| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics |
School of Biological Sciences, Washington State University, Pullman, Washington 99164-4236 USA
Received for publication December 12, 2003. Accepted for publication April 13, 2004.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Key Words: Amorpheae • Fabaceae • floral evolution • ITS • Papilionoideae • phylogeny • trnK/matK
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Floral diversity in Amorpheae is centered on the corolla and androecium. Among species, corollar variation includes differences in number (no petals, one petal, five petals), in fusion between petals, and in differentiation between petals. Androecia also vary in number and fusion of parts. Further variation includes the presence, in some species, of a unique organ arrangement such that the petals appear perched on the staminal column.
Understanding how these forms have evolved requires understanding relationships among lineages in Amorpheae. As a natural group, Amorpheae have only been recognized since Barneby's monograph of four of the genera (Barneby, 1977
). Barneby presented an evolutionary hypothesis for the tribe based largely on characters assumed to be primitive, such as woody habit, chromosome number, and inflorescence structure. This attempt was the first and only to characterize relationships among the genera, but it is not explicitly phylogenetic. Understanding this diversification will also require evaluation of relationships among Amorpheae and other papilionoids. Inferring monophyly for Amorpheae and identifying its near relatives will provide valuable information on the ancestral Amorpheae floral form. Additionally, understanding the evolution of the various Amorpheae floral forms in the context of the papilionoid flower requires an evaluation of the position of Amorpheae in the legume family.
Several recent studies have addressed papilionoid phylogeny; some of these have included exemplar taxa from Amorpheae. Pennington et al. (2001)
used data from the chloroplast trnL (UAA) intron to study relationships among "basal" papilionoids. Two taxa from Amorpheae were included and formed a sister pair with strong support (100% bootstrap value). Unfortunately, resolution among clades was poor; the Amorpheae clade formed part of a large polytomy that included all papilionoid taxa known or inferred to have a 50-kilobase (kb) inversion in the chloroplast (Pennington et al., 2001
). A single taxon from Amorpheae, Amorpha fruticosa, was included in phylogenetic analyses using the chloroplast gene rbcL (Doyle, 1995
; Doyle et al., 1997
; Kajita et al., 2001
). The most recent and most taxon-rich analysis using this gene (Kajita et al., 2001
) places Amorpha in a large polytomy but with 0% bootstrap support. Previously published results using rbcL (Doyle et al., 1997
) indicate Amorpha is sister to a clade containing Aeschynomeneae, Dalbergieae, and Adesmieae (together comprising the dalbergioid clade), but the support for this placement was also extremely weak (<50% jackknife support). In the study of Lavin et al. (2001)
, data from the chloroplast trnK intron (including the matK gene), alone and in combination with nonmolecular data, and data from the trnL intron indicate strong support for Amorpheae as sister to the dalbergioids (bootstrap values of 92% for trnK/matK, 92% for trnK/matK + nonmolecular, and 79% for trnL). However, the focus of the study was relationships among the Dalbergieae, Adesmieae, and Aeschynomeneae. Nine papilionoid outgroup sequences were selected from among hundreds, based on their relevance to the dalbergioids, but the placement of Amorpheae was not specifically addressed. In a study of the tribe Millettieae (Hu et al., 2000), species that may be closely related to Amorpheae were sequenced for trnK/matK but no Amorpheae were included in their published results.
The goals of this study are to test for the monophyly of Amorpheae, to address the placement of Amorpheae in the phylogeny of the Papilionoideae, to infer phylogenetic relationships among Amorpheae, and to discuss floral evolution in the tribe.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
). To complement the chloroplast DNA data, we selected the internal transcribed region of the nuclear ribosomal genes, which has also been shown to be informative among species in papilionoids (Wojciechowski et al., 1993
, 1999
).
Tissue was removed from plants in the field and dehydrated with silica gel or from dried herbarium specimens. Total DNA was isolated using a standard cetyltrimethylammonium bromide (CTAB) protocol (Doyle and Doyle, 1987
). The trnK/matK region was amplified using the primers trnK3914F and psbAR or trnK2R and sequenced using at least six of the following primers: trnK3914F, matK710F, matK1470R, trnK2R, trnK290R, matK820R, matK840R, matK1221F, matK2025F, matK2340R (Appendix Table 1 for sequences and citations in Supplemental Data accompanying the online version of this article). Polymerase chain reaction (PCR) was conducted using the following profile: 4 min at 94°C, 35 cycles of 1 min at 94°C, 1 min at 44°– 48°C, and 3 min at 72°C, followed by a final extension of 5 min at 72°C. The internal transcribed spacer (ITS) region was amplified and sequenced using the primers Nnc18s10 and C26A using a "touchdown" profile: 4 min at 94°C, 5 cycles of 1 min at 94°C, 1 min at 52°C, and 2 min at 72°C, decreasing the annealing temperature by 1° each cycle, followed by 30 cycles using a 48°C annealing temperature, and ending with a final extension of 5 min at 72°C. Cycle sequencing reactions were exposed to 4 min at 94°C, 30 cycles of 1 min at 94°C, 1 min at 44°C (trnK/matK) or 50°C (ITS), and 2 min at 60°C.
Additional sequences outside of Amorpheae were acquired from GenBank and an aligned data set from M. Lavin (web site: http://gemini.ocsc.montana.edu.edu/
mlavin/data/dalbdat.htm). GenBank accession numbers for all outgroups are listed in Hu et al. (2000) or Lavin et al. (2001)
.
Taxon sampling
Within Amorpheae, taxa were chosen for inclusion in the study using several related criteria. Using taxonomy as a guide, we selected species from all eight genera (Appendix Table 2 for voucher specimen information in Supplemental Data accompanying the online version of this article) and attempted to include representatives of subgenera when possible. A second criterion was our initial estimate of the phylogeny, based on preliminary DNA sequence data (McMahon and Hufford, 2000
) and morphological data (Barneby, 1977
). We used this information to increase sampling density for those genera suspected of being non-monophyletic (e.g., Psorothamnus). We also used our preliminary analyses to attempt to span the root of subclades or break up long branches (Graybeal, 1998
). For example, if a single taxon was attached to a relatively deep node in our preliminary analyses, we attempted to include putative near-relatives of that taxon, particularly in the larger genera Dalea and Marina. A third criterion was to include all the major variations in floral morphology in the tribe. Our final analysis was conducted on 54 taxa from Amorpheae (13 of these contain only ITS sequences).
Taxa outside Amorpheae were included based on their position in published phylogenies. Amorpheae is nested in the "50 kb inversion clade" according to data from rbcL (Doyle, 1995
) and the trnL intron (Pennington et al., 2001
). We included all published trnK/matK sequences for papilionoid taxa outside the "50 kb inversion" clade and several published sequences from within the clade, concentrating on the dalbergioids (Lavin et al., 2001
) and sampling more sparsely from tribes included in the trnK/matK study of Millettieae (Hu et al., 2000). Two different sets of outgroups were used. For our analysis in which we investigated the position of Amorpheae, we used 70 outgroups (with a subset of the ingroup). For our analysis of ingroup relationships, we reduced the outgroups to four.
Phylogenetic analyses
Sequences were aligned manually using Se-Al (Rambaut, 1995
). Regions that were difficult to align (all in noncoding regions, generally involving indels of varying lengths) were excluded from the analysis and gaps were treated as missing data. To evaluate sensitivity to alignment, sequences (all ingroup and four selected outgroups) were also aligned using Clustal X (Jeanmougin et al., 1998
) with a variety of gap-opening costs (from five to 25), a gap extension cost of five, and transitions weighted equally with transversions. Parsimony searches were conducted using PAUP* (Swofford, 2000
) using 100 random addition sequence starting trees and tree bisection-reconnection (TBR) branch swapping. Bootstrap analyses (Felsenstein, 1985
) were conducted to study support for clades. Resampling was repeated 500 times; for each data set, two random addition starting trees were swapped using tree bisection-reconnection, holding a maximum of 1000 trees within each replicate. For maximum-likelihood analyses, we first compared models of evolution using Modeltest (Posada and Crandall, 1998
). We used the results of this algorithm in two ways. First, we allowed all parameters in the selected model, GTR+G+I, to be optimized; this search continued until 5000 rearrangements had been considered. Second, to allow a more thorough tree search, we used the parameter values estimated by Modeltest; this search ran to completion. For both maximum likelihood (ML) analyses, we reduced the data set to 38 taxa (from 58 for ITS and 44 for trnK), eliminating taxa that differed from another taxon by only a few autapomorphic changes.
To estimate the placement of Amorpheae in the Papilionoideae, we used 70 taxa outside and 20 taxa from within Amorpheae and conducted parsimony searches as above. We further explored the data in the following ways: First, in order to ensure that we were exploring treespace thoroughly and were not hampered by the large number of taxa, we reduced the taxa to 81 (from an original total of 90) by trimming taxa from each identified subclade. We also reduced it to 39, following the same procedure. To investigate the effect of codon, we downweighted third position sites (Swofford et al., 1996
), and to evaluate sensitivity to transition bias, we applied a step matrix that made transversions twice as costly as transitions. To each, we applied the parsimony search parameters described.
Based on the results from our papilionoid analysis, we were confident that Amorpheae are monophyletic and that the number and selection of outgroups does not affect the rooting of Amorpheae. Therefore, to improve the efficiency of our searches within Amorpheae, we reduced the outgroups to four, and we included all Amorpheae sampled, 41 for trnK/matK and 54 for ITS/5.8S. We analyzed each data set separately and in combination, and we combined data sets using two methods. First, we included only the 40 ingroup taxa for which we had both genes fully sequenced. Second, we included all 55 Amorpheae taxa for which we had ITS or trnK/matK, or both, assigning missing data where needed.
Character evolution was studied using parsimony as implemented in MacClade 4.0 (Maddison and Maddison, 2000
).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Using only the matK gene, our search resulted in 792 most parsimonious trees of length 3346 on one island (Fig. 1). Amorpheae is strongly supported to be monophyletic, with a bootstrap value of 100%. In all of the most parsimonious trees, Amorpheae is placed as sister to the recently recognized dalbergioid clade (Lavin et al., 2001
). However, bootstrap support for this placement is weak. All branches ancestral to the node uniting Amorpheae and dalbergioids are also poorly supported by the data. When we downsized the taxon set, downweighted the third positions, or applied a step matrix weighting transversions more heavily than transitions (2 : 1), our main results did not change. In no case did we recover trees that placed Amorpheae sister to any other group, and in all cases the dalbergioids and the Amorpheae were each strongly supported to be monophyletic.
|
) resulted in the same strict consensus trees. Many of the deep nodes are strongly supported by the data, as indicated by high parsimony bootstrap values. The high bootstrap values, in conjunction with the reconstructed branch lengths (Fig. 2) indicate that there are many synapomorphic changes for several clades. The monophyly of Amorpheae is again supported. The placement of the root is also strongly supported, as indicated by the bootstrap values of 100 for the clades on either side of the root. We will refer to these two clades as amorphoid and daleoid for ease of discussion. Within the amorphoids, sequence divergence provides data to resolve several clades, except for the amorphas and Errazurizia rotundata. Within the daleoids, the data strongly support the paraphyly of Psorothamnus and the inclusion of Dalea filiciformis in the Marina clade. Some further resolution within these clades is achieved, but among some species of Psorothamnus and among many species of Dalea sequence divergence is minimal, resulting in short internodes and no support for resolution within these clades.
|
|
Combining data sets and including only those taxa for which we had both gene regions sequenced, we found 62 most parsimonious trees of length 2544 on one island. When we included the 55 Amorpheae taxa and the four outgroups, assigning missing data where needed, we found 2016 trees of length 2675 on one island. The consensus tree from the smaller taxon set, i.e., only those taxa for which we have both genes sequenced, is a subtree of the consensus using the larger taxon set (Fig. 4), with one exception (Dalea wrightii is placed as sister to D. lanata + D. pogonathera; however, neither this placement nor the one obtained with all taxa is supported strongly by the data). Likelihood analysis of the smaller data set produced nearly the same tree as parsimony, but Errazurizia rotundata attaches closer to the root of the amorphoids (Fig. 4). The parsimony bootstrap values are nearly identical for trees inferred from the two taxon sets, implying that inclusion of the taxa for which we had only one gene did not impact inferences of support.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
, members of Amorpheae belonged to Daleae Hutch. (characterized by four lateral petals raised above the rim of the hypanthium) and to Psoraleeae Benth. emend. Hutch. (characterized by petals inserted on the hypanthial rim) (Hutchinson, 1964
). Barneby united the group based on the presence of several synapomorphies: epidermal glands throughout the plant body; dry, indehiscent fruits that are single-seeded; and terminal inflorescences. This last character, along with leaves that are mostly pinnate, separates the tribe from the mostly trifoliolate Psoraleeae sensu stricto, now understood to be distantly related (Barneby, 1977
; Stirton, 1981
; Turner, 1986
; Doyle et al., 1997
).
The monophyly of Amorpheae is strongly supported by the trnK/matK sequence data, confirming Barneby's hypothesis of Amorpheae as a natural group. This result is also consistent with previous molecular studies in which multiple taxa from Amorpheae were included (Hu et al., 2000; Lavin et al., 2001
; Pennington et al., 2001
). The placement of Amorpheae, however, is not as strongly supported. The data indicate a sister relationship between Amorpheae and the recently circumscribed dalbergioid clade, as was found in the dalbergioid analysis (Lavin et al., 2001
) and is being confirmed in a larger study of papilionoids (M. Wojchiechowski, Arizona State University, and M. Lavin, Montana State University, personal communication).
When we added other taxa to challenge this placement, this relationship was maintained but with poor support (Fig. 1). We tentatively accept the placement, however, because the poor support is from a weak signal, not a strongly conflicting one. Branch lengths, in terms of inferred evolutionary changes in this gene region, are very short for most internodes along the "backbone" of the phylogeny.
Other sources of data may help to place Amorpheae, although it is clear that rbcL (Doyle et al., 1997
), ITS (Lavin et al., 2001
), and trnL (Lavin et al., 2001
; Pennington et al., 2001
) do not provide sufficient resolution to place Amorpheae with confidence. Additional taxa may help for those sources of data that may be, as currently assembled, too variable to address deeper nodes, such as the 5' end of the trnK intron.
Phylogeny of Amorpheae
Our results, parsimony and likelihood analyses alike, indicate a phylogeny for Amorpheae that is consistent with some previously circumscribed taxonomic groups and with some aspects of the general distribution of floral forms (Fig. 5). On one side of the root lie the daleoids (Psorothamnus, Dalea, and Marina), genera in which the papilionoid-appearing flower is prevalent. On the other side of the root lie the amorphoids (Apoplanesia, Amorpha, Eysenhardtia, Errazurizia, and Parryella), genera in which the papilionoid floral form is not seen.
|
are each paraphyletic, whereas the monophyly of the sampled Marina is strongly supported. The two clades of Psorothamnus divide neatly along taxonomic lines. Barneby's (1977)
section Psorothamnus (Rydberg's genus Psorothamnus), based on sessile or nearly sessile flowers, non-exserted pods, and non-spiny inflorescences, is the sister of Dalea + Marina. The clade that is sister to Dalea + Marina + Psorothamnus section Psorothamnus consists of stout shrubs or trees with spiny inflorescences and/or exserted pods. This group of five species corresponds to sections Xylodalea (Rydberg's genus Psorodendron), Capnodendron, and Winnemucca; we will refer to this clade as the Psorodendron clade. These characters, as well as our DNA sequence data, indicate that Psorothamnus should be divided into two genera. A formal set of species transfers will be proposed after further taxonomic study.
The genera Marina and Dalea together form a strongly supported clade. This clade corresponds to Hutchinson's (1964)
conception of Daleeae, which was characterized by the lateral petal attachments positioned above the rim of the hypanthium. However, we have found that some Psorothamnus (McMahon and Hufford, 2002
) also have this condition, although the petals are attached much closer to the hypanthium than in Dalea or Marina. The Psorothamnus that share this trait are mostly in the clade that is sister to Dalea + Marina, with the exceptions of Psorothamnus polydenius (in the sister clade, but petals attach on the rim of the hypanthium) and Psorothamnus spinosus (in the Psorodendron clade, but the petals attach slightly above the rim of the hypanthium).
Marina is supported to be monophyletic. Dalea is also supported to be monophyletic, with the exception of Dalea filiciformis, which is placed as sister to Marina. This unique Dalea was segregated by Barneby (1977)
into the monotypic subgenus Psoropteris and was proposed to be a primitive relic, linking Dalea to Marina and Psorothamnus. This species has loose inflorescences of pedicellate flowers and a laterally compressed pod with two crescent-shaped rows of blister glands (Barneby, 1977
), closely resembling conditions in Marina. Barneby did not include D. filiciformis in Marina because it differs from Marina in several diagnostic characteristics: chromosome number, ovule number, vestiture, and leaf surface features. The latter two are easily considered as synapomorphies of Marina sensu stricto. A single ovule was considered diagnostic of marinas, but we have found several marinas that have two collateral ovules (as in the daleas), although one generally aborts prior to anthesis. Chromosome number is more difficult to accommodate because this placement of D. filiciformis would require homoplasy in chromosome numbers. Daleas have x = 8 or x = 7; all other Amorpheae have x = 10 (Barneby, 1977
). Our inferred phylogeny would require either an independent change from x = 10 to x = 8 in D. filiciformis or a return to x = 10 in the ancestor of the marina clade. Alternatively, the count may be incorrect for D. filiciformis and would be worth revisiting; the reported number is ±8 chromosome pairs, whereas other counts in the groups were decisively reported as either x = 8 or x = 7 (Barneby, 1977
).
Within Dalea, our inferred phylogeny corresponds to portions of Barneby's subgeneric classification. Subgenus Theodora is represented in our study by three species, placed together in a well-supported clade. These three, Dalea mollis, D. mollissima, and D. neomexicana, are similar low-herbaceous desert dwellers with x = 8 (Barneby, 1977
) and corollas loosely enclosing the androecium. Conversely, we found the subgenera Dalea and Parosela to be not monophyletic. Although there are many polytomies in Dalea, all strongly supported clades include members of both subgenera, despite clear morphological differences. Subgenus Dalea is characterized by free petals and/or exserted androecia, whereas subgenus Parosela is characterized by having abaxial petals fused into a conventional papilionoid keel that encloses the androecium.
The amorphoid clade
The amorphoid clade is less well resolved than the daleoid clade, for two reasons. First, among species of Amorpha, there is very little sequence divergence in either ITS or trnK/matK. This genus, currently considered to have 15 species (Wilbur, 1975
), is distributed through the American prairies into the southeastern United States and northern Baja California. Circumscription of species has been difficult, largely because there are few characters to distinguish them; many characters used by Rydberg (1919)
to segregate the group into 23 species were later found to be variable within species across ranges and across seasons (Wilbur, 1975
). Poor morphological differentiation coincides with poor molecular differentiation for the markers we have used and may indicate that Amorpha species evolved recently, although further sampling within Amorpha would be required to establish this. Alternatively, gene flow may be homogenizing the species, but this seems less likely. For example, one species pair that share identical sequences for both trnK/matK and ITS are of widely separated taxa: A. apiculata is restricted to a mountain range in northern Baja California, and its sister, A. georgiana, lives on the coastal plain of the southeastern United States. Also poorly distinguished from the amorphas is the morphologically unique Errazurizia rotundata. This plant combines the lack of petals as in Parryella (occasionally the single petal of Amorpha) with the densely glandular, low tortuous woody habit, and large glandular pod of many Psorothamnus, and the pollen of Parryella (Barneby, 1977
). Tentatively placed with the amorphas in the parsimony analysis, the single occasional banner petal may be significant. This enigmatic plant is only known from a few small populations on the Little Colorado River in northern Arizona. Investigating the possibility of a cross-generic hybrid origin for E. rotundata, as tentatively suggested by Barneby (1977)
, would be valuable.
The second reason for poor resolution among amorphoids is the conflict between trnK/matK and ITS in the placement of the monotypic genera Apoplanesia and Parryella. In the ITS tree, Apoplanesia is positioned as sister to the rest of the amorphoids, although it has many autapomorphic changes and is placed without strong support (Fig. 3). In the trnK/matK tree, Parryella is placed as sister to the other amorphoids with strong support (Fig. 2). The issue is one of rooting the amorphoids; unrooted, they are identical hypotheses, varying in resolution. Combining the data collapses all nodes near the root of the amorphoids, as expected by observing the differences between data sets. For the trnK/matK data set, Parryella is on a relatively long branch (split by the rooting of the amorphoids). For the ITS data set, Apoplanesia is on a long branch. Unfortunately, no extant candidate taxa exist that would allow us to break these longer branches. More dense taxon sampling in Amorpha would help only if it served to identify additional synapomorphies shared with Parryella, an unlikely find using these two molecular markers. Adding more eysenhardtias and the one other Errazurizia, from Chile, may help if they share synapomorphies with Apoplanesia.
The two gene regions could also have different histories with respect to either Apoplanesia or Parryella. If the ITS tree is the correct organismal phylogeny, Apoplanesia as sister to the rest of the amorphoids would require either the retention of multiple chloroplast lineages through several speciation events or a lateral transfer from the ancestor of Errazurizia sensu stricto + Eysenhardtia. Errazurizias live in northern Mexico and Chile, eysenhardtias are distributed throughout Mexico into the southwestern United States, and Apoplanesia lives in southern Mexico through Central America. These overlapping distributions imply that a chloroplast capture may have been possible, but beyond that we cannot infer. Similar arguments pertain to Parryella, as it lives near Errazurizia rotundata in northern Arizona, and amorphas occur throughout most of the United States. Therefore, we cannot rule out the hypothesis that the gene regions differ because they have different histories. However, we require additional evidence to distinguish between this case and that of simply not having enough molecular data to identify the root of the amorphoid clade.
Floral evolution
Although inferring the evolutionary history of floral forms requires detailed morphological work to establish homologies, we can draw general evolutionary conclusions from the phylogenetic distribution of floral forms. Floral diversity arose in the Amorpheae along two themes. These themes each dominate a major clade: amorphoids vary in corolla merosity and have less differentiation among corolla parts than the ancestral form; daleoids vary primarily in the presence and extent of a stemonozone (a novel corolla-androecium synorganization; McMahon and Hufford, 2002
). Flowers in Amorpha have a single petal and monadelphous stamens, fused in a sheath that is open on the adaxial side. The monotypic Parryella has no petals and a monadelphous androecium fused in an entire tube. Eysenhardtia and the monotypic Apoplanesia have five poorly differentiated petals held open, revealing the androecium. Three of four Errazurizia species have flowers similar to Eysenhardtia, but the fourth, Errazurizia rotundata (Wooton) Barneby, has no petals, although it is also reported, rarely, to have one. Most species of the three remaining genera, Marina, Dalea, and Psorothamnus, have flowers with more standard papilionaceous corollas (a banner, two wing petals, and two keel petals, marginally fused to enclose the androecium). Within this group, however, is a unique organ arrangement: the petals of Marina, Dalea, and some Psorothamnus appear perched on the staminal column.
Ancestral floral form
To discuss the evolution of floral form in Amorpheae, we must first discuss the ancestral state for the clade. To do this, we can compare the distribution of general floral form to the phylogeny obtained for Amorpheae and the large set of outgroups (Fig. 1). For this comparison, we will take a broad view and consider flowers either "papilionaceous" or "nonpapilionaceous," although this greatly simplifies papilionoid floral diversity. Nevertheless, "papilionaceous" flowers have features in common: five petals differentiated into keels, wings and banner, a hypanthium, and an androecium that is commonly connate (but not adnate to the corolla, as in the daleoid stemonozone). Using this definition, among Amorpheae, only some Psorothamnus are papilionaceous (McMahon and Hufford, 2002
), but almost all of the outgroups are. A few exceptions occur in the dalbergioids: some or all species of Riedeliella, Etaballia, and Inocarpus (Polhill, 1981
), and Pictetia (Beyra-M. and Lavin, 1999
). However, they are scattered and apparently derived within the dalbergioid clade (Fig. 1 and Lavin et al., 2001
). Some members of the tribe Sophoreae (approximately 69 of a few hundred species) and all members of Swartzieae (approximately 184 species) also possess nonpapilionaceous corollas, but these taxa are not known to be closely related to Amorpheae (Fig. 1). If we consider papilionaceousness to be a character with two states, parsimony analysis on the tree confirms that the nonpapilionaceous state arose separately in all taxa indicated. This is a simplified view of nonpapilionaceous flowers, which can be structurally quite dissimilar. However, if papilionaceousness is in some way homologous, we can infer that the ancestors of the dalbergioid clade, the genistoid clade and its sister, and the 50-kb inversion clade (Fig. 1) all had papilionaceous flowers. Therefore, we infer that the Amorpheae did as well.
Stemonozone evolution
Members of Dalea and Marina and some members of Psorothamnus share a novel corolla– androecium synorganization in which the petals are perched above the rim of the hypanthium, appearing to attach on the staminal column (Barneby, 1977
; McMahon and Hufford, 2002
). We refer to the region below the insertion point of the petals as the "stemonozone" (McMahon and Hufford, 2002
). The height of the stemonozone varies tremendously, from very short in some Psorothamnus to extensive in the prairie clover daleas (Dalea subgenus Dalea section Kuhnistera). Because our sampling is not exhaustive (all Psorothamnus, but only six out of 38 Marina and 24 of approximately 164 Dalea species were included), we cannot yet precisely map stemonozone height onto the phylogeny and fully describe its evolutionary changes. However, with the current phylogeny, we can address whether the various manifestations of stemonozone are homologous among Amorpheae.
Homology is used here to mean similarity by descent (Lankester, 1870
; Donoghue, 1992
; Hufford, 1996
), and we can infer homology if we can establish both similarity and descent. To do this, we ascertain whether a trait is consistent with structural criteria (e.g., Remane, 1952
), establishing similarity, and with phylogenetic criteria, establishing the possibility of descent from a common ancestor (e.g., Patterson, 1988
). Based on morphological data, the stemonozones in several taxa were found to be structurally similar (McMahon and Hufford, 2002
). However, mapping the stemonozone onto the phylogeny (Fig. 6), we find poor support for a single origin of the stemonozone. Dalea and Marina, in which the stemonozone is present as a structure visible to the unaided eye, share a common ancestor and therefore likely share the stemonozone by descent. The distribution of the stemonozone in Psorothamnus, however, is much more complex. There is a minute stemonozone, not visible by the unaided eye but structurally similar to the larger versions, in some members of the clade that is sister to Dalea + Marina: P. scoparius, P. thompsoneae, and P. emoryi (McMahon and Hufford, 2002
). However, it is lacking in both varieties of P. polydenius. In the second clade of Psorothamnus, all but P. spinosus lack a stemonozone. If the stemonozone is phylogenetically homologous, i.e., it was inherited in common from the daleoid ancestor, our phylogenetic hypothesis requires the additional inference of three or four losses, depending on whether P. kingii and P. spinosus are sisters or form a grade. Alternatively, P. thompsoneae may have regained the trait, and/or P. spinosus may have evolved the trait independently. The most parsimonious mapping of the character requires four steps; these mappings include a single origin of the stemonozone only if P. kingii and P. spinosus form a grade, as weakly supported by the ITS data (Fig. 3). However, the matK data and the combined analysis place P. kingii sister to P. spinosus (Fig. 4). Therefore, we tentatively conclude that the daleoid stemonozone (at least in its minute form) has arisen at least twice and has been lost up to two times (Fig. 6).
|
) was based in part on the presence of the stemonozone. It was also based on the idea that the woody species such as Apoplanesia and Eysenhardtia, with their nondifferentiated petals held open (i.e., nonpapilionaceous corollas), reflected the ancestral condition in the tribe (Barneby, 1977
). Our data do not support this hypothesis. We inferred earlier that the ancestral state for the tribe was papilionaceousness, and we also infer that the root of the daleoids was papilionaceous (Fig. 6). Therefore, the corollas in Dalea and Marina can be considered phylogenetically homologous to corollas outside of Amorpheae, and we conclude that they do not represent a reinvention of the papilionaceous corolla. This is consistent with our finding that the daleoid corolla is structurally homologous to other papilionoids (McMahon and Hufford, 2002
). The amorphoids form a clade in which the nonpapilionaceous corolla is derived and takes several forms (one petal, zero petals, five similar petals not enclosing the reproductive organs). Lack of resolution at the root of the amorphoids limits our ability to make precise inferences of the number and direction of trait changes, but it is clear that several changes occurred in this relatively small and molecularly poorly differentiated clade.
Although the daleoids are dominated by papilionaceous corollas and the presence of a stemonozone, there are exceptions. Members of Dalea subgenus Dalea, as circumscribed by Barneby (1977)
, have corollas that are not papilionaceous: a banner is present, but the four lateral petals are highly similar, unfused, and do not enclose the reproductive organs. Our data do not support this subgenus as monophyletic, and these cases are phylogenetically separate from the undifferentiated corollas in the amorphoid clade. Therefore, corolla differentiation has been lost in the Amorpheae at least three times (Fig. 5).
Conclusion
Traditional taxonomy in the tribe Amorpheae has been corroborated, in part, by our sequence data from trnK/matK and ITS/5.8S. For those cases of disagreement, there are sound reasons to accept the results from the molecular data. Errazurizia rotundata, lacking a corolla, is phylogenetically separated by our data from its corolla-bearing congeners. Although it is consistently placed with the amorphas as sister or as part of a polytomy, our data result in low support for these branches. Therefore, we cannot yet exclude the possibility that it belongs with Parryella filifolia, with which it shares floral form; indeed, the basionym for E. rotundata is Parryella rotundata Wooton, indicating its similarity. Dalea and Marina are strongly supported to be monophyletic with the exception Dalea filiciformis; this species is robustly placed by our data as sister to Marina, with which it shares many morphological features. Molecular data provide evidence for paraphyly of Psorothamnus, and division of the genus into two monophyletic genera will fall along previous taxonomic lines. We have also confirmed the monophyly of Amorpheae and further confirm its placement as sister to the dalbergioid clade. This allowed us to infer that an ancestral papilionaceous floral form gave rise to tremendous diversity in the Amorpheae. Although many morphological traits corroborate clades that our molecular data resolved, we also have strong evidence for homoplasy in other traits, leading to the general conclusion of very active floral evolution in this relatively small tribe of legumes.
|
FOOTNOTES |
|---|
2 Present address: Evolution and Ecology, University of California, Davis, CA 95616 USA (e-mail: mmmcmahon{at}ucdavis.edu
)
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Beyra M. A. M. Lavin 1999 Monograph of Pictetia (Leguminosae– Papilionoideae) and review of the Aeschynomeneae. Systematic Botany Monographs 56: 1-93
Donoghue M. J. 1992 Homology. In E. F. Keller and E. A. Lloyd [eds.], Keywords in evolutionary biology, 170–179. Harvard University Press, Cambridge, Massachusetts, USA
Doyle J. J. 1995 DNA data and legume phylogeny: a progress report. In M. Crisp and J. J. Doyle [eds.], Advances in legume systematics, part 7, 11–30. Royal Botanic Gardens, Kew, Richmond, Surrey, UK
Doyle J. J. J. L. Doyle 1987 A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin 19: 11-15
Doyle J. J. J. L. Doyle J. A. Ballenger E. E. Dickson T. Kajita H. Ohashi 1997 A phylogeny of the chloroplast gene rbcL in the Leguminosae: taxonomic correlations and insights into the evolution of nodulation. American Journal of Botany 84: 541-554[Abstract]
Felsenstein J. 1985 Confidence limits on phylogenies: an approach using the bootstrap. Evolution 38: 783-791
Graybeal A. 1998 Is it better to add taxa or characters to a difficult phylogenetic problem?. Systematic Biology 47: 9-17
Hu J.-M. M. Lavin M. Wojciechowski M. J. Sanderson 2000 Phylogenetic systematics of the tribe Millettieae (Leguminosae) based on chloroplast trnK/matK sequences, and its implications for evolutionary patterns in Papilionoideae. American Journal of Botany 87: 418-430
Hufford L. 1996 The morphology and evolution of male reproductive structures of Gnetales. International Journal of Plant Sciences 157: S95-S112[CrossRef][Web of Science]
Hutchinson J. 1964 The genera of flowering plants. Angiospermae. Clarendon Press, Oxford, UK
Jeanmougin F. J. D. Thompson M. Gouy D. G. Higgins T. J. Gibson 1998 Multiple sequence alignment with Clustal X. Trends in Biochemical Sciences 23: 403-405[CrossRef][Web of Science][Medline]
Kajita T. H. Ohashi Y. Tateishi C. D. Bailey J. J. Doyle 2001 rbcL and legume phylogeny, with particular reference to Phaseoleae, Millettieae, and allies. Systematic Botany 26: 515-536[Web of Science]
Lankester E. R. 1870 On the use of the term homology in modern zoology. Annual Magazine of Natural History, ser. 4: 34-43
Lavin M. R. T. Pennington B. B. Klitgaard J. I. Sprent H. C de Lima P. E. Gasson 2001 The dalbergioid legumes (Fabaceae): delimitation of a pantropical monophyletic clade. American Journal of Botany 88: 503-533
Maddison D. R. W. P. Maddison 2000 MacClade 4: analysis of phylogeny and character evolution, version 4.0. Sinauer, Sunderland, Massachusetts, USA
McMahon M. L. Hufford 2000 Phylogeny of Amorpheae (Fabaceae: Papilionoideae) based on DNA sequences from the chloroplast trnK intron, including the matK gene. American Journal of Botany 87: (Supplement) 143-144 (Abstract)
McMahon M. L. Hufford 2002 Morphology and structural homology of corolla-androecium synorganization in the tribe Amorpheae (Fabaceae: Papilionoideae). American Journal of Botany 89: 1884-1898
Patterson C. 1988 Homology in classical and molecular biology. Molecular Biology and Evolution 5: 603-625[Abstract]
Pennington R. T. M. Lavin H. Ireland B. Klitgaard J. Preston J.-M. Hu 2001 Phylogenetic relationships of basal papilionoid legumes based upon sequences of the chloroplast trnL intron. Systematic Botany 26: 537-556[Web of Science]
Polhill R. M. 1981 Papilionoideae. In R. M. Polhill and P. H. Raven [eds.], Advances in legume systematics, part 1, 191–204. Royal Botanic Gardens, Kew, UK
Posada D. K. A. Crandall 1998 Modeltest: testing the model of DNA substitution. Bioinformatics 14: 817-818
Rambaut A. 1995 Se-Al, Sequence alignment program, v1.d1. University of Oxford, Oxford, UK
Remane A. 1952 Die Grundlagen des naturlichen Systems der vergleichenden Anatomie und der Phylogenetik. Geest and Portig, Leipzig, Germany
Rydberg P. A. 1919 Amorpha. North American Flora 24: 26-34
Stirton C. H. 1981 Psoraleeae. In R. M. Polhill and P. H. Raven [eds.], Advances in legume systematics, part 1, 337–343. Royal Botanic Gardens, Kew, UK
Swofford D. L. 2000 PAUP*: phylogenetic analysis using parsimony (*and other methods), versions 4.0b8 and 4.0b10. Sinauer, Sunderland, Massachusetts, USA
Swofford D. L. G. J. Olsen P. J. Waddell D. M. Hillis 1996 Phylogenetic inference. In D. M. Hillis, C. Moritz, and B. K. Mable [eds.], Molecular systematics, 2nd ed., 407–514. Sinauer, Sunderland, Massachusetts, USA
Turner G. W. 1986 Comparative development of secretory cavities in the tribes Amorpheae and Psoraleeae (Leguminosae: Papilionoideae). American Journal of Botany 73: 1178-1192[CrossRef][Web of Science]
Wilbur R. L. 1975 A revision of the genus Amorpha (Leguminosae-Psoraleae). Rhodora 77: 337-409
Wojciechowski M. F. M. J. Sanderson B. G. Baldwin M. J. Donoghue 1993 Monophyly of aneuploid Astragalus (Fabaceae): evidence from nuclear ribosomal DNA-ITS sequences. American Journal of Botany 80: 711-722[CrossRef][Web of Science]
Wojciechowski M. F. M. J. Sanderson J.-M. Hu 1999 Evidence on the monophyly of Astragalus (Fabaceae) and its major subgroups based on nuclear ribosomal DNA ITS and chloroplast DNA trnL intron data. Systematic Botany 24: 409-437[CrossRef][Web of Science]
This article has been cited by other articles:
|
|
 |
|
  R. A. Scherson, R. Vidal, and M. J. Sanderson Phylogeny, biogeography, and rates of diversification of New World Astragalus (Leguminosae) with an emphasis on South American radiations Am. J. Botany, August 1, 2008; 95(8): 1030 - 1039. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. M. McMahon and M. J. Sanderson Phylogenetic Supermatrix Analysis of GenBank Sequences from 2228 Papilionoid Legumes Syst Biol, October 1, 2006; 55(5): 818 - 836. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. L. Citerne, R. T. Pennington, and Q. C. B. Cronk An apparent reversal in floral symmetry in the legume Cadia is a homeotic transformation PNAS, August 8, 2006; 103(32): 12017 - 12020. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
80% are indicated. Likelihood analysis of taxa for which both regions were sequenced resulted in a nearly identical topology. Errazurizia rotundata was placed as indicated by the curving line in the maximum likelihood (ML) tree. Note that this is consistent with the results from ITS (
