru and Chile, and a disjunct enclave with a few representatives is found in southeast Brazil. Since Linnaeus first described the genus in 1770, Calceolaria has been subject to several revisions, most notably by Bentham (1846), Kränzlin (1907), Pennell (1920, 1945, 1951a, b), Molau (1988), and Ehrhart (2000). However, in spite of the considerable attention to the genus in the literature, there are still a number of unresolved questions with respect to its taxonomic placement and evolutionary history. Calceolaria together with Jovellana R. & P. and Porodittia G. Don, is traditionally placed in the tribe Calceolarieae of Scrophulariaceae (Lamiales). Recent molecular studies (Olmstead & Reeves, 1995; Oxelman & al., 1999; Olmstead & al., 2000; Albach & al., 2001) have shown that Scrophulariaceae s.l. is polyphyletic. This raises questions about the placement of groups traditionally referred to Scrophulariaceae, including tribe Calceolarieae. Olmstead & al. (2001) based on sequence data from rbcL, ndhF, and rps2 showed that Calceolarieae are monophyletic and should be removed from the polyphyletic Scrophulariaceae. Calceolarieae branched off near Gesneriaceae but support for its position relative to Gesneriaceae was weak (53% bootstrap support). Because it was strongly supported as monophyletic but at the same time of somewhat uncertain affinity, the tribe was raised to the rank of family (Calceolariaceae), a position taken also by the Angiosperm Phylogeny Group (Stevens, 2001–present). The weak support obtained in the study by Olmstead & al. (2001) does not resolve the exact position of Calceolariaceae relative to other families of Lamiales. Calceolariaceae are characterised by paired-flowered cymes, a bilabiate corolla (in Porodittia abaxial “lip” strongly bilobed), and elaiophores seen as pads of hairs on the inside of the abaxial lip. The number of autapomorphic morphological characters is high in most species. Molau (1988) noted numerous possible parallelisms in the genus and that subgeneric division therefore remains problematic. While Calceolaria and Porodittia are limited to the Andes and adjacent areas, Jovellana, comprising only six species, is represented in both Chile (two species) and New Zealand (four species). Thus, Calceolariaceae is not strictly South American and is considered an Austral-Antarctic element (Cleef, 1979). The latest revision (Molau, 1988) recognised 181 125 Andersson • Phylogeny of Calceolaria species in Calceolaria, the latter being divided into three subgenera (Molau, 1988): Calceolaria (19 sections), Cheiloncos (two sections), and Rosula (three sections). The subgenera and sections of Molau (1988) are listed in Table 1. This work did not include representatives from Chile and Argentina. Two new species have recently been described by Molau (2003). Ehrhart (2000) revised the Chilean representatives of the genus and recognised 50 species. There is yet no modern revision of the Argentinean species. Thus, in addition to the 233 species of Calceolaria recognised in modern revisions, there is an unknown number of Argentinean representatives. Considering the high number of narrow endemics in the genus, more species are likely to be found in the future as more remote areas in the Andes are surveyed. The centre of species richness is found in the middle Andean region (6°S–20°S) where 122 species of Calceolaria, and the genus Porodittia occur (Molau, 1988). Molau (1988) and Ehrhart (2000) expressed different opinions on which subgeneric group retains the most ancestral character state in Calceolaria. Molau (1988) suggested that rosulate herbs in the temperate region were most likely to have done so, and placed them in the subgenus Rosula, while other temperate, more woody species and tiny subshrubs, were placed in subgenus Cheiloncos. All remaining species were placed in the large subgenus Calceolaria (Table 1). Molau also included sect. Micranthera in subgenus Cheiloncos, although Table 1. The subgenera and sections of Calceolaria recognised by Molau (1988). Subgenus Calceolaria Cheiloncos Rosula Section Anacyrta Calceolaria** Englerina Ericoides Integerrimae Lehmannina Lobatae Parvifoliae Perfoliatae Polyclada Revolutae Rotundifoliae* Salicifoliae Scapiflorae* Symplocophylla Teucriifoliae Thamnobia Urticopsis Verticillatae* Micranthera Rugosae*** Bellidifoliae** Corymbosae*** Kremastocheilos*** Type species C. virgata C. pinnata C. engleriana C. ericoides C. pinifolia C. lehmannina C. lobata C. parvifolia C. perfoliata C. brachiata C. revoluta C. rotundifolia C. salicifolia C. scapiflora C. connatifolia C. teucrioides C. rosmarinifolia C. lamiifolia C. angustiflora C. dichotoma C. integrifolia C. bellidifolia C. corymbosa C. uniflora * Section not represented in this study. ** Type species not included in this study. *** Sections included in Calceolaria by Molau (1988), but not revised by him. 126 55 (1) • February 2006: 125–137 the subgenus is not strictly temperate. Most of its representatives (C. dichotoma, C. utricularioides, and C. pumila) occur in the tropical region and only one species in the temperate region (C. tucumana in S. Bolivia). Like Molau, Ehrhart (1997) also believed that temperate species have retained more ancestral character states in Calceolaria, and she placed the main body of these in section Cheiloncos. In addition to sect. Cheiloncos she recognised two more sections of temperate Calceolaria, viz. Tenella (with the only species, C. tenella) and Kremastocheilos (with two species, C. fothergillii and C. uniflora). Witasek (1907) first described the latter section from of a group originally described by Bentham (1846) as Scaposae in his section Eucalceolaria. Ehrhart did not further treat the remaining species, all of which are tropical. Ehrhart further divided section Cheiloncos into two informal groups based on the length of the upper lip of the corolla relative to that of the lower lip. Group “A” is characterised by having an upper lip about 1/3 the length of the lower, and group “B” by having an upper lip at least 1/3 the length of the lower. Differences between Molau’s and Ehrhart’s classifications are shown in Table 2 for the species represented in this study. The aim of this study is to test hypotheses of phylogenetic and systematic relationships within Calceolariaceae, especially the genus Calceolaria. The hypothesis of Molau and Ehrhart that temperate species contain the most ancestral character states in Calceolaria is crucial to an understanding of character evolution and biogeography of the genus and needs verification. Also, the delimitation of clades and the morphological characterization of sections sensu Molau (1988) and Ehrhart (2000) require attention. MATERIAL AND METHODS Selection of taxa. — Based on the results of Oxelman & al. (1999), Albach & al. (2001), and Olmstead & al. (2001), the analysis of relationships of Calceolariaceae (Analysis I) was limited to a selection of Lamiales taxa. Because all the above studies place Oleaceae as sister to all other groups in Lamiales, two species of that family were used as outgroup. The ingroup consists of sixteen representatives from eight families (Tetrachondraceae, Gesneriaceae, Plantaginaceae, Scrophulariaceae s.l., Orobanchaceae, Pedaliaceae, Verbenaceae, and Lamiaceae) plus six representatives of the Calceolariaceae (one species of Jovellana and five species of Calceolaria). For the analyses II, III, and IV, designed to estimate the phylogeny of Calceolariaceae, an ingroup was assembled comprising a total of 87 species (Appendix). Kohleria (Gesneriaceae) was used as outgroup (Olm- 55 (1) • February 2006: 125–137 Table 2. Comparison of Molau’s (1988) and Ehrhart’s (2000) classifications of temperate species. Only species included in the present sample, with exception for C. dichotoma and C. utricularioides, are enumerated. Molau Subgenus Rosula C. biflora C. corymbosa C. pennellii C. umbellata C. uniflora Subgenus Cheiloncos C. dichotoma C. integrifolia C. plectranthifolia C. polifolia C. tucumana C. utricularioides Subgenus Calceolaria Remaining species Ehrhart Section Cheiloncos Group C. biflora A C. corymbosa A C. filicaulis A C. germainii A C. integrifolia B C. pinifolia A C. polifolia B C. polyrhiza A C. thyrsiflora B Section Tenella C. tenella Section Kremastocheilos C. uniflora Remaining species not treated stead & al., 2001). Ingroup taxa were sampled to include representatives from as many as possible of the 22 sections recognised by Molau (1988), as well as the two additional temperate sections mentioned, but not revised by him (Corymbosae and Kremastocheilos). Suitable material from the small sections Rotundifoliae (three species), Scapiflorae (three species), and Verticillatae (two species) could not be obtained. All other sections, sensu Molau (1988) in Calceolaria are represented, including the type species of all sections except Bellidifoliae and Calceolaria (Table 1). Material from the only species of Porodittia was included together with available material of a single South American species of Jovellana. In this study, the work by Molau (1988) is used as the main reference for sampling of Calceolaria species, since his monograph considers species from all sections and known floral regions where Calceolaria is abundant, while he also provides a more complete framework for discussion on sectional delimitations. The monograph of Chilean species by Ehrhart (2000) was used as an additional reference and guide for sampling of extratropical species. The number of sampled species is kept proportional to the occurrence of all known species in each of the floral regions (south, central and north), defined by Molau (Table 3). Information on origin, voucher specimens, and GenBank accession numbers for all taxa included in this study is provided in the Appendix. Selection of loci. — A number of loci (ITS, matK, atpB-rbcL spacer, rps16 intron, rbcL) were surveyed to identify regions likely to yield sufficient resolution for the analysis. The two loci ITS and matK showing the highest level of sequence variation were chosen. The matK region has been shown to have a rate of variability approximately three times that of rbcL (Plunkett & al., Andersson • Phylogeny of Calceolaria 1997), making it suitable for resolution at infrafamilial and infrageneric level (Johnson & Soltis, 1995; Sang & al., 1997; Li & al., 1999; Soltis & al., 2001; Li & Huang, 2002). To provide detailed resolution in Calceolaria, the nuclear ITS region (internal transcribed spacer 1, 5.8S gene, internal transcribed spacer 2) was chosen as it has proved useful in numerous species level studies in plant systematics (Hills & Dixon, 1991; Alvarez & Wendel, 2003). Hence, the combination of ITS and matK data was believed to provide higher resolution than either region separately by addressing different levels in the phylogenetic tree, ITS contributing mainly to distal and matK mainly to basal resolution. DNA extraction, amplification, and sequencing. — Total genomic DNA was isolated from leaves, flowers, or flower buds, mainly using fresh or silica gel dried material. However, herbarium material was used when no other material was available. Extractions were made with DNEasy® Plant Mini Kit (QIAGEN®, Hilden) according to the manufacturers’ standard protocol. Amplification and sequencing were made using the primers specified in Table 4. At no point were any multiple bands encountered in the amplification process, nor were there any such ambiguously called bases in the post-processing of the sequenced strands that would indicate multiple loci in the ITS. Sequencing was conducted on an automated Beckman Coulter CEQ 8000 using Genetic Analysis System Software v5.0 (Beckman Coulter, Fullerton). For ITS, a fragment of approximately 700 bp was sequenced. The sequenced fragment comprised a small part of the 18S exon, the ITS1, the 5.8S exon, the ITS2, and a small part of the 26S exon. Five species of Calceolaria (C. uniflora, C. pennellii, C. vaccinioides, C. dilatata, and C. bicrenata), together with the one species of Jovellana (J. violacea), were sequenced for 90–95% (~1,400 bp) of the matK gene. The last ~75 bp of the 3' end were not covered with Table 3. Comparison of the species sampled in Calceolariaceae to the total number of species with respect to proportion of species occurring in the geographic regions of South America discerned by Molau (1988). Because of the lack of a modern revision of Argentinean species, they are excluded from the total number of species. The number of species present in Argentina can be estimated to be 16–30+ such that the actual number for the South region is likely to be 66–80+. South Middle North Total South region No. of % of total species 50 22 122 52 61 26 233 100 This study No. of % of total species sample 16 19 51 60 18 21 85 100 127 Andersson • Phylogeny of Calceolaria the primers used for sequencing (Table 4). A segment of the sequenced matK strand with the highest amount of variation was identified. For the remaining species, this segment, the approximately 840 bp between the primers 323f and 1189r, was amplified and sequenced. Alignment. — The forward and reverse sequence strands for both ITS and matK sequences were checked and edited using the Sequencher™ software version 4.1 (Gene Codes Corporation, Ann Arbor, Michigan, U.S.A). Sequence strands were assembled and manually adjusted. The final sequences were aligned into separate ITS and matK data matrices using ClustalX version 1.83 (Thompson & al., 1997) and the resulting matrices were manually adjusted in SeaView (Galtier & al., 1996). Gaps were coded as missing data, and all informative characters in the matrices were manually verified against original sequence data in order to eliminate possible misreadings. A total of 88 ITS sequences were obtained, requiring eight gaps in the alignment. Gaps considered carrying phylogenetic information with respect to the ingroup were coded as separate indel characters and added to the end of the matrix using G/C for present/absent, and accounted for 5.8% of the parsimony-informative sites. Forty-nine matK sequences were obtained from the 88 taxa sequenced for ITS, the remaining 39 taxa failing to amplify in spite of various adjustments to the protocols. The alignment of matK sequences did not contain any informative gaps. Phylogenetic analyses. — Four different parsimony analyses were conducted using PAUP* 4.0b10 (Swofford, 2002). One of these (Analysis I) was based on matK data and designed to reveal the placement of Table 4. Amplification (A) and sequencing (S) primers for ITS and matK regions. Underlined bases are modified as compared to the cited reference. Primers ITS ITS11 ITS21 ITS31 ITS41 ITS102 “N1”3 “N2”3 matK 323f2 775f2 1189r2 1168r4 1470r4 trnK3914f5 trnK2R5 A/S Sequence (5´-3´) S S S A/S A A/S A/S TCCGTAGGTGAACCTGCGG GCTGCGTTCTTCATCGATGC GCATCGATGAAGAACGCAGC TCCTCCGCTTATTGATATGC TGCTAACTAGCTATGYGGAG TATGCTTAAAYTCAGCGGGT AACAAGGTTTCCGTAGGTGA S S S S S A A ATTNTCAAATCNTAKCAGAGGGG TCTTGAACGAATNTATTTCTATG CGGCTTACTAATRGGATGCCC TACNATCAATTCATTCAATATTYCC AAGATGTTGAGCGTAAATGA GGGGTTGCTAACTCAACGG AACTAGTCGGATGGAGTAG 1Wojciechowski & al., 1993; 2L. Andersson, pers.comm.; 3Nickrent & al., 1994; 4Johnson & Soltis, 1995; 5Plunkett & al., 1997. The primers by Nickrent & al. were published without designations and the names “N1” and “N2” are used here as a matter of convenience. 128 55 (1) • February 2006: 125–137 Calceolariaceae in Lamiales. The second analysis (Analysis II) included an outgroup, and the 49 species of Calceolariaceae for which both ITS and matK data were available. The third (Analysis III) used only ITS sequences of 87 species of Calceolariaceae. The fourth analysis (Analysis IV) used a combination of ITS and matK sequences and included all 87 species of Calceolariaceae and an outgroup. Thus, in the fourth analysis, matK sequences were lacking for many species and missing data were coded as question marks. For each analysis the same search strategy was applied: a heuristic search with 1000 random addition sequence replicates (RAS) and tree-bisection-reconnection (TBR) branch swapping was performed without a MAXTREE limit. All characters were equally weighted. In analysis III and IV, this quickly resulted in buffer overflow, calling for a change of strategy. Thus, 200,000 RAS were performed (analysis III), saving one most parsimonious tree (MPT) per RAS and limiting search time for each RAS to a maximum of three seconds, saving one MPT if its length was equal to or lower than the ones obtained earlier. These settings were used to limit the search time while maximising the tree space explored. However, this procedure does not guarantee that the globally shortest tree is found, but the trees found should be a close approximation and would thus serve to provide consistency and retention indices. For analysis IV the problem of slow speed in heuristic search methods was circumvented by using only jacknife analysis. Thus, no CI or RI values were obtained in this analysis. Tree support was estimated for each data set using jacknife resampling with parameters to simulate bootstrap resampling as described by Harshman (1994), Farris & al. (1996), and Farris (1998). Analyses were made using the following parameters: 37% deletion (JAC-emulation), 10,000 replicates, 3 RAS, TBR swapping, and a single MPT saved per RAS. Otherwise default settings were used. Support values of 50–69% are here considered “weak”, 70–89% “moderate”, and 90–100% “strong”. RESULTS Analysis I. — The alignment of matK data comprises 1486 characters, of which 286 (19.2%) were parsimony-informative. Three additional indel characters were added to the matrix. The heuristic search was run to completion and resulted in a single MPT of 1104 steps (CI = 0.719, RI = 0.635). This tree is shown in Fig. 1. Basal nodes are strongly supported. Branches representing strongly supported non-Calceolariaceae lineages include the Tetrachondraceae-clade (100%), and Orobanchaceae-clade (100%). Moderately supported lin- 55 (1) • February 2006: 125–137 Andersson • Phylogeny of Calceolaria Olea Oleaceae (Outgroup) 92 98 Tetrachondra 100 100 50 Polypremum Tetrachondraceae 65 71 96 Peltanthera 62 58 70 Streptocarpus 95 99 98 Kohleria Gesneriaceae 86 75 100 97 Jovellana 98 59 Calceolaria uniflora 100 99 68 100 Calceolaria pennellii 78 90 100 Calceolaria vaccinioides 76 100 74 82 61 Calceolaria dilatata 73 Calceolaria bicrenata Calceolariaceae 76 60 60 Veronica 82 59 68 Antirrhinum Plantaginaceae 63 83 Scrophularia 62 83 100 Myoporum Scrophulariaceae 65 75 Orobanche Pedicularis 10 Changes 81 99 100 Sesamum Ajuga C. spruceana (th) C. martinezii (le) C. rosmarinifolia (th) C. fusca (sa) C. connatifolia (sy) C. phaeotricha (sa) C. bicrenata (ur) C. glauca (an) Porodittia triandra C. comosa ssp. comosa (an) C. lobata (lo) C. phaceliifolia (lo) C. rugulosa (an) C. virgata (an) C. penlandii ssp. penlandii (ur) C. maculata (ur) C. ballotifolia (ur) C. chelidonioides (ca) C. mandoniana (ca) C. perfoliata (pe) C. tomentosa (pe) C. lanata (pe) C. calycina (pe) C. pavonii (pe) VIb VIc Clade VII Clade VIII Clade IX Clade X Orobanchaceae Pedaliaceae Verbena Kohleria spicata Outgroup Jovellana violacea C. uniflora (kr) Clade I C. pennellii (be) C. polyrhiza (?) C. tucumana (mi) C. buchtieniana (re) Clade II C. engleriana ssp. engleriana (en) C. myriophylla (pa) C. rupestris (re) Clade III C. boliviana (en) C. oblonga (le) C. sparsiflora (pa) Clade IV C. flexuosa ssp. flexuosa (ur) C. vaccinioides (po) C. atahualpae ssp. atahualpae (po) C. ericoides spp. ericoides (er) C. dentifolia (sa) C. gaultherioides (sa) Clade V C. tetragona ssp. tetragona (sa) C. moyobambae (sa) C. pedunculata ssp. pedunculata (le) C. lavandulifolia (te) C. purpurascens (pe) VIa C. dilatata (pe) Clade VI Jasminum Verbenaceae Lamiaceae Fig. 1. The single most parsimonious tree (1104 steps, CI = 0.7192, RI = 0.6353) obtained from Analysis I (matK data). Numbers at nodes indicate jacknife support values. eages include the Plantaginaceae-ScrophulariaceaeOrobanchaceae-Pedaliaceae-Verbenaceae-Lamiaceaeclade (83%), while weakly supported lineages include the Plantaginaceae-clade (68%) and Gesneriaceae-clade (58%). The Calceolariaceae-clade receive strong support (100%) and its sister group relationship to the Gesneriaceae-clade receives moderate support (75%). Analysis II. — The alignment of Calceolariaceae with both ITS and partial matK sequences comprise 46 species of Calceolaria, one of Jovellana, one of Porodittia, and the outgroup Kohleria spicata. Out of 1,573 characters, 158 (9.8%) were parsimony-informative and 132 (8.2%) of these informative in Calceolaria and Porodittia only. The alignment also include eight additional indel characters in the ITS region. The heuristic search resulted in 42 MPTs of 690 steps (CI = 0.713, RI = 0.697). The strict consensus tree of these (Fig. 2) is well resolved and has only four polytomies in the ingroup. All these occur in distal branches. Strongly supported clades include clade I (92%), and there is strong support (96%) that this clade and C. tucumana are basal to all other species of Calceolaria Fig. 2. Strict consensus tree of the 42 most parsimonious trees (690 steps, CI = 0.7130, RI = 0.6968) from Analysis II (ITS and matK data). Values above branches indicate jacknife support. The solid bar indicates dispersal event across the Huancabamba deflection. Abbreviations within parenthesis indicate section (sensu Molau 1988), (an) Anacyrta, (be) Bellidifoliae, (ca) Calceolaria, (co) Corymbosae, (en) Englerina, (er) Ericoides, (in) Integerrimae, (kr) Kremastocheilos, (le) Lehmannina, (lo) Lobatae, (mi) Micranthera, (pa) Parvifoliae, (pe) Perfoliatae, (po) Polyclada, (re) Revolutae, (ru) Rugosae, (sa) Salicifoliae, (sy) Symplocophylla, (te) Teucriifoliae, (th) Thamnobia, (ur) Urticopsis. Question-marks indicate that the species were not treated by Molau (1988). (including Porodittia). The species of clade I are all of exclusively temperate origin. Other strongly supported groups are clade V (97%), consisting entirely of species from sect. Salicifoliae (C. dentifolia, C. gaultherioides, C. tetragonal, and C. moyobambae); clade VIa (100%) with two species from sect. Perfoliatae (C. purpurascens and C. dilatata); clade VIb (90%) consisting of two species from sect. Thamnobia (C. spruceana and C. rosmarinifolia) and one from sect. Lehmannina (C. martinezii); and clade IX (100%) corresponding to sect. Calceolaria (C. chelidonioides and C. mandoniana). Moderately supported clades include clade VII (76%), consisting of all representatives from sect. Anacyrta (C. glauca, C. comosa, C. rugulosa, and C. virgata), two species from sect. Lobatae (C. lobata, and C. phaceliifolia), and one species from sect. Urticopsis (C. bicrenata). The monotypic Porodittia is also nested with129 Andersson • Phylogeny of Calceolaria in this clade. Clade X is also moderately supported (75%) and includes all species of sect. Perfoliatae except for the two in clade VIa. Clade VIc (74%) consists of species from sects. Salicifoliae (C. fusca and C. phaeotricha) and Symplocophylla (C. connatifolia). Clade V through X together forms a moderately supported clade (82%). Weakly supported clades include clade III (65%) consisting of species from sects. Parvifoliae (C. myriophylla), Revolutae (C. rupestris), and Englerina (C. boliviana). Clade IV (62%) also comprises species from several sections: Lehmannina (C. oblonga), Parvifoliae (C. sparsiflora), Urticopsis (C. flexuosa), and Polyclada (C. vaccinioides and C. atahualpae). Clade VIII (63%) consists of species from sect. Urticopsis (C. penlandii, C. maculata, and C. ballotifolia). Clade VI (68%) consists of clades VIa, VIb, and VIc together with C. lavandulifolia (sect. Teucriifoliae) and C. pedunculata (sect. Lehmannina). Clade II is present in strict consensus but has no jacknife support at the 50% cut off level. It comprises one representative each of the sections Revolutae (C. buchtieniana) and Englerina (the type species, C. englerina). Analysis III and IV. — Analysis III and IV included 85 species of Calceolaria, Porodittia traindra, and three outgroup taxa (two species of Kohleria and one of Jovellana). In analysis III only ITS data were used, whereas analysis IV combined ITS and matK data. In the ITS data set there were 730 characters in the matrix, of which 223 (30.5%) were parsimony-informative. Of these, 142 (19.0%) were informative with respect to the ingroup. The combined ITS and matK alignment consist of 1,583 sites of which 266 (16.8%) were parsimony informative, 177 (11.2%) of these in Calceolaria and Porodittia only. Both data sets also include eight additional parsimony-informative indel characters from the ITS region. The heuristic search in analysis III resulted in 105,677 saved trees with a length of 733 steps. One arbitrarily chosen tree is presented in Fig. 3 (CI = 0.584, RI = 0.751). The jacknife consensus tree from analysis IV (Fig. 4) was largely consistent with the strict consensus tree retrieved from analysis III. The tree in Fig. 4 shows several polytomies. However, although their relationships are largely obscure, a number of clades are supported in both analysis III and IV. Five clades are strongly supported in either, or both of, Analysis III or Analysis IV. Clade VI (100% analysis III/100% analysis IV) comprises two of three species in sect. Micranthera (C. utricularioides and the type species, C. dichotoma). The third species, C. pumila, was not represented in the sample. The two sampled species both occur north of Bolivia. Clade IX (100%/100%) corresponds to sect. Calceolaria (C. rivularis, C. tripartita, 130 55 (1) • February 2006: 125–137 5 changes Kohleria spicata Kohleria rugata Jovellana violacea C. filicaulis ssp. filicaulis C. umbellata C. germainii C. uniflora C. pennellii C. biflora C. polyrhiza C. corymbosa ssp. montana C. filicaulis ssp. luxurians C. thyrsiflora C. integrifolia C. polifolia C. plectranthifolia C. tenella C. tucumana C. pinifolia C. buchtieniana C. parvifolia ssp. parvifolia C. teucrioides C. engleriana ssp. engleriana C. polyclada C. myriophylla C. boliviana C. cypripediiflora C. rupestris C. revoluta C. oblonga C. ericoides spp. ericoides C. barbata C. linearis C. flexuosa ssp. flexuosa C. vulpina C. sparsiflora C. inflexa C. vaccinioides C. atahualpae ssp. atahualpae C. tetragona ssp. tetragona C. moyobambae C. gaultherioides C. dentifolia C. salicifolia ssp. salicifolia C. pedunculata ssp. pedunculata C. lehmannina C. brachiata C. nivalis ssp. nivalis C. semiconnata C. fusca C. microbefaria ssp. fruticosa C. connatifolia C. phaeotricha C. lavandulifolia C. helianthemoides C. purpurascens C. dilatata C. stricta C. rosmarinifolia C. gossypina C. martinezii C. spruceana C. scabra C. dichotoma C. utricularioides C. bicrenata C. sericea C. glauca Porodittia triandra C. variifolia C. comosa ssp. comosa C. lobata C. phaceliifolia C. rugulosa C. virgata C. maculata C. ballotifolia C. obtusa C. lamiifolia C. penlandii ssp. penlandii C. tomentosa C. lanata C. calycina C. pavonii C. perfoliata C. rivularis C. tripartita C. chelidonioides C. mandoniana Fig. 3. One of 105.677 most parsimonious trees (733 steps, CI = 0.5839, RI = 0.7506) resulting from analysis III (ITS data) presented as a phylogram. C. chelidonioides and C. mandoniana). Clade XII (91%/ 89%) consists of two species from sect. Lehmannina (C. pedunculata and the type species, C. lehmannina). Clade XIV (99%/99%) corresponds to clade VIa in analysis II, and clade XV (95%/94%) corresponds to clade VIb, with the addition of C. gossypina (sect. Thamnobia). Clade VIII (85%/92%) corresponds to clade V in analysis II, but with the addition of the type species of sect. Salicifoliae (C. salicifolia). Clade XIII (86%/92%) comprises two species of sect. Teucriifoliae (C. lavandulifolia and C. helianthemoides). Seven clades are moderately supported. Clade III (73%/77%) comprises all species of sect. Ericoides (C. 55 (1) • February 2006: 125–137 90/81 65/62 57/50 73/77 89/89 77/73 69/73 61/60 89/74 -/71 75/77 96/97 100/100 -/87 63/71 51/51 98/98 99/98 85/92 85/87 100/100 -/88 -/60 62/72 77/79 75/78 88/89 98/99 72/72 74/69 60/53 98/82 91/89 -/54 86/92 99/99 80/71 53/62 95/94 -/66 73/70 62/55 60/53 Set I 85/95 98/98 Outgroup Set II 64/54 Kohleria spicata Kohleria rugata Jovellana violacea C. filicaulis ssp. filicaulis (?) C. umbellata (be) C. germainii (?) C. uniflora* (kr) C. pennellii (be) C. biflora (kr) C. polyrhiza (?) C. pinifolia* (in) C. thyrsiflora (in) C. integrifolia* (ru) C. polifolia (ru) C. plectranthifolia (ru) C. tucumana (mi) C. tenella (?) C. corymbosa ssp. montana (co) C. filicaulis ssp. luxurians (?) C. myriophylla (pa) C. buchtieniana (re) C. parvifolia* ssp. parvifolia (pa) C. teucrioides* (te) C. oblonga (le) C. engleriana* ssp. engleriana (en) C. polyclada (po) C. ericoides* spp. ericoides (er) C. barbata (er) C. linearis (er) C. boliviana (en) C. cypripediiflora (le) C. rupestris (re) C. revoluta* (re) C. vulpina (le) C. sparsiflora (pa) C. flexuosa ssp. flexuosa (ur) C. inflexa (po) C. vaccinioides (po) C. atahualpae ssp. atahualpae (po) C. scabra (te) C. dichotoma* (mi) C. utricularioides (mi) C. maculata (ur) C. ballotifolia (ur) C. obtusa (lo) C. lamiifolia* (ur) C. penlandii ssp. penlandii (ur) C. tetragona ssp. tetragona (sa) C. moyobambae (sa) C. dentifolia (sa) C. gaultherioides (sa) C. salicifolia* ssp. salicifolia (sa) C. rivularis (ca) C. tripartita (ca) C. chelidonioides (ca) C. mandoniana (ca) C. perfoliata* (pe) C. tomentosa (pe) C. lanata (pe) C. calycina (pe) C. pavonii (pe) C. bicrenata (ur) C. glauca (an) Porodittia triandra C. variifolia (an) C. sericea (an) C. comosa ssp. comosa (an) C. rugulosa (an) C. virgata* (an) C. lobata* (lo) C. phaceliifolia (lo) C. pedunculata ssp. pedunculata (le) C. lehmannina* (le) C. stricta (sa) C. brachiata* (po) C. lavandulifolia (te) C. helianthemoides (te) C. purpurascens (pe) C. dilatata (pe) C. martinezii (le) C. spruceana (th) C. rosmarinifolia* (th) C. gossypina (th) C. nivalis ssp. nivalis (sa) C. fusca (sa) C. semiconnata (sy) C. microbefaria ssp. fruticosa (sa) C. connatifolia* (sy) C. phaeotricha (sa) Andersson • Phylogeny of Calceolaria Clade III Clade IV Clade V Clade VI Clade VII Clade VIII Clade IX Clade X Clade XI Clade XII Clade XIII Clade XIV Clade XV Clade XVI Fig. 4. The jacknife consensus tree resulting from analysis IV (ITS and matK data). Values above branches indicate jacknife support (analysis III/analysis IV). The thickened branch indicates a dispersal event across the Huancabamba deflection. Abbreviations within parenthesis indicate section (sensu Molau 1988), (an) Anacyrta, (be) Bellidifoliae, (ca) Calceolaria, (co) Corymbosae, (en) Englerina, (er) Ericoides, (in) Integerrimae, (kr) Kremastocheilos, (le) Lehmannina, (lo) Lobatae, (mi) Micranthera, (pa) Parvifoliae, (pe) Perfoliatae, (po) Polyclada, (re) Revolutae, (ru) Rugosae, (sa) Salicifoliae, (sy) Symplocophylla, (te) Teucriifoliae, (th) Thamnobia, (ur) Urticopsis. Question-marks indicate that the species were not treated by Molau (1988). Species that include a type of a section are marked by an asterisk. (C. vulpina), and Parvifolia (C. sparsiflora). It contains a moderately supported (75%/77%) subclade comprising representatives from sect. Polyclada (C. inflexa, C. vaccinioides, and C. atahualpae). Clade X (75%/78%) corresponds to clade X in analysis II and comprises a majority of the species in sect. Parvifoliae (C. tomentosa, C. lanata, C. calycina, C. pavonii, and the type species, C. perfoliata). Clade XI (72%/72%) essentially equals clade VII in analysis II with two more species from sect. Anacyrta (C. sericea and C. variifolia). Clade XVI (73%/70%) comprises species from sect. Salicifoliae (C. nivalis, C. fusca, C. microbefaria, and C. phaeotricha) and Symplocophylla (C. semiconnata and the type species, C. connatifolia). The clades XIII through XVI form a moderately (80%/71%) supported group, in which they appear in a polytomy together with the species C. stricta (sect. Salicifolia) and C. brachiata (type species for sect. Polyclada). Three clades are weakly supported. Clade VII (63%/71%) includes four species of sect. Urticopsis (C. maculata, C. ballotifolia, C. penlandii, and the type species, C. lamiifolia). Nested among these is one species of sect. Lobatae (C. obtusa). A cluster of moderately to strongly supported clades (XII through XVI) appears as a more inclusive one, but with negligible support (-/54%). Clades VI through XVI are similarly united (62%/72%) together with the species C. scabra (sect. Teucriifoliae). The branches of the basal polytomy (Set I) in Fig. 4 are all from temperate regions (sects. Bellidifoliae and Kremastocheilos). The main body of species are separated with weak support (61%/60%), while other temperate representatives (Set II) form an extensive polytomy together with neotropical representatives. These temperate taxa include representatives from sect. Integerrimae (C. thyrsiflora and the type species, C. pinifolia), Rugosae (C. plectranthifolia, C. polifolia, and the type species, C. integrifolia), Micranthera (C. tucumana), and Corymbosae (C. corymbosa). The species C. filicaulis, C. germainii, C. polyrhiza, and C. tenella were not treated by Molau (1988) and therefore have no sectional affiliation in his classification. DISCUSSION barbata, C. linearis, and the type species C. ericoides). It is attached to an extensive polytomy together with four other clades and many unresolved species. Clade IV (77%/73%) includes representatives of sect. Englerina (C. boliviana), Lehmannina (C. cypripediiflora) and Revolutae (C. rupestris and the type species, C. revoluta). Clade V (89%/74%) includes representatives from different sections: Urticopsis (C. flexuosa), Lehmannina The nuclear ITS region, consisting of two non-coding spacers with the 5.8 S gene in between, differs in location and function from the coding plastid maturase K (matK) exon. The ITS region has been widely used in plant and fungal systematics providing information for phylogenetic studies at generic as well as species level because of its rapid evolution (Alvarez & Wendel, 2003). The matK gene, on the other hand, has only recently been 131 Andersson • Phylogeny of Calceolaria used in plant systematics (Johnson & Soltis, 1995; Sang & al., 1997; Li & al., 1999; Soltis & al., 2001; Li & Huang, 2002). It has been shown to have an evolution rate about three times that of rbcL (Plunkett & al., 1997) and thus has the potential to provide information where more slowly evolving coding genes such as rbcL and atpB fail to do so. The proportion of variable sites in ITS in Calceolaria is low (19.0%) compared to groups at similar taxonomic level, (e.g., 43.2% in Tropaeolaceae; Andersson & Andersson, 2000). The amount of variation in matK in Calceolariaceae is much smaller than in ITS, providing only 4.2% parsimony-informative characters in the most variable portion chosen for this study. Nevertheless, the information from matK appears useful at a level different from that where ITS information is most helpful, strengthening support mainly at basal nodes. The overall agreement between the tree topologies obtained from analysis III and IV data, respectively, indicates that there are no major conflicts among data sets, even though support in analysis IV is probably affected by the large amount of missing matK data. When analysed together, ITS and matK jointly provide more information than any of them does separately. The low divergences in the rbcL, matK and in particularly ITS sequences seem to support the idea of recent diversification of the group. This is illustrated by the short branches in the phylogram (Fig. 3). Position of Calceolariaceae in Lamiales. — Analysis I and II strongly support that Calceolariaceae are monophyletic, comprising the three genera Calceolaria, Porodittia, and Jovellana. The traditional association of the group with Scrophulariaceae finds no support in this study. Instead, it suggests a placement as sister group of the Gesneriaceae, but only with moderate support (75%). The placement of Calceolariaceae in Lamiales (Analysis I) is in agreement with results from analyses of Lamiales using ndhF and rbcL data (Oxelman & al., 1999), and 18S rDNA, rbcL, ndhF, and atpB data (Albach & al., 2001). However, the support for placing Calceolariaceae as sister group to Gesneriaceae (75%) is higher than that for the alternative placement inferred by Olmstead & al. (2001). Their results separated Calceolariaceae and Gesneriaceae in succession in the tree with weak bootstrap support (53%), placing Gesneriaceae more distal than Calceolariaceae. Infrageneric relationship in Calceolaria. — Sampling of Calceolaria was not comprehensive, covering only 85 of the 250+ species (≈30%), but considerable effort was made to include as many sections as possible with the number of sampled species kept proportional to the size of the sections. Many strongly to moderately supported clades were identified, but relationships among them remain largely unresolved. All sections 132 55 (1) • February 2006: 125–137 mentioned below refer to Fig. 4 if nothing else is noted. The sections of Calceolaria sensu Molau (1988) were largely delimited on basis of characters derived from growth habit and leaf morphology, while species were primarily delimited by well-defined traits in floral and seed morphology. Thus, the species, often narrow endemics, are easily recognised, but sections are in many cases difficult to demarcate. As previously alluded to, Molau (1988) also remarked that morphological parallelisms are probably common in the genus. A somewhat different light is shed on this assertion by the present results, which indicate that several of Molau’s sections are polyphyletic. Nonetheless, it is also true that some sections are clearly monophyletic. Some groups suggested by the present analyses are difficult to characterise morphologically, showing a wide range of variation. This calls for a thorough re-examination and evaluation of morphological characters in the light of phylogeny. The below discussion is a first step towards a new understanding of infrageneric relationships in Calceolaria. The species sets marked as “Set I” and “Set II” in Fig. 4 include all the sampled temperate species, some of which may reach southernmost Bolivia. However, note that Set I and Set II are strictly artificial groupings, labelled only for aiding the discussion. The species of Set I seem to constitute a paraphyletic grade basal to other species of Calceolaria, but the node setting them off is weakly supported (61%/60%) in analysis III and IV. In analysis II, some species of Set I form a strongly supported monophyletic group that constitutes the sister group of the remainder of the genus Calceolaria. This group is part of Set I in Analyses III and IV, but with substantially lower support. This decrease in support could to be related to the fact that more species are added in Analysis III and IV and that these species are not represented by matK data. Opinions differed between Molau (1988) and Ehrhart (1997) as to which representatives of Calceolaria constitute the most ancestral members of the genus. The results obtained in this study do not unequivocaly answer this question, considering the low support values for basal branches taxa in the larger analyses. There is no evidence to support the assertion that the temperate species form monophyletic groups, and thus no support for recognising them as subgenera or sections. There also is no support for treating the sections Tenella or Kremastocheilos (sensu Ehrhart, 2000) as distinct, nor for the sections in subgenera Rosula and Cheiloncos (sensu Molau, 1988) because all groups are distinctly non-monophyletic. There is, however, some structure that supports suggestions about evolution of morphological traits. Most species of Set I (Fig. 4) seem to have a tendency towards more rosulate growth habit and an upper corolla lip shorter than one third of the lower lip, 55 (1) • February 2006: 125–137 Andersson • Phylogeny of Calceolaria Fig. 5. Flower photographs of Calceolaria (A–D), Jovellana (E), and Porodittia (F). A, Overview of a Calceolaria flower, forced open; B, cross-section of Calceolaria flower showing the saccate lobes. Note how the stigma and stamens does not protrude from the corolla; C, stamens of Calceolaria species with corolla removed. D, dry septicidal Calceolaria capsule with two locules, and a number of scattered seeds. E, inflorescense of Jovella violacea; F, flower of Porodittia. Note the third, partially concealed, stamen behind the stigma. Photos A–D by Stephan Andersson. Photo E by Richard Gautier. Photo F by Alexander Schmidt-Lebuhn. 133 Andersson • Phylogeny of Calceolaria while most representatives of Set II seem to have a tendency towards being more none-rosulate species with an upper corolla lip longer than one third of the lower lip. More extensive studies are called for in order to test if this is a true trend or not. Some of Molau’s sections appear to be monophyletic. The one that is most strongly supported is sect. Calceolaria (clade IX), a group recognised by Bentham (1846) as sect. Aposecos. This clade is characterised by its weedy growth, its preference for moist or wet habitats, and tetraploid karyotype (2n = 32). It includes the most widespread species in the genus (C. tripartita, C. mexicana, and C. chelidonioides), all ranging from southern Mexico or Central America to southern Peru or Bolivia. Members of the section are often found along roadsides and in ditches, rendering them easily accessible and often collected. The specialisation of the group is also hinted in the phylogram (Fig. 3) where it is the only clade to be attached to a conspicuously long branch. The section Ericoides is also monophyletic (clade III), represented by all known species. They are easily recognised as shrubs or subshrubs with short, needle-like leaves and are confined to jalca and páramo vegetation in Ecuador and northern Peru. The species of section Salicifoliae appear in two clades, VIII and XVI, with the former one comprising only such species. The members of this group (Clade VIII) are easily distinguished by their large, dentate leaves, and large flowers. This group was believed by Molau (1988) to be a distinct evolutionary lineage in the section, and his hypothesis is well corroborated here. The Salicifoliae species in clade XVI were also believed to constitute a separate lineage in the section, restricted to the northern region, but in these analyses they are joined by two species from sect. Symplocophylla. Molau believed there to be a close connection between sect. Salicifoliae and sect. Symplocophylla, but he treated them separately. Like those of sect. Salicifoliae, species of sect. Perfoliatae appear in two separate clades (X and XIV). Although they are virtually identical in morphology, these clades are separated geographically; clade X occurs in the middle region and clade XIV in the northern. It is remarkable but indisputable that their distinctive morphology, with very broadly winged petioles connate across the nodes, has actually evolved twice. Winged petioles are, however, not unique to these clades. Though less conspicuous, connate winged petioles are present in several different representatives of the genus. The character state found in the Perfoliatae hence seems to be an extreme elaboration of a more widely distributed trait. Section Micranthera is another example of a section splitting into its geographical components. It was referred by Molau (1988) to the temperate subgenus Chei134 55 (1) • February 2006: 125–137 loncos, although it features three species (C. utricularioides, C. pumila, and C. dichotoma) occurring in the middle region and only one (C. tucumana) in the southern temperate region. In the present analyses, the two sampled species from the middle region form a strongly supported group that is placed together with other species from this region, whereas the single temperate species does not group with these. A splitting of sect. Micranthera according to geographical range is thus indicated. Other clades correspond roughly to the sections of Molau (1988), although some emendations are suggested. Clade VII is made up of a majority of species from section Urticopsis (including the type), but one representative from sect. Lobatae occurs here as well. Clade XI comprises all species of the section Anacyrta (including the type), but species of sect. Urticopsis and Lobatae (including the type) are also nested within this clade, as is the genus Porodittia. Clade XV consists of all representatives of sect. Thamnobia (including the type) together with one species of sect. Lehmannina. The subspecies of Calceolaria filicaulis. — The two subspecies of C. filicaulis do not seem closely related in this study. There are however, doubts on the separation of C. filicaulis from closely related species from the Mamuil Malal area in Chile and adjacent Argentina (Ehrhart, pers. comm.). The present phylogeny emphasises the need for further revision of the C. filicaulis species complex. Position of Porodittia. — Surprisingly, the monotypic genus Porodittia is deeply nested inside Calceolaria, together with representatives from sects. Anacyrta and Lobatae (Clade XI). The relationship with Calceolaria is strongly supported, in spite of the fact that Porodittia is divergent from the rest of Calceolaria in number of stamens (3) and corolla shape (open, more bell-shaped, with a bilobate upper lip), Fig. 5F. However, within Calceolariae the possession of three stamens is not unique to this species. Molau (1988) reported that some species of Calceolaria occasionally have three stamens, although this seems to be an aberration in the species concerned. Based on the present results, the genus Porodittia can no longer be recognised as a separate genus. As a species of Calceolaria, its correct name is Calceolaria triandra (Cavanilles) Vahl. The importance of the Huancabamba deflection. — Molau (1988) stressed the importance of the Huancabamba deflection in northern Peru as a dispersal barrier. He showed that the highest numbers of species and endemics of Calceolaria are found in close proximity to the Huancabamba deflection, making this region and the northern part of the middle Andes the centre of species richness for the genus. The present results strongly corroborate the barrier hypothesis. Although the rele- 55 (1) • February 2006: 125–137 vant node is weakly supported (54%) in analysis IV (Fig. 4), the tree indicates a single dispersal event across the deflection. In addition, the tree from analysis II is also consistent with a single dispersal event, and with slightly higher support (68%). Even if evidence for a single dispersal event is weak, the two clades (XII, and XIII through XVI) occurring exclusively north of the Huancabamba deflection are well supported, thus allowing, at most, two separate dispersal events. From a morphological point of view, however, it is not obvious how to distinguish representatives in the northern region from representatives in the middle region. Conclusions. — The phylogenetic hypotheses resulting from the analyses do not provide sufficient data on which to base a complete revision of the sectional classification proposed by Molau (1988). However, they highlight the difficulties in using morphology for inference of relationships among the species and circumscription of the sections. Many of the sections of Molau (1988) are demonstrably artificial and should be viewed as an aid for identification rather than expressing phylogenetic relationships. The distinctness of Calceolariaceae from Scrophulariaceae is supported by the results. Based on the present phylogeny it may be justifiable to include Calceolariaceae in a larger family Gesneriaceae. However, considering that this study was based on a single gene whereas that of Olmstead & al. (2001) was based on three genes, calls for cautious conclusions. Further and more extensive studies are needed on this subject. The present results call for a more thorough taxon sampling, primarily of southern temperate representatives of the genus, to resolve the relationships between the temperate and neotropical species. While the data support the idea that temperate distribution is ancestral in the genus, it does not provide enough resolution or support to draw conclusion on character evolution and geographical migration. ACKNOWLEDGEMENTS This research has been supported by grants from Helge Ax:son Johnson’s Stiftelse and Stiftelsen Längmanska kulturfonden. I wish to thank Ulf Molau, Henrik Nilsson, Roger Eriksson, and Claes Persson, for support and suggestions on the research, and Vivian Aldén for skill in the laboratory work. I also want to express my gratitude to Alexander Schmidt-Lebuhn for providing me with invaluable material for this study, and Olof Ryding for help and support at the Copenhagen Herbarium. Personal thanks to Josefine Norman, and Mattias Lindholm. This article is in memoriam of Prof. Lennart Andersson at Göteborg University. Andersson • Phylogeny of Calceolaria LITERATURE CITED Albach, D. C., Soltis, P. S., Soltis, D. E. & Olmstead, R. G. 2001. Phylogenetic analysis of Asterids based on sequences of four genes. Ann. Missouri Bot. Gard. 88: 163–212. Alvarez, I. & Wendel, J.F. 2003. Ribosomal ITS sequences and plant phylogenetic inference. Molec. Phylog. Evol. 29: 417–34 Andersson, L. & Andersson, S. 2000. A molecular phylogeny of Tropaeolaceae and its systematic implications. Taxon 49: 721–736. Bentham, G. 1846. Calceolaria. Pp. 204–224 in: de Candolle, A. P. (ed.), Prodromus Systematis Naturalis Regni Vegetabilis 10. Treuttel & Würtz, Paris. Cleef, A. M. 1979. The phytogeographical position of the neotropical vascular páramo flora with special reference to the Colombian Cordillera Oriental. Pp. 175–184 in: Larsen, K. & Holm-Nielsen, L. B. (eds.), Tropical Botany. Academic Press, London. Ehrhart, C. 1997. Zur Cytologie chilensischer CalceolariaArten (Scrophulariaceae). Sendtnera 4: 41–59. Ehrhart, C. 2000. Die Gattung Calceolaria (Scrophulariaceae) in Chile. Biblioth. Bot. 153: 1–283. Farris, J. S. 1998. The future of phylogenetic reconstruction. Zool. Scripta 26: 303–311. Farris, J. S., Albert V. A., Källersjö, M., Lipscomb, D. & Kluge, A. G. 1996. Parsimony jackknifing outperforms neighbor-joining. Cladistics 12: 99–124. Galtier, N., Gouy, M. & Gautier, C. 1996. SeaView and Phylo_win, two graphic tools for sequence alignment and molecular phylogeny. Comput. Applic. Biosci. 12: 543–548. Harshman, J. 1994. The effect of irrelevant characters on bootstrap values. Syst. Biol. 43: 419–424. Hillis, D. M., & Dixon, M. T. 1991. Ribosomal DNA: molecular evolution and phylogenetic inference. Quart. Rev. Biol. 66: 411–446. Johnson, L. A. & Soltis, D. E. 1995. Phylogenetic inference in Saxifragaceae sensu stricto and Gilia (Polemoniaceae) using matK sequences. Ann. Missouri Bot. Gard. 82: 149–175. Kränzlin, F. 1907. Scrophulariaceae-Antirrhinoideae-Calceolaireae. in: Engler, A. (ed.), Das Pflanzenreich, IV. 257C. Leipzig. Li, J., Bogle, A. L. & Klein, A. S. 1999. Phylogenetic relationships in the Hamamelidaceae: Evidence from the nucleotide sequences of the plastid gene matK. Pl. Syst. Evol. 218: 205–219. Li, J. & Huang, H. 2002. Molecular phylogeny and infrageneric classification of Actinidia (Actinidiaceae). Syst. Bot. 27: 408–415. Molau, U. 1988. Scrophulariaceae - Part I. Calceolarieae. Flora Neotropica 47: 1–326. Molau, U. 2003. Two new species of Calceolaria (Scrophulariaceae) from the tropical Andes. Novon 13: 101–103. Nickrent, D. L., Schuette, K. P. & Starr, E. M. 1994. A molecular phylogeny of Arceuthobium (Viscaceae) based on nuclear ribosomal DNA internal transcribed spacer sequences. Amer. J. Bot. 81: 1149–1160. Olmstead, R. G., dePamphilis, C. W., Wolfe, A. D., Young, N. D., Elisons, W. J. & Reeves, P. A. 2001. Disintegration 135 Andersson • Phylogeny of Calceolaria of the Scrophulariaceae. Amer. J. Bot. 88: 348–361. Olmstead, R. G., Kim, K-J., Jansen, R. K. & Wagstaff, S. J. 2000. The phylogeny of the Asteridae sensu lato based on chloroplast ndhF gene sequences. Molec. Phylog. Evol. 16: 96–112. Olmstead, R. G. & Reeves, P. A. 1995. Evidence for the polyphyly of the Scrophulariaceae based on chloroplast rbcL and ndhF sequences. Ann. Missouri Bot. Gard. 82: 176–193. Oxelman, B., Backlund, M. & Bremer, B. 1999. Relationships of the Buddlejaceae s. l. investigated using parsimony jacknife and branch support analysis of chloroplast ndhF and rbcL sequence data. Syst. Bot. 24: 164–182. Pennell, F. W. 1920. Scrophulariaceae of Colombia. I. Proc. Acad. Nat. Sci. Philadelphia 72: 136–187. Pennell, F. W. 1945. The genus Calceolaria in southeastern Peru. Proc. Acad. Nat. Sci. Philadelphia 97: 137–177. Pennell, F. W. 1951a. The united-leaved calceolarias of the northern and middle Andes (Scrophulariaceae). Notul. Nat. Acad. Nat. Sci. Philadelphia 236: 1–3. Pennell, F. W. 1951b. The genus Calceolaria in Ecuador, Colombia and Venezuela. Proc. Acad. Nat. Sci. Philadelphia 103: 85–196. Plunkett, G. M., Soltis, D. E. & Soltis, P. S. 1997. Clarification of the relationship between Apiaceae and Araliaceae based on matK and rbcL sequence data. Amer. J. Bot. 84: 565–580. Sang, T., Crawford, D. J. & Stuessy, T. F. 1997. Chloroplast DNA phylogeny, reticulate evolution, and biogeography of Paeonia (Paoniaceae). Amer. J. Bot. 84: 1120–1136. Soltis, D. E., Kuzoff, R. K., Mort, M. E., Zanis, M., Fishbein, M., Hufford, L., Koontz, J. & Arroyo, M. K. 2001. Elucidating deep-level phylogenetic relationships in Saxifragaceae using sequences for six chloroplastic and nuclear DNA regions. Ann. Missouri. Bot. Gard. 88: 669–693. Stevens, P. F. 2001–present. Angiosperm Phylogeny Website. Version 4, May 2003 [and more or less continuously updated since]. http://www.mobot.org/MOBOT/research/AP web/ Swofford, D. L. 2002. PAUP*. Phylogenetic Analysis Using Parsimony (*and Other Methods). Version 4. Sinauer Associates, Sunderland, Massachusetts. Thompson, J. D., Gibson, T. J., Plewniak, F., Jeanmougin, F. & Higgins, D. G. 1997. The ClustalX windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucl. Acids Res. 25: 4876–4882. Vuilleumier, F. 1969. Pleistocene speciation in birds living in the high Andes. Nature 223: 1179–1180. Witasek, J. 1907. Über Kränzlins Bearbeitung der “Scrophulariaceae-Antirrhinoideae-Calceolarieae” in Englers “Pflanzenreich”. Österr. Bot. Zeitschr. 57: 217–230, 259–265. Wojciechowski, M. F., Sanderson, M. J., Baldwin, B. G. & Donoghue, M. J. 1993. Monophyly of aneuploid Astragalus (Fabaceae): Evidence from nuclear ribosomal DNA internal transcribed spacer sequences. Amer. J. Bot. 80: 711–722. 136 55 (1) • February 2006: 125–137 55 (1) • February 2006: 125–137 Andersson • Phylogeny of Calceolaria Appendix. List of accessions. Voucher information, origin, and GenBank accession numbers are given for all species included in this study. Sequences downloaded from GenBank are labelled with GenBank as origin. Species, origin, voucher, ITS acc. no, matK acc. no. Ajuga decumbens Thunb., GenBank, -, -, AF315299; Antirrhinum majus L., GenBank, -, -, AF375189; Calceolaria atahualpae Kränzl. ssp. atahualpae, Bolivia, Ayopaya, Beck & Seidel 14454 (GB), AJ579420, AJ579813; C. ballotifolia Kränzl., Peru, Amazonas, Schmidt-Lebuhn 474 (GB), AJ579434, AJ579812; C. barbata Molau, Peru, Hualgayoc, Molau & Eriksen 3362 (GB), AJ579409, -; C. bicrenata R. & P., Peru, Cajamarca, Becker & Terrones 995 (GB), AJ579422, AJ580491; C. biflora Lam., Argentina, Chubut, Hjerting & al. 75-152 (C), AJ579390, –; C. boliviana (Rusby) Pennell, Bolivia, La Paz, Schmidt-Lebuhn 593 (GB), AJ579404, AJ579815; C. brachiata Sodiro ex Kränzl., Ecuador, Cotopaxi, Steiner 234 (GB), AJ579462, -; C. buchtieniana Kränzl., Bolivia, Murillo, Moraes 686 (GB), AJ579411, AJ579816; C. calycina Benth., Peru, Amazonas, Schmidt-Lebuhn 481 (GB), AJ579444, AJ579802; C. chelidonioides HBK, Peru, Huánuco, Schmidt-Lebuhn 527 (GB), AJ579441, AJ579796; C. comosa Pennell ssp. comosa, Peru, Amazonas, Schmidt-Lebuhn 482 (GB), AJ579427, AJ579803; C. connatifolia Pennell, Peru, Amazonas, Schmidt-Lebuhn 473 (GB), AJ579469, AJ579823; C. corymbosa Ruiz & Pav. ssp. montana, Bolivia, Mendoza, Böcher & al. 1322 (C), AJ579393, -; C. cypripediiflora Kränzl., Peru, Cuzco, Molau 752 (GB), AJ579405, -; C. dentifolia Edwin, Peru, Amazonas, Schmidt-Lebuhn 477 (GB), AJ579450, AJ579794; C. dichotoma Lam., Ecuador, Loja, Harling & Andersson 18585 (GB), AJ579437, -; C. dilatata Benth., Ecuador, Cotopaxi, Løjtnant & Molau 13950 (GB), AJ579464, AJ580492; C. engleriana Kränzl. ssp. engleriana, Peru, Cuzco, Molau & Öhman 1602 (GB), AJ579401, AJ579817; C. ericoides Vahl ssp. ericoides, Ecuador, Pichincha, Molau & Öhman 1138 (GB), AJ579408, AJ579825; C. filicaulis Clos. ssp. filicaulis, Argentina, Neuquén, Böcher & al. 1811 (C), AJ579386, -; C. filicaulis Clos. ssp. luxurians (Witasek), Chile, Metropolitana, Rahn & Ødum 4790 (C), AJ579394, -; C. flexuosa R. & P. ssp. flexuosa, cultivated, C, Molau s.n., AJ579415, AJ579793; C. fusca Pennell, Ecuador, Loja, Øllgaard & al. 74155 (GB), AJ579466, AJ579824; C. gaultherioides Molau, Peru, Amazonas, Molau & Eriksen 3410 (GB), AJ579451, AJ579798; C. germanii W.t., Argentina, Neuquén, Böcher & al. 1579 (C), AJ579387, -; C. glauca R. & P., Peru, Ancash, Schmidt-Lebuhn 519 (GB), AJ579423, AJ579805; C. gossypina Benth., Ecuador, Salasaca, Franquemont 180A (GB), AJ579459, -; C. helianthemoides HBK, Ecuador, Loja, Elleman 91661 (GB), AJ579461, -; C. inflexa R. & P., Peru, Huánuco, Molau 777 (GB), AJ579418, -; C. integrifolia Murr. in L.., Chile, Santiago, Steward 34 (C), AJ579396, -; C. lamiifolia HBK, Ecuador, Carchi, Schmidt-Lebuhn 360 (GB), AJ579435, -; C. lanata HBK, Cultivated, GBG, S. Andersson 3 (GB), AJ579447, AJ579826; C. lavandulifolia HBK, Ecuador, Azuay, Harling 27232 (GB), AJ579460, AJ579818; C. lehmannina Kränzl., Ecuador, Carchi, Løjtnant & al. 12589 (GB), AJ579454, -; C. linearis R. & P., Peru, Yungay, Molau & Eriksen 3481 (GB), AJ579410, -; C. lobata Cav., Bolivia, La Paz, Luteyn & Dorr 13783 (GB), AJ579428, AJ579791; C. maculata Edwin, Peru, Cajamarca, Molau 891 (GB), AJ579433, AJ579810; C. mandoniana Kränzl., Bolivia, La Paz, Schmidt-Lebuhn 596 (GB), AJ579442, AJ579808; C. martinezii Kränzl., Ecuador, Tungurahua, Schmidt-Lebuhn 392 (GB), AJ579456, AJ579800; C. microbefaria Kränzl. ssp. fruticosa (Pennell) Molau, Ecuador, Pichincha, Holm-Nielsen & Øllgaard 24344 (GB), AJ579468, -; C. moyobambae Kränzl., Peru, Amazonas, Schmidt-Lebuhn 478 (GB), AJ579449, AJ579827; C. myriophylla Kränzl., Peru, Cusco, SchmidtLebuhn 588 (GB), AJ579403, AJ579828; C. nivalis HBK ssp. nivalis, Ecuador, Loja, Harling 25373 (GB), AJ579465, -; C. oblonga R. & P., Peru, Oxapampa, Schmidt-Lebuhn 552 (GB), AJ579414, AJ579821; C. obtusa Molau, Ecuador, Azuay, Harling & al. 15074 (GB), AJ579432, -; C. parvifolia Wedd., Bolivia, La Paz, Gutte 86 (GB), AJ579412, -; C. pavonii Benth., Peru, Amazonas, Schmidt-Lebuhn 459 (GB), AJ579445, AJ579814; C. pedunculata Molau ssp. pedunculata, Ecuador, Pichincha, Schmidt-Lebuhn 355 (GB), AJ579453, AJ579809; C. penlandii Pennell ssp. penlandii, Ecuador, Napo, Molau & Eriksen 2173 (GB), AJ579436, AJ579807; C. pennellii, cultivated, GBG, S. Andersson 1 (GB), AJ579389, AJ580488; C. pefoliata L.f., Ecuador, Napo, Løjtnant & Molau 12857 (GB), AJ579443, AJ579797; C. phaceliifolia Edwin, Peru, Ancash, Schmidt-Lebuhn 501 (GB), AJ579429, AJ579831; C. phaeotricha Molau, Ecuador, Morona, Molau & al. 2929 (GB), AJ579470, AJ579830; C. pinifolia CAV., Argentina, San Juan, Pedersen 15255 (C), AJ579392, -; C. plectranthifolia Walp., Cultivated, GBG, Molau s.n. (GB), AJ579398, -; C. polifolia Hook., Chile, Santiago, Steward 32 (C), AJ579397, -; C. polyclada Kränzl., Bolivia, Tarija, Killeen 2712, AJ579402, -; C. polyrhiza, Cultivated, GBG, Molau s.n. (GB), AJ579391, AJ579822; C. purpurascens (Kränzl.) Molau, Ecuador, Pichincha, Molau & al. 2441 (GB), AJ579463, AJ579792; C. revoluta Pennell, Peru, Cuzco, Molau 760 (GB), AJ579407, -; C. rivularis Kränzl., Bolivia, Ayopaya, Beck & Seidel 14535 (GB), AJ579440, -; C. rosmarinifolia Lam., Ecuador, Azuay, Harling & Ståhl 27035a (GB), AJ579457, AJ579804; C. rugulosa Edwin, Peru, Piura, Schmidt-Lebuhn 409 (GB), AJ579430, AJ579795; C. rupestris Molau, Peru, Tayacaja, SchmidtLebuhn 560 (GB), AJ579406, AJ579811; C. salicifolia R. & P. ssp. salicifolia, Peru, Huánuco, Molau 825 (GB), AJ579452, -; C. scabra R. & P., Peru, Junin, Molau 802 (GB), AJ579421, -; C. semiconnata Pennell, Ecuador, Loja, Harling & Andersson 22050 (GB), AJ579467, -; C. sericea Pennell, Ecuador, Pichincha, Schmidt-Lebuhn 334 (GB), AJ579426, -; C. sparsiflora Kuntze, Bolivia, Murillo, Beck 3019 (GB), AJ579417, AJ579829; C. spruceana Kränzl., Ecuador, Tungurahua, Schmidt-Lebuhn 390 (GB), AJ579458, AJ579799; C. stricta HBK, Ecuador, Loja, Harling & Andersson 14086 (GB), AJ579455, -; C. tenella Poepp & Randl, Chile, Ozorno, Rahn 4659 (C), AJ579400, -; C. teucrioides Griseb., Bolivia, Tarija, Molau 710 (GBG), AJ579413, -; C. tetragona Benth. ssp. tetragona, Peru, Amazonas, Schmidt-Lebuhn 471 (GB), AJ579448, AJ579801; C. thyrsiflora Grah., Chile, Santiago, Steward 23 (C), AJ579395, -; C. tomentosa R. & P., Peru, Amazonas, Schmidt-Lebuhn 420 (GB), AJ579446, AJ579820; C. tripartita R. & P., Ecuador, Loja, Jørgensen & al. 5 (GB), AJ579439, -; C. tucumana Descole, Bolivia, Mendoza, Molau 708 (GB), AJ579399, AJ579806; C. umbellata Weddell, Argentina, Tutumán, Bothmer & Hjerting 6341 (C), AJ579384, -; C. uniflora Lam., Cultivated, GBG, S. Andersson 2 (GB), AJ579388, AJ580489; C. utricularioides Benth., Peru, Cajamarca, Molau & al. 1800(I) (GB), AJ579438, -; C. vaccinioides Kränzl., Bolivia, La Paz, Luteyn & Dorr 13485 (GB), AJ579419, AJ580490; C. variifolia Edwin, Ecuador, Loja, Schmidt-Lebuhn 405 (GB), AJ579425, -; C. virgata R. & P., Peru, Amazonas, Schmidt-Lebuhn 483 (GB), AJ579431, AJ579819; C. vulpina Kränzl., Peru, Hacienda Huari, Lourteig 3138 (GB), AJ579416, -; Jasminum fluminense Vell., GenBank, -, -, AJ388202, AJ388272; Jovellana violacea (Cav.) G. Don, Cultivated, GBG, S. Andersson 4 (GB), AJ579385, AJ580487; Kohleria spicata (Kunth) Oerst., Cultivated, GBG, S. Andersson 5 (GB), AJ579471, AJ580486; K. rugata (Scheidw.) L.P. Kvist & L.E. Skog, Genbank, Zimmer & al., AY047075, -; Myoporum acuminatum R. Br., GenBank, -, -, AJ429347; Olea europaea L., GenBank, Erixon & Bremer 34 (UPS), -, AJ429335; Orobanche uniflora L., GenBank, -, -, AF051996; Pedicularis sylvatica L., GenBank, -, -, AF489959; Peltanthera floribunda Benth., GenBank, Hammel 19855 (MO), -, AJ429330; Polypremum procumbens L., GenBank, Struwe 1000 (UPS), -, AJ429351; Porodittia triandra (Cav.) G. Don, Peru, Ancash, Molau & Eriksen 3486 (GB), AJ579424, AJ579832; Scrophularia arguta Sol. ex Aiton, GenBank, -, -, AJ429349; Sesamum indicum L., GenBank, Erixon & Bremer 43 (UPS), -, AJ429340; Streptocarpus caulescens Vatke, GenBank, -, -, AJ429331; Tetrachondra patagonica Skottsb., GenBank, Martinsson & Swenson 314 (UPS), -, AJ429352; Verbena hybrida Groenl. & Rumpler, GenBank, -, -, AF315302; Veronica arvensis L., GenBank, -, -, AF052003. 137