| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics and Phytogeography |
2Section of Evolution and Ecology, Division of Biology, University of California, Davis, California 95616 USA; 3Jepson Herbarium and Department of Integrative Biology, University of California, Berkeley, California 94720-2465 USA
Received for publication August 5, 2005. Accepted for publication December 2, 2005.
ABSTRACT
We present a maximum likelihood tree of 41 PgiC sequences for the monophyletic Stephanomeria, with 10 perennial and six annual species, widely distributed in western North America and exemplary of different speciation processes. The phylogenetic analysis represents the first use of PgiC sequences for Compositae. The annual species were originally delimited by biosystematic studies that provided evidence of their reproductive compatibility and chromosome structural homology. The perennial species are highly distinctive in morphology and have not been examined similarly. The PgiC tree provides more resolution than our previous ITS/ETS tree and reflects both past and ongoing hybridization and/or incomplete lineage sorting. Two major PgiC clades were resolved in Stephanomeria. One clade contains the genes from the annual species plus the perennial, insular endemic S. guadalupensis, which appears closely related to a monophyletic S. virgata. Stephanomeria exigua is not monophyletic. The second clade includes the genes from all other sampled perennial species and a monophyletic subclade of four genes from two annual species. The results are compared to previous studies, also using PgiC, of Clarkia (Onagraceae). Both molecular systematic and biosystematic approaches are essential to discern the very different courses of evolution in these two, well-studied genera of western North America.
Key Words: Asteraceae • biosystematics • Cichorieae • Compositae • PgiC • phylogeny • speciation • Stephanomeria
The western North American genus Stephanomeria Nutt., a member of the Cichorieae tribe of Compositae, contains 10 perennial and six annual species, and constitutes a well-supported, monophyletic group based on analysis of ITS and ETS nuclear rDNA sequences (Lee et al., 2002
). The species are widely distributed and are commonly encountered in many habitats from the western high plains to the California coast and from the Canadian border to the northern states of Mexico (Gottlieb, 2005). The ITS/ETS study clearly identified the clade of all stephanomerias with 100% bootstrap support and provided a sound basis for excluding from the genus two formerly problematic species, Munzothamnus blairii (Munz & I. M. Johnst.) P. H. Raven (= S. blairii Munz & I. M. Johnst.) and Pleiacanthus spinosus (Nutt.) Rydb. (= S. spinosa Nutt.). Within Stephanomeria, the study was less informative, placing all of the annual species and five perennial species into a single, largely unresolved clade. Four of those five perennial species were placed in a well-supported subclade, and two other perennials, the montane S. lactucina A. Gray and the coastal S. cichoriacea A. Gray, which differ in many morphological features, had an unexpected sister relationship. Because of the lack of resolution, previous results from intensive biosystematic studies on the annual species (Gottlieb, 1971
, 1972
; Gallez and Gottlieb, 1982
) could not be integrated with the sequence-based ITS/ETS analysis. To remedy this, we undertook the present study based on sequences of the nuclear gene PgiC. Stephanomeria becomes one of the first plant genera in which species relationships have been studied by biosystematic points of view, typical of the field from about 1950 to 1975, as well as by electrophoretic analysis of allozymes and phylogenetic analyses of ITS/ETS sequences and a well-characterized nuclear gene encoding protein.
The biosystematic studies of the annual stephanomerias were carried out because their morphological diversity had made it impossible to delimit species (Ferris, 1960
, p. 574). Species were not distinguished until strong correlations were discovered that linked populations into biological groups on the basis of shared morphological features and karyotypes, moderate to high fertility of experimental hybrids, and the number and structural homologies of chromosomes (Gottlieb, 1971
, 1972
). The most dissimilar annuals are the diploids S. exigua Nutt. and S. virgata Benth., both common and conspicuous in the summer flora of central and southern California. Many of the traits that distinguish them appear in different combinations in two other annual species. One is S. elata Nutt., their allotetraploid derivative, proved by cytogenetic analysis of experimental triploid hybrids between S. elata and both S. exigua and S. virgata (Gottlieb, 1969
). These three species form a classic polyploid complex in which a morphologically intermediate tetraploid blurs the recognition of differences between its diploid parents. The other intermediate species is the diploid S. diegensis Gottlieb, also derived from hybridization between S. exigua and S. virgata but without change in chromosome number (Gallez and Gottlieb, 1982
). Thus, the taxonomic complexity of the annual species appeared to be the consequence of the presence of two morphologically intermediate taxa, one diploid and one tetraploid, that had evolved following hybridization between a pair of divergent diploid species. Each derived species had a different mix of parental features and few if any novel ones and both grew in the same region as the parents. The parental species and the derived species appeared to intergrade because it was not previously known how to recognize them individually.
The current taxonomy assigns five subspecies to S. exigua and two to S. virgata (Gottlieb, 1972
). Within each species, the subspecies share numerous morphological features and have the same karyotype. Experimental hybrids within either species are significantly more fertile than hybrids between the species. The annual stephanomerias also include two other species, S. malheurensis Gottlieb and S. paniculata Nutt. Stephanomeria malheurensis is known only from a single site in eastern Oregon, and much evidence has been accumulated that it originated from the population of S. exigua subsp. coronaria (Greene) Gottlieb that grows at the same site (Gottlieb, 1973
, 1977a
, 1978a
, b
, 1979
; Gottlieb and Bennett, 1983
; Brauner and Gottlieb, 1987
, 1989
; Lee et al., 2002
). Stephanomeria paniculata, found from northern California to Oregon, Washington, and Idaho, may also have arisen from S. exigua subsp. coronaria because they share many morphological features and have relatively high reproductive compatibility (Gottlieb, 1969
).
All the annual species have a taproot and are similar in vegetative features; they are distinguished primarily by features of the cypselae and pappi. In contrast, the perennial species differ greatly in vegetative characters, for example, size, type of perennating stems (caudex or rhizome), aerial stem/branch architecture, leaf margins, and persistence and size of cauline leaves at anthesis. The perennials are broadly allopatric and are usually associated with different plant communities. Most occupy distinctive habitats; for example, S. fluminea Gottlieb grows only on raised cobble benches in the shifting gravel beds of creeks and rivers in northwestern Wyoming (Gottlieb, 1999
), S. guadalupensis Brandegee, endemic to Guadalupe Island, Mexico, grows in cracks in basaltic cliffs as well as on gravelly hillsides, S. pauciflora (Torr.) Nelson populates sandy washes in desert scrub communities and sites in the short-grass plains east of the Rocky Mountains, and S. lactucina is found in the yellow pine and red fir forests of the Sierra Nevada of California and north to the Oregon Cascades. All the perennials are diploid except S. parryi A. Gray, a tetraploid found in the Upper Sonoran zone of southeastern California and adjacent regions. In general, the perennial species are easy to identify. Reproductive relationships among them have not been investigated, and there has been little speculation about their phylogenetic relationships from a biosystematic perspective. The greater distinctiveness of the perennial stephanomerias and the widespread distribution of some of them suggested that they might be older as a group than the annual species.
At the outset of the present study, several questions were of interest. (1) Are the annual species as a group monophyletic and did they evolve from a particular perennial species? (2) What are the relationships among the perennial species? (3) Are the species S. exigua and S. virgata monophyletic? (4) Can particular subspecies of S. exigua or S. virgata be identified as parents of S. diegensis?
In this study, we address these and other questions by examining sequences of PgiC. PgiC encodes the cytosolic isozyme of phosphoglucose isomerase and is a very well characterized single-copy gene (Thomas et al., 1992
, 1993
; Gottlieb and Ford, 1997
, 2002
) that has proved valuable in phylogenetic studies of Clarkia Pursh (Gottlieb and Ford, 1996
; Ford and Gottlieb, 1999
, 2002
, 2003
) and related genera of Onagraceae (V. Ford and L. Gottlieb, unpublished data). For the present study, we initially designed "universal" primers to amplify large portions of the PgiC gene and later developed primers based directly on sequences from Stephanomeria.
MATERIALS AND METHODS
Species
Within Stephanomeria, seven of the 10 perennial species and three of the six annual species were examined (Appendix). The two tetraploid species, the annual S. elata and the perennial S. parryi, found in the southwestern deserts, were not included. Diploid annuals that were not examined are the self-pollinating species S. paniculata and S. malheurensis, whose relationship to S. exigua subsp. coronaria has already been well studied (cited in Introduction), and S. exigua subsp. carotifera (Hoover) Gottlieb, one of five subspecies of S. exigua. The two diploid perennials that were not examined are S. monocephala Moran, a rare species known from one or two sites in the Sierra San Pedro Mártir (Baja California), and S. thurberi A. Gray, from upland sites in the southwestern USA. Tissue of S. monocephala and S. thurberi was unavailable, but both were included in the previous ITS/ETS study (Lee et al., 2002
). Outgroup species were selected on the basis of the previously mentioned ITS/ETS study of Stephanomeria and a second ITS/ETS study that examined all primarily western North American genera of Cichorieae (Lee et al., 2003
). These studies showed 100% bootstrap support for a clade consisting of Stephanomeria together with Munzothamnus P.H. Raven, Rafinesquia Nutt., and Pleiacanthus (Nutt.) Rydb., all with a base chromosome number of n = 8, and Prenanthella (A. Gray) Rydb. with n = 7. Munzothamnus, Pleiacanthus, and Prenanthella are monotypic, and Rafinesquia is ditypic. PgiC sequences were obtained from the annual R. californica Nutt. and the perennial M. blairii.
|
; Ford and Gottlieb, 1999
). In general, the CTAB protocol was followed by an additional cleanup treatment with the Qiaex II protocol (Qiagen, Valencia, California, USA).
DNA fragments were amplified by PCR initially using a series of degenerate "universal" primers that were designed against highly conserved exon sequences of PgiC genes from Clarkia, Arabidopsis Heynhold (Thomas et al. 1993
), several other species in Onagraceae (V. Ford and L. Gottlieb, unpublished data) and EMBL sequences of maize, yam, and rice. PgiC has 23 exons and 22 introns; the structure was diagrammed in Thomas et al. (1993)
. The primers amplified regions between exons 11 and 16 and between 16 and 21 and are AA11F (TTY GCN TTY TGG GAY TGG GT) in exon 11, AA16F (ATG GAR AGY AAY GGN AAR GG) and its reverse complement AA16R in exon 16, and AA21RM (CCC CAY TGR TCR AAI GAR TTD ATI CCC CA) and yamv (TCI ACI CCC CAI TGR TCA AAI GAR TTI AT), both in exon 21. In the latter two primer sequences, which are partially overlapping, I represents inosine. AA11F is identical to a primer used in Terauchi et al. (1997)
, and yamv is similar to another primer described in that study. The primers against exons 11 and 16 appear to be "universal" in the sense that they also amplify PgiC fragments from a wide array of angiosperms including Aristolochia californica Torr. (Aristolochiaceae), Astragalus praelongus E. Sheld. (Leguminosae), and Spinacia oleracea L. (Chenopodiaceae) as judged by sequence analysis of the amplified fragments. Concentrations of the primers in the PCR reaction mix were increased 5–10 fold to compensate for their degeneracy (24–64 fold).
Once we had accumulated and sequenced PgiC fragments from several species of Stephanomeria, new nondegenerate primers SE15F (GCT CTT GAG AAG CTT GCT CC) in exon 15 and SE16R (CCT TTC CCA TTA CTC TCC ATG C) in exon 16 were designed. These Stephanomeria-specific primers were more consistently successful and made it possible to obtain the complete region between exons 11 and 21 by amplifying and combining two overlapping fragments between exons 11 to 16 and 15 to 21. One additional PCR primer, per15F (GCT TGC TCC CCA TAT TCA G), was designed specifically to obtain clones of allele 2A in S. exigua subsp. coronaria (seco2A). Four other Stephanomeria-specific primers in exons 13, 14, 18, and 20 were designed to use in sequencing: SE13F (GGT GCT AGA AGC ATT GAT CAG C), SE14R (CCA TAC GCT CAA CAA ACC TAG AAG), SE18F (CAC AAC CGG ATG CTC TCG), and SE20F (ATA GGC CAT CCC TTA GCC).
The PCR-amplified fragments were separated on agarose gels and excised with a razor blade. The DNA fragments were recovered with the ZymoClean Kit (Zymo Research, Orange, California, USA) and cloned with kits from Invitrogen (Carlsbad, California, USA). Sequencing was done as described elsewhere (Lee et al., 2002
), or by Davis Sequencing Co. (Davis, California, USA), or the DNA Sequencing facility of the Division of Biology, University of California at Davis, California. Two independent clones were sequenced for each fragment, and discrepancies (presumed PCR artifacts) were resolved by sequencing a third clone (available for 55 of 65 cloned fragments; standard ambiguity codes were used when a third clone was not available). Both strands of each fragment were sequenced. Accession numbers of all sequenced fragments are reported in the Appendix.
Sequence alignment and phylogenetic analysis
Sequences were aligned by eye, one exon or intron at a time. There were no indels in exons. The alignment of introns for all sequences was generally straightforward. One 152-nt region in intron 18 of the sequence from Munzothamnus was difficult to align and was treated as an insertion. Alignments obtained using ClustalW (Thompson et al., 1994
) were unsatisfactory because the program does not handle large insertions well and does not recognize that the splice sites at the ends of each intron are homologous. A preliminary analysis of exon data showed that there was very little phylogenetic information in the exons. Thereafter, the analysis was based on the complete aligned intron and exon data. The aligned matrix has 41 sequences and 6178 nt sites (Appendix S1, see Supplemental Data accompanying online version of this article).
Phylogenetic relationships were assessed using maximum likelihood (ML) and maximum parsimony (MP) methods performed using PAUP* version 4.0 (Swofford, 2001
) with gaps treated as missing. The most appropriate model for maximum likelihood analysis was determined to be the HKY + G model by Modeltest v. 3.6 (Posada and Crandall, 1998
) with the hierarchical likelihood ratio test. The ML search employed the HKY + G model, estimated all parameters, initiated swapping on the most parsimonious trees, and allowed unlimited tree-bisection-reconnection (TBR) branch swapping, with all other search conditions at default values. Bootstrapped maximum likelihood analysis consisted of 1000 replicates. Each replicate employed the HKY + G model, the parameters previously estimated by the ML search, initiated swapping on the NJ tree determined for that replicate, and allowed unlimited TBR branch swapping, with other options as before. A single ML tree was identified for all replicates except seven, for which two ML trees were obtained for each, and three others for which three trees were obtained for each.
Short ML analyses were performed with one sequence (sgua) or three sequences (sdie1, sdie2, sdie3) deleted to test hypotheses concerning the relationships of particular taxa. These ML searches were performed as before but cut short at 50 000 s. Bootstrapping consisted of 200 replicates and was otherwise as described in the previous paragraph, using the parameters determined from the corresponding ML tree.
Parsimony analysis consisted of a heuristic search with 10 replicate random additions of sequences, 10 trees held at each addition cycle, TBR branch swapping, and other default options. All replicates found the same 24 trees. Bootstrapped parsimony analysis consisted of 1000 replicates, each employing the same options as the initial parsimony search.
RESULTS
Overview of PgiC genes
The species examined have a single PgiC gene, a fact already known from previous electrophoretic studies of allozyme variation (Gottlieb, 1975
, 1977b
; Gallez and Gottlieb, 1982
). Thirty-nine PgiC genes were sequenced from Stephanomeria and one each from outgroups Rafinesquia and Munzothamnus (Appendix). Twenty-eight of the sequences, including the two from outgroups, were between exons 11 and 21 (approximately 3600 nt) and 13 sequences were between exons 15 and 21 (approximately 1800 nt). The exons between 11 and 21 all have exactly the same lengths as the corresponding exons in the PgiCs previously examined in Clarkia (Gottlieb and Ford, 1996
), Arabidopsis (Thomas et al., 1993
), and yam (Terauchi et al., 1997
), and all introns in the region occupy exactly the same positions as in the previously studied genes. The sequences between exons 11 and 21 have 697 exon nt, and those between exons 15 and 21 have 454 exon nt.
The inferred amino acid sequences of PgiC in these species are very highly conserved, and 30 of them are identical. In the non-identical sequences, nearly all amino acid substitutions are private (i.e., found only in single genes). Seven sequences have one private substitution, one (seco2a) has two, and mbla has three. Of the two non-private amino acid substitutions, one is shared by two genes, seco2c and seex, and the other is shared by three genes from two taxa in Stephanomeria (slac, sten1, sten2) plus Munzothamnus.
The exon sequences of PgiC are somewhat less conserved than the amino acids due to the occurrence of silent substitutions, although four pairs and one trio of sequences were identical in exons. Among the total of 697 exon positions, 612 are constant and 46 show only private variants. Thirty-nine of the third codon positions were "informative for parsimony," but a trial parsimony search limited to 150000 most parsimonious trees produced a consensus tree (not shown) that was completely unresolved except for four clades of two genes, three clades of three genes, and one clade of five genes, most of these groupings consisting of conspecific sequences. Thus, exon sequences provided little phylogenetically useful data.
All introns in these sequences have the 5'-GT...AG-3' boundaries typically found in plant PgiCs. Among the 10 introns examined, five are short with fewer than 105 bases, three are intermediate with 375–490 bases, and two are long with 588–680 bases. Intron 19 with 62 nt was invariant in length among the genes. The alignment of the genes from Stephanomeria, Rafinesquia, and Munzothamnus revealed nine large insertions from 128 to 405 nt in length, six deletions of more than 100 nt in length and numerous smaller indels, predominantly in three of the five longer introns, numbers 11, 18, and 20. Also, a 152-nt segment in intron 18 of Munzothamnus was treated as an insertion because it was difficult to align. Overall, the 10 large insertions increased the aligned length of the intron sequences by 2423 nucleotides to a total of 5481 nt. Of this total, with gaps treated as missing data, 4334 sites were invariant, 567 showed private differences only, and 580 were "informative for parsimony," more than 14 times the number in exons. Thus, the introns provided nearly all the phylogenetically useful information for this study.
Although most indels were private or shared by closely related sequences, there was evidence of repeated insertions and deletions of possibly related sequences in intron 11 in all the Stephanomeria genes but not in the outgroup sequences. These sequences are flanked by the same or similar direct repeats and inverted repeats, suggesting control by a particular transposon. Most of the nine large insertions were flanked by direct and/or inverted repeats, also suggestive of transposon activity. There were also many repetitions of variants of AGCT in intron 14. Other than those in intron 11, all the major insertions and deletions occur in the annual taxa or Munzothamnus, except an insertion of 283 nt flanked by 17-nt direct repeats in intron 12 of S. fluminea. However, this finding may simply reflect the larger number of sequences sampled from annuals. Although many indels were evidently shared by related sequences, not all could be so clearly interpreted. Gaps were treated as missing data and no attempt was made to code up the indels for phylogenetic analysis.
Maximum likelihood tree
The maximum likelihood tree for the 41 PgiC sequences from Stephanomeria and the two outgroups with estimated branch lengths and bootstrap support for each clade is displayed in Fig. 1. Both outgroup rooting, based on results of Lee et al. (2002)
, and midpoint rooting of the ML tree place the PgiC sequence from Munzothamnus as sister to all the others, and the PgiC from Rafinesquia as sister to all genes from Stephanomeria. The midpoint rooting reflects the fact that, in pairwise comparisons, the gene from Rafinesquia differs about as much from those of Stephanomeria as the most diverged genes of Stephanomeria differ from each other, while the gene from Munzothanmus differs about 2–3 fold more (data not shown). Congruence between the midpoint rooting and the consensus of most parsimonious ITS/ETS trees (Lee et al., 2002
) provides additional support for the removal of Munzothamnus from Stephanomeria.
|
Within clade A, none of the three annual species is represented by a single monophyletic clade that enjoys high bootstrap support. The clade V of sequences from S. virgata has a modest 60% support, and the sequences from each of its two subspecies, subsp. pleurocarpa (Greene) Gottlieb (VP) and subsp. virgata (VV), are polyphyletic. A heterozygous individual of VV provided one allele, svvi3a, that is assigned to a clade of genes from VP while the other, svvi3b, is in the clade with the other genes from VV. Two genes from VP, svpl1a and svpl4b, form a separate clade that is sister to the rest of V while two other genes from the same populations, svpl1b and svpl4a, are assigned to the main clade of VP. The clade D of sequences from S. diegensis has only 43% support because equal bootstrap support (44%) was given to a conflicting clade consisting of two of the three sequences from S. diegensis (sdie2, sdie3) plus the two sequences from S. exigua subsp. deanei (MacBride) Gottlieb (ED).
The sequences from S. exigua do not form a clade. Three of the subspecies, subspp. macrocarpa Gottlieb (EM), coronaria (EC), and exigua (EE), form one clade, although with only 72% support. The sequences from subsp. deanei (ED) are in a weakly supported clade with S. diegensis, S. virgata, and S. guadalupensis. The possibility that the position of ED was distorted by the presence of the sequences from S. diegensis (which might be its derivative, see Discussion) was tested by deleting D from the analysis. The resulting ML tree (not shown) has the same topology as shown in Fig.1 with clade D deleted. The bootstrap support for the clade ED + G + V increased to 88%, suggesting that subsp. deanei is really more closely related to S. virgata than to the other subspecies of S. exigua.
Also within clade A, the PgiC of the perennial S. guadalupensis (G) is sister to clade V of S. virgata, with ML bootstrap support value of 89%. In addition to the data used for ML analysis, the relationship between the species is further supported by a landmark 6-nt insertion in intron 20 shared only by them. If sequence sgua is deleted from the analysis, the resulting tree has the same topology shown in Fig.1 with G deleted, and the bootstrap support for clade V increases from 60 to 86%. This increase suggests that the low support for clade V in the full analysis resulted from the similarity of the sequences of S. virgata and S. guadalupensis.
Within clade P, there is a strongly supported subclade (100% bootstrap) that includes the genes from four perennials: S. tenuifolia (Raf.) H. M. Hall, S. fluminea, S. runcinata Nutt., and S. pauciflora. There is moderate support for a sister relationship between S. fluminea and S. pauciflora and also for a sister relationship between them and S. runcinata. The trio in turn is sister to S. tenuifolia (T). The relationship of the quartet to S. cichoriacea and S. lactucina is not clear. There is moderate support for a sister relationship between S. cichoriacea and the previously mentioned clade 4A of genes from annuals.
Many "landmarks" distinguish the A and P clades, including 50 nt substitutions and eight indels. Curiously, 43 of the substitutions and all of the indels occur in intron 20. Their concentration in intron 20 could be a chance result of its great length, c. 590 to 700 nt, depending on the sequence, but also this concentration could have arisen if they resulted from coordinated rather than independent events. Alternatively, upstream differences between sequences in the A and P lineages may have been obscured by crossing over among alleles in an ancestral gene pool. Two pairs of alleles that were obtained from single heterozygous individuals (seco2a, seco2b from S. exigua subsp. coronaria and svvi2a, svvi2b from S. virgata subsp. virgata) have members in both the A and P clades.
Parsimony analysis
The strict consensus of the 24 most parsimonious trees (not shown) has several unresolved trichotomies but is compatible with the ML tree (Fig. 1), except that the pair of sequences from S. exigua subsp. macrocarpa (EM) appears as sister to all the other sequences in clade A. The same clades receive the strongest bootstrap support in both methods of analysis; i.e., parsimony bootstrap gives 98–100% support to every clade that has 98–100% support from likelihood bootstrap analysis and vice versa. Clades that received higher support from likelihood analysis include EM + EC + EE (72 vs. 30), EC + EE (97 vs. 89), G + V (89 vs. 46), V (60 vs. 40), the subclade of V consisting of all sequences other than svpl1a and svpl4b (87 vs. 47), VV (94 vs. 89) and P (97 vs. 93). Parsimony analysis gave slightly higher support to svvi2b + svvi2e (92 vs. 88) and svpl1b + svpl2 + svpl3 + svpl4a (77 vs. 65) as well as several subclades of the P clade: 4A + C (90 vs. 83), T (90 vs. 85), sflu + spau (87 vs. 83), srun + sflu + spau (97 vs. 87).
Branch lengths
Unsurprisingly, a likelihood-ratio test for rate constancy of molecular evolution across lineages (Felsenstein, 1988
; see Sanderson, 1998
) showed that divergence in the ML PgiC tree is not clocklike; i.e., the likelihood of the tree as calculated without the constraint of clocklike divergence is significantly greater than the likelihood calculated with a molecular clock enforced (-ln L = 19022.64 vs. 19118.15, df = 39, G = 191, P < 0.001). Although there is evident variation in branch length among sequences from both annuals and perennials, it is notable that branch lengths for perennials overall are shorter than those for annuals. It is particularly noticeable that the branch length to the perennial S. guadalupensis is very short even though it is located among the sequences of annuals in clade A. Also, the four sequences (clade 4A) from annual species located within the perennial clade P are conspicuously long.
Informal inspection of the aligned sequences suggests that the long branch to seco2a reflects a particularly large number of private gene substitutions, 58, larger than in any other sequence except rcal, with 69 private substitutions, and mbla, with 159. It should be noted that some branch lengths may be slightly misrepresented because about a third of the genes were not sequenced in the region between exons 11 and 15.
DISCUSSION
The maximum likelihood tree derived from PgiC sequences provides considerably more resolution than the ITS/ETS tree (Lee et al., 2002
), but there are many inconsistencies between the two. We start our discussion by taking the results of the PgiC study at face value and discussing their implications for specific issues of interest. The availability of the two sequence-based studies and the earlier biosystematic studies provides an unusual opportunity to assess the usefulness of molecular systematics vs. biosystematics for understanding species divergence and species relationships.
The PgiC tree suggests that Stephanomeria consists of two monophyletic clades, the annual species plus the perennial S. guadalupensis in one clade and all other perennial species in the other. None of the annual species appears to have evolved from any extant perennial, and annuality probably originated only once in this genus or may even be the ancestral condition. The growth form of the common ancestor of stephanomerias is unknown and may remain so because the other four genera in the "Stephanomeria clade" are equally divided between annuals and perennials; their relationship to each other and to the other clades of North American Cichorieae are not yet well understood (Lee et al., 2003
). The possibility that the annual and perennial species might be sister clades was not revealed by the ITS/ETS study (Lee et al., 2002
). That study showed a clade of two perennial species as sister group of a mostly unresolved assemblage of the remaining perennial species and all of the annuals. Future analysis of additional, independent genes may provide definitive conclusions.
In the PgiC tree, the branch lengths to the perennial species, including the one to S. guadalupensis, are shorter than those to the annual species, as in some other angiosperm lineages (e.g., Andreasen and Baldwin, 2000). This may reflect the fact that the perennials pass through fewer generations than the annuals in the same length of time (but see Whittle and Johnston, 2003
) or other biological differences between annuals and perennials (cf. Andreasen and Baldwin, 2000). The ITS/ETS parsimony analysis did not provide estimates of branch lengths, but pairwise comparisons showed less divergence among the annual sequences, with an average of 11 differences (maximum 20), than among the perennials (average 31, maximum 49). This inconsistency will be discussed further below.
Perennial species
The PgiC tree suggests strongly that the perennial S. guadalupensis evolved from annual ancestors because it is well embedded among the annuals in clade A and is particularly close to S. virgata, a species with many populations located along the southern Californian coast. Bootstrap support for their sister relationship is 89%. If this resolution accurately reflects phylogeny, then the perenniality of S. guadalupensis must have evolved independently of that of all other perennial stephanomerias. The species has a woody caudex, and its wood anatomy resembles that of many other species of rosette plants that have evolved woodiness after colonizing oceanic islands (Carlquist, 1960
; see Baldwin et al., 1998
). This possibility was not suggested by the ITS/ETS tree, which placed S. guadalupensis and another Baja Californian perennial, S. monocephala (not studied here), in unresolved positions among the perennial species but outside the well-supported ITS/ETS clade that contained all the annuals and five perennials (S. fluminea, S. pauciflora, S. runcinata, S. tenuifolia, and S. thurberi) (Lee et al., 2002
). Reconciling the incongruent placements of S. guadalupensis in the PgiC and ITS/ETS trees will require another line of phylogenetic data.
The six other perennial species form clade P (with the inclusion of subclade 4A, discussed below). Among them, S. tenuifolia, S. fluminea, S. runcinata, and S. pauciflora had a strongly supported relationship (100% bootstrap). The former three species were also linked in a well-supported clade (with S. thurberi, not studied here) in the previous ITS/ETS study. Each of the four species has few florets per head, a key characteristic distinguishing them from the other perennials. The former three species perennate via creeping rhizomes and have fully plumose pappus bristles, whereas S. pauciflora grows from a deeply seated caudex and has partially plumose pappus elements. With such different morphologies, even the moderate support (83%) for a sister relationship between any of the former three species, in this case S. fluminea, and S. pauciflora was unexpected. The ITS/ETS study had S. fluminea as sister to S. runcinata with very high support (98%). Representatives of three populations of the particularly widespread and variable S. tenuifolia (discussed in Lee et al., 2002
) formed a single PgiC clade although only moderately supported. Both S. lactucina and S. cichoriacea have fully plumose pappus bristles but the former is rhizomatous, while the latter has a caudex. The relationship of these two species on the PgiC tree is not strongly supported but is not in significant disagreement with the unexpected conclusion of the ITS/ETS study that they are sister species (91% support). There has not been any morphological phylogenetic study of stephanomerias, and nothing is known about the genetic complexity of hypogeal stem structure or other differences among the species nor how they might be modified. Much firmer evidence of the relationships of the perennial species is required before these or other morphological characters can be "mapped" onto a well-supported tree.
Annual species
The annual species were the subject of biosystematic studies (Gottlieb, 1971
, 1972
) that identified the biological boundaries of individual species, thereby ending taxonomic confusion in the group. The species were delimited primarily by making a very large number of interpopulation hybrids, with no presumption regarding to which species a population might belong, and then correlating their fertility with the extent of morphological similarity, karyotype, chromosome number, and structural homology. The studies also showed that past hybridization between S. exigua and S. virgata had given rise to the diploid S. diegensis and the tetraploid S. elata. When these studies were done, questions about phylogenetic relationships and degrees of closeness were rarely articulated. It was sufficient to make use of correlated similarities and differences along with evidence of the extent of reproductive isolation, when available, to recognize species.
1. Stephanomeria virgata
For the present study, six PgiC sequences were obtained from S. virgata subsp. pleurocarpa (VP) and 10 from subsp. virgata (VV). Thirteen of the sequences form a single clade (V). The other three were placed in subclade 4A of clade P and are discussed separately below. The relatively low support (60%) for clade V results from the similarity of the sequences from S. virgata to the one from S. guadalupensis. This was suggested by the 89% support for the combined G + V clade and further demonstrated by deleting the sequence of S. guadalupensis from the analysis, after which support for clade V rose to 86%. Thus S. virgata likely constitutes a monophyletic group among the annuals. There is little support for a relationship of sgua to particular sequences from V (at most 3.5% support for sgua with one subset of V; not shown). Hence, S. virgata and S. guadalupensis may be sister species, but the origin of S. guadalupensis from within S. virgata is a possibility. This putative relationship was not apparent from the ITS/ETS sequences, as noted above.
Within clade V, six of the seven genes from subsp. virgata (VV) form a well-supported subclade, but the seventh (svvi3a) was placed in one of the two subclades of subsp. pleurocarpa (VP). This result is consistent with numerous observations (Gottlieb, 1969
) that the subspecies hybridize and exchange genes wherever they are sympatric. However, although gene flow occurs between the subspecies, their genes are not broadly intermixed, consistent with their recognition as distinct taxa. The basal position of two genes (svpl1a, svpl4b) from VP within V might indicate that subsp. virgata was derived from within subsp. pleurocarpa but, given the relatively limited sampling and ongoing hybridization and gene flow, this conclusion seems premature. In this case, the biosystematic evidence of the subspecies' reproductive compatibility clearly guides interpretation of the molecular tree. Without this evidence, the PgiC tree would likely be interpreted as demonstrating incomplete lineage sorting in the two subspecies of S. virgata, which in fact is only one of two possibilities. Thus, the "pleurocarpa-like" svvi3a gene might only recently have been brought into subsp. virgata from subsp. pleurocarpa or, alternatively, it might be a relictual gene that arose prior to the divergence of the subspecies.
2. Stephanomeria exigua
Four of the five subspecies of S. exigua were sampled in the present study. Each subspecies except subsp. coronaria was monophyletic, but the number of sequences sampled was insufficient to test for gene flow among the subspecies as predicted by field observations of hybridization (Gottlieb, 1971
). Subspecies coronaria and exigua form a very strongly supported subclade (bootstrap value 97%) that is weakly tied (bootstrap value 72%) to subsp. macrocarpa. Subspecies deanei is placed in a separate clade along with both S. diegensis and S. virgata, although with weak support (62%). Thus, the genes sampled from S. exigua do not form a monophyletic clade that corresponds to the species that was identified from the biosystematic studies. The ITS/ETS study showed strong support for a clade consisting of subspp. exigua and deanei and S. diegensis, but insignificant support for any other relationship among the subspecies of S. exigua.
The inclusion of subsp. exigua within the clade of sequences of subsp. coronaria is particularly interesting because subsp. coronaria varies morphologically toward subsp. exigua, particularly in the populations from the eastern slopes of the Sierra Nevada in California and north into eastern Oregon and southwestern Idaho. Subspecies exigua is native to the nearby Mojave Desert and south to the Sonoran desert. The resemblance is most evident in the architectural arrangement of the heads along the branches. In central and southern California, the heads of subsp. coronaria are borne singly or clustered on very short peduncles, but in eastern California and to the north and east the heads are on long, sometimes branched peduncles that resemble the paniculiform arrays along the branches of subsp. exigua. This convergent morphology has been presumed to result from hybridization between the subspecies where they make contact at intermediate and low elevations in the Sierra and other mountains, but direct evidence is lacking. Additional sequences are required to provide more complete evidence regarding their relationship and hybridization.
3. Stephanomeria diegensis
The annual S. diegensis is thought to have arisen following hybridization between S. exigua and S. virgata (Gallez and Gottlieb, 1982
; Gottlieb, 1971
). At the time of the cited studies, it was not apparent which subspecies of S. exigua and S. virgata were the best candidates for parentage. The origin of S. diegensis may have been made possible because S. exigua and S. virgata are similar genetically even though they differ in a number of morphological and other features. Their genetic identity (I = 0.81, based on electrophoretic variability at 20 loci encoding allozymes) is significantly higher than that of the average pair of diploid plant species (I = 0.67; Gallez and Gottlieb, 1982
). A high genetic identity suggests that their hybrid progenies may contain a higher proportion of viable recombinants than is typical in progenies of interspecific hybrids.
In principle, because S. diegensis is diploid, the PgiC study could detect evidence of both parents only if the sequenced region of the PgiC gene that became established in S. diegensis contained portions of the genes from both parents; i.e., was recombinant. That would be improbable in the first place and, even then, perhaps recognizable only by comparison to gene sequences sufficiently similar to those from the actual parental individuals. Molecular evidence for the hybridity of S. diegensis will eventually emerge from analyses of multiple genes when some of them are placed in clades with genes of one parent and others are placed in clades with genes of the other parent. Thus, appropriate evidence will reflect additivity of parental genes rather than intermediacy.
The PgiC tree shows S. diegensis in a weakly supported (62%) trichotomy with both ED (S. exigua subsp. deanei) and V + G (S. virgata and S. guadalupensis). Repeating the analysis without the sequences from S. diegensis yielded a ML tree with unchanged topology (less clade D) but increased bootstrap support (88%) for the clade ED + V + G. The test was made to see whether the PgiC genes from S. diegensis might have recombinant features that misleadingly linked ED and V into a single clade. Instead, it appears that when the genes of S. diegensis are removed from the analysis, the case that subsp. deanei is more closely related to S. virgata than to S. exigua is strengthened.
If the PgiC of S. diegensis was inherited from its S. virgata parent, then that parent was actually an ancestral species antedating the divergence of S. virgata and S. guadalupensis. Alternatively, if the PgiC of S. diegensis was inherited from its S. exigua parent, then that parent was subsp. deanei. The possibility that subsp. deanei is one parent of S. diegensis is also concordant with the placement of S. diegensis along with subspp. deanei and exigua in a very strongly supported clade (98% bootstrap) in the ITS/ETS tree.
Occurrence of genes from annual species in the perennial clade
Four PgiC genes from annual species are more closely related to the PgiC genes of perennial species than to other genes from annual species. Three of these are from S. virgata subsp. virgata and one from S. exigua subsp. coronaria. They form subclade 4A in clade P (100% support). Their presence suggests either incomplete lineage sorting since the divergence of the annual and perennial lineages or, alternatively, is evidence of hybridization and introgression from the perennial lineage into annual species. However, hybrids between annual and perennial stephanomerias have never been found. Because the four perennial-type genes in the annuals have a common origin, any putative hybridization must have occurred before the divergence of S. exigua subsp. coronaria and S. virgata, a time corresponding to the base of the A clade. On either hypothesis, it appears that a "perennial-type" PgiC lineage has been retained in the annual species for a long time.
It is difficult to discriminate between incomplete lineage sorting vs. ancient hybridization. In fact, if the early populations carrying the A and P PgiC lineages experienced occasional hybridization as now observed between S. exigua and S. virgata, the two hypotheses may not be significantly different. One line of inquiry could involve identification and sequencing of other genes linked to PgiC. If the A and P lineages diverged before the hybridization event, the hybridization hypothesis predicts that introgressed perennial-type PgiC genes would have been linked to perennial-type alleles of other genes, a situation for which evidence might still be obtainable. However, if sufficient time has passed since such putative hybridization, there might be no evidence of such a linkage association unless it was maintained by conferring a fitness advantage. Also, finding such a linkage group does not definitively rule out the "incomplete lineage sorting" hypothesis because the same linkage group may have also occurred in the ancestral stephanomerias.
Both hypotheses predict that PgiC genes assignable to clade P will be discovered in additional individuals of both S. virgata and S. exigua and possibly even S. diegensis. They are not expected to be rare because the four genes of clade 4A were found in a rather small sample. Such additional genes may all belong to clade 4A. However, the discovery of a second "perennial-type" lineage (within clade P but disjunct from clade 4A) would be consistent with either hypothesis. In either case, it would be an interesting result, increasing our knowledge of the diverse outcomes that may be observed when gene lineages are retained while species and populations diverge.
Speciation in Stephanomeria and Clarkia
Biosystematic studies and phylogenetic analyses of PgiC from the annual species of Stephanomeria provide a very different view of species divergence and relationships than discovered by similar analyses carried out in Clarkia, a genus that contains only annual species (Lewis, 1953
, 1962
, 1973
; Lewis and Lewis, 1955
; Gottlieb and Ford, 1996
; Ford and Gottlieb, 2003
). In Stephanomeria, the annual species are not completely reproductively isolated by genetic or chromosomal structural differences and are able to form hybrids that have low but not zero fertility wherever they grow sympatrically. The taxa show an increasing gradation of divergence from subspecies to species, measured as morphological difference, karyotypic difference, chromosomal structural difference, or fertility of experimental F1 hybrids. The origin of species in Stephanomeria evidently involved much genetic divergence but, on the basis of the electrophoretic analysis of allozyme variation (Gallez and Gottlieb, 1982
), the divergence was limited and did not prevent interspecific hybridization from producing derivative species.
Lewis amalgamated into Clarkia five morphologically and chromosomally diverse species groups originally recognized as genera, along with a few species transferred from other genera. The species were placed into sections on the basis of shared chromosome numbers and morphological similarity (32 of the 42 species are diploid). The species are completely reproductively isolated (except C. concinna and C. breweri). Natural hybridization is infrequent and, at the diploid level, has not played any role in their divergence. Very closely related species in the same section often look very much alike (and have the same allozymes in similar frequencies), yet differ strikingly in chromosome number and structural homology (seen as reciprocal translocations and inversions). Many diploid species in the group are thought to have arisen rapidly and abruptly from a parental species and, initially, with no adaptations not already present in their parents.
The phylogenetic trees of PgiC genes from Clarkia and the annual species of Stephanomeria reflect these differences. In Clarkia, it was possible to construct trees for two PgiC genes because the locus was duplicated at or near the origin of the genus. All species have retained the PgiC1 gene and many have an active or identifiable pseudo PgiC2. Both phylogenetic trees had concordant topology and placed the sections (and, for the most part, all the species assigned by Lewis to each of the sections) into a series of monophyletic clades, many enjoying very high bootstrap support, especially on the PgiC1 tree. Each species was represented in these studies by a single PgiC1 gene and, when available, a single PgiC2. The genetic separation of the species suggested that sampling one sequence per taxon would be sufficient and the concordance of the trees for the two genes justified this procedure.
In Stephanomeria, the phylogenetic tree of the annual species is much more complex, the species and subspecies are not clearly monophyletic, and bootstrap support for many clades is lower. Species divergence among annual stephanomerias at both the morphological and molecular levels appears to be very much slowed by past and ongoing hybridization and introgression. This may have different consequences for single copy genes like PgiC vs. multicopy regions such as the ITS and ETS. For PgiC, hybridization presumably offsets the pruning of gene lineages by lineage sorting, occasionally introducing new alleles to a species (or subspecies) or restoring alleles that had been lost. Nucleotide substitutions and indels would continue to accumulate in each lineage, so distances between typical alleles of diverging taxa would continue to increase, but some alleles would be closely related to those of another taxon. Recombination with transferred alleles could create reticulate relationships seemingly crossing taxon boundaries and complicating the attempt to represent sampled genes on a bifurcating tree. The lower bootstrap values in Stephanomeria compared to Clarkia likely reflect these biological differences although they may also in part be an artifact of the different style of sampling, with more representatives per taxon in the Stephanomeria sample.
For the multicopy ITS and ETS regions, the ongoing hybridization in annual stephanomerias might result in "correction" of sequences of the recurrent parent by newly introgressed sequences (or vice versa). In this case, substitutions and indels characterizing one lineage might be obliterated by the introduction of another lineage, suppressing the development of phylogenetically informative differences between species or subspecies. This might account for the low amount of difference and lack of phylogenetic information among the ITS/ETS sequences sampled from annual species. Because "correction" processes for multicopy genes can act fairly quickly, it should be possible to test this possibility by making artificial F1, F2, and later generation hybrids between S. exigua and S. virgata or between subspecies of either to examine changes in the ITS/ETS sequences present (e.g., Fuertes Aquilar, Roselló and Neito Feliner, 1999
). Such analysis may be complicated by differences in the karyotypes of the two species; in S. exigua two chromosomes, a large and a small one, carry a nucleolar-organizing region, but in S. virgata and S. diegensis these regions are found on two small chromosomes (Gottlieb, 1971
). The segregation of these chromosomes will likely affect correction processes in interspecific crosses.
Molecular systematics and biosystematics
The PgiC phylogenetic tree of Stephanomeria has provided valuable information about relationships within the group that were not and could not have been detected by the original biosystematic studies. Most obviously, it showed that the annual species descended from a common annual ancestor, in keeping with the ITS/ETS findings (Lee et al., 2002
) but substantially extending those results, which were not conclusive about relationships of the annuals. Whether the annual habit substantially predates the most recent common ancestor of the annual species remains unknown, based on both equivocal reconstruction of the ancestral habit for Stephanomeria from the PgiC tree topology as well as conflict between the PgiC and ITS/ETS tree with respect to relationships among the perennials. Second, it showed that S. guadalupensis, a perennial species endemic to Guadalupe Island, Mexico, may have evolved within the clade of annuals and, if so, probably from a common ancestor with S. virgata. Conflict between the placements of S. guadalupensis in the PgiC tree and in the ITS/ETS tree (Lee et al., 2002
) may reflect ancient hybridization or lineage sorting or both. Third, the PgiC tree indicated that four of the perennial species, each with many distinctive morphological features, form a monophyletic clade, in agreement with previous molecular phylogenetic results (Lee et al., 2002
). Fourth, it showed that there is a clade of perennial-type genes obtained from the genomes of the annual species that either predates the separation of the extant annual and perennial species or represents the residue of past hybridization. Fifth, it confirmed that there is some gene flow between the subspecies of S. virgata, consistent with the absence of reproductive isolation between them, but that S. virgata is probably a monophyletic group. Sixth, it showed that the subspecies of S. exigua may not form a monophyletic group.
The evidence of hybridization and incomplete lineage sorting emphasize the need for multiple lines of molecular evidence to understand species relationships and show the value of getting enough evidence from each sequenced gene or region to determine whether they support significantly different phylogenies. The incongruities between the PgiC and ITS/ETS phylogenies should not be seen as problems standing in the way of a neat characterization of phylogenetic relationships among species, but rather as phenomena of importance in their own right, illustrating the fact that divergence is often not a neat process and different genes reflect that process in different ways.
Biosystematic studies provided a consistent framework for the delimitation of the annual species of Stephanomeria by sorting populations into groups based on morphological and karyotypic similarity and hybrid fertility. Such considerations of reproductive isolation and the possibility of gene flow constitute one of the most important differences between biosystematics and molecular systematics. Biosystematics was grounded in an evolutionary context that attempted to assess whether taxa are genetically independent, whereas present-day molecular systematics is concerned with the history of divergence. The former asks whether certain plants, regardless of the extent of their similarity or difference, represent different species. The latter is not directly concerned with speciation per se and, in fact, often avoids the issue by dealing with well-diverged taxa above the species level. However, Baldwin (1995)
and Verne Grant (1998
, 2003
) have offered different perspectives on the importance of integrating molecular systematics and biosystematics at higher taxonomic levels.
The different points of view are typified by the information now available regarding S. exigua. Biosystematic studies identified many similarities and reproductive compatibilities among the groups that were designated as subspecies of S. exigua, including subsp. deanei. All appeared equally different from S. virgata. Now, the analysis of PgiC sequences suggests subsp. deanei may be more closely related to S. virgata than to the other subspecies of S. exigua. If supported by other molecular evidence, this conclusion presumably would mean that S. virgata arose from a common ancestor with subsp. deanei, but has diverged sufficiently that neither morphology nor crossing experiments now reveal that relationship. One phylogenetic point of view would argue that subsp. deanei would merit species status because of its closer relationship to S. virgata. On the other hand, the biosystematic view is that subsp. deanei remains genetically interconnected with the other subspecies of S. exigua regardless of the fate of S. virgata and therefore should retain its subspecific designation, just as S. exigua subsp. coronaria retains its status despite being the parent of S. malheurensis. Clearly, both biosystematics and molecular phylogeny are essential to understand plant evolution and speciation.
FOOTNOTES
1 The authors thank M. L. Sanderson and M. M. McMahon for discussion of phylogenetic methods and for access to computing resources. This research was supported in part by the Lawrence R. Heckard Endowment Fund of the Jepson Herbarium, U.C. Berkeley. 
4 Author for correspondence (e-mail: bbaldwin{at}berkeley.edu
)
5 Present address: Korea Research Institute of Bioscience and Biotechnology, Daejeon 305-333, South Korea
LITERATURE CITED
Andreasen K Baldwin B. G. 2001. Unequal evolutionary rates between annual and perennial lineages of checker mallows (Sidalcea, Malvaceae): evidence from 18S-26S rDNA internal and external transcribed spacers. Molecular Biology and Evolution 18: 936-944.
Baldwin B. G. 1995. A new prospect for California botany: integrating biosystematics and phylogenetics. Madroño 42: 154-167.
Baldwin B. G Crawford D. J Francisco-Ortega J Kim S.-C Sang T Stuessy T. F. 1998. Molecular phylogenetic insights on the origin and evolution of oceanic island plants. In D. E. Soltis, P. S. Soltis, and J. J. Doyle [eds.], Molecular systematics of plants II: DNA sequencing, 410–441. Kluwer, Boston, Massachusetts, USA.
Brauner S Gottlieb L. D. 1987. A self-compatible plant of Stephanomeria exigua subsp. coronaria (Asteraceae) and its relevance to the origin of its self-pollinating derivative Stephanomeria malheurensis. Systematic Botany 12: 299-304.[CrossRef][ISI]
Brauner S Gottlieb L. D. 1989. Response to selection for time of bolting in Stephanomeria exigua subsp. coronaria and implications for the origin of S. malheurensis Asteraceae). Systematic Botany 14: 516-524.[CrossRef][ISI]
Carlquist S. 1960. Wood anatomy of Cichorieae (Compositae). Tropical Woods 112: 65-91.
Felsenstein J. 1988. Phylogenies from molecular sequences: inference and reliability. Annual Review of Genetics 22: 521-565.[CrossRef][ISI][Medline]
Ferris R. S. 1960. Illustrated flora of the Pacific states, vol. IV, Bignoniaceae to Compositae Stanford University Press Stanford, California, USA.
Ford V. S Gottlieb L. D. 1999. Molecular characterization of PgiC in a tetraploid plant and its diploid relatives. Evolution 53: 1060-1067.[CrossRef][ISI]
Ford V. S Gottlieb L. D. 2002. Single mutations silence PgiC2 genes in two very recent allotetraploid species of Clarkia. Evolution 56: 699-707.[CrossRef][ISI][Medline]
Ford V. S Gottlieb L. D. 2003. Reassessment of phylogenetic relationships in Clarkia sect. Sympherica. American Journal of Botany 90: 284-292.
Fuertes Aquilar J Rosselló J. A Nieto Feliner G. 1999. Nuclear ribosomal DNA (nrDNA) concerted evolution in natural and artificial hybrids of Armeria (Plumbaginaceae). Molecular Ecology 8: 1341-1346.[CrossRef][Medline]
Gallez G. P Gottlieb L. D. 1982. Genetic evidence for the hybrid origin of the diploid plant Stephanomeria diegensis. Evolution 36: 1158-1167.[CrossRef][ISI]
Gottlieb L. D. 1969. The role of hybridization in the evolution of the annual species of Stephanomeria (Compositae) Ph.D. dissertation, University of Michigan Ann Arbor, Michigan, USA.
Gottlieb L. D. 1971. Evolutionary relationships in the outcrossing diploid annual species of Stephanomeria (Compositae). Evolution 25: 312-329.[CrossRef][ISI]
Gottlieb L. D. 1972. A proposal for classification of the annual species of Stephanomeria (Compositae). Madroño 21: 463-481.
Gottlieb L. D. 1973. Genetic differentiation, sympatric speciation and the origin of a diploid species of Stephanomeria. American Journal of Botany 60: 545-553.[CrossRef][ISI]
Gottlieb L. D. 1975. Allelic diversity in the outcrossing plant Stephanomeria exigua subsp. carotifera. Evolution 29: 213-225.[CrossRef][ISI]
Gottlieb L. D. 1977a. Genotypic similarity of large and small individuals in a natural population of the annual plant Stephanomeria exigua subsp. coronaria. Journal of Ecology 65: 127-134.[CrossRef]
Gottlieb L. D. 1977b. Phenotypic variation in Stephanomeria exigua ssp. coronaria (Compositae) and its recent derivative species "malheurensis.". American Journal of Botany 64: 873-880.[CrossRef][ISI]
Gottlieb L. D. 1978a. Allocation, growth rates and gas exchange in seedlings of Stephanomeria exigua ssp. coronaria and its recent derivative S. malheurensis. American Journal of Botany 65: 970-977.[CrossRef][ISI]
Gottlieb L. D. 1978b. Stephanomeria malheurensis, a new species from Oregon. Madroño 25: 44-46.
Gottlieb L. D. 1979. The origin of phenotype in a recently evolved species. In O. T. Solbrig, S. Jain, G. B. Johnson, and P. H. Raven [eds.], Topics in plant population biology, 264–286. Columbia University Press, New York, New York, USA.
Gottlieb L. D. 1999. A new species of Stephanomeria from northwestern Wyoming. Madroño 46: 58-60.
Gottlieb L. D. 2004. Rethinking classic examples of recent speciation in plants. New Phytologist 161: 71-82.[CrossRef][ISI]
Gottlieb L. D. In press. Stephanomeria. In Flora of North America Editorial Committee Flora of North America, vol. 19 Oxford University Press New York, New York, USA.
Gottlieb L. D Bennett J. P. 1983. Interference between individuals in pure and mixed cultures of Stephanomeria malheurensis and its progenitor. American Journal of Botany 70: 276-284.[CrossRef][ISI]
Gottlieb L. D Ford V. S. 1996. Phylogenetic relationships among the sections of Clarkia (Onagraceae) inferred from the nucleotide sequences of PgiC. Systematic Botany 21: 45-62.
Gottlieb L. D Ford V. S. 1997. A recently silenced, duplicate PgiC locus in Clarkia. Molecular Biology and Evolution 14: 125-132.[Abstract]
Gottlieb L. D Ford V. S. 2002. The 5' leader of plant PgiC has an intron: the leader shows both the loss and maintenance of constraints compared with introns and exons in the coding region. Molecular Biology and Evolution 19: 1613-1623.
Grant V. 1998. Primary classification and phylogeny of the Polemoniaceae, with comments on molecular cladistics. American Journal of Botany 85: 741-752.[Abstract]
Grant V. 2003. Incongruence between cladistics and taxonomic systems. American Journal of Botany 90: 1263-1270.
Lee J Baldwin B. G Gottlieb L. D. 2002. Phylogeny of Stephanomeria and related genera (Compositae-Lactuceae) based on analysis of 18S-26S nuclear rDNA ITS and ETS sequences. American Journal of Botany 89: 160-168.
Lee J Baldwin B. G Gottlieb L. D. 2003. Phylogenetic relationships among the primarily North American genera of Cichorieae (Compositae) based on analysis of 18S-26S nuclear rDNA ITS and ETS sequences. Systematic Botany 28: 616-626.[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 Lewis M. E. 1955. The genus Clarkia. University of California Publications in Botany 20: 241-392.
Posada D Crandall K. A. 1998. Modeltest: testing the model of DNA substitution. Bioinformatics 14: 817-818.
Sanderson M. 1998. Estimating rate and time in molecular phylogenies: beyond the molecular clock?. In D. E. Soltis, P. S. Soltis and J. J. Doyle [eds.], Molecular systematics of plants II: DNA sequencing, 242–264.
Swofford D. L. 2001. PAUP*: phylogenetic analysis using parsimony (*and other methods), version 4.068 Sinauer Sunderland, Massachusetts, USA.
Terauchi R Terachi T Miyashita N. T. 1997. DNA polymorphism at the PgiC locus of a wild yam, Dioscorea tokoro. Genetics 147: 1899-1914.[Abstract]
Thomas B. R Ford V. S Pichersky E Gottlieb L. D. 1993. Molecular characterization of duplicate cytosolic phosphoglucose isomerae genes in Clarkia and comparison to the single gene in Arabidopsis. Genetics 135: 895-905.[Abstract]
Thomas B. R Laudencia-Chingcuanco D Gottlieb L. D. 1992. Molecular analysis of the plant gene encoding cytosolic phosphoglucose isomerase Plant Molecular Biology 19: 745-757.[CrossRef][ISI][Medline]
Thompson J. D Higgins D. G Gibson T. J. 1994. Clustal W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Research 22: 4673-4680.
Whittle C.-A Johnston M. O. 2003. Broad-scale analysis contradicts the theory that generation time affects molecular evolutionary rates in plants. Journal of Molecular Evolution 56: 223-233.[CrossRef][ISI][Medline]
| ||||||||||||||||||||||||||||||||||||
