| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics and Phytogeography |
University of Missouri–St. Louis, Department of Biology, 8001, Natural Bridge Road, St. Louis, Missouri 63121-3892 USA; and Missouri Botanical Garden, Center for Conservation and Sustainable Development, P.O. Box 299, St. Louis, Missouri 63166-0299 USA
Received for publication June 24, 2005. Accepted for publication November 18, 2005.
ABSTRACT
The tribe Bignonieae (Bignoniaceae) is a large and morphologically diverse clade of neotropical lianas. Despite being a conspicuous component of the neotropical flora, the systematics of the tribe has remained uncertain due to confusing patterns of morphological variation within the group. Chloroplast (ndhF) and nuclear (PepC) DNA sequences were used here to reconstruct the phylogeny of Bignonieae. Individual analyses of ndhF and PepC were highly similar to one another, yet localized differences in the placement of six species suggests some conflict between data sets. Combined analyses result in trees that are consistent with those from the individual analyses and provide greater support for the suggested relationships. This phylogeny provides important new insights into the systematics of the tribe. It identifies 21 strongly supported species groups, eight of which broadly correspond to currently recognized genera. In addition, each of these 21 species groups is supported by morphological synapomorphies. The consistency between morphological and molecular data suggests that the current phylogeny provides a solid framework for a formal revision of the generic-level classification and for addressing other aspects of the biology of Bignonieae.
Key Words: Bignoniaceae • Bignonieae • generic taxonomy • lianas • ndhF • PepC • phylogeny
The tribe Bignonieae (Bignoniaceae) is a large and morphologically diverse clade of neotropical lianas. The tribe contains all neotropical, lianescent Bignoniaceae with septicidal capsules, as well as several shrubby species and Bignonia capreolata L., a climbing species from the southeastern United States (Lohmann, 2004
). The group is widely distributed in the neotropics, occurring in Central America, Amazonia, the Atlantic forests of eastern Brazil, and the open dry forests and savannas of Argentina, Bolivia, Brazil, and Paraguay. Bignonieae are distinguished by a distinctive wood anatomy in which the phloem is arranged in four to 32 discontinuous wedges (Fig. 1E) and by opposite, compound leaves that often have the terminal leaflet replaced by a tendril (Fig. 1H–J). They are also known for their showy flowers, which attract a wide array of pollinators: bees, wasps, butterflies, hawkmoths, hummingbirds, and bats (Gentry, 1974
; Fig. 1K–O). Their capsular fruits vary widely in size and shape and contain seeds that are usually winged and wind-dispersed, or less frequently corky and water-dispersed (Gentry, 1983b
; Fig. 1P–U).
|
, 1976
, 1979
). As currently circumscribed, Bignonieae includes nearly half of the genera (47 of 104) and more than one third of the species (360 of 860) in Bignoniaceae (Gentry, 1973b
; Fischer et al., 2004
). This diversity is not uniformly distributed across Bignonieae; 30 of the currently recognized genera have fewer than four species, while four large genera (i.e., Arrabidaea, Adenocalymma, Anemopaegma, and Memora) account for more than 250 species (Lohmann, in press). Furthermore, the genera often lack diagnostic features, present overlapping patterns of morphological variation and are difficult to identify. Complicating this issue, the most comprehensive taxonomic treatments of the tribe date from the late 19th century, with the authors recognizing between 22 (Bentham and Hooker, 1876
) and 46 (Schumann, 1894
) genera. These treatments were often based on narrow generic concepts, and commonly described new genera that were often small and not easily distinguished morphologically. For example, Pichon (1945)
described 11 new genera, most of which were segregates of Arrabidaea. Even though there has been no comprehensive treatment of the tribe since the treatment of Schumann (1894)
, there have been numerous revisions of specific genera and comprehensive regional floras (e.g., Urban, 1916
; Melquior, 1927
; Gomes, 1949
, 1956
: Gentry, 1973a
, 1982
, 1983a
; Lohmann and Pirani, 1996
; Lohmann and Hopkins, 1999
; Lohmann et al., 2002
).
Alwyn Gentry was the last to make a significant contribution to the study of Bignonieae. His work ranged from systematics to ecology, evolution, and biogeography (see Miller et al., 1996
). Rather than tying his generic concept exclusively to the characters used by previous workers (e.g., seed wing length and infloresence branching pattern), Gentry considered a wide range of morphological, anatomical, pollen, and field characteristics as generic characters. For example, "pseudostipule" shape, wood anatomy, and tendril type were suggested as potential characters for a generic-level classification by Gentry (1979)
. Even though Gentry did not produce a formal revision of the tribe, his broader generic concept resulted in a considerable reduction in the number of genera (Gentry, 1973b
, 1976
, 1979
). However, as Gentry himself noted, several of these genera either lack diagnostic features or are diagnosed by conflicting characters that make them difficult to identify in the field and in the herbarium.
Understanding the taxonomy and diversity of Bignonieae has remained difficult because of the lack of a comprehensive phylogeny for the tribe. To date, the only explicitly phylogenetic study is that of dos Santos (1995)
, who sampled 13 species for the chloroplast gene ndhF. Despite its limited scope, this study indicated that Arrabidaea, the largest genus in the tribe, was polyphyletic. For the present study, I used DNA sequences from both nuclear and chloroplast genomes. Like dos Santos (1995)
, I used the chloroplast gene ndhF, which encodes a subunit of the chloroplast NADH dehydrogenase enzyme (Sugiura, 1992
; Catalán et al., 1997
). The phylogenetic utility of this marker at the generic level has been demonstrated in several angiosperm families (e.g., Olmstead and Sweere, 1994
; Olmstead and Reeves, 1995
; Scotland et al., 1995
; Giussani et al., 2001
). I also analyzed sequences from a nuclear gene encoding phosphoenolpyruvate carboxylase (PepC), an enzyme involved primarily in the photosynthetic fixation of CO2 in C4 and CAM plants but with other housekeeping functions (Matsuoka and Minami, 1989
; Lepiniec et al., 1993
). In many lineages the enzyme appears to be encoded by a small multigene family, with individual genes approximately 7 kb in length and consisting of 10 exons and nine introns (Matsuoka and Minami, 1989
; Panstruga et al., 1995
). Although not widely used for phlyogeny reconstruction, the utility of this marker has been investigated in several studies. For example, sequences from the fourth intron of PepC have been used to investigate species relationships within Tamarix (Tamaricaceae; Gaskin and Schaal, 2002
), Gaertnera (Rubiaceae; Malcomber, 2002
) and Moringa (Moringiaceae; Olson, 2002
). In addition, Helfgott and Mason-Gamer (2004)
used two sections of PepC sequence (i.e., intron 1 and the region spanning exons 2–4) to address the origins of the North American members of polyploidy genus Elymus (Poaceae) and more generally the phylogeny of Triticeae. In each case, PepC provided phylogenetically useful levels of variation; indeed Malcomber (2002)
identified two copies in Gaertnera, both of which were useful for phylogeny reconstruction.
Here I investigate the phylogeny of Bignonieae using separate and combined analyses of ndhF and PepC DNA sequences for almost one third of the currently recognized species. The gene trees resulting from these analyses provide a comprehensive and consistent picture of the phylogeny of Bignonieae. The implications of these trees for the generic-level taxonomy of Bignonieae are discussed, both in terms of the groups recognized and the morphological characters used to define them.
MATERIALS AND METHODS
Taxon sampling
Given the size and taxonomic complexity of Bignonieae, I developed a strategy to sample the group broadly. Rather than rely on the current generic classification, I divided Bignonieae into a series of species groups based on putative morphological synapomorphies. This strategy resulted in the subdivision of many of the larger, more polymorphic genera, whereas smaller genera were often grouped together. To assess the predictive value of the morphologically defined species groups, I initially chose 70 species, representing each of the species groups and outgroups for the molecular data sets. Based on a set of preliminary molecular trees and morphological data, I predicted the phylogenetic position of an additional 52 species that were sequenced and included in the analyses. The final analyses included two or three species from each morphological group, representing morphological extremes and an intermediate from each group. In total, I sampled 46 of the 47 traditionally recognized genera (see Fischer et al., 2004
) and one third of the currently recognized species (119 of 360 species). The monotypic Perianthomega, a putative member of Bignonieae (Gentry, 1980
) was also included; only the monotypic Macrantisiphon (presently Bignonieae) and Spingiphila tetramera A.H. Gentry (currently not placed in any tribe but possibly a member of Bignonieae) were not sampled. Based on Spangler and Olmstead (1999)
and R. Olmstead et al. (University of Washington, unpublished data), two members of the tribe Tecomeae were included as outgroups, Podranea ricasoliana (Tanfani) Sprague and Tecoma capensis (Thunb.) Lindl. Voucher information, collection localities, and GenBank numbers for DNA sequences are summarized in the Appendix.
|
) or via a large-scale cesium chloride extraction protocol (Sambrook et al., 1989). For herbarium specimens, DNA was extracted using a modified CTAB protocol (S. Stefanovic, University of Washington, unpublished data), and further purified with a GeneClean III kit (BIO 101, Vista, California, USA).
PCR amplification, cloning, and sequencing
The chloroplast ndhF gene was amplified in 50 µl reactions containing 5 µl 10x reaction buffer, 5 µl 25 mM MgCl2, 2 µl dNTP (10 µM solution), 1 µl of each primer (10 µM solutions), 1.0 µl BSA, 0.5 µl 100% dimethyl sulfoxide, and 2 µl undiluted DNA template. A hot-start PCR was used, with 0.5 µl Taq polymerase (Promega, Madison, Wisconsin, USA) added after an initial 4 min at 94°C. Thermocycling conditions were: 40 cycles of 1 min at 94°C, 1 min at 48°C, and 2 min at 68°C, with a final 10 min extension at 68°C. Typically, the locus was amplified as two overlapping fragments using primer pair 5F/1318R and 972F/3R (Table 1, Fig. 2). In some cases, alternative primer pairs were used to amplify a series of smaller fragments.
|
|
, two fragments, distinguished by length (approximately 400 bp and 750 bp), were often co-amplified in Bignonieae. Similarity searches suggested that both fragments represented PepC genes. However, preliminary analyses indicated that the smaller fragment was not useful for reconstructing the phylogeny of Bignonieae; thus, only the larger fragment was further characterized. Amplification products were separated by agarose gel electrophoresis, the larger band excised, and purified using the QIAquick Gel extraction kit (Qiagen, Valencia, California, USA). Purified PCR fragments were cloned into pGEM-T Easy vectors (Promega) following the manufacturer's instructions, except that reaction volumes for ligations and transformations were halved. Plasmid preparations were screened using PCR and the M13 primers (Promega). To investigate the possibility that paralogous copies were co-amplified, PCR products from up to three positive clones were sequenced for each accession. Amplification products were purified using the QIAquick PCR purification kit (Qiagen). Automated sequencing was performed using the ABI Prism BigDye Terminator Cycle Sequencing Ready Reaction kit (Applied Biosystems, Foster City, California, USA) and extension products analyzed using an ABI Prism 377 automated DNA sequencer. Amplification primers were also used for DNA sequencing, except that the plasmid primers T7 and SP6 (Promega) were used instead of PepC-4F and PepC-5R.
Sequence alignment
Consensus sequences were assembled from forward and reverse sequences using Sequencher 3.0 (Gene Codes, Ann Arbor, Michigan, USA). Preliminary multiple sequence alignments for ndhF and PepC were obtained using the progressive alignment procedure implemented in ClustalX, with subsequent visual inspection and manual adjustment.
Congruence testing and tree statistics
Incongruence between data sets with identical taxa (104 species) was tested using the Templeton test (Templeton, 1983
) and the incongruence length difference test (ILD; Farris et al., 1994
) as implemented in PAUP*, version 4.0b10 (Swofford, 2002
). I compared the individual data sets using ILD tests. These tests used 1000 replicates with heuristic search settings that matched those of the initial maximum parsimony (MP) analyses, except that MULTREES was turned off. For Templeton tests, test and rival trees were 50% majority rule consensus trees from MP bootstrap analyses.
Phylogenetic analyses
Data sets were analyzed using MP, maximum likelihood (ML), and Bayesian approaches.
For MP searches, all uninformative characters were excluded. Preliminary MP analyses indicated that unconstrained heuristic searches would result in large numbers of trees. Therefore, MP searches were conducted using a reverse constraint approach in PAUP*, version 4.0b10. Initial searches used closest stepwise addition (retaining 10 trees after each replicate), tree bisection reconnection (TBR) branch swapping, MULTREES in effect, equal weighting of all characters, and MAXTREES set to 10000 trees. A consensus of trees from this initial search was then used as a reverse constraint (retaining trees that were not compatible with the constraint) for a search with 10000 random addition replicates (again retaining 10 trees at each step), and TBR branch swapping on. To search for trees not consistent with the constraint, MULTREES was turned off for this second search. This procedure was repeated, using the strict consensus of trees from all previous searches as the constraint, until no further MP trees were found. Trees from all searches were then summarized as a single strict consensus in order to summarize all MP trees (Catalán et al., 1997
). Non-parametric bootstrap analyses (200 replicates) used simple taxon addition, with TBR branch swapping, MULTREES in effect, and MAXTREES set to 500.
For ML searches, a best-fit model of nucleotide substitution and model parameters for the ndhF, PepC, and combined molecular data sets were determined using ModelTest 3.04 (Posada and Crandall, 1998
). In each case, the GTR + I + G was identified as the most appropriate model. Heuristic ML searches were conducted using PAUP* 4.0b10 with TBR branch swapping and randomly generated starting trees. Nonparametric bootstrap analyses used 100 replicates with single taxon addition and TBR branch swapping.
Bayesian analyses were performed with MrBayes 3.01b4 (Ronquist and Huelsenbeck, 2003
) using GTR + I + G substitution model. Searches consisted of three independent runs, each with four simultaneous chains. Default settings were used for the heating scheme (i.e., three "heated" chains, and one "cold" chain) as well as for priors on the rate matrix (0–100), branch lengths (0–10), gamma shape parameter (0–10), and the proportion of invariable sites (0–1). Base frequency parameters were modeled by a Dirichlet distribution and an uninformative prior was used for the tree topology. Each Markov chain was initiated with a random tree and run for 3000000 generations, sampling every 1000 generations. Likelihood values were monitored graphically to determine stationarity. After discarding the burn-in, trees sampled from the cold chains of all three searches were summarized as a single consensus tree. Posterior probabilities (p.p.) were used to evaluate the support of all nodes (Ronquist and Huelsenbeck, 2003
).
RESULTS
Phylogenetic analyses of ndhF
The ndhF sequences for the traditional Bignonieae (119 species), Perianthomega, and outgroup species ranged in length from 2089 to 2116 nucleotides (nt). The resulting aligned data matrix contained 2125 nucleotide positions, including 728 variable and 402 parsimony informative positions. Although gaps were inferred, all were consistent with the reading frame; the high similarity among sequences resulted in a largely unambiguous alignment. In two cases, gaps were consistent with topologies based on nucleotide substitutions. A 3-nt insertion supported a clade containing four Mansoa species and a 6-nt insertion differentiated Tecoma capensis, one of the two outgroups. The remaining gaps occur between aligned positions 1565 and 1579; several independent and overlapping events may have occurred in this one region since insertions of 6, 9, or 15 nt are inferred. Gaps were not coded as phylogenetic characters.
The initial MP search resulted in 10000 trees, each of 1366 steps with a consistency index (CI) of 0.4832 and retention index (RI) of 0.7201. A second search, which used the strict consensus of these initial trees as a reverse constraint, found only longer trees; the initial consensus tree was therefore considered the best MP topology. Parsimony bootstrap analyses provided moderate to strong support for many relationships within Bignonieae (Fig. 3). The ML search resulted in a single tree (–lnL = 15143.10714) that was consistent with that found using MP. Furthermore, ML bootstrapping provided support values similar to those from MP analyses (Fig. 3). Similarity in log-likelihood scores and p.p. supporting congruent nodes suggest that the three independent Bayesian searches had converged on similar topologies. Following exclusion of the burn-in, trees sampled from all three cold chains were combined; the consensus of these 7593 trees was similar to those identified by MP and ML. Support values for this topology are shown in Fig. 3.
|
Within Bignonieae, the PepC locus varied in length from 651 to 777 nt. The aligned matrix contained 869 nt positions; 571 varied and of these 383 were parsimony informative. The initial MP search recovered 10 000 trees, each 1517 steps long with CI = 0.4450 and RI = 0.7343. A subsequent constrained search found an additional 1130 MP trees. A strict consensus of these 11130 trees was considered the MP tree (Fig. 4). Parsimony bootstrapping provided moderate to strong support for many relationships within Bignonieae. The ML search produced a single tree with –lnL of 11349.24440. This topology was consistent with that recovered by MP and received similar levels of bootstrap support (Fig. 4). Correlation between log-likelihoods and p.p. from all three Bayesian searches suggests they had converged on similar topologies. The consensus of trees remaining after discarding the burn-in (combined sample contained 8703 trees) was very similar to that recovered by MP and ML analyses and with moderate to strong support for many relationships.
|
Congruence among individual data sets
Visual inspection of trees from analyses of the individual data sets indicates broad similarities among topologies. Numerous well-supported species groups appear in all three trees, and the relationships among these clades are also highly similar. Differences between the two gene trees were limited to the placement of individual accessions within well-supported species groups. Specifically, the positions of Memora adenophora Sandwith, Glaziovia bauhinioides Bureau ex Baill., Arrabidaea selloi (Spreng.) Sandwith, Arrabidaea revillae A.H. Gentry, Arrabidaea inaequalis (DC. ex Split.) K. Schum., and Lundia virginalis DC. differed. However, the alternative placements for these taxa were poorly supported in one or both gene trees.
Despite visual similarity among trees, the ILD tests suggest that these data sets are in conflict with one another (P = 0.001). However, to test the influence of the six highly localized differences in species placement, I repeated the tests with these taxa excluded. Removing these taxa resulted in a P = 0.133, suggesting that these six localized differences had substantial impact on the overall level of incongruence. Consistent with the ILD, Templeton tests using trees containing 104 species also suggest incongruence between the molecular data sets. However, when the 98 species trees are considered, the test cannot reject the ndhF tree as a worse representation of the PepC data (see Table 2).
|
The combined molecular data set contained 2994 characters, including 1299 variable and 785 parsimony informative positions. The first MP search recovered 10000 trees (2624 steps, CI = 0.4661, RI = 0.7318). A subsequent constrained search found only longer trees (2625 steps); therefore, the initial strict consensus was considered to be the best parsimony topology. Bootstrap analyses provided moderate to strong support for many of the clades in this topology (Fig. 5). The ML search led to a single tree with a –lnL of 25379.14169. As in the individual analyses, this topology was consistent with that found by MP; bootstrap support for many nodes was also similar (Fig. 6). Similarity in log-likelihoods and p.p. supporting congruent nodes suggests convergence among the three Bayesian searches. After discarding burn-in, samples from all three cold chains were combined; the consensus of 8733 trees was highly similar to the MP and ML topologies. Support values are shown in Fig. 6. Since outgroups were not included in this analysis; the tree was rooted based on results from the ndhF analysis.
|
|
In this study, I used two individual data sets and a combined molecular data set to investigate phylogenetic relationships in the tribe Bignonieae. Generally, these gene trees provide a consistent picture of the phylogeny of Bignonieae, with many relationships being strongly supported by bootstrapping and posterior probabilities. In the following sections I describe the individual gene trees and discuss the implications for understanding the generic classification of Bignonieae.
Gene trees
Parsimony, likelihood, and Bayesian analyses of the individual molecular data sets resulted in broadly similar topologies. With minor exceptions, these analyses identified the same 21 species groups, most of which were well supported by bootstrapping (
80%) and posterior probabilities (100%). Even though there are some differences in the relationships among species groups, these were not well supported in either analysis. Not surprisingly, trees from the combined molecular analysis were very similar to the individual trees. The 21 species groups were again recovered, often with increased support; however, relationships among clades remained weakly supported. Given the overall similarity of the molecular trees I discuss their general structure, focusing on results from the combined analysis. Unless otherwise stated, the support value ranges reported reflect the results of all analyses. For specific values the reader should refer to the relevant figure (i.e., Figs. 3, 4, or 5).
Outgroup taxa and the putative Bignonieae, Perianthomega vellozoi, were only included in the analyses of ndhF. In this tree, Perianthomega is strongly supported (94% MP bootstrap support, 93% ML bootstrap support, and 99% p.p.) as sister to the remainder of Bignonieae (Fig. 3). This analysis also identifies a basal split between two well-supported clades (support values of 95–100%). The smaller of these two groups contains the traditional genera Sampaiella, Memora, Adenocalymma, Neojobertia, and Gardnerodoxa, and is here called the SMANG clade. The second major group is much larger, containing the bulk of the species diversity in the tribe. This lineage is referred to as the Core Bignonieae. Even though outgroups were not included in PepC and combined analyses, these are also consistent with the division of Bignonieae into two major lineages.
Relationships within the SMANG clade were well resolved and supported in PepC and combined molecular analyses. These analyses suggest two distinct lineages. A clade containing Gardnerodoxa and Neojobertia (82% MP bootstrap support, and 100% p.p. in combined analyses) is sister to one including Adenocalymma, Memora, and Sampaiella (support values of 91–100%). This latter group is referred to as the volcano-gland clade due to the presence of a characteristic gland type on the calyx. Within this lineage, Memora is monophyletic (100% support in the combined analyses), whereas Sampaiella appears nested within Adenocalymma.
The Core Bignonieae contains two large and four smaller well-supported clades. Relationships among clades are often more weakly supported. One of the larger clades contains taxa characterized by stems that have phloem wedges in multiples of four in cross section, and is here called the multiples of four clade. This lineage received 100% support with MP and ML bootstrapping and 100% Bayesian p.p. in combined analyses. The second large clade contains Arrabidaea plus several related genera (99–100% support values). It is referred to as the Arrabidaea and allies clade. The four smaller clades of the Core Bignonieae are, (1) a clade containing Dolichandra, Melloa, Macfadyena, and Parabignonia that is characterized by a unique tendril type (referred to as the cat's-claw clade); (2) a monophyletic Stizophyllum; (3) a monophyletic Martinella; and, (4) a monophyletic Pleonotoma. In the combined analyses, each of these four clades received 100% support with MP and ML bootstrapping and 100% Bayesian p.p.
Molecular analyses suggest that the multiples of four clade contains four well-supported lineages; however, relationships among these sub-clades remain uncertain. In the combined analyses, a clade containing Amphilophium, Haplolophium, Glaziovia, Distictis, Pithecoctenium, and Distictella is very strongly supported (support values of 100%). This clade corresponds to the tribe Pithecoctenieae of Melquior (1927)
. Also well supported (values of 89–100%) is the mimetic clade, so-called because these species are thought to not reward pollinators and instead use mimicry or deception (Gentry, 1974
). This lineage contains the monotypic genera Bignonia, Phryganocydia, Potamoganos and Saritaea, the ditypic Roentgenia and Mussatia, one species of Tanaecium (T. nocturnum), as well as Cydista and Clytostoma. The third clade, which has support values 85–100%, includes the monophyletic Anemopaegma and Pyrostegia. The final group received 100% support as measured by bootstrapping and Bayesian p.p. and contains most representatives of Mansoa. The exception is M. kerere (Aubl.) A.H. Gentry, which was placed sister to Leucocalantha in analyses of ndhF.
The Arrabidaea and allies clade contains two major lineages. The first is strongly supported in the combined analyses (95–100% support values) and consists of the monotypic Ceratophytum, Periarrabidaea, and Pseudocatalpa, the ditypic Spathicalyx and Paragonia, one species of Tanaecium (T. crucigerum Seem.), and four species of Arrabidaea (A. affinis A.H. Gentry, A. bilabiata (Sprague) Sandwith, A. revillae, and A. selloi). This lineage is characterized by axillary buds with bud scales that resemble a bromeliad and is therefore referred to as the bromeliad clade. The second clade, support for which varied depending on type of analysis (59% MP bootstrap support, 72% MP bootstrap support and 100% p.p. in combined analyses), contains five subgroups, (1) a monophyletic Lundia; (2) a monophyletic Tynanthus; (3) a clade containing Cuspidaria, Pyrostegia cinerea Bureau ex K. Schum., and several Arrabidaea species (Cuspidaria s.l. clade); (4) a clade containing most species of Arrabidaea (including the type species A. rego) as well as the monotypic genera Piriadacus and Fridericia (true Arrabidaea clade); and, (5) a clade consisting of Xylophragma and Arrabidaea harleyi A.H. Gentry (Xylophragma s.l. clade). Each of these five lineages is well supported; 100% support values in combined analyses except the Cuspidaria s.l. clade, which had 86% and 89% bootstrap support in MP and ML tests, respectively.
Taxonomic implications
The generic-level taxonomy of Bignonieae has remained uncertain largely due to the conflicting patterns of morphological variation in the group (Gentry, 1973b
, 1976
, 1979
). Previous taxonomic treatments have varied in the selection of diagnostic characters and ultimately in the generic limits of the groups described. Indeed, even though Gentry (1976
, 1979
) considerably reduced the number of genera, he acknowledged that many of these groups lacked diagnostic characters and remained difficult to identify. The present study provides clear evidence that the current generic classification is inadequate, and provides a solid framework for addressing taxonomic problems in Bignonieae.
The phylogeny presented here provides a clear test for the monophyly of previously recognized genera. Of the 21 species groups consistently identified in all three analyses, six are consistent with traditionally recognized genera, with Anemopaegma, Lundia, Martinella, Pleonotoma, Stizophyllum, and Tynanthus all forming well-supported monophyletic groups. In addition, Pyrostegia and Mansoa are, with minor exceptions, monophyletic; with a single species falling outside of the core clade of each genus. In several other cases, traditional genera are included within a broader clade, either as a monophyletic sub-group (e.g., Memora in the volcano gland clade) or as a paraphyletic assemblage (e.g., Haplolophium in the Pithecoctenieae clade). Often these larger clades have previously been suggested on the basis of morphology—for example, the cat's-claw clade (Gentry, 1972
, 1973b
; dos Santos, 1995
), the Pithecoctenieae clade (Melquior, 1927
), and the true Arrabidaea clade (Gentry, 1972
; dos Santos, 1995
). The most obvious example of non-monophyly of a traditionally recognized genus is Arrabidaea. In this case, taxa previously placed within the genus fall into one of four species groups (e.g., the true Arrabidaea, Xylophragma s.l., Cuspidaria s.l., and bromeliad clades), each of which also includes representatives of other traditionally recognized genera. It is perhaps not surprising that phylogenetic analyses indicate non-monophyly since Arrabidaea was used as taxonomic "trash can" for taxa with uncertain affinities in the past.
Besides clarifying the monophyly of traditional genera, phylogenetic analyses also suggest several unsuspected relationships—for example, the mimetic clade, the bromeliad clade, the pairing of Mansoa kerere and Leucocalantha, and the association of Sampaiella trichochlada Bureau & K. Schum. with Adenocalymma and Memora. Even though not previously recognized, each of these groups has several potential synapomorphies. For example, taxonomic studies had suggested that Sampaiella trichochlada was closely related to Arrabidaea and Cuspidaria (Sandwith, 1968
; Silva-Castro et al., 2004
). However, consistent with the molecular phylogeny, it shares several morphological characters with Adenocalymma and Memora, including foliaceous, elliptic, and stipitate prophylls, complex glands on the calyx (Rivera, 2000
), and articulated petioles and petiolules (a unique feature in the Bignonieae). Likewise, although the Cuspidaria s.l. clade contains representatives of several previously recognized genera (i.e., Arrabidaea, Cuspidaria, and Pyrostegia), these species are united by morphological synapomorphies (e.g., pollen in tetrads and midribs bordered by a pair of ridges). In addition, there are several characters that are shared by Pyrostegia cinerea and Cuspidaria but which differ from the remainder of Pyrostegia. Specifically, Pyrostegia cinerea and Cuspidaria both have cylindrical stems, interpetiolar glands, and simple tendrils whereas other Pyrostegia species have hexagonal stems, lack such glands, and have trifid tendrils. A similar pattern can be seen for Mansoa kerere and Leucocalantha. The link between these two species is supported by at least eight characters, all of which differ from other Mansoa species (e.g., wood anatomy, stem and bark structure, and inflorescence form).
In addition to offering new insights into relationships within Bignonieae, these analyses provide a means to evaluate the morphological characters previously used to construct classification schemes for the tribe. Even though there is some overlap between traditionally recognized genera and the species groups identified here, it is clear that traditional genera do not often correspond to clades. Likewise the morphological groups I used to direct the initial sampling did not always correspond to species groups identified in phylogenetic analyses. However, based on the two-stage sampling process it was possible to refine the groups recognized and the characters used, while still being able to assess the predictive value of the general scheme. At this second stage of sampling, morphological synapomorphies for each of the species groups were identified. In each case, morphological characters uniquely define the clade (Table 3). These characters were then used to predict the positions of an additional 50 species within the molecular phylogeny. Subsequent molecular analyses confirmed the placement of all 50 taxa based on morphology, indicating that the morphological synapomorphies identified are indeed predictive of the placement of taxa.
|
, who suggested several vegetative characters as potential generic level characters. Indeed, a number of the features he noted do help to diagnose groups identified in the molecular phylogeny (e.g., axillary bud prophylls, tendril type, branchlet section, and the distribution of phloem wedges).
Future prospects
The molecular analyses presented here provide the first comprehensive view of the phylogeny of Bignonieae. Gene trees from separate and combined analyses provide a consistent picture of the evolutionary history of the tribe and identify 21 strongly supported species groups. In addition, each of these clades is distinguished by one or more putative morphological synapomorphies. Given the consistency between molecular data and morphological characters, and the apparent predictive value of the morphological synapomorphies here identified, major changes to the overall picture are not expected. Indeed, based on morphology the remaining Bignonieae species can be assigned without difficulty to one of the 21 species groups. That is, species not currently placed in the molecular phylogeny do not have novel character combinations that might be suggestive of unsampled diversity. It seems then that we can be confident of this general framework.
Although these analyses confirm some aspects of the existing generic classification (i.e., the monophyly of Anemopaegma, Lundia, Martinella, Pleonotoma, Stizophyllum, and Tynanthus), it is clear that the current generic classification needs to be revised. Indeed, the phylogenetic framework described here is the basis of a new classification scheme in which each of the 21 species groups is formally recognized at the generic level (Lohmann, in press). In addition to providing a framework for revising the taxonomy, the phylogeny also offers the opportunity to address many other aspects of Bignonieae biology (Lohmann, 2003
). In particular, studies of morphological character evolution are of great interest. Such studies allow us to better understand the patterns of morphological variation in Bignonieae, as well as provide an important model for addressing general questions about the use of morphology in phylogeny reconstruction (L. Lohmann et al., Universidade de São Paulo, unpublished data). The current framework also allows us to investigate patterns of geographical distribution and ecological diversity. These studies provide insights into the origins of broad-scale biogeographic patterns within the Bignonieae and allow us to make broader inferences about the neotropical flora as a whole. Further, we can begin to address how ecology and morphology correlate with habitat, and ultimately how this may have influenced the diversification of Bignonieae (L. Lohmann et al., Universidade de São Paulo, unpublished data). Although the current phylogeny is an important step forward, further work is needed to improve both resolution and support for relationships among species groups, and to more fully understand relationships within these clades. It will also be important to pursue comparative studies of development, ecological interactions, and distribution patterns in order to understand in more detail patterns of evolution in Bignonieae.
FOOTNOTES
The author thanks T. Kellogg, P. Stevens, P. Raven, M. Richardson, C. Taylor, B. Loiselle, S. Malcomber, R. Olmstead, R. Winkworth, F. Zapata, and an anonymous reviewer for comments on the manuscript; S. Malcomber, A. Doust, H. Cota, J. Barber, C. Engineer, and S. Kleweis for assistance in the lab; J. Gaskin for providing unpublished PepC primers; J. Lombardi, L. Galleto, R. Martins da Silva, L. Arroyo, and R. Evans for leaf material; and G. Stiehr, P. Sweeney, R. Winkworth, and M. Belstein for help with data analyses. This work was completed as part of the author's Ph.D. dissertation at the University of Missouri–St. Louis and Missouri Botanical Garden. Funding was provided by the Conselho de Auxilio à Pesquisa (CAPES/Brazilian Government), the University of Missouri–St. Louis, the Missouri Botanical Garden, a Compton Fellowship, the International Center of Tropical Ecology, the National Science Foundation (Dissertation Improvement Grant 73052), the American Society of Plant Taxonomists, the Botanical Society of America, the Federated Garden Clubs of Missouri, and Idea Wild. 
2 Author for correspondence (e-mail: llohmann{at}usp.br
; current address: Universidade de São Paulo, Instituto de Biociências, Departamento de Botânica, C.P. 11461, CEP 05422-970, São Paulo, SP, Brazil; phone: +55-11-3091-7595
LITERATURE CITED
Bentham G Hooker J. D 1876 Genera plantarum, vol. 2 L. Reeve and Co London, England
Catalán P Kellogg E. A Olmstead R. G 1997 Phylogeny of Poaceae subfamily Pooideae based on chloroplast ndhF gene sequences. Molecular Phylogenetics and Evolution 8: 150-166[CrossRef][ISI][Medline]
Doyle J. A Doyle J. L 1987 A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemistry Bulletin 19: 11-15
dos Santos G 1995 Wood anatomy, chloroplast DNA, and flavonoids of the tribe Bignonieae (Bignoniaceae). Ph.D. dissertation University of Reading, Reading, UK
Farris J. S Källersjö M Kluge A. G Bult C 1994 Testing significance of incongruence. Cladistics 10: 315-319[CrossRef][ISI]
Fischer E Theisen I Lohmann L. G 2004 Bignoniaceae. In J. W. Kadereit [ed.] The families and genera of vascular plants, vol. 7 9-38 Springer-Verlag Heidelberg, Germany
Gaskin J. F Schaal B 2002 Hybrid Tamarix widespread in U.S. invasion and undetected in native Asian range. Proceedings of the National Academy of Sciences, USA 99: 11256-11259
Gentry A. H 1972 An eco-evolutionary study of the Bignoniaceae of southern Central America. Ph.D. dissertation Washington University, St. Louis Missouri, USA
Gentry A. H 1973a Bignoniaceae. Flora of Panamá 60: 781-977
Gentry A. H 1973b Generic delimitations of Central American Bignoniaceae. Brittonia 25: 226-242[CrossRef][ISI]
Gentry A. H 1974 Coevolutionary patterns in Central American Bignoniaceae. Annals of the Missouri Botanical Garden 61: 728-759[CrossRef][ISI]
Gentry A. H 1976 Studies in Bignoniaceae. 19. Generic mergers and new species of South American Bignoniaceae. Annals of the Missouri Botanical Garden 63: 46-80[CrossRef][ISI]
Gentry A. H 1979 Additional generic mergers in Bignoniaceae. Annals of the Missouri Botanical Garden 66: 778-787[CrossRef][ISI]
Gentry A. H 1980 Bignoniaceae, part I, Tribes Crescentieae and Tourretieae. Flora Neotropica Monograph 25: 1-131
Gentry A. H 1982 Bignoniaceae. Flora de Veracruz 24: 1-222
Gentry A. H 1983a Bignoniaceae. Flora de Venezuela 8: 7-433
Gentry A. H 1983b Dispersal and distribution in Bignoniaceae. Sonderbaende des Naturwissenschaftlichen Vereins in Hamburg 7: 303-314
Gentry A. H Tomb A. S 1979 Taxonomic implications of Bignoniaceae palynology. Annals of the Missouri Botanical Garden 66: 889-895
Giussani L. M Cota-Sánchez J. H Zuloaga F. O Kellogg E. A 2001 A molecular phylogeny of the grass subfamily Panicoideae (Poaceae) shows multiple origins of C4 photosynthesis. American Journal of Botany 88: 1993-2012
Gomes J. C. Jr. 1949 Contribuição ao conhecimento das Bignoniáceae Brasileiras. I. Sampaiella J.C. Gom., nov. gen. Rodriguésia 12: 107-111
Gomes J. C. Jr. 1956 Bignoniaceae Brasilienses novae. Arquivos do Serviço Florestal 10: 199-205
Helfgott D. M Mason-Gamer R. J 2004 The evolution of North American Elymus (Triticeae, Poaceae) allotetraploids: evidence from phosphoenolpyruvate carboxylase gene sequences. Systematic Botany 29: 850-861[CrossRef][ISI]
Lepiniec L Keryer E Philippe H Gadal P Cretin C 1993 Sorghum phosphoenolpyruvate carboxylase gene family: structure, function and molecular evolution. Plant Molecular Biology 21: 487-502[CrossRef][ISI][Medline]
Lohmann L. G 2003 Phylogeny, classification, morphological diversification and biogeography of Bignonieae (Bignoniaceae, Lamiales). Ph.D. dissertation University of Missouri–St. Louis St. Louis, Missouri, USA
Lohmann L. G 2004 Bignoniaceae. In N. Smith, S. A. Mori, A. Henderson, D. Wm. Stevenson, and V. Heald [eds.] Flowering plants of the neotropics 51-53 Princeton University Press, Princeton, New Jersey, USA
Lohmann L. G In press A new generic classification of Bignonieae (Bignoniaceae) based on molecular phylogenetic data and morphological synapomorphies. Annals of the Missouri Botanical Garden
Lohmann L. G Brown J Mori S 2002 Bignoniaceae. In S. A. Mori, G. Cremers, C. A. Gracie, J. J. Granville, S. V. Heald, M. Hoff, and J. D. Mitchell [eds.] Guide to the vascular plants of central French Guiana, part 2, Dicotyledons 118-139 New York Botanical Garden Press, New York, New York, USA
Lohmann L. G Hopkins M. J. G 1999 Bignoniaceae. In J. E. L. S. Ribeiro, M. J. G. Hopkins, A. Vicentini, C.A. Sothers, M. A. da Costa, J. M. de Brito, M. A. D. de Souza, L. H. P. Martins, L. G. Lohmann, P. A. C. L. Assunção, E. da Pereira, C. F. da Silva, M. R. Mesquita, and L. C. Procópio [eds.] Flora da Reserva Ducke: guia de identificação de uma floresta de terra-firme na Amazônia Central 608-623 INPA/DIFD Manaus, Brazil
Lohmann L.G Pirani J. R 1996 Tecomeae (Bignoniaceae) da Cadeia do Espinhaço, Minas Gerais e Bahia, Brasil. Acta Botânica Brasilica 10: 103-138
Malcomber S. T 2002 Phylogeny of Gaertnera Lam. (Rubiaceae) based on multiple DNA markers: evidence of a rapid radiation in a widespread, morphologically diverse genus. Evolution 56: 42-57[ISI][Medline]
Matsuoka M Minami E.-I 1989 Complete structure of the gene for phosphoenolpyruvate carboxylase from maize. European Journal of Biochemistry 181: 593-598[ISI][Medline]
Melquior H 1927 Der natürliche Formenkreis der Pithecocteniinae innerhalb der Familie der Bignoniaceae. Repertorium Specierum Novarum Regni Vegetabilis. Beihefte 46: 71-82
Miller J. S Barkley T. M Iltis H. H Lewis W. H Forero E Plotkin M Phillips O Rueda R 1996 Alwyn Howard Gentry, 1945–1993: a tribute. Annals of the Missouri Botanical Garden 83: 433-460[ISI]
Panstruga R Seiler A Hirsch H.-J Kreuzaler F 1995 Genomic structure of a phosphoenopyruvate carboxylase gene from potato (Solanum tuberosum). Plant Physiology 109: 1126
Pichon M 1945 Notes sur les Bignoniacées. Bulletin de la Société Botanique de France 93: 121-123
Olmstead R. G Reeves P. A 1995 Evidence for the polyphyly of the Scrophulariaceae based on chloroplast rbcL and ndhF sequences. Annals of the Missouri Botanical Garden 82: 176-193[CrossRef][ISI]
Olmstead R. G Sweere J. A 1994 Combining data in phylogenetic systematics: an empirical approach using three molecular data sets in the Solanaceae. Systematic Biology 43: 467-481[CrossRef][ISI]
Olson M. E 2002 Combining data from DNA sequences and morphology for a phylogeny of Moringaceae (Brassicales). Systematic Botany 27: 55-73
Posada D Crandall K. A 1998 Modeltest: testing the model of DNA substitution. Bioinformatics 14: 817-818
Rivera G. L 2000 Nectarios extranupciales florales en especies de Bignoniaceae de Argentina. Darwiniana 38: 1-10
Ronquist F Huelsenbeck J. P 2003 MrBayes3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19: 1572-1574
Sandwith N. Y 1968 Contributions to the flora of tropical America: LXXVI. Notes on Bignoniaceae: XXIX. Arrabidaea in Martius's ‘Flora Brasiliensis' and subsequently. Kew Bulletin 22: 403-420[CrossRef]
Schumann K 1894 Bignoniaceae. In A. Engler, and K. Prantl [eds.] Die natürlichen Pflanzenfamilien, vol. 4 189-252 Engelmann Leipzig, Germany
Scotland R. W Sweere J. A Reeves P. A Olmstead R. G 1995 Higher level systematics of the Acanthaceae determined by chloroplast DNA sequences. American Journal of Botany 82: 266-275[CrossRef][ISI]
Silva-Castro M. M Giulietti A. M dos Santos F. A. R 2004 Reestablishment of the genus Sampaiella J.C. Gomes (Bignoniaceae). Revista Brasileira de Botanica 27: 781-785
Spangler R Olmstead R 1999 Phylogenetic analysis of Bignoniaceae based on the cpDNA gene sequences rbcL and ndhF. Annals of the Missouri Botanical Garden 86: 33-46[CrossRef][ISI]
Sugiura M 1992 The chloroplast genome. Plant Molecular Biology 19: 149-168[CrossRef][ISI][Medline]
Swofford D. L 2002 PAUP*: phylogenetic analysis using parsimony (*and other methods), version 4.0b10 Sinauer Sunderland, Massachusetts, USA
Templeton A. R 1983 Phylogenetic inference from restriction endonuclease cleavage site maps with particular reference to the evolution of humans and apes. Evolution 37: 221-244[CrossRef][ISI]
Urban I 1916 Über Ranken und Pollen der Bignoniaceen. Berichte der Deutschen Botanischen Gesellschaft 34: 728-758
This article has been cited by other articles:
|
|
 |
|
  R. D. C. Ortiz, E. A. Kellogg, and H. V. D. Werff Molecular phylogeny of the moonseed family (Menispermaceae): implications for morphological diversification Am. J. Botany, August 1, 2007; 94(8): 1425 - 1438. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. G. Ghebretinsae, M. Thulin, and J. C. Barber Relationships of cucumbers and melons unraveled: molecular phylogenetics of Cucumis and related genera (Benincaseae, Cucurbitaceae) Am. J. Botany, July 1, 2007; 94(7): 1256 - 1266. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Duangjai, B. Wallnofer, R. Samuel, J. Munzinger, and M. W. Chase Generic delimitation and relationships in Ebenaceae sensu lato: evidence from six plastid DNA regions Am. J. Botany, December 1, 2006; 93(12): 1808 - 1827. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
