| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics |
Section of Evolution and Ecology, Division of Biology, University of California, 1 Shields Ave., Davis, California 95616 USA
Received for publication April 16, 2002. Accepted for publication September 24, 2002.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
interpreted the parentage of C. delicata and other evidence of intersectional hybridization to mean that the diploid sections of the genus, though highly diverse, were closely related and should be maintained in the single genus Clarkia. Here we assess phylogenetic relationships among the species of sect. Sympherica and related species by analyzing the nucleotide sequences of PgiC1 and PgiC2, a pair of paralogous genes that encode the cytosolic isozyme of phosphoglucose isomerase (EC 5.3.1.9). The major results were the following: (1) C. unguiculata and both genomes of C. delicata are within a well-defined "Sympherica" clade; thus, C. delicata should not be considered an intersectional hybrid; (2) C. heterandra belongs in the clade and is closely related to C. unguiculata; and (3) on the evidence of PgiC1, C. dudleyana is not in the clade and is not closely related to C. heterandra.
Key Words: Clarkia • DNA sequence data • Onagraceae • PgiC • phosphoglucose isomerase
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
, 1973
; Lewis and Lewis, 1955
; Gottlieb and Weeden, 1979
; Gottlieb, 1986
; Sytsma and Gottlieb, 1986a
; Sytsma, Smith, and Gottlieb, 1990
; Gottlieb and Ford, 1996
). The genus contains 42 species of which 32 are diploid, seven tetraploid, and three hexaploid. Lewis and Lewis (1955)
erected seven sections to accommodate seven groups of diploid species and related polyploids, each group defined by shared morphological and cytological traits, as well as three monotypic sections for allopolyploid species considered derived by intersectional hybridization. The intersectional polyploids were taxonomically critical, in their view, because they linked together morphologically diverse diploid sections and justified their inclusion in the single genus Clarkia.
The largest section by far is sect. Sympherica Holsinger & Lewis (formerly sect. Peripetasma Lewis & Lewis), which presently contains eight diploid and one allotetraploid species (Holsinger and Lewis, 1986
). The section was divided into three morphologically distinguishable, diploid subsections and a fourth, tetraploid subsection by Lewis and Lewis (1955)
, and this division was maintained after the nomenclatural change (Table 1). Several diploid species within sect. Sympherica have provided important evidence about the origin of reproductive isolation and morphological divergence in annual plants. For example, C. lingulata Lewis & Lewis originated recently, apparently by rapid and abrupt chromosomal reorganization from a geographically peripheral population of C. biloba (Dur.) Nels. & Macbr., and is one of the very few examples of a recent natural origin of a diploid plant species (Roberts and Lewis, 1955
; Lewis and Roberts, 1956
; Lewis, 1962
, 1973
; Gottlieb, 1974
).
|
) following the demonstration on the basis of shared restriction sites in chloroplast DNA that it was closely related to C. dudleyana (Abrams) Macbr. in sect. Sympherica (Sytsma and Gottlieb, 1986a
, b
). Lewis and Raven noted several similarities between C. heterandra and sect. Sympherica, including chromosome number of n = 9, sepals that remain fused at anthesis, and stamens in two series, the inner series being smaller. However, these traits also characterize other sections, for example, Phaeostoma and Fibula. They were reluctant to place C. heterandra in sect. Sympherica because in several other respects (the appearance of young seedlings, the straight axis of the inflorescence in bud, the width and venation of leaves) it has greater morphological similarity to C. unguiculata in sect. Phaeostoma. Thus, C. heterandra was placed in a new monotypic section Heterogaura (Lewis and Raven, 1992
).
Section Sympherica contains an allotetraploid species, C. similis Lewis & Ernst, that arose following hybridization between members of two subsections, C. epilobioides (Nutt.) Nels. & Macbr. (subsect. Micranthae) and C. modesta Jepson (subsect. Lautiflorae) (Lewis and Lewis, 1955
). Clarkia similis was assigned to a monotypic subsect. Prognatae. Clarkia epilobioides is also a parent of the allotetraploid C. delicata (Abrams) Nels. & Macbr., the other parent being C. unguiculata Lindl. from sect. Phaeostoma (Lewis and Lewis, 1955
). Clarkia delicata was placed in a monotypic sect. Connubium and is one of the allotetraploids cited by Lewis and Lewis to demonstrate the relatedness of diverse species groups in Clarkia.
More recently, phylogenetic relationships among the sections of Clarkia have been inferred using the nucleotide sequences of PgiC genes from numerous diploid species (Gottlieb and Ford, 1996
). PgiC encodes the cytosolic isozyme of phosphoglucose isomerase (EC 5.3.1.9) and was duplicated in the ancestral stock of the genus giving rise to paralogous genes PgiC1 and PgiC2 (Ford, Thomas, and Gottlieb, 1995
). Parsimony analysis, using Oenothera mexicana Spach as outgroup, placed all the PgiC sequences in two groups corresponding to the two duplicate genes and provided a very strongly supported tree of species relationships (Gottlieb and Ford, 1996
). The PgiC duplication and the presence of unicellular papillae on the stigmatic surface (Heslop-Harrison, 1990
; Hoch et al., 1993
) are the only identified autapomorphies for the genus.
PgiC1 and PgiC2 have identical structures of 23 exons encoding proteins (568–571 amino acids) and 22 introns in identical positions, but are readily distinguishable in sequence, differing about 4% in exons and 12% in introns (Thomas et al., 1993
). The two genes assort independently (Gottlieb, 1977
; Gottlieb and Weeden, 1979
; Weeden and Gottlieb, 1979
). PGIC enzymes are dimeric proteins and the subunits encoded by both genes associate in vivo in all possible combinations of homo- and heterodimers, their number dependent on allelic state at each of the two loci (Gottlieb, 1977
; Gottlieb and Weeden, 1979
). About half of the diploid clarkias have two PGIC enzymes and half have only one. Sequence analysis showed that PgiC1 is expressed in all the species, and PgiC2 is the gene that has been silenced (Gottlieb and Ford, 1996
, 1997
).
All but one of the diploid species in sect. Sympherica have two PGIC isozymes encoded by the two PgiC genes (Gottlieb, 1977
; Gottlieb and Weeden, 1979
); the exception is C. rostrata Davis, which has a PGIC1 isozyme but no PGIC2 because its coding gene was silenced (Gottlieb and Ford, 1996
). Clarkia heterandra (then Heterogaura heterandra) was first thought to have only a single isozyme (Gottlieb and Weeden, 1979
). However, following demonstration of a close relationship between this species and C. dudleyana (Systma and Gottlieb, 1986b
), additional populations of C. heterandra were examined by electrophoresis and evidence for two PGIC isozymes was obtained (L. D. Gottlieb, unpublished data).
PgiC genes have also been sequenced from allotetraploid species of Clarkia. The first one examined, Clarkia gracilis (Piper) Nels. & Macbr., in sect. Rhodanthos (Fisch. & Mey.) Raven, was found to have two PgiC1 genes and one PgiC2 (Ford and Gottlieb, 1999
). More recently, for a study of gene silencing, we sequenced PgiC genes from C. delicata and C. similis and their three diploid parents, two of which are presently placed in sect. Sympherica (Ford and Gottlieb, 2002
). Clarkia delicata and C. similis each have two PgiC1s and two PgiC2s, as expected, but not all are expressed (Ford and Gottlieb, 2002
). This latter study stimulated the present molecular analysis of the taxonomic relationships of the species assigned to sect. Sympherica.
In this paper, we report newly obtained sequences of PgiC genes, both PgiC1 and PgiC2, from three additional species of the section and from C. heterandra. We use these sequences together with those from the two allotetraploid species and their diploid parents to assess overall phylogenetic relationships in the section. The phylogenetic analysis reveals a number of unexpected relationships and suggests that the sectional assignment of several species should be reconsidered.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Sequence information from the following species was also used: C. lewisii Raven & Parnell, C. rostrata, C. williamsonii (Dur. & Hilg.) Lewis & Lewis, C. xantiana Gray, C. mildrediae (Heller) Lewis & Lewis, C. arcuata (Kell.) Nels. & Macbr., C. franciscana Lewis & Raven, C. concinna (Fisch. & Mey.) Greene subsp. automixa Bowman, and Oenothera mexicana (Gottlieb and Ford, 1996
); C. gracilis subsp. gracilis (Ford and Gottlieb, 1999
); C. epilobioides, C. modesta, C. unguiculata, C. similis, and C. delicata (Ford and Gottlieb, 2002
). Of the nine species in sect. Sympherica (Table 1), only C. cylindrica (Jeps.) Lewis & Lewis is not included. Only sequences from population 148b of C. epilobioides and population 9610 of C. similis were used (Ford and Gottlieb, 2002
). Only PgiC2 of C. gracilis was used (Ford and Gottlieb, 1999
).
DNA isolation and sequencing
Genomic DNAs were prepared from seedling leaves of single individuals as described (Ford and Gottlieb, 1999
). Templates for genomic sequencing were obtained by polymerase chain reaction (PCR) using primers directed against various combinations of exons and two conserved regions upstream of the start codon. Each of the longer sequences was assembled from a number of overlapping templates. Because PCR generally yielded fragments from two PgiC genes, the fragments were separated on agarose gels, excised singly or as a group with a razor blade, DNA recovered with the ZymoClean Kit (Zymo Research, Orange, California, USA) and cloned with kits from Invitrogen (Carlsbad, California, USA). Both strands were sequenced by Davis Sequencing, Davis, California, USA. Two independent clones were sequenced for each fragment and conflicts (presumed PCR artifacts) were resolved by sequencing a third clone. For C. heterandra, C. lingulata, and C. biloba subsp. biloba, sequences were obtained from two loci but only one allele at each locus. (Clarkia heterandra is predominantly self-pollinating and highly homozygous; C. lingulata and C. biloba are outcrossing but the clones obtained for each locus appeared to represent one allele.) For C. dudleyana, two alleles were found at each locus; two were sequenced for PgiC2 but only one for PgiC1.
For C. lingulata, C. dudleyana, and C. heterandra, both PgiC genes were sequenced starting from 400 to 200 nucleotides (nt) upstream of the translation start. The sequences of PgiC1 and PgiC2 of C. lingulata extend to exon 21 (5344 and 5140 nt, respectively), PgiC1, PgiC2a, and PgiC2b of C. dudleyana extend to exon 16 (4471, 4094, and 4070 nt, respectively), PgiC1 of C. heterandra extends to exon 21 (5282 nt), and PgiC2 extends to exon 23 (5860 nt). For C. biloba subsp. biloba, shorter sequences from exon 10 to exon 14 were sufficient, 1366 and 1110 nt for PgiC1 and PgiC2, respectively. European Molecular Biology Laboratory (EMBL) accession numbers are AJ437270-78. Accession numbers of other sequences used in the present analysis are X89384-97, Y14130, AJ302021-24, AJ311744-52, AJ307690, AJ307692, AJ307694, AJ307696, AJ307698, AJ307700, AJ307704, AJ307705, AJ307707, AJ307709.
Sequence analysis
Sequences were aligned as described (Ford and Gottlieb, 1999
). The aligned matrix can be accessed at http://ajbsupp.botany.orgv90/. The aligned data included 7856 sites, excluding 12 insertions totaling 4709 sites that were not informative for parsimony (see Results). Bootstrapped parsimony analyses were performed with PAUP* v4.0 (Swofford, 2001
) using all data (exons, introns, 5' leader region, 3' untranslated region) with gaps treated as missing. Significance tests comparing user-defined trees were performed by PAUP* using the Templeton (Wilcoxon signed-rank) test (command "pscores /nonparamtest = yes test details = yes"); the output lists all sites that contribute different numbers of steps to the test trees.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
). Phylogenetic analysis (below) justifies this procedure and confirms that the PgiC1s and PgiC2s form two distinct clades. Pairwise differences between the four PgiC1s and the four PgiC2s average 4.2% in exons (range 2–7.3%) and 12.6% in introns (9–14.5%) (data not shown).
Phylogeny of PgiC genes
The phylogenetic analysis includes sequences of both PgiC1 and PgiC2 from all species in sect. Sympherica (except C. cylindrica) plus C. heterandra, C. unguiculata, and C. delicata. To place these in context, we include all sequences from the initial analysis of sectional relationships in Clarkia (Gottlieb and Ford, 1996
) and PgiC2 of the allotetraploid C. gracilis, the only PgiC2 known from sect. Rhodanthos (Ford and Gottlieb, 1999
).
Figure 1 shows the consensus phylogenetic tree of PgiC genes resulting from parsimony analysis of 1000 bootstrap replicates. Only clades with >80% bootstrap support are reported. The tree consists of two major clades corresponding to PgiC1 and PgiC2 plus the outgroup PgiC of O. mexicana. The PgiC1 subtree provides at least 90% bootstrap support at most nodes; one node has 81% support and there are two trichotomies. The PgiC2 subtree provides weaker support at many nodes, particularly within the clade of species from sect. Sympherica, but the two subtrees are consistent.
|
To understand whether parsimony analysis of relationships within sect. Sympherica is affected by the inclusion of species from other sections we also constructed bootstrap 80% majority rule consensus trees for each gene excluding the sequences from sects. Myxocarpa, Rhodanthos, and Eucharidium as well as O. mexicana. In these analyses, C. xantiana served as outgroup because it is sister to the rest of the species used (Fig. 1). There were 367 and 351 informative sites for PgiC1 and PgiC2, respectively. The topologies and support for each node were the same as the corresponding subtrees of Fig. 1 with one minor difference (see Results, point 5). Thus, both genes provide about the same number of informative sites, and the lower bootstrap values on the PgiC2 tree are not explained on this basis.
The PgiC1 tree suggests certain conclusions that we describe separately below: (1) C. epilobioides, C. modesta, and C. unguiculata are more closely related to their tetraploid derivatives than to any other diploid species studied; (2) C. unguiculata, C. delicata, and C. heterandra fall within a well-defined Sympherica clade; (3) C. unguiculata and C. heterandra are sisters among the diploid species studied; (4) C. dudleyana is not in the Sympherica clade and is not closely related to C. heterandra; (5) the present subsectional divisions of sect. Sympherica are supported (except regarding C. dudleyana); the C. heterandra/C. unguiculata subclade is separate from them; C. biloba and C. lingulata are sister species.
Point 1: Clarkia epilobioides, C. modesta, and C. unguiculata are more closely related to their tetraploid derivatives than to any other diploid species studied
PgiC1 of C. epilobioides is more closely related to its tetraploid derivatives PgiC1epi of C. delicata and PgiC1epi of C. similis than to PgiC1 from any diploid species shown. Likewise, PgiC1 of C. modesta is most closely related to PgiC1mod of C. similis and PgiC1 of C. unguiculata is most closely related to PgiC1ung of C. delicata (Fig. 1).
The PgiC1 result shows that the origins of both tetraploid species are more recent than any diploid speciation events involving the three parent species (but see Discussion regarding C. unguiculata). The origin of C. delicata may be more recent than that of C. similis because of the closer relationship of PgiC1 of C. epilobioides to its derivative in the former species.
The PgiC2 tree supports the close relationship of C. epilobioides to both its tetraploid derivatives but is not informative regarding the relationship between C. unguiculata and C. delicata. Clarkia modesta is closely related to C. similis, but the tree does not show that they are more closely related to each other than to two other diploid species (Fig. 1).
Point 2: The "Sympherica" clade includes Clarkia unguiculata, C. delicata, and C. heterandra
The PgiC1 tree provides 98% bootstrap support for what we term the Sympherica clade ("Sympherica" in Fig. 1). This group includes all species presently assigned to the section except C. dudleyana. It also includes C. heterandra, C. unguiculata, and both PgiC1 genes of C. delicata. The sister to this clade is C. williamsonii, with 100% bootstrap support. Thus, C. williamsonii, representing sect. Godetia, forms a natural and well-defined boundary between the Sympherica clade and the more basal sections of the PgiC1 tree. On the evidence of Fig. 1, there is no more narrowly defined subclade that includes the species of sect. Sympherica but excludes C. heterandra and C. unguiculata.
The PgiC2 subtree provides 99% bootstrap support for a larger Sympherica clade that would include all the species of the PgiC1 Sympherica clade, plus C. dudleyana. Clarkia williamsonii, like other species of sect. Godetia, lacks an expressed PgiC2 (no pseudoPgiC2 has been found) and does not appear on the PgiC2 tree. Clarkia unguiculata and C. heterandra and both PgiC2 genes from C. delicata are included in the PgiC2 Sympherica clade. However, the lack of resolution within this clade and the absence of information from sect. Godetia makes the PgiC2 evidence less useful than the PgiC1 evidence.
Clarkia unguiculata was included in sect. Phaeostoma by Lewis and Lewis (1955)
, and therefore we expected the species to appear as sister to C. xantiana. This relationship is clearly excluded by both the PgiC1 and PgiC2 subtrees. The matter was specifically tested by placing the PgiC1 genes from C. unguiculata and C. delicata (PgiC1ung) next to C. xantiana in a user-defined tree: this tree was 53 steps longer, highly significant by Templeton's test (P < 0.0001). There are 60 sites favoring the tree shown in Fig. 1 (i.e., requiring fewer steps for this tree) and only seven that favor the alternative.
Point 3: Clarkia heterandra and C. unguiculata are sister species among the diploid species studied
The PgiC1 tree has 99% bootstrap support for a sister relationship between C. heterandra and C. unguiculata. This result is neither supported nor contradicted by the PgiC2 subtree. We tested an alternative PgiC1 tree that breaks the sister relationship by placing C. heterandra and C. unguiculata/C. delicata (PgiC1ung) as two legs of a trichotomy, with C. similis/C. modesta/C. lingulata/C. biloba as the third leg. This tree is 13 steps longer, significantly longer by Templeton's test (P < 0.001). The test identifies 13 sites that affect the relative length of the two trees. At each of them, the PgiC1s of C. heterandra and C. unguiculata and PgiC1ung of C. delicata all have the same base and differ from all other sequences. Thus, these 13 sites support the placement of C. heterandra, C. unguiculata, and PgiC1ung of C. delicata as a clade over any alternative placement. There were no sites favoring the alternative tree tested.
Point 4: Clarkia dudleyana is not a member of the Sympherica clade
The PgiC1 tree shows C. dudleyana in a disjunct position outside the Sympherica clade. The PgiC2 tree, which includes two alleles from C. dudleyana, places them within a large clade that includes all other species from sect. Sympherica, but does not unite them with other species of subsect. Lautiflorae (the present taxonomic home of the species). Because there are no PgiC2 sequences representing species of sect. Godetia, the PgiC2 tree is not incompatible with the PgiC1 tree.
To evaluate the evidence regarding the placement of the PgiC1 of C. dudleyana we compared five PgiC1 trees (Fig. 2A–E). The first three are the possibilities consistent with the consensus tree (Fig. 1). The fourth places C. dudleyana as a basal member of the Sympherica clade. The fifth places C. dudleyana with sect. Godetia, represented by C. williamsonii. Tree A is the shortest (2567 steps), with trees B and C only three and four steps longer, not significant by Templeton's test. Trees D and E are each 27 steps longer, significantly longer (P < 0.0001). There are 29 sites favoring any of trees A, B, or C over either tree D or E, whereas only two sites favor D over A, B, and C, and two other sites favor E over the top three. On this evidence, the relative positions of C. dudleyana and C. xantiana cannot be determined, but both are basal to C. williamsonii and cannot be included in the Sympherica clade.
|
The combined result of the analyses of the two genes suggests that C. williamsonii is sister to the Sympherica clade as defined for PgiC1, C. dudleyana is basal to C. williamsonii, and C. xantiana is basal to C. dudleyana.
This result contradicts the sister relationship between C. dudleyana and C. heterandra previously identified by restriction site analysis of cpDNA (Sytsma and Gottlieb, 1986a
, b
). To explicitly test this conclusion, we also examined the effect of removing PgiC1 of C. heterandra from its position in Fig. 1 and placing it next to PgiC1 of C. dudleyana and vice versa. The former change increases tree length by 46 steps and the latter by 35 steps, both highly significant by Templeton's test (P < 0.0001).
Point 5: The present subsectional divisions of sect. Sympherica are supported (except regarding Clarkia dudleyana); the C. heterandra/C. unguiculata subclade is separate from them; and C. biloba and C. lingulata are sister species
The PgiC1 Sympherica clade has three major subclades, each with 99–100% bootstrap support, shown boxed in Fig. 1. One (A) corresponds to subsects. Sympherica and Micranthae and the tetraploid derivatives of Micranthae, the second (B) consists of C. heterandra and C. unguiculata and its tetraploid derivative, and the third (C) corresponds to subsect. Lautiflorae (but without C. dudleyana) and its tetraploid derivative (Table 1). As expected (see Discussion), there is 100% support for the sister relationship of C. biloba and C. lingulata within subsect. Lautiflorae.
Templeton's test was used to compare the PgiC1 tree shown, with subclades B and C as sisters (2569 steps), to alternative trees with subclades A and C as sisters (2572 steps) or A and B as sisters (2573 steps). Only eight sites affect the relative lengths of the three trees. Five favor (i.e., require the fewest steps or changes on) the first tree, two favor the second tree, and one favors the third tree. The first tree is not significantly superior to the other two (for tree 1 vs. tree 2, P = 0.2568, for tree 1 vs. tree 3, P = 0.1025). Thus, the 81% bootstrap support shown for the clade in Fig. 1 is not sufficient and the three major subclades are presently best regarded as a trichotomy. In fact, the bootstrap 80% majority-rule consensus tree constructed using only the PgiC1 sequences from the Sympherica clade plus C. williamsonii, C. dudleyana, and C. xantiana showed the three major subclades as a trichotomy.
The PgiC2 tree gives weaker support to the relationship of subsects. Sympherica and Micranthae; the Lautiflorae are divided into two clades, one as for the PgiC1 tree and the other of C. dudleyana only; and the C. heterandra/C. unguiculata clade is not apparent at all. No further analysis of PgiC2 data was attempted.
Gaps in exons
A codon duplication in exon 22 was previously observed in all PgiC2s plus PgiC1 of C. concinna, C. lewisii, and C. rostrata, a taxonomic distribution requiring three separate events (three duplications, or a duplication, a loss, and a second duplication or gene conversion; Gottlieb and Ford, 1996
). This duplication also occurs in PgiC1 of C. epilobioides and its derivatives C. delicata PgiC1epi and C. similis PgiC1epi, but not in C. delicata PgiC1ung or C. similis PgiC1mod. Exon 22 of the PgiC1 genes of the other diploids in the Sympherica clade was not sequenced, not being needed for phylogenetic analysis. These results suggest that the occurrence of the duplication in PgiC1 genes of sect. Sympherica is limited to the clade consisting of C. lewisii, C. rostrata, and C. epilobioides. All PgiC2s have the exon 22 codon duplication, but C. heterandra has three copies of this codon, implying a fourth event unique to this gene. The only other insertion known in exons of expressed Clarkia PgiC genes is the duplication of a trinucleotide that overlaps two codons in exon 23 of PgiC1 of C. rostrata (Gottlieb and Ford, 1996
).
Gaps in introns
Many small insertions and deletions (generally less than 20 nt) occur in PgiC introns. All PgiC2s share an approximately 200-nt deletion in intron 12 relative to all PgiC1s and PgiC of the outgroup Oenothera. There are other large unique deletions, e.g., in intron 10 of C. modesta PgiC2, a 180-nt deletion not shared by the closely related C. similis PgiC2mod, and large deletions in intron 16 of both PgiC1 and PgiC2 of C. heterandra.
Some insertions are flanked by direct and/or inverted repeats, apparently the remnants of transposable elements (Gottlieb and Ford, 1996
). Examples are a 30-nt insertion in intron 10 of C. modesta PgiC1 that involves a 16-nt direct repeat and a 56-nt insertion in intron 10 of C. unguiculata PgiC1 flanked by a 10-nt direct repeat. For all except two of these transposon-like elements, each is present in only a single diploid species (and its tetraploid derivatives, if any). The 12 longest, ranging from 131 to 857 nt and totaling 4709 nt, were omitted from the aligned data set because they are unwieldy and not informative for parsimony. The two exceptions are a 239-nt insertion in intron 16 of PgiC2 of C. lewisii and C. epilobioides (the presence of this insertion in the pseudoPgiC2 of C. rostrata could not be evaluated because the intron has not been sequenced) and a 210-nt insertion in intron 19 of PgiC2 of C. epilobioides and C. unguiculata. The latter insertion includes seven informative sites at which C. epilobioides PgiC2, C. delicata PgiC2epi, and C. similis PgiC2epi differ from C. unguiculata PgiC2 and C. delicata PgiC2ung, but adding these sites to the PAUP* infile did not change the results.
Eight of the long insertions belong to a single family characterized by a 9-nt target site and conserved approximately 105-nt inverted repeats. Examples are in PgiC2s of C. epilobioides, C. unguiculata, and C. heterandra, in addition to previously known cases in C. lewisii, C. xantiana, and C. mildrediae (Gottlieb and Ford, 1996
). No examples occur in PgiC1s, but a copy is located in the 5' untranslated region of the lis (linolool synthase, AF067602) gene of C. concinna. This family will be described separately. PgiC2 of C. epilobioides is unique in having insertions from this family in introns 12, 16, 18, and 19 as well as an unrelated transposon-like insertion in intron 4, increasing its overall length by about 1500 nt compared to other PgiC genes.
Insertions that are unique to one or to two genes have no direct effect in a parsimony analysis when gaps are treated as missing. However, transposon excision may result in changes in adjacent sequences that obscure their phylogenetic relationships. All except two of the observed transposon insertions are in PgiC2 genes. It may be that a greater frequency of transposition events in introns of PgiC2 genes contributed to the lower bootstrap support for nodes on the PgiC2 tree.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Clarkia unguiculata and C. delicata are in the Sympherica clade
The close relationship shown here between C. unguiculata and species of the Sympherica clade was anticipated by a restriction site analysis of cpDNA (Sytsma, Smith, and Gottlieb, 1990
). In that study, sect. Sympherica was represented by C. rostrata and C. biloba and sect. Phaeostoma by C. xantiana and C. unguiculata. There was 100% bootstrap support for pairing the two species from sect. Sympherica and 59% support for placing C. unguiculata basal to that pair rather than with C. xantiana. The latter species appeared to be more closely related to C. bottae from sect. Fibula.
The fact that the genomes of the morphologically dissimilar C. unguiculata (sect. Phaeostoma) and C. epilobioides (sect. Sympherica) were united in the tetraploid C. delicata led Lewis and Lewis (1955)
to consider the tetraploid species as a link between the two sections and justified including both in the single genus Clarkia. However, the evidence from PgiC suggests a simpler interpretation: C. unguiculata is relatively closely related to C. epilobioides and is better accommodated in an enlarged sect. Sympherica. Consequently, Clarkia delicata, like C. similis, would be an intrasectional hybrid originating within the Sympherica clade.
Lewis and Lewis (1955)
placed C. unguiculata with C. xantiana in sect. Phaeostoma. The two species show "important differences" in a number of morphological traits, both vegetative and floral, that led Holsinger (1985a)
to define separate subsections for them. Clarkia unguiculata is closely related to C. exilis, C. tembloriensis, and C. springvillensis (Holsinger, 1985b
), leading us to predict that DNA sequences of their PgiC1 and PgiC2 genes would place them also in the Sympherica clade.
Clarkia heterandra is in the Sympherica clade and is most closely related to C. unguiculata
Although C. heterandra was formally placed in the monotypic sect. Heterogaura (Lewis and Raven, 1992
), the primary evidence supporting the transfer of this species from Heterogaura to Clarkia came from a study that showed its close relationship to species in sect. Sympherica (Sytsma and Gottlieb, 1986a
). That study was based on restriction site analysis of chloroplast DNA of all eight species in sect. Sympherica and showed with 98% bootstrap support that C. heterandra is most closely related to C. dudleyana. The present data from PgiC1 show, with 99% bootstrap support, that C. heterandra is more closely related to C. unguiculata than to any species presently in sect. Sympherica. The previous study (Sytsma and Gottlieb, 1986a
, b
) did not examine C. unguiculata. The PgiC1 tree provides 98% bootstrap support for a Sympherica clade that includes C. heterandra, C. unguiculata, and all species presently in sect. Sympherica except C. dudleyana (discussed below). This result confirms the major conclusion of Sytsma and Gottlieb (1986a
, b)
that Heterogaura heterandra is a Clarkia.
Clarkia heterandra resembles C. unguiculata more than it does C. dudleyana with respect to the appearance of young seedlings, the width and venation of leaves, and the straight axis of the inflorescence in bud, observations invoked by Lewis and Raven (1992
) to justify placing C. heterandra in a monotypic section rather than in sect. Sympherica. The reduced floral petals of C. heterandra are also more suggestive of the limb of the petals of C. unguiculata than the large fan-shaped petals of C. dudleyana. The similarity of the two species now supports the close relationship between them shown by the PgiC1 data and no longer presents an obstacle to the inclusion of C. heterandra in sect. Sympherica.
Clarkia lingulata and C. biloba are sister species
Clarkia lingulata, with nine pairs of chromosomes and known from only a pair of populations in the Merced River canyon of the Sierra Nevada of central California, has long been regarded as derived from the more widespread C. biloba, with eight pairs of chromosomes. The original evidence reflected their near morphological identity (differing only in the shape of the flower petal) and the homology of the extra chromosome of C. lingulata to parts of two chromosomes of C. biloba (Lewis and Roberts, 1956
; Lewis, 1962
). The very close relationship between them was later confirmed by their high genetic similarity revealed by isozyme analysis (Gottlieb, 1974
) and detection of their close sister relationship in the phylogenetic tree based on cpDNA (Sytsma and Gottlieb, 1986a
). It would now be appropriate to test whether multiple PgiC alleles (or alleles at other genes) of C. lingulata form a subclade within a clade of alleles from both species.
The three major Sympherica subclades are consistent with taxonomic subsections
One subclade consists of subsects. Sympherica and Micranthae and indicates a particularly close relationship between them. The relationship of C. epilobioides, the only member of subsect. Micranthae, to other diploid species had been considered uncertain on morphological grounds (Lewis and Lewis, 1955
), but the cpDNA study (Sytsma and Gottlieb, 1986a
, b
) demonstrated its relationship to subsect. Sympherica. That study placed C. rostrata as sister to C. epilobioides, an arrangement that was also supported by both species having single cytosolic and chloroplast isozymes of 6-phosphogluconate dehydrogenase (Odrzykoski and Gottlieb, 1984
), but would make subsect. Sympherica paraphyletic. The present study is in better agreement with the taxonomy and shows that subsect. Sympherica is probably monophyletic, although the PgiC genes of C. cylindrica also in this subsection have yet to be sequenced.
The second major subclade contains C. heterandra and C. unguiculata. The third corresponds to subsect. Lautiflorae, but without C. dudleyana. All three subclades also include appropriate derived subgenomes of C. similis and C. delicata.
Does the Sympherica clade include C. dudleyana?
The PgiC1 tree unequivocally excludes C. dudleyana from the Sympherica clade. There is 98% bootstrap support for a Sympherica clade that includes C. heterandra, C. unguiculata, and all species presently in sect. Sympherica except C. dudleyana and 100% support for a larger clade including sect. Sympherica and sect. Godetia, represented by C. williamsonii, that still excludes C. dudleyana. The inclusion of C. dudleyana in the Sympherica clade derived from PgiC2 does not contradict this conclusion because the PgiC2 tree does not include sect. Godetia (which has no PgiC2).
This result is in direct contradiction to the study (Sytsma and Gottlieb, 1986a
, b
) based on restriction site analysis of chloroplast DNA of C. heterandra plus all eight species in sect. Sympherica, with C. amoena (Leh.) Nels. & Macbr. and C. xantiana as outgroups, that placed C. dudleyana well within the Sympherica clade. One possible explanation for this discrepancy is that the 1986 study would have showed a different result had other species, such as C. unguiculata and C. williamsonii, been included. Bootstrap parsimony analysis of PgiC1 data for the species of the 1986 study, less C. cylindrica, with C. franciscana substituted for its close relative C. amoena, provides 100% support for a clade consisting of C. lewisii, C. rostrata, C. epilobioides, C. modesta, C. lingulata, C. biloba, and C. heterandra, excluding C. dudleyana (not shown). This analysis suggests that the discordance between the two data sets is not an artifact of the choice of species used in each study.
Alternatively, the different results may simply reflect a difference in data quality; a restriction site can be changed by a substitution in any of the nucleotides recognized by the restriction enzyme, and, consequently, the absence of a site may occur because of different substitutions.
A third possibility is that C. dudleyana may be a hybrid species whose chloroplast DNA (derived only from the maternal parent) and nuclear DNA have different phylogenetic histories.
Finally, the sequenced PgiC1 allele from C. dudleyana may have diverged from the alleles sampled from the other species in sect. Sympherica before the divergence of the species. This hypothesis requires that populations in the early Sympherica line carried a number of PgiC1 lineages, one of which was by chance maintained only in C. dudleyana while another gave rise to the alleles of the Sympherica clade. A finding of other, Sympherica-type alleles in C. dudleyana would confirm this hypothesis, but a failure to find such alleles would not reject it. Sequencing of other, unlinked genes would be a more certain way to test the hypothesis.
There are other suggestions that C. dudleyana needs further study. The species appears to include two groups of populations, one in the Sierra Nevada and the other in southern California, that are reproductively isolated from each other and morphologically separable (Lewis and Lewis, 1955
). The two alleles of PgiC2 that we sequenced were from a population in the Sierra Nevada and were surprisingly different: 4.6% in introns (comparable to the difference between C. modesta and C. lingulata) and 1.4% in exons. They also differed by a 75-nt deletion in intron 7 and a 78-nt deletion in intron 9 of one allele and a 92-nt deletion in intron 10 of the other. By contrast, two alleles of C. lewisii PgiC2 differed by only 0.3% in introns and 0.5% in exons (Thomas et al., 1993
). Clarkia dudleyana appears to show reproductive barriers between some individuals of the same population as well as between populations (Lewis and Lewis, 1955
), and, if the barriers reduce genetic recombination, this would tend to increase the dissimilarity of alleles. More extensive examination of C. dudleyana will be required to better understand this species before its relationships to other species can be meaningfully assessed.
The tetraploid species Clarkia delicata and C. similis have a recent origin
The close relationship of the orthologous genes in C. epilobioides, C. modesta, and C. unguiculata and their tetraploid derivatives was already apparent from pairwise comparisons (Ford and Gottlieb, 2002
), but the present phylogenetic analysis of PgiC1 shows that the origin of both tetraploid species is more recent than any diploid speciation events involving these taxa and the other species studied. The inclusion of C. cylindrica (subsect. Sympherica) in the study would probably not change this conclusion, because morphological and cpDNA evidence strongly suggest it is more closely related to C. lewisii than to C. epilobioides. However, C. unguiculata may prove to be closer to one or more of the species C. tembloriensis, C. springvillensis, and C. exilis than to C. delicata.
The recent origin of these tetraploid species is not surprising given their morphological uniformity and limited distribution. However, the role that C. delicata was assigned as a hybrid between sects. Sympherica and Phaeostoma suggested that it might have originated earlier in the history of those sections, before they had diverged much. The fact that both tetraploid species are hybrids between two of the three major subclades indicates that the diploid species in sect. Sympherica retained the ability to form tetraploids successfully well after their divergence.
Evidence from PgiC1 vs. PgiC2
Most of the phylogenetic conclusions of this paper depend primarily on the evidence of the PgiC1 sequences, which provided very high bootstrap support for nearly all clades. Evidence from PgiC2 did not prove to be conclusive regarding many relationships, even though this gene provided about the same number of sites informative for parsimony as PgiC1. The lack of strong bootstrap support for many clades of the PgiC2 tree was unexpected because in our prior analysis of sectional relationships in Clarkia (Gottlieb and Ford, 1996
) the PgiC1 and PgiC2 trees were equally informative, subject to the limitation that some of the species lacked an expressed PgiC2 or obtainable pseudoPgiC2. Presumably, among the more closely related species in a single section, the information from PgiC2 was more obscured by random resemblances or by transposon activity. The PgiC2 tree is still useful, providing an independent source of information supporting some of the results of the PgiC1 tree.
The circumscription of sect. Sympherica
The sectional taxonomy of Clarkia was determined by studies of morphology, cytology, crossing relationships, and field studies (Lewis and Lewis, 1955
). However, despite a wealth of evidence, only two sections were characterized by autapomorphies: sect. Myxocarpa, in which the surface of the immature capsules of all the species develops a soft and slimy surface prior to drying and seed release, and sect. Eucharidium, in which the flowers of both species have a long floral tube, trilobed petals, and only four stamens. Section Sympherica is an aggregation of three morphologically well-defined diploid subsections and a derived allotetraploid subsection. The nine species assigned to it are recognized by particular combinations of traits that are also found elsewhere in Clarkia.
For example, in sect. Sympherica, the rachis of the inflorescence is recurved in bud and may appear wilted, as also seen in sect. Myxocarpa and in subsect. Flexicaules of sect. Rhodanthos. In sect. Sympherica, the sepals remain united and deflexed to one side at anthesis, but the trait is also present in all of the other sections except Godetia. The stamens in sect. Sympherica are in two dissimilar series, but so are those in sects. Fibula and Phaeostoma. The immature capsules in sect. Sympherica are 4- or 8-grooved or ribbed or neither (C. epilobioides), but immature capsules with similar ribs or grooves are found elsewhere. The petals of species of sect. Sympherica are highly diverse: oblanceolate to obovate, or fan-shaped, or bilobed, and all of these shapes (except bilobed found only in C. biloba) are also seen in other sections.
The PgiC evidence suggests C. heterandra, C. unguiculata, and C. delicata should be transferred to sect. Sympherica. These additions would increase the already great morphological diversity of the section. For example, the rachis of the inflorescence of these species is straight, unlike other species of the section. But, both C. heterandra and C. delicata are the only members of their sections, and C. unguiculata and its relatives have already been placed in their own subsection of sect. Phaeostoma (Holsinger, 1985a), so their removal to sect. Sympherica would not subvert any close relationships previously determined on morphological or cytological analysis.
Because it is likely that C. exilis, C. springvillensis, and C. tembloriensis, the relatives of C. unguiculata, would also have to be transferred to the section, it would then have 15 species, of which 13 are diploid, 41% of the 32 diploid species in the genus. In view of the large size and morphological diversity of such an expanded sect. Sympherica and the lack of a defining autapomorphy, it might be preferable to elevate some of the subsections to sectional status. On the contrary, it might also be argued that the expanded sect. Sympherica should be retained as a unit because the intersubsectional hybrids C. similis and C. delicata each combine genomes from two of the three major subclades. This, of course, is the same reasoning employed by Lewis and Lewis (1955)
to show relationships among sections, now applied at the level of subsections.
The suggestion that C. dudleyana might have to be removed from sect. Sympherica does seem unexpected. On the evidence of the traits Lewis and Lewis (1955)
used to define sections, such as rachis of inflorescence recurved, and anthers in two dissimilar series, the membership of C. dudleyana in sect. Sympherica seems appropriate, and they described it as having a clear resemblance to C. biloba and C. modesta. Lewis and Lewis attempted (p. 329) to hybridize the species with members of sect. Sympherica but were only able to obtain a single hybrid with C. biloba and the individual produced no flowers. They did produce sterile hybrids with C. unguiculata (sect. Phaeostoma, but now seeming to belong with sect. Sympherica) and with the hexaploid C. purpurea (sect. Godetia). Possible new homes for the species include sects. Fibula, Godetia, and Phaeostoma, which have chromosome number n = 9 and some morphological characters like C. dudleyana. But we have already noted that C. dudleyana is quite poorly known and requires study to make sense of the substantial amount of hybrid sterility found both within and between its populations before a taxonomic reassessment is appropriate.
In previous studies, PgiC sequence evidence has proved consistent with morphological and cytological evidence at the sectional (Gottlieb and Ford, 1996
) and subsectional (Ford and Gottlieb, 1999
) levels. Yet, although the phylogenetic evidence from PgiC1 is very strong, taxonomic decisions are not generally based on single genes. A vast amount of information, perhaps 200 or more published papers, has been accumulated about Clarkia since the monograph by Lewis and Lewis in 1955
. Few other genera of wild plants have been examined so intensively. Ironically, it is because we know so much about Clarkia that it seems inappropriate to rush to judgment. Thus, the PgiC1 phylogenetic tree is perhaps best regarded as an hypothesis motivating a thorough reexamination of the diverse morphological traits that appear in many different combinations in different species.
|
FOOTNOTES |
|---|
2 Author for reprint requests (ldgottlieb{at}ucdavis.edu
)
|
LITERATURE CITED |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONLITERATURE CITED |
|---|
Ford V. S. L. D. Gottlieb 2002 Single mutations silence PgiC2 genes in two very recent allotetraploid species of Clarkia. Evolution 56: 699-707[CrossRef][ISI][Medline]
Ford V. S. B. R. Thomas L. D. Gottlieb 1995 The same duplication accounts for the PgiC genes in Clarkia xantiana and C. lewisii (Onagraceae). Systematic Botany 20: 147-160[CrossRef][ISI]
Gottlieb L. D. 1974 Genetic confirmation of the origin of Clarkia lingulata. Evolution 28: 244-250[CrossRef][ISI]
Gottlieb L. D. 1977 Evidence for duplication and divergence of the structural gene for phosphoglucose isomerase in diploid species of Clarkia. Genetics 86: 289-308
Gottlieb L. D. 1986 Genetic differentiation, speciation and phylogeny in Clarkia (Onagraceae). In K. Iwaksuki, P. H. Raven, and W. J. Bock [eds.], Modern aspects of species, 145–160. University of Tokyo Press, Tokyo, Japan
Gottlieb L. D. V. S. Ford 1996 Phylogenetic relationships among the sections of Clarkia (Onagraceae) inferred from the nucleotide sequences of PgiC. Systematic Botany 20: 45-62
Gottlieb L. D. V. S. Ford 1997 A recently silenced, duplicate PgiC locus in Clarkia. Molecular Biology and Evolution 14: 125-132[Abstract]
Gottlieb L. D. N. F. Weeden 1979 Gene duplication and phylogeny in Clarkia. Evolution 33: 1024-1039[CrossRef][ISI]
Heslop-Harrison Y. 1990 Stigma form and surface in relation to self-incompatibility in the Onagraceae. Nordic Journal of Botany 10: 1-19
Hoch P. C. J. V. Crisci H. Tobe P. E. Berry 1993 A cladistic analysis of the plant family Onagraceae. Systematic Botany 18: 31-47
Holsinger K. E. 1985a Taxonomic and nomenclatural notes on Clarkia sect. Phaeostoma Lewis and Lewis (Onagraceae). Taxon 34: 704-706[CrossRef][ISI]
Holsinger K. E. 1985b A phenetic study of Clarkia unguiculata Lindley (Onagraceae) and its relatives. Systematic Botany 10: 155-165[CrossRef][ISI]
Holsinger K. E. H. Lewis 1986 Description of a new section and subsection in Clarkia (Onagraceae). Annals of the Missouri Botanical Garden 73: 491-494[CrossRef][ISI]
Lewis H. 1953 The mechanism of evolution in the genus Clarkia. Evolution 7: 1-20[CrossRef][ISI]
Lewis H. 1962 Catastrophic selection as a factor in speciation. Evolution 16: 257-271[CrossRef][ISI]
Lewis H. 1973 The origin of diploid neospecies in Clarkia. American Naturalist 107: 161-170[CrossRef][ISI]
Lewis H. M. E. Lewis 1955 The genus Clarkia. University of California Publications in Botany 20: 241-392
Lewis H. P. H. Raven 1992 New combinations in the genus Clarkia (Onagraceae). Madroño 39: 163-169
Lewis H. M. R. Roberts 1956 The origin of Clarkia lingulata. Evolution 10: 126-138[CrossRef][ISI]
Odrzykoski I. J. L. D. Gottlieb 1984 Duplications of genes coding 6-phosphogluconate dehydrogenase in Clarkia (Onagraceae) and their phylogenetic implications. Systematic Botany 9: 479-489[CrossRef][ISI]
Roberts M. R. H. Lewis 1955 Subspeciation in Clarkia biloba. Evolution 9: 445-454[CrossRef][ISI]
Swofford D. L. 2001 PAUP*: phylogenetic analysis using parsimony (*and other methods), version 4.0b8. Sinauer, Sunderland, Massachusetts, USA
Sytsma K. J. L. D. Gottlieb 1986a Chloroplast DNA evolution and phylogenetic relationships in Clarkia sect. Peripetasma (Onagraceae). Evolution 40: 1248-1261[CrossRef][ISI]
Sytsma K. J. L. D. Gottlieb 1986b Chloroplast DNA evidence for the origin of the genus Heterogaura from a species of Clarkia (Onagraceae). Proceedings of the National Academy of Sciences, USA 83: 5554-5557
Sytsma K. J. J. F. Smith L. D. Gottlieb 1990 Phylogenetics in Clarkia (Onagraceae): restriction site mapping of chloroplast DNA. Systematic Botany 15: 280-295[CrossRef][ISI]
Thomas B. R. V. S. Ford E. Pichersky L. D. Gottlieb 1993 Molecular characterization of duplicate cytosolic phosphoglucose isomerase genes in Clarkia and comparison to the single gene in Arabidopsis. Genetics 135: 895-905[Abstract]
Weeden N. F. L. D. Gottlieb 1979 Distinguishing allozymes and isozymes of phosphoglucose isomerases by electrophoretic comparison of pollen and somatic tissues. Biochemical Genetics 17: 287-296[CrossRef][ISI][Medline]
This article has been cited by other articles:
|
|
 |
|
  V. S Ford, J. Lee, B. G Baldwin, and L. D Gottlieb Species divergence and relationships in Stephanomeria (Compositae): PgiC phylogeny compared to prior biosystematic studies Am. J. Botany, March 1, 2006; 93(3): 480 - 490. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. D. Hoggard, P. J. Kores, M. Molvray, and R. K. Hoggard The phylogeny of Gaura (Onagraceae) based on ITS, ETS, and trnL-F sequence data Am. J. Botany, January 1, 2004; 91(1): 139 - 148. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||
