| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Department of Botany and Institute of Cell and Molecular Biology, University of Texas, Austin, Texas 78713
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: Berberidaceae • chloroplast DNA phylogeny • parsimony • restriction sites
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The Berberidaceae comprise 17 genera (or 13 broadly defined genera), which exhibit extremely diverse morphologies. The only feature uniting the heterogeneous genera is the gynoecium of a single carpel, a condition that has been interpreted as pseudomonomerous and representing two or three fused carpels (Chapman, 1936). The flowers are predominantly trimerous but occasionally tetramerous (Epimedium), or sometimes the perianth is lacking (Achlys). Anthers mostly open by two valves or by longitudinal slits in Nandina, Podophyllum, Dysosma and Sinopodophyllum. Fruits are of various types such as berries, capsules, or bladders that rupture by the enlarging seeds. Base chromosome number is also diverse with x = 10, 8, 7, and 6.
Extensive systematic studies of floral anatomy (Terabayashi, 1985a, b), palynology (Kosenko, 1980; Nowicke and Skvarla, 1981), serology (Jensen, 1973), and karyology (Miyaji, 1930; Moore, 1963; Kuroki, 1965, 1967, 1968, 1970) have been performed to understand evolutionary relationships among the genera of the Berberidaceae. As a result, the close affinity of some generic pairs (i.e., Berberis-Mahonia, Jeffersonia-Plagiorhegma, Epimedium-Vancouveria, and Caulophyllum-Leontice) and monophyly of the Podophyllum group (i.e., Diphylleia, Dysosma, Podophyllum, and Sinopodophyllum) appears indisputable. However, there was substantial disagreement regarding relationships among these groups. Meacham (1980) recognized four major groups in the family that can be identified by fruit type and chromosome number. More recently, Loconte and Estes (1989) reanalyzed the morphological characters and proposed a new classification for the family (Table 1), which is largely congruent with Meacham's except for the position of Bongardia. The two modern classification schemes proposed for the family are radically different from the traditional systems, which are also very different from each other (Table 1).
|
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Restriction site maps of the chloroplast genome for 25 species representing 16 genera of the Berberidaceae and two species from two putatively related families (Ranunculaceae, Lardizabalaceae) were constructed (Table 2). A detailed restriction site map of Mahonia higginsae cpDNA (Kim and Jansen, 1994), which was constructed by single and double enzyme digestion, was used as a reference map to determine relative positions of the restriction sites in other cpDNAs.
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Phylogenetic analysis of the complete data set
A total of 524 variable restriction sites was mapped, 501 of which were phylogenetically informative for the 27 taxa compared (copies of the data matrix are available upon request from the first author). Wagner parsimony analysis generated 14 equally parsimonious trees with 924 steps, which had a consistency index (CI) of 0.56 (excluding uninformative characters) and retention index (RI) of 0.84. Except for the unresolved interspecific relationships in the Berberis/Mahonia clade most groups were well resolved (Fig. 1). There was strong support for a close phylogenetic relationship of several generic pairs, including Epimedium/Vancouveria, Jeffersonia/Plagiorhegma, Leontice/Gymnospermium, and Berberis/Mahonia. Among the genera for which multiple species were examined, Berberis, Mahonia, and Diphylleia were not monophyletic. The most important feature of the tree is the strong support for the monophyly of the four chromosomal groups with the base number (x) of 10, 8, 7, and 6. However, interchromosomal group relationships were either unresolved (x = 10 and 7) or very poorly supported (x = 8 and 6).
|
|
Wagner analyses
Wagner parsimony generated two equally parsimonious trees 907 steps long, which had a CI of 0.55 (excluding autapomorphies) and RI of 0.73. The only difference between the two trees is the position of the Berberis group (i.e., Berberis, Mahonia, and Ranzania) and Nandina (Nandina is basal in one tree, whereas the Berberis group is basal in the other, Fig. 3). Each of the four chromosomal groups (x = 10, 8, 7, 6) forms a distinct clade. The first (x = 10) included only Nandina. The x = 8 clade, which had high bootstrap (100%) and decay (over 5) values, consisted of the three genera Caulophyllum, Leontice, and Gymnospermium. Within this group, Caulophyllum was the sister group to the Gymnospermium and Leontice clade. The x = 7 group included the closely related woody genera Berberis and Mahonia, which formed a strong monophyletic group with their herbaceous sister genus, Ranzania. The largest chromosomal group in the family (x = 6) was also well supported (97% bootstrap value and decay index of over 5). Within this clade, four lineages were observed with all groups containing multiple genera having 100% bootstrap support: (1) Jeffersonia and Plagiorhegma, (2) Diphylleia, Podophyllum, Dysosma, and Sinopodophyllum, (3) Epimedium and Vancouveria, and (4) Achlys. Among the four lineages, Jeffersonia and Plagiorhegma formed the sister group to the other three lineages, and Diphylleia was the sister to the Podophyllum, Sinopodophyllum and Dysosma group (73% bootstrap value and decay value of 1). However, relationships among the chromosomal groups were poorly resolved (or supported) (Fig. 3).
|
Dollo analyses
Dollo parsimony generated a single tree of 1124 steps with a CI (excluding autapomorphies) of 0.44 and RI of 0.87 (Fig. 5). The most noticeable feature of the Dollo tree was the resolution and strong support for relationships among the four chromosomal groups. Nandina (x = 10) formed a monophyletic group with the x = 8 clade (80% bootstrap value and decay index of 5 or greater) and the x = 7 and 6 clades were grouped together with 79% bootstrap support and a decay value >5. Within the x = 6 clade, the Epimedium group (i.e., Epimedium and Vancouveria) was sister to the Achlys and the Podophyllum group.
A single minimum-length Dollo tree (131 steps with the CI of 0.61 and RI of 0.93) was also generated when only the IR region was analyzed. The tree (not shown) was identical to the tree based on the entire chloroplast genome except for less resolution in the Podophyllum group. In general, however, the degree of support for the relationships was lower in the IR region tree.
Weighted analyses
Weighted parsimony was performed using a wide range of weights in favor of gains over losses. In an initial analysis, with the ancestral states designated as "0" (sensu Holsinger and Jansen, 1993), the weights were gradually increased from 1.1:1 until the Dollo tree was obtained. When weights of 1.1:1 to 1.3:1 were applied, a single tree was obtained that was identical to one of the two equally parsimonious Wagner trees (i.e., the Berberis group is basal; Fig. 6a). At a weight of 1.4:1, two trees were observed, which were basically identical to the two Wagner trees except for a very minor difference in the Podophyllum group (Fig. 6b, c). With weights up to 1.9:1, one of the two weighted trees (1.4:1) was obtained (Fig. 6c). At a weight of 2.0:1 four equally parsimonious trees were generated, three of which were different from the previous trees (Fig. 6d–f). The differences were caused by the novel positions of the Berberis and Epimedium groups. The Berberis group (x = 7) was placed near the x = 6 group in two trees (Fig. 6e, f). A sister-group relationship between Achlys and the Epimedium group was observed for the first time using this weight (Fig. 6d, f). With higher weights only two additional trees were found (Fig. 6g, h) until the single stabilized Dollo tree was obtained (Fig. 5) at weights of 4.9:1 and higher. The switched position of Epimedium group and Achlys began at a weight of 4.1:1 (Fig. 6h). Among the seven novel tree topologies identified in weighted parsimony analyses, the tree in Fig. 6g was produced using the widest range of weights (2.1:1 to 4.0:1).
|
The same weighted parsimony search was performed for the IR region (trees are not shown). With ancestral states of "0," a tree identical to the single most parsimonious Wagner tree (Fig. 4) was obtained at weights up to and including 1.4:1. At a weight of 1.5:1 two additional trees were found in which the Berberis group was basal. In one of these trees, the x = 8 group, surprisingly, was not monophyletic. Only one of these three trees occurred when weights of 1.6:1 to 1.8:1 were applied. A novel tree was identified at a weight of 1.9:1 (the joining of Nandina and the Berberis group). Nandina moved in and out of the Berberis group without any other changes in the tree topology until a weight of 3.0:1 was used, which generated a topology that grouped Nandina and the Caulophyllum group. At higher weights from 3.1:1 to 3.9:1 no new tree topologies were found. The stabilized Dollo tree was obtained at weights of 4.0:1 and higher.
When the ancestral states were designated as "?," only a single tree topology was observed up to a weight of 1.9:1 (tree is not shown but it is similar to Fig. 6a). The Dollo tree was found among the five equally parsimonious trees generated with weight of 2.0:1. However, the Dollo tree did not occur with higher weights (up to 100:1).
Comparison of the trees based on three different parsimony approaches
The trees based on the three different parsimony approaches exhibited substantial congruence. The monophyly of the four chromosomal groups (x = 10, 8, 7, and 6) is supported strongly by nearly all parsimony methods. Few exceptions are found in the weighted parsimony analyses when only the IR region was used, and when extremely high and unreasonable weights (e.g., 100:1) were applied with ancestral states as "?" (not shown). Other than the four chromosomal groups, some genera, which have traditionally been considered as closely allied, were also recognized as strong monophyletic groups in all parsimony methods (i.e., Epimedium-Vancouveria, Jeffersonia-Plagiorhegma, Mahonia-Berberis, Leontice-Gymnospermium, and the Podophyllum group).
Three major differences were observed among the trees based on the three methods (see Figs. 3, 5, 6). First, the relationship within the Podophyllum group showed marked differences in all three analyses. The sister-group relationship of Podophyllum and Sinopodophyllum observed in the Wagner trees was not present in weighted trees, and the former genus groups with Dysosma in the Dollo trees. Second, the Podophyllum group formed a sister group to the Epimedium group in the Wagner trees, but to the Achlys group in the Dollo tree. In weighted trees, both relationships were observed depending on the weights used. Furthermore, at a weight of 2.0:1 the sister group relationship between Achlys and Epimedium group was also observed (Fig. 6d, f). The third and most significant difference concerns the relationships among the four chromosomal groups. Nandina,which had an uncertain position (either basal or sister group to the x = 6, 8 group; Fig. 3) in the Wagner tree, formed a monophyletic group with the Leontice clade (the x = 8 group) in the Dollo tree (Fig. 5). The x = 6 group occurred with the x = 8 clade in Wagner trees, but with the x = 7 group in Dollo trees. A high level of support for the monophyly of the x = 6 and x = 7 groups was observed in the Dollo tree (bootstrap value of 79, decay index >5). In the weighted parsimony analyses, an additional topology for the chromosomal groups was obtained (Fig. 6g). The x = 10 group was basal and the x = 8 clade was the sister to the x = 7 and 6 group. This tree is also the one that persisted through the widest range of weights (from 2.1:1 to 4.0:1). The same relationship among chromosomal groups was seen in a Wagner tree based on IR region only (Fig. 4).
Maximum likelihood analyses
To choose among the competing topologies, the likelihoods of four competing trees (Fig. 7 for simplified trees) were compared. These four trees represent the different relationships observed for the chromosomal groups in the three different parsimony analyses. The highest likelihood (Ln = -5182.93) was obtained for one of the weighted trees (Fig. 7c). The next most likely topology is the Wagner tree with the x = 10 group basal (Ln = -5200.24; Fig. 7a). This value is 17.3 units lower than the previous one. The next most likely topology is the other Wagner tree with the x = 7 group basal (Ln = -5204.90; Fig. 7b). The Dollo tree has the lowest likelihood of all topologies examined (Ln = -5208.77, Fig. 7d). The Kishino and Hasegawa (1989) statistical test indicated that the Dollo tree is significantly worse than the weighted tree.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
In the present study 15.5% of the variable sites were detected in the IR region (81 out of 524 sites). If restriction site changes occur at the same rate across all regions of chloroplast genome, one would expect ~19.2% (i.e., 25 out of 130 kb) of the change should be detected in the IR. Thus the value observed in this study is ~3.7% lower than expected. Furthermore, if the sites in the single-copy regions that were excluded because of length variation were considered, the percentage of changes in the IR region would be even lower.
Only 34.6% of the potentially informative site changes in the IR were homoplastic in Wagner analyses (Fig. 2b). The value increased to 50.7% when the single-copy regions were included (Fig. 2a). If only the single-copy regions were considered the percentages were even higher (57.8%; not shown). The CI and RI of the "IR region only" were also higher in both Wagner and Dollo analyses, indicating that, in general, the level of homoplasy is significantly lower in the IR region. This result is in contrast to a similar study in the tribe Lactuceae of the Asteraceae (Whitton, Wallace, and Jansen, 1995) in which the level of homoplasy in the IR was not necessarily lower than the single-copy regions. A possible explanation for these contrasting results is that the relationship between the level of homoplasy and the amount of divergence in cpDNA may vary at different taxonomic levels. Thus, in comparisons at higher taxonomic levels or among more divergent chloroplast genomes, homoplasy could be substantially lower in less variable regions (IR in this case).
Phylogenetic trees using only the IR region were very similar to the ones based on the entire chloroplast genome (compare Figs. 3 and 4). However, the amount of support (especially the decay value) was lower in the IR trees. This is presumably due to the smaller number of characters in IR region. Although tree topologies from the two data sets did not differ much, the lower level of homoplasy in the IR region suggests that if enough characters are generated (e.g., by using more enzymes), restriction site analysis of this region would be useful for addressing systematic questions at greater evolutionary depth (Downie and Palmer, 1992, 1994; Manos, Nixon, and Doyle, 1993).
Choosing among the competing tree topologies
Restriction site data pose special problems for phylogenetic analyses because of unequal probability of gains and losses of sites (Holsinger and Jansen, 1993; Olmstead and Palmer, 1994). Because of the much higher probability of site losses than gains (DeBry and Slade, 1985), Dollo parsimony, which prohibits parallel site gains, has sometimes been preferred over Wagner parsimony (Jansen et al., 1990). However, Albert et al. (1992) criticized the use of Dollo parsimony because it is too strict and biologically unrealistic. Instead, they suggested weighted parsimony, which places different weights on restriction site gains and losses. Holsinger and Jansen (1993) recommended that all three parsimony methods be used for restriction site data. They also recommended that the best estimate of the phylogeny be chosen by comparing likelihoods of the competing topologies.
Four major topologies were produced from the analysis of restriction site data using the three parsimony methods. These four trees differ primarily in the relationships among the four chromosomal groups (Fig. 7). Comparison of the likelihoods of the topologies revealed that the weighted tree (Figs. 6g, 7c) has the highest likelihood. It differs by 17.3 units from the next most likely topology, which is a Wagner tree. A difference of 17.3 units in log-likelihood means that for the given restriction site data, the weighted tree is 3.3 x 10 (=e) times more likely to indicate the true relationships than the Wagner tree. Similarly, the data are 1.7 x 10 (=e) times more likely given the topology in the weighted tree than the one in the Dollo tree. It is notable that the Dollo tree, which has the highest bootstrap and decay values for monophyletic groups, has the lowest likelihood value. The Kishino and Hasegawa (1989) test also indicated that the Dollo tree is significantly worse than the weighted tree, while the Wagner trees are not.
Although there are many other weighted trees, only the tree in Fig. 6g was selected for likelihood tests because all others are either identical to Wagner or Dollo trees or very similar to the one selected. Indeed, the difference in likelihoods between the one selected and another weighted tree (Fig. 6e) is only 0.86 units. This indicates that the two trees do not deviate significantly from each other compared to the differences observed in the comparisons of the major topologies. Thus, the weighted tree in Fig. 6g will be used in the discussion of character evolution of the Berberidaceae. It is noteworthy that this tree topology is very similar to the Wagner tree based on only the IR region (Fig. 4).
Chromosomal group relationships
The most important result of the restriction site analyses is the strong support of the four chromosomal groups in the Berberidaceae. This result is also congruent with trees generated in the recent studies based on nonmolecular data (Meacham, 1980; Loconte and Estes, 1989). However, relationships among the four chromosomal groups were not resolved or strongly supported in most of the trees. The only exception is the cpDNA tree reconstructed by Dollo parsimony (Fig. 5) in which two major clades of the family (x = 10, 8 and x = 7, 6) are strongly supported. However, the maximum likelihood tests indicate that the Dollo topology has the lowest likelihood. Various combinations of relationships among the chromosomal groups such as (10(7,(8,6))) or (7(10,(8,6))), (10(8,(7,6))) are also frequently observed in topologies from the three different parsimony methods. However, all other combinations (such as 6(7,(10,8)), which were observed in the rbcL tree [Kim and Jansen, 1996]) were not present. Recent molecular systematic studies based on partial sequences of the nuclear encoded gapA gene (Adachi et al., 1995) support the relationships found in the Dollo tree (Fig. 5). However, the groups in the gapA tree were also poorly supported as indicated by low bootstrap values (32–54%). Poor resolution and/or support and substantial incongruence found among the different data sets indicate that the four chromosomal lineages of the Berberidaceae may have radiated from their ancestral stock in relatively short evolutionary time.
Intergeneric relationships
The monotypic genus Nandina is the only member of the x =10 clade. The phylogenetic position of this genus has been the most controversial issue in the systematics of the Berberidaceae (Jensen, 1973; Cronquist, 1981). In most classifications the genus is either excluded from the family or positioned as a basal member (Meacham, 1980; Cronquist, 1981; Loconte and Estes, 1989; Table 1). However, the most recent sequence data from gapA (Adachi et al., 1995), rbcL (Kim and Jansen, 1996), and matK (J. Adachi et al., unpublished data) suggested that Nandina is not basal, but it is related to Caulophyllum (a member of the x = 8 group), although support for this grouping was relatively weak. The two genera also share similar vegetative branching pattern and paniculate inflorescences (Nickol, 1995). In the cpDNA trees the position of Nandina varies depending on the parsimony method employed. The genus is either basal (in one of the Wagner trees and weighted trees) or forms a monophyletic group with the x = 8 genera, or in some cases it is nested inside the family and forms a sister group to the x = 8 and 6 group (Fig. 6b). Thus, any conclusions concerning the taxonomic position of Nandina should be delayed until additional data are available. However, one can conclude with considerable confidence that Nandina does belong in the Berberidaceae and should not be given separate familial status.
The x = 8 group, corresponding to the tribe Leonticeae sensu Kosenko (1980), includes the three genera Caulophyllum, Leontice, and Gymnospermium. Loconte and Estes (1989) concluded that Leontice is more closely related to Gymnospermium because both genera have tubers, an indeterminate inflorescence, and an excalyculate flower. In this study, the monophyly of Leontice and Gymnospermium is evident, supporting the close relationship (Loconte and Estes, 1989). Bongardia, a genus that is often allied with these three genera, was not available for this analysis. However, rbcL data (Kim and Jansen, 1995, 1996) indicated that the genus forms a sister group to Achlys and Epimedium in the x = 6 group.
Restriction site data strongly support a clade consisting of Ranzania and Berberis/Mahonia. This group is also supported by several nonmolecular characters, including sensitive stamens (Kumazawa, 1937), pollen wall structure (Nowicke and Skvarla, 1981), and a base chromosome number of 7 (Loconte and Estes, 1989). This clade was recognized in recent morphological trees (Meacham, 1980; Loconte and Estes, 1989), and rbcL (Kim and Jansen, 1995, 1996) and gapA (Adachi et al., 1995) gene trees. Berberis and Mahonia comprise ~520 species that are distributed predominantly in the Northern Hemisphere and South America (Ahrendt, 1961). Thus, most berberidaceous species belong to these two closely related woody genera. A close phylogenetic relationship of these woody genera has been suggested based on several different characters (Dermen, 1931; Jensen, 1973; Terabayashi, 1978; Nowicke and Skvarla, 1981; Kim and Jansen, 1994). Monophyly of Berberis or Mahonia was not supported in the cpDNA restriction site trees. Ahrendt (1961) postulated that Berberis was derived from Mahonia by reducing leaflets to produce spines. More extensive sampling of both genera is needed to resolve generic circumscriptions. Berberis and Mahonia were often placed close to Epimedium and Vancouveria (Airy Shaw, 1973; Stearn, 1985). However, these two generic groups are nested in different chromosomal groups, indicating a remote phylogenetic relationship.
The ten herbaceous genera with base chromosome number of x = 6 form the largest generic clade in the Berberidaceae. The monophyly of this group was also proposed recently by Loconte and Estes (1989) and Kim and Jansen (1995, 1996). Two pollen characters (striate exine and discontinuous endexine; Loconte and Estes, 1989) also support this clade. Although monophyly of these genera was very weakly supported in the rbcL tree (Kim and Jansen, 1995, 1996), the amount of support in this study is very high. The basal position of Jeffersonia/Plagiorhegma within the x = 6 group is also supported by the rbcL tree, which is in conflict with the terminal position of these genera in morphological trees of Meacham (1980) and Loconte and Estes (1989). However, the plesiomorphic wood anatomy of Jeffersonia indicates that the genus diverged relatively early from other Berberidaceae (S. Carlquist, personal communication). Excluding Jeffersonia and Plagiorhegma, the rest of the genera of the x = 6 clade formed a core group, which was observed in both rbcL and ITS (internal transcribed spacer regions of nuclear ribosomal DNA) trees (Kim and Jansen, 1996). Strong coherence of the core genera was suggested only by the molecular data. Critical reexamination of previous morphological characters is required to explain this new relationship. Diphylleia formed a strong monophyletic group with Podophyllum and its allied (or congeneric) genera, Sinopodophyllum and Dysosma. Relationships among the four genera were not resolved by cpDNA restriction site data because of low levels of variation. The monophyly of the Diphylleia is not supported from this study. This is a rather surprising result because the genus is markedly different from the other three genera in anther dehiscence (longitudinal slits in Diphylleia vs. valvate in the other three genera). The inclusion of Diphylleia chinensis and additional species of Dysosma is needed to assess the monophyly of Diphylleia. The different positions of Achlys and the Epimedium group in Dollo and Wagner trees indicate the uncertain position of these genera. This is one of the major incongruences between the rbcL/ITS (Kim and Jansen, 1996) and the cpDNA restriction site trees. Achlys and the Epimedium group were monophyletic in rbcL and ITS trees. Occasional grouping of these genera in some of weighted trees indicates that restriction site data cannot unequivocally resolve relationships among the core x = 6 genera. Inclusion of Bongardia is needed for direct comparisons of the restriction site tree with the rbcL and ITS trees.
Character evolution
The independently derived cpDNA phylogeny provides opportunities to discuss evolution of other characters in the Berberidaceae. The distribution of some characters, which have been important in the systematics of the family, is plotted on the weighted tree (Fig. 8).
|
The cpDNA tree suggests that the ancestral base chromosome number for the Berberidaceae is 10, which is in agreement with Miyaji's idea that there has been reduction of chromosome number in the family. High base chromosome numbers were also reported from two outgroup genera, Hydrastis (x = 13) and Glaucidium (x = 10) of the Ranunculaceae. These genera were sometimes included in the Berberidaceae (Kumazawa, 1930; Miyaji, 1930; Dahlgren, 1980) or thought to be intermediate between the Berberidaceae and Ranunculaceae (Kumazawa, 1937; Tobe and Keating, 1985). Recent molecular studies of Ranunculales using rbcL, atpB, and 18S nuclear ribosomal DNA sequences (Hoot and Crane, 1995) suggested that these two genera are basal in the Ranunculaceae. Thus, if the Berberidaceae and Ranunculaceae arose from common ancestral stock as has been suggested by molecular data (Hoot and Crane, 1995; Kim and Jansen, 1995, 1996), a base chromosome number of x = 10 is more parsimonious for the Berberidaceae. Critical evaluation of the chromosomal characters, including the examination of more genera, should provide valuable insights into the evolution and phylogenetic relationships among the four chromosomal groups.
Another important character in the Berberidaceae is the number of fused carpels, which was examined in detail by Chapman (1936). Genera in the family have a gynoecium of either two or three fused carpels, whereas Nandina has both conditions. This character was often rejected (Cronquist, 1981) or ignored (Meacham, 1980; Loconte and Estes, 1989) in systematic comparisons of the family. However, the presence of a gynoecium of two fused carpels provides additional support for the monophyly of the x = 6 group, if the polymorphism reported in Nandina is interpreted as the ancestral condition.
Several features of the flowers, fruits, and seeds have evolved more than two times (Fig. 8). For example, anthers dehisce by longitudinal slits in Nandina and the Podophyllum group (except for Diphylleia), genera that are very remotely positioned in different clades in cpDNA tree. However, critical examination of this character indicates that longitudinal slits may have originated several times in the family (Kumazawa, 1937; M. Nickol, personal communication). The different fruit types may have also originated multiple times. Fruit type has often been considered one of the more important taxonomic characters in the Berberidaceae (Meacham, 1980). Indeed, Bongardia has been allied to Leontice in most classification systems because they share bladder-like fruits (as well as other vegetative characters). However, rbcL and ITS sequence data (Kim and Jansen, 1996) suggested that these genera are remotely related, and that Bongardia is more closely allied to Epimedium/Achlys, which have capsules dehiscing by a single slit. Finally, the woody habit and aril of the seed have also evolved more than two times. The different origins of the woody habit were already discussed by Shen (1954) based on comparative wood anatomy.
|
FOOTNOTES |
|---|
2 Author for correspondence, current address: Department of Biology, Hallym University, Chuncheon 200-702, South Korea.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Airy Shaw, H. K.1973J. C. Willis' a dictionary of the flowering plants and ferns, 8th ed. Cambridge University Press, Cambridge.
Ahrendt, L. W. A.1961Berberis and Mahonia: a taxonomic revision. Journal of Linnaean Society of London, Botany 57: 1–410.
Albert, V. A., B. D. Mishler, and M. W. Chase.1992Character-state weighting for restriction site data in phylogenetic reconstruction, with an example from chloroplast DNA. In P. S. Soltis, D. E. Soltis, and J. J. Doyle [eds.], Molecular systematics of plants, 369–403. Chapman and Hall, New York, NY.
Bremer, B., and R. K. Jansen.1991Comparative restriction site mapping of chloroplast DNA implies new phylogenetic relationships within Rubiaceae. American Journal of Botany 78: 198–213. [CrossRef][ISI]
Bremer, K.1988The limits of amino acid sequence data in angiosperm phylogenetic reconstruction. Evolution 42: 795–803. [CrossRef][ISI]
Chapman, M.1936Carpel morphology of the Berberidaceae. American Journal of Botany 23: 340–348. [CrossRef][ISI]
Cronquist, A.1981An integrated system of classification of flowering plants. Cambridge University Press, Cambridge.
Dahlgren, R. M. T.1980A revised system of classification of the angiosperms. Journal of Linnaean Society of London, Botany 80: 91–124.
DeBry, R. W., and N. A. Slade.1985Cladistic analysis of restriction endonuclease cleavage maps within a maximum-likelihood framework. Systematic Zoology 34: 21–34. [CrossRef]
Dermen, H.1931A study of chromosome number in two genera of Berberidaceae: Mahonia and Berberis. Journal of Arnold Arboretum 12: 281–287.
Donoghue, M. J., R. G. Olmstead, J. F. Smith, and J. D. Palmer.1992Phylogenetic relationships of Dipsacales based on rbcL sequences. Annals of the Missouri Botanical Garden 79: 333–345.
Downie, S. R., and J. D. Palmer.1992Restriction site mapping of the chloroplast DNA inverted repeat: a molecular phylogeny of the Asteridae. Annals of the Missouri Botanical Garden 79: 266–283. [CrossRef][ISI]
———, and ———.1994A chloroplast DNA phylogeny of the Caryophyllales based on structural and inverted repeat restriction site variation. Systematic Botany 19: 236–252. [CrossRef][ISI]
Doyle, J. J., and J. L. Doyle.1987A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemical Bulletin 19: 11–15.
Farris, J. S.1970Methods for computing Wagner trees. Systematic Zoology 19: 83–92. [CrossRef]
Felsenstein, J.1985Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783–791. [CrossRef][ISI]
———1991Phylogeny inference package (PHYLIP). version 3.4. University of Washington, Seattle, WA.
Holsinger, K. E., and R. K. Jansen.1993Phylogenetic analysis of restriction site data. Methods in Enzymology 224: 439–455. [ISI][Medline]
Hoot, S. B.1995Phylogeny of the Ranunculaceae based on preliminary atpB, rbcL and 18S nuclear ribosomal DNA sequence data. Plant Systematics and Evolution (Supplement) 9: 241–251.
———, and P. R. Crane.1995Inter-familial relationships in the Ranunculidae based on molecular systematics. Plant Systematics and Evolution (Supplement) 9: 119–131.
Jansen, R. K., K. E. Holsinger, H. J. Michaels, and J. D. Palmer.1990Phylogenetic analysis of chloroplast DNA restriction data at higher taxonomic levels: an example from the Asteraceae. Evolution 44: 2089–2099. [CrossRef][ISI]
———, and J. D. Palmer.1987Chloroplast DNA from lettuce and Barnadesia (Asteraceae): structure, gene localization, and characterization of a large inversion. Current Genetics 11: 553–564.
———, and ———.1988Phylogenetic implications of chloroplast DNA restriction site variation in the Mutisieae (Asteraceae). American Journal of Botany 75: 753–766. [CrossRef][ISI]
Jensen, U.1973The interpretation of comparative serological results. In G. Bendz and J. Santesson [eds.], Nobel symposium 25: chemistry in botanical classification, 217–227. Academic Press, New York, NY.
Johansson, J. T., and R. K. Jansen.1993Chloroplast DNA variation and phylogeny of the Ranunculaceae. Plant Systematics and Evolution 187: 29–49. [CrossRef][ISI]
Kawano, S., and M. Ihara.1967Chromosome morphology of Caulophyllum robustum (Podophyllaceae) and its systematic implications. Journal of Japanese Botany 42: 1–7.
Kim, K.-J., B. L. Turner, and R. K. Jansen.1992Phylogenetic and evolutionary implications of interspecific chloroplast DNA variation in Krigia (Asteraceae-Lactuceae). Systematic Botany 17:449–469.
Kim, Y.-D. and R. K. Jansen.1994Characterization and phylogenetic distribution of a structural change in chloroplast DNAs of Berberidaceae. Plant Systematics and Evolution 193: 107–114. [CrossRef][ISI]
———, and ———.1995Phylogenetic implications of the chloroplast DNA variation from the Berberidaceae. Plant Systematics and Evolution (Supplement) 9: 341–349.
———, and ———.1996Phylogenetic implications of rbcL and ITS sequence variation in the Berberidaceae. Systematic Botany 21: 381–396. [CrossRef][ISI]
Kishino, H., and M. Hasegawa.1989Evaluation of the maximum likelihood estimate of the evolutionary tree topologies from DNA sequence data, and the branching order in Hominoidea. Journal of Molecular Evolution 29: 170–179. [CrossRef][ISI][Medline]
Kosenko, V. N.1980Comparative palynomorphological study of the family Berberidaceae. 2. Morphology of the pollen grains of the genera Gymnospermium, Leontice, Caulophyllum, Bongardia, Epimedium, Vancouveria, Achlys, and Jeffersonia. Botanicheskii zhurnal (Moscow & Leningrad) 65: 1412–1423.
Kumazawa, M.1930Morphology and biology of Glaucidium palmatum Sib. et Zucc. with notes of affinities to the allied genera Hydrastis, Podophyllum, and Diphylleia. Journal of the Faculty of Science, University of Tokyo, Section III, Botany 2: 345–380.
———.1937Ranzania japonica, its morphology, biology and systematic affinities. Japanese Journal of Botany 9: 55–70.
Kuroki, Y.1965Chromosome study in three species of Berberidaceae. Memoirs of Ehime University, Section II (Science), Series B. (Biology) 5: 19–24.
———.1967Chromosome study in seven species of Berberidaceae. Memoirs of Ehime University, Section II (Science), Series B. (Biology) 5: 175–181.
———.1968Chromosome study in five species of Berberidaceae. Memoirs of Ehime University, Section II (Science), Series B. (Biology) 6: 11–16.
———.1970Chromosome study in four species of Berberidaceae. Memoirs of Ehime University, Section II (Science), Series B. (Biology) 6: 63–69.
Loconte, H., and J. R. Estes.1989Phylogenetic systematics of Berberidaceae and Ranunculales (Magnoliidae). Systematic Botany 14: 565–579. [CrossRef][ISI]
Mabberley, D. J.1990The plant-book. Cambridge University Press, Cambridge.
Maddison, D. R.1991The discovery and importance of multiple islands of most parsimonious trees. Systematic Zoology 40: 315–328. [CrossRef]
Manos, P. S., K. C. Nixon, and J. J. Doyle.1993Cladistic analysis of restriction site variation within the chloroplast DNA inverted repeat region of selected Hamamelididae. Systematic Botany 18: 551–562. [CrossRef][ISI]
Meacham, C. A.1980Phylogeny of the Berberidaceae with an evaluation of classifications. Systematic Botany 5: 149–172.
Miyaji, Y.1930Beitrage zur Chromosomenphylogenie der Berberidaceen. Planta 11: 650–659. [CrossRef]
Moore, R. J.1963Karyotype evolution in Caulophyllum. Canadian Journal of Genetics and Cytology 5: 384–388.
Nickol, M. G.1995Phylogeny and inflorescence of Berberidaceae—a morphological survey. Plant Systematics and Evolution [Suppl.] 9: 327–340.
Nowicke, J. W., and J. J. Skvarla.1981Pollen morphology and phylogenetic relationships of the Berberidaceae. Smithsonian Contributions to Botany 50: 1–30.
Olmstead, R. G., H. J. Michaels, K. M. Scott, and J. D. Palmer.1992Monophyly of the Asteridae and identification of their major lineages inferred from DNA sequences of rbcL. Annals of the Missouri Botanical Garden 79: 249–265.
———, and J. D. Palmer.1992A chloroplast DNA phylogeny of the Solanaceae: subfamilial relationships and character evolution. Annals of the Missouri Botanical Garden 79: 346–360. [CrossRef][ISI]
———, and ———.1994Chloroplast DNA systematics: a review of methods and data analysis. American Journal of Botany 81: 1205–1224. [CrossRef][ISI]
Palmer, J. D.1986Isolation and structural analysis of chloroplast DNA. Methods in Enzymology 118: 167–186. [ISI]
Ranker, T. A., D. E. Soltis, P. S. Soltis, and A. J. Gilmartin.1990Subfamilial phylogenetic relationships of the Bromeliaceae: evidence from chloroplast DNA restriction site variation. Systematic Botany 15: 425–434. [CrossRef][ISI]
Shen, Y.1954Phylogeny and wood anatomy of Nandina. Taiwania 5: 85–92.
Soltis, D. E., P. S. Soltis, T. G. Collier, and M. L. Edgerton.1991Chloroplast DNA variation within and among genera of the Heuchera group (Saxifragaceae): evidence for chloroplast transfer and paraphyly. American Journal of Botany 78: 1091–1112. [CrossRef][ISI]
———,———, and B. G. Milligan.1992Intraspecific chloroplast DNA variation: Systematic and phylogenetic implications. In P. S. Soltis, D. E. Soltis, and J. J. Doyle [eds.], Molecular systematics of plants, 117–150. Chapman and Hall, New York, NY.
———, D. R. Morgan, A. Grable, P. S. Soltis, and R. Kuzoff.1993Molecular systematics of Saxifragaceae sensu stricto. American Journal of Botany 80: 1056–1081. [CrossRef][ISI]
Soreng, R. J., J. I. Davis, and J. J. Doyle.1990A phylogenetic analysis of chloroplast DNA restriction site variation in Poaceae subfam. Pooideae. Plant Systematics and Evolution 172:83–97.
Stearn, W. T.1985Berberidaceae. In V. H. Heywood [ed.], Flowering plants of the world, 45–46. Croom Helm, London.
Sugiura, M., K. Shinozaki, N. Zaita, M. Kusuda, and M. Kumano.1986Clone bank of the tobacco (Nicotiana tabacum) chloroplast genome as a set of overlapping restriction endonuclease fragments: mapping of eleven ribosomal protein genes. Plant Science 44: 211–216. [CrossRef][ISI]
Swofford, D. L.1993PAUP: phylogenetic analysis using parsimony, version 3.1.1. Smithsonian Institution, Washington, DC.
Sytsma, K. J., and J. F. Smith.1992Molecular systematics of Onagraceae: examples from Clarkia and Fuchsia. In P. S. Soltis, D. E. Soltis, and J. J. Doyle [eds.], Molecular systematics of plants, 295–323. Chapman Hall, New York, NY.
Terabayashi, S.1978Studies in morphology and systematics of Berberidaceae. II. Floral anatomy of Mahonia japonica (Thunb.) DC. and Berberis thunbergii DC. Acta Phytotaxonomica Geobotanica 29: 106–118.
———.1985aThe comparative floral anatomy and systematics of the Berberidaceae. I. Morphology. Memoirs of the Faculty of Science, Kyoto University, Series of Biology 10: 73–90.
———.1985bThe comparative floral anatomy and systematics of the Berberidaceae. II. Systematic considerations. Acta Phytotaxonomica Geobotanica 36: 1–13.
Tobe, H., and R. C. Keating.1985The morphology and anatomy of Hydrastis (Ranunculaceae): systematic reevaluation of the genus. The Botanical Magazine, Tokyo 98: 291–316. [CrossRef]
Whitton, J., R. S. WAllace, and R. K. Jansen.1995Phylogenetic relationships and patterns of character change in the tribe Lactuceae (Asteraceae) based on chloroplast DNA restriction site variation. Canadian Journal of Botany 73: 1058–1073.
Wolfe, K. H., W.-H. Li, And P. M. Sharp.1987Rates of nucleotide substitution vary greatly among mitochondrial, chloroplast, and nuclear DNAs. Proceedings of the National Academy of Sciences, USA 84: 9054–9058.
This article has been cited by other articles:
|
|
 |
|
  J. H. Williams and W. E. Friedman The four-celled female gametophyte of Illicium (Illiciaceae; Austrobaileyales): implications for understanding the origin and early evolution of monocots, eumagnoliids,and eudicots Am. J. Botany, March 1, 2004; 91(3): 332 - 351. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. A. Wattier, A. L. Davidson, B. A. Ward, and C. A. Maggs cpDNA-RFLP in Ceramium (Rhodophyta): intraspecific polymorphism and species-level phylogeny Am. J. Botany, July 1, 2001; 88(7): 1209 - 1213. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||
