| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
0 Department of Biology, Boise State University, 1910 University Drive, Boise, Idaho 83725 USA
Received for publication April 29, 1999. Accepted for publication December 7, 1999.
ABSTRACT
Generic relationships within Episcieae were assessed using ITS and ndhF sequences. Previous analyses of this tribe have focussed only on ndhF data and have excluded two genera, Rhoogeton and Oerstedina, which are included in this analysis. Data were analyzed using both parsimony and maximum-likelihood methods. Results from partition homogeneity tests imply that the two data sets are significantly incongruent, but when Rhoogeton is removed from the analysis, the data sets are not significantly different. The combined data sets reveal greater strength of relationships within the tribe with the exception of the position of Rhoogeton. Poorly or unresolved relationships based exclusively on ndhF data are more fully resolved with ITS data. These resolved clades include the monophyly of the genera Columnea and Paradrymonia and the sister-group relationship of Nematanthus and Codonanthe. A closer affinity between Neomortonia nummularia and N. rosea than has previously been seen is apparent from these data, although these two species are not monophyletic in any tree. Lastly, Capanea appears to be a member of Gloxinieae, although C. grandiflora remains within Episcieae. Evolution of fruit type, epiphytic habit, and presence of tubers is re-examined with the new data presented here.
Key Words: cladistics • DNA sequencing • Episcieae • Gesneriaceae • ITS • maximum-likelihood • ndhF.
Cladistic analyses of DNA sequences of single genes can provide useful insights into the systematics and evolution of plants (Chase et al., 1993
; Baldwin et al., 1995
; Smith et al., 1997b
), however, single genes may reflect gene trees and not species trees (Avise et al., 1983
; Doyle, 1992
; Maddison, 1995
). To maximize the effectiveness of molecular data, it is best to examine DNA sequences from several independent sources (Donoghue, 1994
; de Queiroz, Donoghue, and Kim, 1995
; Hillis, 1995
; Miyamoto and Fitch, 1995
; Maddison, 1997
; Soltis et al., 1998
). Most recently this approach has meant examining the sequences of genes from complementary (and thus independent) portions of the genome, most notably from the chloroplast and nuclear genomes (Seelanen, Schnabel, and Wendel, 1997
; Rodman et al., 1998
; Baum, Small, and Wendel, 1998
; Soltis et al., 1998
). Although there are instances where data from different genetic sources have resulted in no immediate resolution of phylogenetic relationships (Mason-Gamer and Kellogg, 1996), many cases indicate congruence (Seelanen, Schnabel, and Wendel, 1997
; Munro and Linder, 1998
; Rodman et al., 1998
) or at least partial congruence where discrepancies can be further examined for additional resolution or for another biological explanation (Graham et al., 1998
; Lutzoni, 1998
).
The Gesneriaceae have provided several useful examples for comparing data from different sources (Smith et al., 1996
). The Gesneriaceae comprise ~2500–3700 species in 120–147 genera, distributed primarily in the tropics with a few temperate species in Europe, China, and Japan (Heywood, 1978
; Burtt and Wiehler, 1995
). The majority of species in Gesneriaceae are herbaceous perennials, but others are annuals, shrubs, lianas, or trees. Many species (20%) are epiphytic, and Gesneriaceae rank among the top ten plant families in terms of absolute numbers of epiphytic taxa (Madison, 1977
; Kress, 1986
). Leaves are opposite in the majority of the subfamily, but anisophylly, leading to an apparent alternate arrangement following abscission of the smaller leaf, is common. The family is divided into two subfamilies with Gesnerioideae found almost exclusively in the neotropics and Cyrtandroideae almost exclusively in the Old World (Burtt and Wiehler, 1995
; Smith et al., 1997b
). The Gesnerioideae are divided further into six tribes and 60 genera (Burtt and Wiehler, 1995
; Smith et al., 1997b
).
The Episcieae are one of the largest tribes in Gesnerioideae and comprise 17 genera (Table 1). Morphologically they are distinct among other Gesnerioideae in their nodal anatomy (Wiehler, 1983
), a character not used in a cladistic analysis of morphology (Smith, 1996
). Episcieae are characterized by a three-trace trilacunar node with split lateral bundles, superior ovaries, and most members have chromosome counts of x = 9 [x = 8 in Codonanthe (Mart.) Hanst. and Nematanthus Schrader]. This combination of character states is unknown among other neotropical members of the family. Within Gesneriaceae, the tribe Episcieae has been examined more fully using the chloroplast gene ndhF (Smith and Carroll, 1997
). These data have provided support for the monophyly of this tribe, in agreement with earlier molecular (Smith et al., 1997b
) and morphological data (Smith, 1996
).
|
) and is distinguished by the dehiscence of the fleshy fruit walls to reveal a brightly colored interior pericarp that contrasts with the dull, often purple-black mass of seeds and funiculi (Wiehler, 1983
).
Although previous cladistic studies within Episcieae based on ndhF sequences have provided important resolution regarding phylogenetic relationships and evolution of morphological features, many of the clades have remained poorly resolved or weakly supported (Smith and Carroll, 1997
). For example, resolution of relationships among and within several taxa that have been termed "waste-basket" genera were not resolved in the earlier analysis (Smith and Carroll, 1997
). Among these were Columnea L., and Alloplectus Mart. In his earlier revision of Episcieae, Wiehler (1983)
described several new genera and rearranged the position of many species among those already described. One group that received a major rearrangement was the Columnea alliance. Wiehler (1973, 1983)
split this group, which initially comprised three genera (Columnea, Alloplectus and Drymonia Mart.), into eight, adding Corytoplectus Oerst., Pentadenia (Planch.) Hanst., Trichantha Hook., Dalbergaria Tuss., and Bucinellina Wiehler, the latter four divided from Columnea s. l. This split has remained controversial, and more thorough examinations of these species on the basis of morphology (Kvist and Skog, 1993
; Smith, 1994
; Smith and Sytsma, 1994a
) and chloroplast DNA (cpDNA) restriction site analysis (Smith and Sytsma, 1994b, c
) indicated that Columnea s.l. was best retained as a single genus and that the segregate genera of Wiehler (1983)
were not monophyletic. A test of this classification system using ndhF sequences not only was unable to resolve the relationships of groups within Columnea s. l., but could not provide adequate monophyletic delimitations of Alloplectus, Drymonia, Columnea, Corytoplectus, and a Neomortonia Wiehler species included in the analysis (Smith and Carroll, 1997
).
Another unexpected relationship from the cladistic analysis of ndhF sequences regarding Episcieae was the paraphyly of Paradrymonia Hanst. (Smith and Carroll, 1997
). The previous study sampled only three species within this genus, one of which, P. densa, was included because of its unusual characteristics. Paradrymonia densa did not form a monophyletic group with the remaining two species of the genus and caused a paraphyletic relationship. Because of the limited sampling, it is difficult to determine whether the paraphyly of this genus was due to sampling, limited resolving power of the ndhF sequences, or whether P. densa does indeed represent an independent lineage from the remainder of the genus.
All but two genera of Episcieae have a chromosome count of n = 9. These are Codonanthe and Nematanthus and share a count of n = 8. These are the only genera in the neotropical subfamily Gesnerioideae to have a count of n = 8; both have centers of diversity in Brazil (Wiehler, 1983
) and Nematanthus is endemic there (Chautems, 1984, 1988
). Hybrids have been generated between the two (Wiehler, 1977
), implying close phylogenetic affinity. Therefore it was surprising to find that the species of Codonanthe and Nematanthus used in the ndhF analysis were not only not a monophyletic group, but were not closely related (Smith and Carroll, 1997
).
The present analysis seeks to resolve many of these questions that remain from the previous molecular analysis by examining a second source of molecular data. These data were gathered with the goal of (1) resolving relationships not resolved with previous data, (2) strengthening support for areas of congruence between the two data sets, and (3) providing an additional, non-cpDNA source of data in an attempt to determine whether resolved relationships from previous analyses reflect only gene trees or species trees. Additionally, it was the goal of this study to include representatives of the two genera not previously sampled in the analysis, the Guyana endemic Rhoogeton Leeuw. and the Central American Oerstedina Wiehler.
The sources of data for this analysis are the nuclear ribosomal RNA intercistronic transcribed spacer regions (ITS 1 and 2) and ndhF. The ndhF gene has been used extensively within Gesneriaceae (Smith et al., 1997a, b, 1998
; Smith and Carroll, 1997
; Smith and Atkinson, 1998
). The use of ITS sequences for systematic studies has provided a valuable tool to resolve relationships at the species and generic levels (Baldwin et al., 1995
; Möller and Cronk, 1997
) where the higher levels of nucleotide substitution rates found in these regions do not create extensive homoplasy.
MATERIALS AND METHODS
The gene sequences used in this analysis were generated by thermal cycle sequencing (Innis et el., 1988
) of previously amplified regions. The ITS regions and ndhF gene were amplified in two overlapping sections (primers 5P–2G and 3P–8P; Möller and Cronk, 1997
; positions 1–1350, and 972–2044, respectively) from genomic DNA isolated from fresh, frozen, or silica-gel-dried material for most species (Smith et al., 1992
). Amplification and sequencing procedures followed that of Smith et al. (1997b)
used for other members of Gesneriaceae. Alignments of both ITS and ndhF regions were done by hand since the sequence divergence was not overly high among taxa compared and alignment by hand was easy and straightforward.
Most ndhF sequences were from previously published data (Smith and Carroll, 1997
), but additional sequences were obtained for Rhoogeton and Oerstedina, which were not included in the previous analysis. DNA for Rhoogeton and Oerstedina was obtained from herbarium specimens (Savolainen et al., 1995
) using the DNEasy Plant miniprep kits (Qiagen, Valencia, California, USA) following the manufacturer's instructions. The ndhF gene was amplified in two overlapping fragments using primers 172–1350R and 972–2044R. The first fragment is smaller than those reported previously since amplifying the DNA from herbarium specimens required successive amplifications using internal primers. Initial amplifications followed procedures for DNA as described elsewhere (Smith et al., 1997b
) using primers 1 and 1350R for the first half and 803 and 2044R for the second half. Subsequent amplifications required the use of 172 and 972 as forward primers, although the same reverse primers resulted in successful amplifications.
This analysis focused on the relationships within Episcieae, therefore nearly all genera within the tribe and several species within some of the larger genera were sampled for both ITS and ndhF data sets. Two to five species each were used to represent the larger or potentially nonmonophyletic genera (Columnea, Episcia Mart., Alloplectus, and Paradrymonia). Two of the three species of Neomortonia were included in this analysis to test the monophyly of this small, but morphologically diverse genus. The species used in the analysis, voucher information, and GenBank accession numbers are included in Table 1. As some taxa have only ndhF or ITS data available, the taxon sampling for each analysis differed slightly. Forty-two species had sequences for ITS only, 39 for ndhF only, and 33 had sequences for both regions. Table 1 indicates those taxa for which only one data set was available. The data matrix contains 1.71% and 20.83% (including all indels) missing cells for ndhF and ITS, respectively, based on total sequence alignments.
Outgroups were selected to root the tree based on recent morphological and molecular analyses of tribal relationships within Gesneriaceae (Smith, 1996
; Smith et al., 1997b
). The most appropriate outgroups for Episcieae were genera representing sister tribes Gesnerieae and Gloxinieae.
Phylogenetic analysis
Phylogenetic divergence was reconstructed using PAUP version 4.0d64 to implement parsimony (Farris, 1970
; Farris, Kluge, and Eckardt, 1970
; Swofford and Maddison, 1987
) and maximum likelihood estimates (MLE). Parsimony has been used utilized in systematic studies and has provided numerous insights into plant evolution. With molecular data it is possible to form hypotheses on the specific rates of nucleotide substitutions and to generate trees that best fit these substitution rate models (maximum-likelihood). Both methodologies are appealing, and it would be valuable to empirically compare the methods with regard to their resulting trees. A previous analysis comparing parsimony to different models of molecular evolution in Gesneriaceae (tribes Beslerieae and Napeantheae) produced highly congruent trees (Smith, 2000
), therefore it was a goal of this study to make a similar comparison on a different group of taxa.
In this analysis, trees were generated using the general heuristic option. To search for islands of equally parsimonious trees (Maddison, 1991
), the search strategy of Olmstead and Palmer (1994)
was implemented searching for 1000 trees each in five subsequent analyses with the nearest neighbor interchange (NNI) search option in effect and mulpars "off." Each of the results from the five NNI searches was used as the starting tree(s) for a search with tree bisection reconnection (TBR) and mulpars "on." All analyses were conducted with indels treated as missing data since all indels were either small (one to three base pairs) or were autapomorphic. This search strategy was used for several combinations of data sets, (1) ITS only, (2) ndhF only, (3) ITS and ndhF with all taxa from both analyses, (4) ITS and ndhF with only taxa common to both analyses, and (5) ITS and ndhF with only taxa common to both analyses and Rhoogeton excluded. The partition homogeneity test also was examined by partitioning the data into ITS and ndhF for all taxa common to both data sets and a second analysis that excluded Rhoogeton. The partition homogeneity test used 1000 replicates with the NNI search option and mulpars "off." Attempts to use TBR and mulpars "on" resulted in only four replicate searches after 24 h.
Maximum likelihood estimate trees were generated for (1) ITS data, (2) ndhF data, (3) combined ITS and ndhF data with taxa in common both with, and (4) without Rhoogeton. Maximum likelihood searches were conducted using the heuristic search option with TBR and mulpars "on." Under the MLE option, the model of Hasegawa, Kishino, and Yano (1985)
was used, which allows for unequal nucleotide frequencies and differential rates for transitions and transversions. The assumed nucleotide frequencies were estimated from the data: A = 0.26661, 0.27607, 0.26620, 0.26649, C = 0.17570, 0.20148, 0.17637, 0.17579, G = 0.019341, 0.21641, 0.19365, 0.19317, and T = 0.36427, 0.30603, 0.36379, 0.36455, respectively for ITS, ndhF, and ITS combined with ndhF with and without Rhoogeton.
Branch support analysis was performed to examine trees that were six or fewer steps longer than the most parsimonious trees (Bremer, 1988, 1994
; Donoghue et al., 1992
). This type of analysis provides an indication of the robustness of the data by determining which clades persist in a consensus tree as parsimony is relaxed. Clades that persisted in strict consensus trees six steps beyond the most parsimonious trees were examined using the constraints option to search for the shortest tree that did not contain that clade. Bootstrap analysis (Felsenstein, 1985
) was performed using 100 replicates with TBR and mulpars "off."
RESULTS
Amplification of the ITS regions resulted in fragments of ~450 bp each. Full sequences of ITS 1 and 2 were obtained from both strands using all four primers except in regions close to the primers where only a single strand was sequenced. ITS sequences were not obtained for Capanea grandiflora, Oerstedina, and several other species for which ndhF sequences had been obtained (Table 1). ITS sequences were readily aligned by hand, initiating small insertions and deletions (indels) for complete alignments. Indels were not included in any of the phylogenetic analyses due to their small size. All indels were coded as missing data. The combined ITS 1 and 2 regions resulted in a total of 306 parsimony-informative characters. Maximum parsimony analysis of these data resulted in three trees of 1822 steps each, consistency index (CI) = 0.46, and retention index (RI) = 0.54 (Fig. 1).
|
). Paradrymonia is monophyletic and although Neomortonia nummularia is still close to Episcia as it was with ndhF data alone, it is sister to a more inclusive clade (Fig. 1). Nematanthus and Codonanthe are a monophyletic group as would be expected on the basis of their shared chromosome count of n = 8. However, the close sister-group relationship of Codonanthe and Codonanthopsis Mansf. seen with ndhF data (Smith and Carroll, 1997
) is still present with ITS data. Lastly, Rhoogeton, not included in earlier analyses, is within Gloxinieae, designated as the outgroup. Rhoogeton itself was designated as part of the ingroup in the analysis.
To minimize computational time, the MLE analysis of the ITS data used only taxa that were common to both ndhF and ITS data sets (Fig. 2). The -ln likelihood was 7782.33033. In general the MLE tree is similar to the parsimony tree with the exception that both Neomortonia species are sequential sister groups to the Columnea/Alloplectus/Drymonia clade and Corytoplectus is sister to this more inclusive clade (Fig. 2). Alsobia punctata is not within the Alloplectus/Drymonia clade as it was with parsimony (Fig. 1) but is instead in a clade with Cobananthus Wiehler, and Rufodorsia Wiehler as was seen with ndhF data previously (Smith and Carroll, 1997
).
|
) with the inclusion of Rhoogeton and Oerstedina and the use of only a subset of the Gloxinieae and Gesnerieae as outgroups. The parsimony analysis resulted in 47 trees of 2196 steps each, CI = 0.46, and RI = 0.37. One of these trees is presented in Fig. 3, and the strict consensus of all trees is indicated by dashed lines. This consensus tree is much less resolved than the previous results from ndhF analyses (Smith and Carroll, 1997
), possibly in part due to a smaller outgroup sampling and a subsequently more complete search by PAUP (only six most parsimonious trees were found in the earlier, but larger analysis with numerous outgroups and a greater taxon sampling from other tribes). This tree places both Rhoogeton and Oerstedina within Episcieae, and although the position of Rhoogeton is not resolved, Oerstedina is sister to Rufodorsia.
|
) in that Columnea is not monophyletic and contains Alloplectus, Drymonia, and Neomortonia rosea. Likewise, Neomortonia and Paradrymonia are not monophyletic and Capanea grandiflora is within the Episcieae.
|
The combined ITS and ndhF data set was analyzed in three different ways since sequences for both ITS and ndhF were not available for all taxa. The analyses differed in that the first used all taxa with data scored as missing for the gene not sequenced, the second analysis used only taxa with both sequences, and the third used only taxa in common, but with Rhoogeton excluded. The first analysis resulted in 17 trees of 4151 steps each, CI = 0.46, and RI = 0.54. The strict consensus of this tree was less resolved than either ITS or ndhF data analyzed singly (tree not shown). A decay analysis resulted in only eight clades persisting in a consensus tree two steps longer than the most parsimonious tree, and these were clades that are well supported and congruent to both ITS and ndhF analyses (marked with asterisks in Fig. 1). A bootstrap analysis indicated only six clades with support >75%. The inclusion of Rhoogeton in the combined analysis resulted in a large amount of homoplasy and the combined data set tree that includes Rhoogeton is the least resolved tree of all data sets analyzed in this study. The second analysis, using only taxa with sequences for both ITS and ndhF, resulted in ten trees of 3698 steps each, CI = 0.48, and RI = 0.34. The consensus of these trees is much more resolved (Fig. 5) than the consensus tree from the first combined data analysis, and clades have better support in terms of decay and bootstrap values. The third analysis with Rhoogeton removed resulted in a single tree of 3467 steps, CI = 0.49, and RI = 0.51 (Fig. 5). This tree is the most well supported in terms of bootstrap, branch lengths, and decay indices. Columnea is delimited as a monophyletic genus separate from Drymonia, Alloplectus, and Neomortonia rosea (Fig. 5). Paradrymonia is monophyletic, and Nematanthus and Codonanthe are in a monophyletic clade.
|
DISCUSSION
Comparison of data sets
Due to the poor level of resolution for the ndhF parsimony tree (Fig. 3), any comparison of this tree to the ITS parsimony tree (Fig. 1) is mostly irrelevant beyond discussion of Rhoogeton (below) since few other discrepant clades have any support. A comparison of the two MLE trees (Figs. 2, 4) indicates a high degree of similarity. For example, the Columnea alliance clade (Columnea, Alloplectus, Drymonia, Neomortonia, and Corytoplectus; marked with the letter C in Figs. 2 and 4) is similar between the two trees in terms of generic composition, although Columnea is monophyletic with the ITS data. In both trees, Corytoplectus is sister to the remainder of the clade and Neomortonia nummularia is sister to the remainder of the clade excluding Corytoplectus (Figs. 2, 4). The sister to the Columnea alliance clade in the ndhF-MLE analysis is a clade that consists of Episcia, Alsobia Hanst., Cobananthus, Oerstedina, and Rufodorsia (Fig. 4). The ITS tree does not include all of these genera as a single monophyletic clade sister to the Columnea alliance, but instead is represented by two clades (Fig. 2). The first consists of Episcia alone and is sister to the Columnea alliance clade, and the second consists of Alsobia, Rufodorsia, and Cobananthus. The latter clade is sister to the inclusive Episcia and Columnea alliance clades. The most likely explanation for this difference is the greater sampling within Alsobia and inclusion of Oerstedina in the ndhF analysis. Sister to these taxa in both trees is a clade composed of Nematanthus, Codonanthe, and Codonanthopsis, although Rhoogeton and Paradrymonia densa are included in this clade in the ndhF tree (Fig. 4). Lastly, both trees have the Nautilocalyx Linden ex Hanst./Chrysothemis Dcne./Paradrymonia clade as the sister to the remainder of the tribe, although the ndhF tree also includes Capanea grandiflora (see below) and lacks Paradrymonia densa (see below).
Despite some minor differences between the ITS and ndhF analyses, the overall results are similar. The main discrepancy is in the position of Rhoogeton as either a member of Episcieae or Gloxinieae.
Comparison of MLE and parsimony
For the most part the single tree obtained from each of the MLE analyses is one of the trees obtained from the parsimony analysis (Figs. 1–5). In a few instances the relationships are different. For example, with the ITS analysis, the two Neomortonia species are distant from each other with parsimony (Fig. 1), whereas they are successive sister species in the MLE analysis (Fig. 2). Similarly, Nautilocalyx/Chrysothemis is sister to Paradrymonia in the ITS-MLE analysis (Fig. 2), but is sister to Episcia with parsimony (Fig. 1). Although a cursory comparison of Figs. 3 and 4 seems to indicate numerous differences between the MLE and parsimony trees, a more careful analysis reveals that the discrepancies are only in areas of weak or no support in the parsimony tree, most being in clades that collapse in a strict consensus of all trees (Fig. 3). Therefore, the differences between these trees is largely due to the arbitrary selection of the tree represented in Fig. 3 and not any aspect of data analysis.
Combined data sets
The ndhF and ITS data sets provide different degrees of resolution and for the most part congruent results where both are resolved. The partition homogeneity test indicates that there is significantly different phylogenetic signal between the two data sets (P = 0.001). However, this discrepancy is eliminated when Rhoogeton is removed from the analysis and the partition homogeneity test instead indicates a P value of 0.065. Therefore, combining the data sets may be the best solution since the discrepancy can be traced to a single taxon. The position of Rhoogeton in the combined analysis must be viewed with skepticism. The uncertain position of Rhoogeton in the combined analysis results in a large amount of homoplasy and as a result, the combined data set tree that includes Rhoogeton in the analysis is the least resolved tree of all data sets analyzed in this study. This would be expected when incongruent data are combined (Bull et al., 1993
; Huelsenbeck, Bull, and Cunninghan, 1996
). However, when the single discrepant species is removed, the combined data tree is fully resolved, has higher consistency and retention indices, and the support for clades is higher based on bootstrap and decay values (Fig. 5). Therefore this combined data tree can be used as a basis for evolutionary analyses within Episcieae and is likely to be the best estimate of phylogenetic relationships within this tribe, although the position of Rhoogeton remains unresolved.
Although the basis for the incongruency between ITS and ndhF data for Rhoogeton cannot be fully resolved here, one possibility seems likely. Rhoogeton is isolated geographically from most other Gesneriaceae. Although other species can be found in the same general area, Rhoogeton is restricted to bases of water falls where the plants are continually washed in the mist and no other Gesneriaceae are present. Therefore hybridization is extremely unlikely to account for the incongruency between ITS and ndhF sequences in Rhoogeton. The species of Rhoogeton used in this analysis was R. viviparus, named because of the small plantlets that form from the edges of the leaves (Leeuwenberg, 1958
). This genus has been poorly studied, therefore nothing is known regarding the relative rates by which this species reproduces sexually vs. asexually. If asexual reproduction is exclusive or predominant, it is possible that somatic mutations have accumulated in ITS sequences such that the individual sampled has diverged to the point that the sequence no longer reflects its relationships to other taxa. The ndhF gene may not have undergone such an extensive sequence evolution due to functional constraints. If this were true, the ITS would appear discrepant to other sequences, such as it is in this analysis.
Phylogenetic relationships within Episcieae
The discrepant position of Rhoogeton
The greatest discrepancy between the ITS and ndhF data is the position of Rhoogeton (Figs. 1–5) and is confirmed to be the primary source of incongruence for a significant difference between the two data sets. A higher (and not significant) P value of 0.065 is obtained when Rhoogeton is excluded from the partition homogeneity test. This genus has posed problems to previous investigators, largely due to its reduced vegetative structure and thus lack of synapomorphies to ally it to other taxa. Two species of Rhoogeton are found on moist, moss-covered rocks at the base of waterfalls in Guyana where they are regularly sprayed with mist. The plants are tuberous, acaulescent herbs with orange-to-red tubular corollas, a superior ovary, and an enlarged dorsal nectary gland (Leeuwenberg, 1958
). A third species, R. panamensis Wiehler, is endemic to Panama and Costa Rica and differs by having white corollas (Wiehler, 1992, 1995
). Rhoogeton was placed in Episcieae by Wiehler (1983)
on the basis of its superior ovary and single large nectary gland. It has remained there, primarily due to a paucity of collections, inability to maintain plants in cultivation, and a lack of scientific investigation. The inclusion of ndhF and ITS sequences unfortunately does not provide strong evidence for its placement either within or outside of Episcieae (Figs. 1–5) since the ndhF data place Rhoogeton firmly within Episcieae and ITS data place it equally firmly outside of Episcieae and within Gloxinieae. The latter hypothesis cannot be fully examined here since additional tribes such as Sinningieae, which are also tuberous, were not sampled with ITS data. The position of Rhoogeton within Episcieae is maintained by ndhF data in studies that sample all genera of the subfamily Gesnerioideae (Smith, unpublished data), implying that it is not a limitation of sampling that causes its affinity to Episcieae.
Placement of Rhoogeton in either tribe has support from morphological data. The superior ovary and enlarged dorsal nectar gland are indicative of Episcieae (Wiehler, 1983
). In contrast, the presence of a tuber, corolla shape, and calyx venation are all similar to Lembocarpus Leeuw. (Gloxinieae), an endemic genus of French Guiana and Surinam. Further studies using additional molecular regions and morphological data are necessary to resolve the position of this genus.
The Columnea alliance
In a previous analysis of Episcieae using only ndhF sequences, the monophyly of several genera was not resolved (Smith and Carroll, 1997
). One of the larger groups of genera consisted of Columnea, Alloplectus, Drymonia, Corytoplectus, and Neomortonia (Smith and Carroll, 1997
). This group of genera is herein referred to as the Columnea alliance. The generic delimitations of these genera and their relationships to each other are more fully resolved with ITS data (Figs. 1, 2). With the ITS data, Columnea is clearly delimited as a monophyletic group that does not include closely related genera such as Alloplectus, Drymonia, or Neomortonia rosea (Figs. 1, 2). However these latter taxa are not well defined, mainly due to the polyphyly of the Drymonia species and inclusion of Alsobia within this alliance based on parsimony analysis (Fig. 1). The discrepancy may be due to a limited sampling of species within Alloplectus and Drymonia (two each), both of which are moderate to large genera within Gesnerioideae. The sister relationship of Drymonia and Alloplectus is supported by ndhF data as well as morphological characters such as the "display fruit" found in both genera (Wiehler, 1983
). Additional support for the monophyly of Drymonia comes from morphology since this genus has a unique anther arrangement within Gesneriaceae (Wiehler, 1983
).
Previous phylogenetic work based on ndhF sequences had concluded that Neomortonia was not monophyletic (Smith and Carroll, 1997
) and that the only morphological characters that held the species together were a laterally "somewhat compressed orange berry" and absence of stolons (Wiehler, 1983
). Based on earlier ndhF sequence analysis (Smith and Carroll, 1997
) Neomortonia rosea was proposed to be a member of the Columnea alliance, if not a species of Columnea, and N. nummularia was thought to be a species of Episcia. However, the MLE of the ITS and ndhF data and both parsimony and MLE analyses of the combined ndhF and ITS sequences indicate that the two species may be more closely related than indicated previously (Figs. 2, 4, 5). Additional sequence data may yet provide evidence that this is a monophyletic genus, and therefore its exact taxonomic status cannot be resolved at this point.
Corytoplectus was an unresolved member of the Columnea alliance based on ndhF data alone (Smith and Carroll, 1997
; Fig. 3). However, with the parsimony analysis of ITS data this genus is sister to the Episcia/Nautilocalyx/Chrysothemis clade (Fig. 1) This position is unusual in that Corytoplectus has had a long taxonomic history of relationship and classification as a member of the Columnea alliance and no affinity to Episcia.
In contrast, the MLE analysis of ITS data places Corytoplectus as sister to the Columnea alliance (Fig. 2). This latter relationship is more in accordance with the traditional taxonomy for this genus as well as the ndhF data. The combined analysis excluding Rhoogeton is in agreement with the MLE analysis of the ITS data (Figs. 2, 5). Additional support for the placement of Corytoplectus in the Columnea alliance is its berry fruit, which most members of the Columnea alliance share. Therefore this genus is best viewed as a member of the Columnea alliance.
Paradrymonia
Paradrymonia was polyphyletic in previous (Smith and Carroll, 1997
) and present (Figs. 3, 4) analyses of ndhF sequences. The lack of monophyly is due to the inclusion of P. densa, which has long creeping stems, uncharacteristic for the remainder of the genus. The lack of monophyly for this clade based on ndhF sequences alone has several alternative explanations: (1) Paradrymonia densa should be reconsidered a separate genus, (2) ndhF sequences have low resolving power and are not capable of fully resolving the monophyly of Paradrymonia, or (3) greater taxon sampling is necessary since only three species have been sampled. The ITS data resolve this issue in that all three Paradrymonia species form a monophyletic group (Figs. 1, 2), implying that the low resolving power of ndhF sequences at this taxonomic level was the main problem in resolving the relationships of this genus and not limited taxon sampling.
The n = 8 clade
An unexpected result from previous ndhF analyses of Episcieae was the distant relationship of Codonanthe and Nematanthus (Smith and Carroll, 1997
). Both of these genera have centers of diversity in Brazil and Nematanthus is endemic there (Wiehler, 1983
; Chautems, 1984, 1988
). In addition these are the only members of Gesnerioideae to have a chromosome count of n = 8, and artificial intergeneric hybrids have been generated between the two (Wiehler, 1977
). The ndhF data, however, placed both taxa widely apart and implied that Codonanthopsis (n = 9) was the sister to Codonanthe (Smith and Carroll, 1997
). These results imply that chromosome numbers may have general implications for phylogenetic relationships, but for more specific relationships, they were less reliable.
In the present analysis, two different species of Nematanthus and an additional species of Codonanthe have been added. All new species were collected from the wild or from cultivated material that was originally collected from the wild. With the wild-collected species included, it is apparent that the n = 8 species form a monophyletic group based either on ndhF or ITS data regardless of analysis (Figs. 1–5). Codonanthopsis is still sister to Codonanthe and is embedded within the n = 8 clade. Further studies of chromosomal rearrangements may clarify this, but at present, it appears that Codonanthopsis represents a reversal to the n = 9 state from within the n = 8 clade.
Oerstedina
This small Central American genus was not included in previous analyses of Episcieae (Smith and Carroll, 1997
) as material was not available and attempts at extracting DNA from herbarium specimens had not been made at that time. Unfortunately ITS sequences were not attainable using the herbarium specimen DNA, although the ndhF gene was amplified and sequenced with success. Oerstedina was described by Wiehler (1977)
as similar to Rufodorsia but different by having (1) larger corollas that (2) lack red coloration on the back, and (3) pointed berries. The ndhF data unambiguously place Oerstedina as sister to Rufodorsia in both the MLE and parsimony analyses (Figs. 3, 4) with one of the more strongly supported clades in the parsimony analysis (decay of 8, bootstrap of 99; Fig. 3). Therefore the molecular data with support from morphological characters imply a close sister-group relationship between these two genera.
Capanea
In a previous cladistic study of ndhF sequences within Gloxinieae and Gesnerieae (Smith and Atkinson, 1998
) Capanea grandiflora was placed within Episcieae which was the designated outgroup for the analysis. Although this genus has traditionally been placed within Gloxinieae (Wiehler, 1983
; Burtt and Wiehler, 1995
), its placement within Episcieae was strongly supported based on ndhF sequences (Smith and Atkinson, 1998
). Its position within Episcieae seemed justified in that Capanea was the only epiphytic member of Gloxinieae and seemed more appropriately placed amid the remaining epiphytic members of Gesnerioideae in Episcieae (Smith and Atkinson, 1998
). The inclusion of C. affinis in both ITS and ndhF sequence data in this analysis does not concur with the position of Capanea in Episcieae and instead places Capanea affinis within Gloxinieae (Figs. 1–5). However, the ndhF data still place C. grandiflora within Episcieae (Figs. 3, 4). ITS sequences were not obtained for C. grandiflora despite numerous attempts via direct sequencing of PCR products. The unreadable ITS sequences imply that multiple copies of the ITS region may exist for this individual. One possible explanation is that a hybridization event may have carried the chloroplast genome and an ITS copy from a member of Episcieae into this individual of Capanea. Intergeneric hybrids are known for Gesneriaceae (Wiehler, 1968, 1977
; Worley, 1979
) but are not common in the wild. A cross between Capanea and any member of Episcieae would be unlikely due to discrepant chromosome numbers (n = 8 or 9 in Episcieae, n = 13 in Capanea), therefore the exact nature of the discrepant ndhF sequence for Capanea grandiflora cannot be resolved with the data available. The position of C. grandiflora as a species in this present analysis must be viewed with skepticism until further data can resolve the discrepancy. In contrast, the position of C. affinis in Gloxinieae is in accordance with the ITS and ndhF data presented here (Figs. 1–5) and is supported by chromosome numbers (n = 13 in Gloxinieae) and morphological data such as nectary structure and nodal anatomy (Wiehler, 1983
).
Origin of morphological characters
A previous phylogenetic analysis of Episcieae examined the origin of several morphological characters mapped onto the cladogram produced from ndhF data (Smith and Carroll, 1997
). With a more fully resolved and well-supported tree based on the combined data (Fig. 5), it is worthwhile to re-examine those characters on the present cladogram.
Fruit type
Three fruit types are found within Episcieae: (1) capsules, common to the majority of the subfamily, (2) berries, known only from Gesnerioideae outside the Episcieae in Besleria L. and the tribe Coronanthereae, and (3) the "display fruit" (Wiehler, 1983
). The berries of Besleria and Coronanthereae are structurally different, however, and are unlikely to be homologous to the berries of Episcieae. The display fruit is a fleshy dehiscent fruit where the interior of the pericarp is brightly colored, either red, pink, white, or yellow, and contrasts with the mass of seeds and funiculi it reveals when split open, which are usually black or blue-black (Wiehler, 1983
).
All outgroup genera have capsular fruits (Fig. 5), therefore the capsular fruit is assumed to the ancestral condition for Episcieae. Other molecular-based cladistic studies of tribal relationships in Gesneriaceae have indicated that the earliest lineages in the tribe either have only capsular fruits (Smith et al., 1997b
) or that the capsular fruit is ancestral to the tribe (Smith, 2000
). The only exception to this is tribe Coronanthereae which has fleshy fruits that are structurally different from all other fleshy fruits in Gesnerioideae (Wiehler, 1983
).
Earlier analyses indicated that berries evolved a minimum of four times and display fruits three times (Smith and Carroll, 1997
). In the present analysis, both of these fruit types have one less origin each due to the monophyly of Codonanthe, Codonanthopsis and Nematanthus, where there is a single origin for both fruit types (Fig. 5). The other origins are the Columnea alliance and Rufodorsia/Oerstedina for berry fruits, and Alloplectus/Drymonia for display fruits (Fig. 5). As found previously (Smith and Carroll, 1997
), the display fruit is always found in clades with berry fruits (Fig. 5). Poor resolution within the ndhF tree prevented any conclusion as to whether one fruit type preceded the other as an intermediate stage. In the present tree, the Columnea alliance clade provides evidence that the berry preceded the display fruit since the Alloplectus/Drymonia clade is embedded in an otherwise berry-fruited clade (Fig. 5). However, the Nematanthus/Codonanthe clade seems to imply that the display fruit preceded the berry since only Codonanthe has berries. However, some species of Codonanthe also have display fruits, and this trait may vary within individual plants of the same species (Wiehler, 1983
). This implies that the berry evolved within Codonanthe. Further cladistic analyses within these genera will be necessary to fully resolve this, but if true, this implies that berries and display fruits have evolved following different pathways and processes of evolution within Episcieae.
Epiphytic habit
Since the Episcieae contain the majority of epiphytic species in the subfamily Gesnerioideae it is useful to examine this character on the tree. Epiphytism here is considered a complex character, being the suite of characters that permit plants to grow on trunks and branches of other plants (Smith and Carroll, 1997
). As before, the most parsimonious explanation for the origin of epiphytism (including facultative epiphytes) is a single gain at the base of the tribe. Poor resolution in the ndhF tree did not permit an accurate estimate on the number of reversals to the terrestrial condition, which was estimated at five or six (Smith and Carroll, 1997
). The present tree clearly indicates a reversal to the terrestrial condition four times with a possible fifth reversal depending on where Rhoogeton is placed in the tree.
The ambiguous position of Rhoogeton does not allow for an accurate estimate for the origin of the terrestrial condition, however its proximity to Nautilocalyx and Chrysothemis implies that it may be a part of this clade and that a single reversal to the terrestrial condition could be hypothesized for these three genera (Fig. 5).
Tubers
Additional support for the placement of Rhoogeton with Nautilocalyx and Chrysothemis comes from examining the origin of tubers. Tubers are known from other tribes in both Gesnerioideae and Cyrtandroideae, but within Episcieae are known only from Drymonia, Nautilocalyx, Chrysothemis, Paradrymonia, and Rhoogeton. The close proximity of Rhoogeton to Nautilocalyx and Chrysothemis implies that the tuber may be viewed as a synapomorphy to unite these three genera and would indicate only three origins for the tuberous condition within Episcieae.
Summary
The results from this analysis indicate that although ndhF data alone do not strongly resolve relationships within Episcieae, these sequences are not incongruent with ITS data and when combined provide a well-resolved and well-supported phylogenetic estimate of these genera. The only discrepancy between the data sets lies in the position of Rhoogeton and the inclusion of this species in both data sets results in incongruency between the two. Rhoogeton is well supported in the Episcieae on the basis of ndhF sequences but is equally well supported in the Gloxinieae with ITS data. Therefore its phylogenetic position within the subfamily Gesnerioideae remains unresolved with these data alone. MLE and parsimony were both found to produce similar results with MLE usually finding one of several most parsimonious trees. The addition of the ITS data to the analysis provides resolution of several relationships within Episcieae. The Columnea alliance is supported and the monophyly of Columnea confirmed. The two Neomortonia species used in this analysis are more closely related to each other with ITS sequences combined with ndhF than with either data sets alone (Smith and Carroll, 1997
). Paradrymonia is most likely monophyletic, and the previous doubts on the monophyly of this genus (Smith and Carroll, 1997
) are most likely the results of the limited resolving power of ndhF at this taxonomic level. Both Codonanthe and Nematanthus are in a monophyletic group due to the inclusion of additional species of Nematanthus. This grouping makes better phylogenetic sense than previous studies that had separated the genera (Smith and Carroll, 1997
) since they are the only members of Gesnerioideae to have a chromosome count of n = 8. The present study also added the ndhF sequence of Oerstedina to the analysis that was placed as sister to Rufodorsia, a result in congruence with morphological data. Capanea affinis is best considered a member of Gloxinieae.
Morphological characters were examined on the combined data tree indicating that berry fruits evolved independently three times and display fruits evolved twice. The ancestral state for the tribe is assumed to be epiphytic with four reversals to the terrestrial condition. If Rhoogeton is sister to the Nautilocalyx/Chrysothemis clade, then tubers would have three origins within the tribe.
Future work on Episcieae will necessitate resolving the position of Rhoogeton. This will require the inclusion of additional data from both chloroplast and nuclear genes. Additionally the inclusion of other species from this genus would enhance the analysis since other species are not known to have extensive vegetative reproduction.
FOOTNOTES
1 The author thanks the following in support for this project: NSF grant DEB-9317775, a grant from the American Gloxinia and Gesneriad Society, and a grant from Boise State University Faculty Research Associates Program; L. E. Skog and J. Boggan of the Smithsonian Institution, R. Dunn, D. Turley, R. Stewart, and M. Peixoto for leaf material; James Puckett and Michael Gerdes for laboratory assistance; M. Peixoto and A. Chautems for assistance in Brazil; and Dr. W. M. M. Eddie and an anonymous reviewer for comments on the manuscript. 
2 E-mail: jsmith14{at}email.boisestate.edu
LITERATURE CITED
Avise, J. C., J. F. Shapiro, S. W. Daniel, C. F. Aquadro, and R. A. Lansman. 1983 Mitochondrial DNA differentiation during the speciation process in Peromyscus Molecular Biology and Evolution 1: 38–56.[Abstract]
Baldwin, B. G., M. J. Sanderson, J. M. Porter, M. F. Wojciechowski, C. S. Campbell, and M. J. Donoghue. 1995 The ITS region of nuclear ribosomal DNA: a valuable source of evidence of angiosperm phylogeny. Annals of the Missouri Botanical Garden 82: 247–277.[CrossRef][ISI]
Baum, D. A., R. L. Small, and J. F. Wendel. 1998 Biogeography and floral evolution of baobabs (Adansonia, Bombacaceae) as inferred from multiple data sets. Systematic Biology 47: 181–207.[CrossRef][ISI][Medline]
Bremer, K. 1988 The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution 42: 795–803.[CrossRef][ISI]
———. 1994 Branch support and tree stability. Cladistics 10: 295–304.[CrossRef][ISI]
Bull, J. J., J. P. Huelsenbeck, C. W. Cunningham, D. L. Swofford, and P. J. Waddell. 1993 Partitioning and combining data in phylogenetic analysis. Systematic Biology 42: 384–397.[CrossRef]
Burtt, B. L., and H. Wiehler. 1995 Classification of the family Gesneriaceae. Gesneriana 1: 1–4.
Chase, M. W., et al. 1993 Phylogenetics of seed plants: an analysis of nucleotide sequences from the plastid gene rbcL. Annals of the Missouri Botanical Garden 80: 528–580.[CrossRef][ISI]
Chautems, A. 1984 Révision taxonomique d'un genre de Gesneriaceae endémique du Brésil: Nematanthus Schrader. Candollea 39: 297–300.
———. 1988 Révision taxonomique et possibilités d'hybridations de Nematanthus Schrader (Gesneriaceae), genre endémique de la fôret côtière brésilienne. Thesis, Botanique 112: 1–126.
De Queiroz, A., M. J. Donoghue, and J. Kim. 1995 Separate versus combined analysis of phylogenetic evidence. Annual Review of Ecology and Systematics 26: 657–681.[ISI]
Donoghue, M. J. 1994 Progress and prospects in reconstructing plant phylogeny. Annals of the Missouri Botanical Garden 81: 405–418.[CrossRef][ISI]
———, R. G. Olmstead, J. F. Smith, and J. D. Palmer. 1992 Phylogenetic relationships of Dipsacales based on rbcL sequences. Annals of the Missouri Botanical Garden 79: 333–345.
Doyle, J. J. 1992 Gene trees and species trees: molecular systematics as one-character taxonomy. Systematic Botany 17: 144–163.[CrossRef][ISI]
Farris, S. J. 1970 Methods for computing Wagner trees. Systematic Zoology 19: 83–92.[CrossRef]
———, A. G. Kluge, and M. J. Eckardt. 1970 A numerical approach to phylogenetic systematics. Systematic Zoology 19: 172–191.[CrossRef]
Felsenstein, J. 1985 Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783–791.[CrossRef][ISI]
Graham, S. W., J. R. Kohn, B. R. Morton, J. E. Eckenwalder, and S. C. H. Barrett. 1998 Phylogenetic congruence and discordance among one morphological and three molecular data sets from Pontederiaceae. Systematic Biology 47: 545–567.[CrossRef][ISI][Medline]
Hasegawa, M., H. Kishino, and T. Yano. 1985 Dating of the human-ape splitting by a molecular clock. Journal of Molecular Evolution 22: 412–419.
Heywood, V. H. 1978 Flowering plants of the world. Prentice Hall, Englewood Cliffs, New Jersey, USA.
Hillis, D. M. 1995 Approaches for assessing phylogenetic accuracy. Systematic Biology 44: 3–16.[CrossRef]
Huelsenbeck, J. P., J. J. Bull, and C. W. Cunningham. 1996 Combining data in phylogenetic analysis. Trends in Ecology and Evolution 11: 145–186.[CrossRef]
Innis, M. A., K. B. Myambo, D. H. Gelfand, and M. A. D. Brow. 1988 DNA sequencing and direct sequencing of polymerase chain reaction-amplified DNA. Proceedings of the National Academy of Sciences, USA 85: 9436–9440.
Kress, W. J. 1986 The systematic distribution of vascular epiphytes: an update. Selbyana 9: 2–22.
Kvist, L. P., and L. E. Skog. 1993 The genus Columnea (Gesneriaceae) in Ecuador. Allertonia 6: 327–400.
Leeuwenberg, A. J. M. 1958 The Gesneriaceae of Guiana. Acta Botanica Neerlandica 7: 291–444.
Lutzoni, F. M. 1997 Phylogeny of lichen- and non-lichen-forming Omphalinoid mushrooms and the utility of testing for combinability among multiple data sets. Systematic Biology 46: 373–406.[CrossRef][ISI][Medline]
Madison, M. 1977 Vascular epiphytes: their systematic occurrence and salient features. Selbyana 2: 1–13.
Maddison, D. R. 1991 The discovery and importance of multiple islands of most-parsimonious trees. Systematic Zoology 40: 315–328.[CrossRef]
Maddison, W. P. 1995 Phylogenetic histories within and among species. In P. C. Hoch and A. G. Stevenson [eds.], Experimental and molecular approaches to plant biosystematics. Monographs in Systematics, vol. 53, 273–287. Missouri Botanical Garden, St. Louis, Missouri, USA.
———. 1997 Gene trees in species trees. Systematic Biology 46: 523–536.[CrossRef]
Mason-Gamer, R. J., and E. A. Kellogg. 1996 Testing for phylogenetic conflict among molecular data sets in tribe Triticeae. Systematic Biology 45: 524–545.[CrossRef][ISI]
Miyamoto, M. M., and W. M. Fitch. 1995 Testing species phylogenies and phylogenetic methods with congruence. Systematic Biology 44: 64–76.
Möller, M., and Q. C. B. Cronk. 1997 Origin and relationships of Saintpaulia (Gesneriaceae) based on ribosomal DNA internal transcribed spacer (ITS) sequences. American Journal of Botany 84: 956–965.[Abstract]
Munro, S. L., and H. P. Linder. 1998 The phylogenetic position of Prionium (Juncaceae) within the order Juncales based on morphological and rbcL sequence data. Systematic Botany 23: 43–56.
Olmstead, R. G., and J. D. Palmer. 1994 Chloroplast DNA systematics: a review of methods and data analysis. American Journal of Botany 81: 1205–1224.[CrossRef][ISI]
Rodman, J. E., P. S. Soltis, D. E. Soltis, K. J. Sytsma, and K. G. Karol. 1998 Parallel evolution of glucosinolate biosynthesis inferred from congruent nuclear and plastid gene phylogenies. American Journal of Botany 85: 997–1006.[Abstract]
Savolainen, V., P. Cuénod, R. Spichiger, M. D. P. Martinez, M. Crèvecoeur, and J. F. Manen. 1995 The use of herbarium specimens in DNA phylogenetics: evaluation and improvement. Plant Systematics and Evolution 197: 87–98.[CrossRef][ISI]
Seelanen, T., A. Schnabel, and J. F. Wendel. 1997 Congruence and consensus in the cotton tribe (Malvaceae) Systematic Botany 22: 259–290.[CrossRef][ISI]
Smith, J. F. 1994 Systematics of Columnea section Pentadenia and section Stygnanthe (Gesneriaceae). Systematic Botany Monographs 44: 1–89.
———. 1996 Tribal relationships within the Gesneriaceae: A cladistic analysis of morphological data. Systematic Botany 21: 497–513.[CrossRef][ISI]
———. 2000 A phylogenetic analysis of tribes Beslerieae and Napeantheae (Gesneriaceae): parsimony and maximum-likelihood analyses of ndhF sequences. Systematic Botany (in press).
———, and S. Atkinson. 1998 Phylogenetic analysis of the tribes Gloxinieae and Gesnerieae (Gesneriaceae): Data from ndhF Sequences. Selbyana 19: 122–131.
———, K. D. Brown, C. L. Carroll, and D. S. Denton. 1996 Tribal relationships within the Gesneriaceae: an assessment of three data sets. American Journal of Botany 83: 192 (Abstract).[CrossRef][ISI]
———, ———, ———, and D. S. Denton. 1997a Familial placement of Cyrtandromoea, Titanotrichum, and Sanango: three problematic genera of the Lamiales. Taxon 46: 65–74.[CrossRef][ISI]
———, and C. L. Carroll. 1997 Phylogenetic relationships of the Episcieae (Gesneriaceae) based on ndhF sequences. Systematic Botany 22: 713–724.[CrossRef][ISI]
———, M. E. Kresge, M. Møller, and Q. C. Cronk. 1998 The African violets (Saintpaulia) are members of Streptocarpus subgenus Streptocarpella (Gesneriaceae): Combined evidence from chloroplast and nuclear ribosomal genes. Edinburgh Journal of Botany 55: 1–11.
———, and K. J. Sytsma. 1994a Evolution in the Andean epiphytic genus Columnea (Gesneriaceae): part I. morphological variation. Systematic Botany 19: 220–235.[CrossRef][ISI]
———, and ———. 1994b Evolution in the Andean epiphytic genus Columnea (Gesneriaceae): part II. chloroplast DNA restriction site variation. Systematic Botany 19: 317–336.[CrossRef][ISI]
———, and ———. 1994c Molecules and morphology: congruence of data in Columnea (Gesneriaceae). Plant Systematics and Evolution 194: 37–52.
———, ———, J. S. Shoemaker, and R. L. Smith. 1992 A qualitative comparison of total cellular DNA extraction protocols. Phytochemical Bulletin 23: 2–9.
———, J. C. Wolfram, K. D. Brown, C. L. Carroll, and D. S. Denton. 1997b Tribal relationships in the Gesneriaceae: evidence from DNA sequences of the chloroplast gene ndhF. Annals of the Missouri Botanical Garden. 84: 50–66.[CrossRef][ISI]
Soltis, D. E., P. S. Soltis, M. E. Mort, M. W. Chase, V. Savolainen, S. B. Hoot, and C. M. Morton. 1998 Inferring complex phylogenies using parsimony: an empirical approach using three large DNA data sets for angiosperms. Systematic Biology 47: 32–42.[CrossRef][ISI][Medline]
Swofford, D. W., and W. P. Maddison. 1987 Reconstructing ancestral character states under Wagner parsimony. Mathematical Biosciences 87: 199–229.[CrossRef][ISI]
Wiehler, H. 1968 x Koellikohleria, a new hybrid genus (Gesneriaceae). Baileya 16: 29–34.
———. 1973 One hundred transfers from Alloplectus and Columnea (Gesneriaceae). Phytologia 27: 306–308.
———. 1977 New genera and species of Gesneriaceae from the Neotropics. Selbyana 2: 67–132.
———. 1983 A synopsis of the neotropical Gesneriaceae. Selbyana 6: 1–249.
———. 1992 New species of Gesneriaceae from the Neotropics. Phytologia 73: 220-241.
———. 1995 New species of Gesneriaceae from the Neotropics (II). Gesneriana 1: 29–97.
Worley, P. J. 1979 x Niphimenes (Niphaea x Achimenes), a new hybrid combination in the Gesneriaceae. Selbyana 5: 214–215.
This article has been cited by other articles:
|
|
 |
|
  Q. C. B. Cronk, M. Kiehn, W. L. Wagner, and J. F. Smith Evolution of Cyrtandra (Gesneriaceae) in the Pacific Ocean: the origin of a supertramp clade Am. J. Botany, June 1, 2005; 92(6): 1017 - 1024. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K.-O. Yoo, P. P. Lowry II, and J. Wen Discordance of chloroplast and nuclear ribosomal DNA data in Osmorhiza (Apiaceae) Am. J. Botany, June 1, 2002; 89(6): 966 - 971. [Abstract] [Full Text] [PDF]
|
|
|
|
