Chloroplast Phylogeny Indicates that Bryophytes Are Monophyletic
http://www.100md.com
分子生物学进展 2004年第10期
* National Institute for Basic Biology, Okazaki, Japan; Department of Biology, Utah State University; Faculty of Science, Shizuoka University, Shizuoka, Japan; The Jackson Laboratory, Bar Harbour, Maine; || Center for Gene Research, Nagoya University, Nagoya, Japan; ? Department of Biology, Faculty of Science, Toyama University, Toyama, Japan; # Division of Functional Genomics, Advanced Science Research Center, Kanazawa University, Kanazawa, Japan; ** Division of Life Science, Graduate School of Natural Science and Technology, Kanazawa University, Kanazawa, Japan; and Department of Molecular Biomechanics, Sokendai, Okazaki, Japan
E-mail: mhasebe@nibb.ac.jp.
Abstract
Opinions on the basal relationship of land plants vary considerably and no phylogenetic tree with significant statistical support has been obtained. Here, we report phylogenetic analyses using 51 genes from the entire chloroplast genome sequences of 20 representative green plant species. The analyses, using translated amino acid sequences, indicated that extant bryophytes (mosses, liverworts, and hornworts) form a monophyletic group with high statistical confidence and that extant bryophytes are likely sisters to extant vascular plants, although the support for monophyletic vascular plants was not strong. Analyses at the nucleotide level could not resolve the basal relationship with statistical confidence. Bryophyte monophyly inferred using amino acid sequences has a good statistical foundation and is not rejected statistically by other data sets. We propose bryophyte monophyly as the currently best hypothesis.
Key Words: land plants ? bryophytes ? codon usage ? nucleotide composition ? LogDet
Introduction
More than 300,000 green plant species now cover the Earth's surface, and these plants are descended from a common ancestor that lived during the Ordovician, more than 470 MYA (Kenrick and Crane 1997a, 1997b; Kenrick 2000). Recent morphological and molecular phylogenetic studies have revealed that the closest relative of land plants is a group of freshwater green algae (reviewed in Graham and Wilcox [2000] and Chapman and Waters [2002]). However, the earliest branches of the land plant phylogeny remain unclear.
The oldest fossil records of land plants are spore tetrads from the mid-Ordovician (Gray 1993) that are thought to be sporopollenin-bearing meiospores and are suggested to be similar to spores of liverworts because of lamellae in the spore walls. Fossil plant fragments containing similar spores, recently found in Ordovician deposits from Oman, appear to be from embryophytes (Wellman, Osterloff, and Mohiuddin 2003). However, the affinity of these spores to extant liverworts is uncertain because no associated megafossils have been found, and similar tetrad spores are also produced by mosses (Gray 1993). A cladistic analysis of several key morphological characters supports the basal branching of liverworts (Mishler and Churchill 1984), whereas other cladistic analyses of ultrastructural, biochemical, and developmental characters (Garbary and Renzaglia 1998; Renzaglia et al. 2000) support a basal position of hornworts rather than liverworts. Data from antheridial development and the complex process of spermatogenesis (Garbary, Renzaglia, and Duckett 1993; Renzaglia et al. 2000; Renzaglia and Garbary 2001) support a monophyly of extant bryophytes (mosses, liverworts, and hornworts). Results using molecular data are also controversial. Analyses using DNA sequences of multiple genes suggest a basal position for hornworts (Nishiyama and Kato 1999), although the branching patterns are only weakly supported. Absence of group II introns in three mitochondrial genes in both liverworts and green algae supports the liverwort basal hypothesis (Qiu et al. 1998), although repeated intron losses have been observed in land plant lineages (Qiu et al. 1998) and the possibility of parallel evolution cannot be excluded. The consensus of these conflicting results yields an unresolved polytomy at the base of land plants. Determining these relationships would not only resolve a critical area of the land plant phylogeny, but it would also be important for understanding the characteristics of the earliest land plants and their subsequent evolution.
Combined analyses of DNA sequences from multiple loci have proven useful in inferring deep phylogenetic relationships (Qiu et al. 1999; Pryer et al. 2001), as predicted by simulation studies (Cummings, Otto, and Wakeley 1995). With the accumulation of genome information, the entire chloroplast genome sequences of 15 seed plants, one liverwort (Marchantia polymorpha), four green algae, three red algae, and one cryptophyte have been reported in the GenBank RefSeq division. Because these published data do not cover all major lineages of land plants, we determined the entire chloroplast genome sequences of two pteridophytes, Psilotum nudum (accession number: NC_003386) and Adiantum capillus-veneris (accession number: AY178864; Wolf et al. 2003), a hornwort Anthoceros angustus (accession number: AB086179; Kugita et al. 2003), and a moss, Physcomitrella patens (accession number: AP005672; Sugiura et al. 2003). We analyzed 51 genes that are found in every chloroplast genome of the sampled land plants and chlorophytes (26,937 bp / 8,979 amino acid sites in total), to infer the phylogeny of land plants and to examine the utility of the amino acid versus nucleotide sequence analyses.
Materials and Methods
Coding sequences were extracted from the annotated sequences using bioruby (http://bioruby.org/). Chloroplast genome sequences were determined for a moss (Physcomitrella patens), a hornwort (Anthoceros angustus), and two pteridophytes (Adiantum capillus-veneris and Psilotum nudum). In addition to these four taxa, 16 green plant chloroplast DNA genome sequences in the DNA database (table 1) were used for phylogenetic analyses. The nucleotide sequences of 51 genes that are present in all the chloroplast genomes of green plants were individually aligned using the program fftnsi in MAFFT (Katoh et al. 2002) version 3.85. The 51 genes are atpA, atpB, atpE, atpF, atpH, atpI, petA, petB, petD, petG, psaA, psaB, psaC, psaI, psaJ, psbA, psbB, psbC, psbD, psbE, psbF, psbH, psbI, psbJ, psbK, psbL, psbM, psbN, psbT, psbZ, rbcL, rpl2, rpl14, rpl16, rpl20, rpl36, rpoB, rpoC1, rpoC2, rps2, rps3, rps4, rps7, rps8, rps11, rps12, rps14, rps18, rps19, ycf3, and ycf4. Unambiguously aligned regions were selected for further analyses and concatenated. The overlapping region of psbD and psbC was excluded. The data matrix is available online as Supplementary Material. The selected regions included 26,937 nucleotide sites. Because RNA editing is abundant in Anthoceros angustus (Kugita et al. 2003), we used cDNA sequence of A. angustus. The amino acid matrix comprised 20 taxa and 8,979 amino acid sites. Pairwise maximum-likelihood (ML) distances were calculated using ProtML (Adachi and Hasegawa 1996) and a Neighbor-Joining (NJ) tree was obtained with Njdist (Adachi and Hasegawa 1996). A local rearrangement search was performed with the NJ tree as the starting tree. Local bootstrap probabilities were calculated with resampling using the estimated log-likelihood method (Adachi and Hasegawa 1996). A user tree was prepared for each of the 105 possible bifurcating topologies among six groups of taxa: outgroups (O), mosses (M), liverworts (L), hornworts (H), monilophytes (F), and seed plants (S). The monilophytes include ferns, psilotophytes (whisk ferns), and equisetophytes (horsetails) (Kenrick and Crane 1997b, p. 228; Pryer et al. 2001; Judd et al. 2002, pp. 162–168), and in our study they are represented by Adiantum capillus-veneris and Psilotum nudum. A. capillus-veneris and P. nudum are forced to form a clade, and the topologies within seed plants and within the outgroup were constrained as the ML tree found in the local rearrangement search (fig. 1). The estimated log-likelihood under the JTT model (Jones, Taylor, and Thornton 1992), adjusted with the empirical amino acid frequency (JTT-F model) of each site under the 105 topologies, was calculated using ProtML (Adachi and Hasegawa 1996). The log-likelihood under the JTT model (Jones, Taylor, and Thornton 1992) was calculated for each gene to take account of rate differences among genes. The likelihood under a discrete gamma model of rate heterogeneity among sites (Yang 1994) was calculated with PAML version 3.13d (Yang 1997). The P values for the bootstrap and approximately unbiased (AU) test (Shimodaira 2002) were calculated using the estimated log-likelihood with CONSEL (Shimodaira and Hasegawa 2001). Maximum parsimony analyses were performed with PAUP* version 4.0b10 (Swofford 1998). The most parsimonious trees were recovered with a heuristic search strategy employing tree bisection and reconnection (TBR) with 1,000 random addition sequence replications. Bootstrap tests were performed with 1,000 replications with 10 replications of random addition sequence in each run. Analyses to overcome compositional heterogeneity with paralinear/LogDet distances (Lake 1994; Lockhart et al. 1994) followed Nishiyama and Kato (1999).
Table 1 Accession Numbers and Codon Usage for Leucine in Chloroplast Genomes
FIG. 1.— Phylogenetic relationship of representative lineages in land plants inferred from a maximum-likelihood analysis using 8,979 amino acid residues for 51 genes from 20 chloroplast genomes. Local bootstrap probabilities are shown on the branches. The horizontal branch lengths are proportional to the estimated number of amino acid substitutions per site. This tree is rooted using all the taxa other than embryophytic land plants. Higher taxonomic categories and common names are indicated.
Results
Heterogenetity in base composition at all codon positions is recognized as a problem for phylogenetic analyses of chloroplast genes (Steel, Lockhart, and Penny, 1993; Nishiyama and Kato 1999). We observed such compositional differences across entire chloroplast genomes; GC composition ranged from 12.9% in the third codon of Marchantia polymorpha to 38.3% in Adiantum capillus-veneris. We also found differences in codon usage for the first codon positions of leucine (table 1). Fifty-five percent of the leucine codons start with U in Nephroselmis olivacea compared to 26% in Chaetosphaeridium globosum. Most conventional methods of phylogenetic analysis are statistically consistent (i.e., reach the true topology with infinite data) only when the sequences evolved under a stationary Markov process (Gu and Li 1996). The large differences in the base composition indicate that the evolutionary process is not stationary. Therefore, conventional methods may give positively misleading results with high bootstrap probabilities (Lockhart et al. 1994). To overcome this problem, we applied three approaches: (1) we eliminated codon usage differences by analyzing amino acids sequences; (2) we removed the sites exhibiting compositional heterogeneity, i.e., all the leucine codons and third codon positions; and (3) we used LogDet/paralinear distances (Lake 1994; Lockhart et al. 1994; Steel 1994; Gu and Li 1996), which are consistent under nonstationary processes.
Analysis of Deduced Amino Acid Sequence Data
A local rearrangement search starting with the NJ tree (Saitou and Nei 1987) under the JTT-F model (Jones, Taylor, and Thornton 1992; Adachi and Hasegawa 1996) found an ML tree (ln likelihood = –89478.09; fig. 1). The three bryophytes, Marchantia polymorpha, Physcomitrella patens, and Anthoceros angustus, formed a monophyletic group with high local bootstrap probability (99%). Previously well-recognized monophyletic groups, the eudicots, monocots, angiosperms, seed plants, and land plants (reviewed in Judd et al. 2002, pp. 153–183), each formed a monophyletic group with high local bootstrap probability. Vascular plants formed a monophyletic group, though the bootstrap value was low (55%). This contrasts previous studies with high bootstrap support for vascular plants (Hedderson, Chapman, and Cox 1998; Nickrent et al. 2000). To test whether other possible trees have a significantly less likelihood than the ML tree, the bootstrap test and AU test (Shimodaira 2002) were used for all 105 possible topologies among seed plants, monilophytes, hornworts, liverworts, mosses, and outgroups. We constrained the topology within seed plants and within outgroup taxa. In the bootstrap test, topology 1 (fig. 2), which had the same topology as the ML tree, had 52.8% bootstrap support. Topology 2 (fig 2), which had the second highest likelihood (ln = –89480.91), occupied 44.3% of the bootstrap replicates. Bryophytes were also monophyletic in topology 2, but the bryophyte clade was nested within vascular plants, and monilophytes were basal among land plants. Other topologies were significantly less likely. The AU test showed that all topologies except topology 2 were significantly less likely than topology 1 at the 5% level (table 2; table S1 in the online Supplementary Material).
FIG. 2.— The best three alternative topologies found in the analyses of 105 trees. S, seed plants; F, monilophytes (Adiantum capillus-veneris plus Psilotum nudum); H, a hornwort (Anthoceros angustus); L, a liverwort (Marchantia polymorpha); M, a moss (Physcomitrella patens); O, outgroup taxa.
Table 2 Statistical Test for the 105 Topologies Under Various Models of Rate Heterogeneity
To consider rate heterogeneity among genes and sites, we performed separate gene analyses, in which each gene is assumed to have evolved independently ("independent gene" in table 2). This was performed under the JTT model using ProtML. We also performed a discrete gamma approximation with eight categories ("gamma 8 category" in table 2) using PAML (Yang 1997). The log-likelihood of every tree was calculated. These analyses also supported topology 1 as the ML tree (table 2). We also obtained the same topology when we used the genomic sequence of Anthoceros angustus to deduce the amino acid sequence (data not shown), indicating that RNA editing does not cause serious problems for these phylogenetic analyses. We also tested 20 data sets, each one omitting one taxon from the matrix. Although the monophyly of vascular plants changed when an outgroup taxon or Adiantum capillus-veneris was removed, the bryophyte monophyly was robust against taxon removal (data not shown).
Analysis of Nucleotide Sequence Data
We found a single most parsimonious tree (14,487 steps) from a TBR search based on nucleotide sequence data omitting third codon positions. The liverwort Marchantia polymorpha was resolved as the earliest diverging lineage in land plants, but bootstrap support for the phylogenetic relationships among bryophytes was low (fig. 3). When all the leucine codons and third codon positions were excluded, Marchantia polymorpha and Physcomitrella patens formed a monophyletic group (fig. 4). Phylogenetic relationships between this moss + liverwort clade and other taxa did not have high bootstrap support.
FIG. 3.— The most parsimonious tree using nucleotides of the first and second codon positions. 4,228 of 17,958 sites were parsimony-informative. Tree length = 14,487. Consistency index excluding uninformative characters = 0.52. Retention index = 0.62. Bootstrap probabilities are indicated on branches. Branch lengths are proportional to the number of substitutions on the branches by ACCTRAN optimization.
FIG. 4.— The most parsimonious tree excluding the nucleotide sites coding leucine and third codon positions. 2,731 of 14,860 sites were parsimony-informative. Tree length = 8,998. Consistency index excluding uninformative characters = 0.63. Retention index = 0.65. Bootstrap probabilities are indicated on branches. Branch lengths are proportional to the number of substitutions on the branches by ACCTRAN optimization.
Because LogDet/paralinear distances assume a uniform rate among sites (Lake 1994; Lockhart et al. 1994; Steel 1994; Gu and Li 1996), we used only the fourfold degenerate sites, of which there were 2,695 across taxa. An NJ tree was calculated using paralinear/LogDet distances (fig. 5). Again, the bootstrap values for the basal branches were low.
FIG. 5.— Neighbor-Joining tree using LogDet distance based on 2,695 fourfold degenerate sites. The number on each branch indicates a bootstrap probability (%) estimated by 10,000 replicates. Branch lengths are proportional to the LogDet distance.
Discussion
The monophyly of extant bryophytes was inferred by our analyses using translated amino acid sequence data of currently available chloroplast genomes. Analyses using nucleotide sequences did not resolve phylogenetic relationships of the three bryophytes with statistical confidence. Amino acid and nucleotide data each have their own merits and drawbacks. Amino acid sequences have frequently been used to infer deep phylogeny because they avoid problems with saturation of silent substitution and differential GC content (Russo, Takezaki, and Nei 1996; Nishiyama and Kato 1999; Simmons, Ochoterena, and Freudenstein 2002 and references therein). But, Simmons, Ochoterena, and Freudenstein (2002) criticized the use of amino acid translation, showing that nucleotide data outperform deduced amino acid data in a phylogenetic analysis of angiosperms, which diverged during the last 200 Myr (Judd et al. 2002, pp. 175–176). However, analyses using nucleotide sequences may not be robust when nucleotide compositions are apparently different among taxa, because phylogenetic analyses usually assume a stationary Markov process (Lockhart et al. 1994). In an attempt to remove characters with different nucleotide compositions, we excluded third codon positions (fig. 3) or the nucleotide sites coding leucine and third codon positions (fig. 4). We also compensated for differences in nucleotide composition by LogDet transformation (fig. 5). Neither the removal of characters nor the LogDet transformation resolved the relationships of extant bryophytes with high statistical support, whereas resolution with high statistical support was achieved with amino acid translation. These results might be explained if a uniform rate of change does not describe evolution of the fourfold degenerate sites. These results may also suggest that nucleotide composition bias can increase in proportion to divergence time among taxa. Nucleotide sequences including the third codon position data are more likely to be useful for inferring phylogeny among taxa that diverged relatively recently, such as angiosperms, which diverged within 200 Myr (Judd et al. 2002, pp. 175–176). If compositional heterogeneity is not extreme and the spatial pattern of substitution is simple, it would follow that nucleotide data will outperform deduced amino acid data because the former has more phylogenetic information than the latter. However, when sequences become highly diverged, the spatial substitution pattern may become more complex and problematic for distance-based correction formulae (Steel, Huson, and Lockhart 2000). The main lineages of land plants probably diverged as many as 470 MYA (Kenrick and Crane 1997a, 1997b; Kenrick 2000).
Phylogenetic analyses using DNA sequences suggest that extant bryophytes are paraphyletic (Lewis, Mishler, and Vilgalys 1997; Duff and Nickrent 1999; Nishiyama and Kato 1999; Nickrent et al. 2000). However, only the MP-3Ti analysis by Nickrent et al. (2000) provided evidence of strong statistical support. Moreover, monophyly of bryophytes was not rejected in the Kishino-Hasegawa (1989) test by Nickrent et al. (2000). The sister relationship of mosses and liverworts received relatively strong support in recent analyses using multiple genes (Nishiyama and Kato 1999; Nickrent et al. 2000). Marchantia appears on a basally diverging branch in land plants with high bootstrap probability in a combined analysis of atpB, rbcL, nad5, and the small subunit rRNA gene (Karol et al. 2001). However, that study used DNA sequences of chloroplast genes without attempting to compensate for the nucleotide composition bias, which means that the bootstrap probabilities may not be a good estimator of the reliability of branches and tree topologies (Lockhart et al. 1994; Swofford et al. 2001; Felsenstein 2003, pp. 272–274). Therefore, the extant bryophyte monophyletic hypothesis inferred in this study stands as the best one statistically supported by molecular sequence data.
Phylogenetic studies of land plant relationships using morphological characters are also incongruent. Most studies suggest that extant bryophytes are paraphyletic (Mishler and Churchill 1984, 1985; Renzaglia et al. 2000), whereas monophyly of extant bryophytes and a sister relationship of mosses and liverworts were inferred using characters of spermatogenesis (Garbary, Renzaglia, and Duckett 1993; Renzaglia et al. 2000; Renzaglia and Garbary 2001). Analyses of spermatogenesis characters also supported monilophytes involving ferns, psilotophytes (whisk ferns), and equisetophytes (horsetails), which are supported by DNA sequence data (Nickrent et al. 2000; Pryer et al. 2001). If our inference of monophyletic bryophytes reflects the true relationships, this would suggest that spermatogenesis characters have retained more phylogenetic signal than other morphological characters.
While our hypothesis was supported with high statistical confidence with presently available data, the bryophyte monophyly hypothesis should be tested with more data and further analysis. Current analyses include only one sample per group from mosses, liverworts, and hornworts and lack a major lineage of land plants, lycophytes. Some major lineages of bryophytes and lycophytes should be included based on previous phylogenetic analyses (Lewis, Mishler, and Vilgalys 1997; Garbary and Renzaglia 1998; Hedderson, Chapman, and Cox 1998; Nickrent et al. 2000; Renzaglia et al. 2000). Sphagnum, Takakia, Andreaea, and Andraeobryum from mosses and Treubia, Haplomitrium, Blasia, one simple thalloid, and one leafy liverwort from liverworts are candidate representatives. In addition to one homosporous and one heterosporous lycophyte, equisetophytes should be targeted for vascular plants. Furthermore, data from Charales may be a better outgroup to land plants than Coleochaetales (Turmel et al. 2002). Mitochondrial and nuclear genome data are also candidate sources of phylogenetic information that could provide an independent test of results based on chloroplast genome data.
The fossil record of plants older than rhyniophytes includes cuticles (Edwards, Duckett, and Richardson 1995) and spore tetrads (Gray 1993). These fossils have been considered to be related to extant liverworts (Edwards, Duckett, and Richardson 1995; Taylor 1995). Our results are not in line with this interpretation because we infer that liverworts do not descend from a basal node in bryophytes (fig. 2). Our results, however, are consistent with a "bryophyte-like ancestor" hypothesis (Graham and Gray 2001), whereby the above fossils represent a common ancestor of the three extant groups of bryophytes.
Supplementary Material
The complete table for statistical tests among 105 topologies is available online as a supplementary table. The nucleotide data matrix is available online in nexus format.
Acknowledgements
We thank two anonymous reviewers, Brent Mishler, Hiroyuki Akiyama, Hironori Deguchi, and members of the Hasebe laboratory for discussion and comments on the manuscript. Work on Anthoceros angustus has been published under the name of A. formosae. Thanks also to Carol Rowe for help sequencing the Adiantum chloroplast genome. The computations were partly done in the Computer Lab of the National Institute for Basic Biology. The research was supported by National Science Foundation grant #DEB-0228432 to P.G.W. and grants by MEXT and JSPS. Adiantum capillus-veneris, Anthoceros angustus, Physcomitrella patens, and Psilotum nudum were sequenced by P.G.W., R.B.S, and M.H.; M.K. and K.Y.; M.S. and C.S.; and T.Y. and K.Y, respectively. Correspondence in general should be addressed to M.H. (e-mail: mhasebe@nibb.ac.jp). Correspondence and requests for materials for chloroplast DNAs should be addressed to P.G.W. (Adiantum, e-mail: wolf@biology.usu.edu), K.Y. (Anthoceros, e-mail: sckyosi@ipc.shizuoka.ac.jp), M.S. (Physcomitrella, e-mail: sugita@gene.nagoya-u.ac.jp), and K.Y. (Psilotum, e-mail: yamada@sci.toyama-u.ac.jp).
References
Adachi, J., and M. Hasegawa. 1996. Molphy Version 2.3: programs for molecular phylogenetics based on maximum likelihood. The Institute of Statistical Mathematics, Tokyo, Japan.
Chapman, R. L., and D. A. Waters. 2002. Green algae and land plants—an answer at last? J. Phycol. 38:237–240.
Cummings, M. P., S. P. Otto, and J. Wakeley. 1995. Sampling properties of DNA sequence data in phylogenetic analysis. Mol. Biol. Evol. 12:814–822.
Duff, R. J., and D. L. Nickrent. 1999. Phylogenetic relationships of land plants using mitochondrial small-subunit rDNA sequences. Am. J. Bot. 86:372.
Edwards, D., J. G. Duckett, and J. B. Richardson. 1995. Hepatic characters in the earliest land plants. Nature 374:635–636.
Felsenstein, J. 2003. Inferring phylogenies. Sinauer Associates, Sunderland, Mass.
Garbary, D. J., and K. S. Renzaglia. 1998. Bryophyte phylogeny and the evolution of land plants: evidence from development and ultrastructure. Pp. 45–63 in J. W. Bates, N. W. Ashton, and J. G. Duckett, eds. Bryology for the twenty-first century. Maney Publishing and British Bryological Society, Leeds, United Kingdom.
Garbary, D. J., K. S. Renzaglia, and J. G. Duckett. 1993. The phylogeny of land plants—a cladistic analysis based on male gametogenesis. Plant Syst. Evol. 188:237–269.
Graham, L. E., and J. Gray. 2001. The origin, morphology, and ecophysiology of early embryophytes: neontological and paleontological perspectives. Pp. 140–158 in P. G. Gensel and D. Edwards, eds. Plants invade the land: evolutionary and environmental perspectives. Columbia University Press, New York.
Graham, L. K., and L. W. Wilcox. 2000. The origin of alternation of generations in land plants: a focus on matrotrophy and hexose transport. Phil. Trans. R. Soc. Lond. B 355:757–766.
Gray, J. 1993. Major Paleozoic land plant evolutionary bio-events. Palaeogeogr. Palaeoclimatol. Palaeoecol. 104:153–169.
Gu, X., and W. H. Li. 1996. Bias-corrected paralinear and LogDet distances and tests of molecular clocks and phylogenies under nonstationary nucleotide frequencies. Mol. Biol. Evol. 13:1375–1383.
Hedderson, T. A., R. Chapman, and C. J. Cox. 1998. Bryophytes and the origins and diversification of land plants: new evidence from molecules. Pp. 65–77 in J. W. Bates, N. W. Ashton, and J. G. Duckett, eds. Bryology for the twenty-first century. Maney Publishing and British Bryological Society, Leeds, United Kingdom.
Jones, D. T., W. R. Taylor, and J. M. Thornton. 1992. The rapid generation of mutation data matrices from protein sequences. Comput. Appl. Biosci. 8:275–282.
Judd, W. S., C. S. Campbell, E. A. Kellogg, P. F. Stevens, and M. J. Donoghue. 2002. Plant systematics: a phylogenetic approach. Sinauer Associates, Sunderland, Mass.
Karol, K. G., R. M. McCourt, M. T. Cimino, and C. F. Delwiche. 2001. The closest living relatives of land plants. Science 294:2351–2353.
Katoh, K., K. Misawa, K. Kuma, and T. Miyata. 2002. MAFFT: a novel method for rapid multiple sequence alignment based on fast Fourier transform. Nucleic Acids Res. 30:3059–3066.
Kenrick, P. 2000. The relationships of vascular plants. Phil. Trans. R. Soc. Lond. B 355:847–855.
Kenrick, P., and P. R. Crane. 1997a. The origin and early evolution of plants on land. Nature 389:33–39.
———. 1997b. The origin and early diversification of land plants: a cladistic study. Smithsonian Institution Press, Washington, D.C.
Kishino, H., and M. Hasegawa. 1989. Evaluation of the maximum likelihood estimate of the evolutionary tree topologies from DNA sequence data, and the branching order in Hominoidea. J. Mol. Evol. 29:170–179.
Kugita, M., A. Kaneko, Y. Yamamoto, Y. Takeya, T. Matsumoto, and K. Yoshinaga. 2003. The complete nucleotide sequence of the hornwort (Anthoceros formosae) chloroplast genome: insight into the earliest land plants. Nucleic Acids Res. 31:716–721.
Lake, J. A. 1994. Reconstructing evolutionary trees from DNA and protein sequences: paralinear distances. Proc. Natl. Acad. Sci. USA 91:1455–1459.
Lewis, L. A., B. D. Mishler, and R. Vilgalys. 1997. Phylogenetic relationships of the liverworts (Hepaticae), a basal embryophyte lineage, inferred from nucleotide sequence data of the chloroplast gene rbcL. Mol. Phylogenet. Evol. 7:377–393.
Lockhart, P. J., M. A. Steel, M. D. Hendy, and D. Penny. 1994. Recovering evolutionary trees under a more realistic model of sequence evolution. Mol. Biol. Evol. 11:605–612.
Mishler, B. D., and S. P. Churchill. 1984. A cladistic approach to the phylogeny of the "bryophytes." Brittonia 36:406–424.
———. 1985. Transition to a land flora: phylogenetic relationships of the green algae and bryophytes. Cladistics 1:305–328.
Nickrent, D. L., C. L. Parkinson, J. D. Palmer, and R. J. Duff. 2000. Multigene phylogeny of land plants with special reference to bryophytes and the earliest land plants. Mol. Biol. Evol. 17:1885–1895.
Nishiyama, T., and M. Kato. 1999. Molecular phylogenetic analysis among bryophytes and tracheophytes based on combined data of plastid coded genes and the 18S rRNA gene. Mol. Biol. Evol. 16:1027–1036.
Pryer, K. M., H. Schneider, A. R. Smith, R. Cranfill, P. G. Wolf, J. S. Hunt, and S. D. Sipes. 2001. Horsetails and ferns are a monophyletic group and the closest living relatives to seed plants. Nature 409:618–622.
Qiu, Y. L., Y. Cho, J. C. Cox, and J. D. Palmer. 1998. The gain of three mitochondrial introns identifies liverworts as the earliest land plants. Nature 394:671–674.
Qiu, Y. L., J. Lee, F. Bernasconi-Quadroni, D. E. Soltis, P. S. Soltis, M. Zanis, E. A. Zimmer, Z. Chen, V. Savolainen, and M. W. Chase. 1999. The earliest angiosperms: evidence from mitochondrial, plastid and nuclear genomes. Nature 402:404–407.
Renzaglia, K. S., R. J. T. Duff, D. L. Nickrent, and D. J. Garbary. 2000. Vegetative and reproductive innovations of early land plants: implications for a unified phylogeny. Phil. Trans. R. Soc. Lond. B 355:769–793.
Renzaglia, K. S., and D. J. Garbary. 2001. Motile gametes of land plants: diversity, development, and evolution. Crit. Rev. Plant Sci. 20:107–123.
Russo, C. A. M., N. Takezaki, and M. Nei. 1996. Efficiencies of different genes and different tree-building methods in recovering a known vertebrate phylogeny. Mol. Biol. Evol. 13:525–536.
Saitou, N., and M. Nei. 1987. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4:406–425.
Shimodaira, H. 2002. An approximately unbiased test of phylogenetic tree selection. Syst. Biol. 51:492–508.
Shimodaira, H., and M. Hasegawa. 2001. CONSEL: for assessing the confidence of phylogenetic tree selection. Bioinformatics 17:1246–1247.
Simmons, M. P., H. Ochoterena, and J. V. Freudenstein. 2002. Amini acid vs. nucleotide charactors: challenging preconceived notions. Mol. Phylogenet. Evol. 24:78–90.
Steel, M. 1994. Recovering a tree from the leaf volourations it generates under a Markov model. Appl. Math. Lett. 7:19–23.
Steel, M. A., D. Huson, and P. J. Lockhart. 2000. Invariable site models and their use in phylogeny reconstruction. Syst. Biol. 49:225–232.
Steel, M. A., P. J. Lockhart, and D. Penny. 1993. Confidence in evolutionary trees from biological sequence data. Nature 364:440–442.
Sugiura, C., Y. Kobayashi, S. Aoki, C. Sugita, and M. Sugita. 2003. Complete chloroplast DNA sequence of the moss Physcomitrella patens: evidence for the loss and relocation of rpoA from the chloroplast to the nucleus. Nucleic Acids Res. 31:5324–5331.
Swofford, D. L. 1998. PAUP*: phylogenetic analysis using parsimony (*and other methods). Version 4. Sinauer Associates, Sunderland, Mass.
Swofford, D. L., P. J. Waddell, J. P. Huelsenbeck, P. G. Foster, P. O. Lewis, and J. S. Rogers. 2001. Bias in phylogenetic estimation and its relevance to the choice between parsimony and likelihood methods. Syst. Biol. 50:525–539.
Taylor, W. A. 1995. Spores in earliest land plants. Nature 373:391–392.
Turmel, M., M. Ehara, C. Otis, and C. Lemieux. 2002. Phylogenetic relationships among streptophytes as inferred from chloroplast small and large subunit rRNA gene sequences. J. Phycol. 38:364–375.
Wellman, C. H., P. L. Osterloff, and U. Mohiuddin. 2003. Fragments of the earliest land plants. Nature 425:282–285.
Wolf, P. G., C. A. Rowe, R. B. Sinclair, and M. Hasebe. 2003. Complete nucleotide sequence of the chloroplast genome from the leptosporangiate fern Adiantum capillus-veneris L. DNA Res. 10:59–65.
Yang, Z. 1994. Maximum likelihood phylogenetic estimation from DNA sequences with variable rates over sites: approximate methods. J. Mol. Evol. 39:306–314.
———. 1997. PAML: a program package for phylogenetic analysis by maximum likelihood. Comput. Appl. Biosci. 13:555–556.(Tomoaki Nishiyama*, Paul )
E-mail: mhasebe@nibb.ac.jp.
Abstract
Opinions on the basal relationship of land plants vary considerably and no phylogenetic tree with significant statistical support has been obtained. Here, we report phylogenetic analyses using 51 genes from the entire chloroplast genome sequences of 20 representative green plant species. The analyses, using translated amino acid sequences, indicated that extant bryophytes (mosses, liverworts, and hornworts) form a monophyletic group with high statistical confidence and that extant bryophytes are likely sisters to extant vascular plants, although the support for monophyletic vascular plants was not strong. Analyses at the nucleotide level could not resolve the basal relationship with statistical confidence. Bryophyte monophyly inferred using amino acid sequences has a good statistical foundation and is not rejected statistically by other data sets. We propose bryophyte monophyly as the currently best hypothesis.
Key Words: land plants ? bryophytes ? codon usage ? nucleotide composition ? LogDet
Introduction
More than 300,000 green plant species now cover the Earth's surface, and these plants are descended from a common ancestor that lived during the Ordovician, more than 470 MYA (Kenrick and Crane 1997a, 1997b; Kenrick 2000). Recent morphological and molecular phylogenetic studies have revealed that the closest relative of land plants is a group of freshwater green algae (reviewed in Graham and Wilcox [2000] and Chapman and Waters [2002]). However, the earliest branches of the land plant phylogeny remain unclear.
The oldest fossil records of land plants are spore tetrads from the mid-Ordovician (Gray 1993) that are thought to be sporopollenin-bearing meiospores and are suggested to be similar to spores of liverworts because of lamellae in the spore walls. Fossil plant fragments containing similar spores, recently found in Ordovician deposits from Oman, appear to be from embryophytes (Wellman, Osterloff, and Mohiuddin 2003). However, the affinity of these spores to extant liverworts is uncertain because no associated megafossils have been found, and similar tetrad spores are also produced by mosses (Gray 1993). A cladistic analysis of several key morphological characters supports the basal branching of liverworts (Mishler and Churchill 1984), whereas other cladistic analyses of ultrastructural, biochemical, and developmental characters (Garbary and Renzaglia 1998; Renzaglia et al. 2000) support a basal position of hornworts rather than liverworts. Data from antheridial development and the complex process of spermatogenesis (Garbary, Renzaglia, and Duckett 1993; Renzaglia et al. 2000; Renzaglia and Garbary 2001) support a monophyly of extant bryophytes (mosses, liverworts, and hornworts). Results using molecular data are also controversial. Analyses using DNA sequences of multiple genes suggest a basal position for hornworts (Nishiyama and Kato 1999), although the branching patterns are only weakly supported. Absence of group II introns in three mitochondrial genes in both liverworts and green algae supports the liverwort basal hypothesis (Qiu et al. 1998), although repeated intron losses have been observed in land plant lineages (Qiu et al. 1998) and the possibility of parallel evolution cannot be excluded. The consensus of these conflicting results yields an unresolved polytomy at the base of land plants. Determining these relationships would not only resolve a critical area of the land plant phylogeny, but it would also be important for understanding the characteristics of the earliest land plants and their subsequent evolution.
Combined analyses of DNA sequences from multiple loci have proven useful in inferring deep phylogenetic relationships (Qiu et al. 1999; Pryer et al. 2001), as predicted by simulation studies (Cummings, Otto, and Wakeley 1995). With the accumulation of genome information, the entire chloroplast genome sequences of 15 seed plants, one liverwort (Marchantia polymorpha), four green algae, three red algae, and one cryptophyte have been reported in the GenBank RefSeq division. Because these published data do not cover all major lineages of land plants, we determined the entire chloroplast genome sequences of two pteridophytes, Psilotum nudum (accession number: NC_003386) and Adiantum capillus-veneris (accession number: AY178864; Wolf et al. 2003), a hornwort Anthoceros angustus (accession number: AB086179; Kugita et al. 2003), and a moss, Physcomitrella patens (accession number: AP005672; Sugiura et al. 2003). We analyzed 51 genes that are found in every chloroplast genome of the sampled land plants and chlorophytes (26,937 bp / 8,979 amino acid sites in total), to infer the phylogeny of land plants and to examine the utility of the amino acid versus nucleotide sequence analyses.
Materials and Methods
Coding sequences were extracted from the annotated sequences using bioruby (http://bioruby.org/). Chloroplast genome sequences were determined for a moss (Physcomitrella patens), a hornwort (Anthoceros angustus), and two pteridophytes (Adiantum capillus-veneris and Psilotum nudum). In addition to these four taxa, 16 green plant chloroplast DNA genome sequences in the DNA database (table 1) were used for phylogenetic analyses. The nucleotide sequences of 51 genes that are present in all the chloroplast genomes of green plants were individually aligned using the program fftnsi in MAFFT (Katoh et al. 2002) version 3.85. The 51 genes are atpA, atpB, atpE, atpF, atpH, atpI, petA, petB, petD, petG, psaA, psaB, psaC, psaI, psaJ, psbA, psbB, psbC, psbD, psbE, psbF, psbH, psbI, psbJ, psbK, psbL, psbM, psbN, psbT, psbZ, rbcL, rpl2, rpl14, rpl16, rpl20, rpl36, rpoB, rpoC1, rpoC2, rps2, rps3, rps4, rps7, rps8, rps11, rps12, rps14, rps18, rps19, ycf3, and ycf4. Unambiguously aligned regions were selected for further analyses and concatenated. The overlapping region of psbD and psbC was excluded. The data matrix is available online as Supplementary Material. The selected regions included 26,937 nucleotide sites. Because RNA editing is abundant in Anthoceros angustus (Kugita et al. 2003), we used cDNA sequence of A. angustus. The amino acid matrix comprised 20 taxa and 8,979 amino acid sites. Pairwise maximum-likelihood (ML) distances were calculated using ProtML (Adachi and Hasegawa 1996) and a Neighbor-Joining (NJ) tree was obtained with Njdist (Adachi and Hasegawa 1996). A local rearrangement search was performed with the NJ tree as the starting tree. Local bootstrap probabilities were calculated with resampling using the estimated log-likelihood method (Adachi and Hasegawa 1996). A user tree was prepared for each of the 105 possible bifurcating topologies among six groups of taxa: outgroups (O), mosses (M), liverworts (L), hornworts (H), monilophytes (F), and seed plants (S). The monilophytes include ferns, psilotophytes (whisk ferns), and equisetophytes (horsetails) (Kenrick and Crane 1997b, p. 228; Pryer et al. 2001; Judd et al. 2002, pp. 162–168), and in our study they are represented by Adiantum capillus-veneris and Psilotum nudum. A. capillus-veneris and P. nudum are forced to form a clade, and the topologies within seed plants and within the outgroup were constrained as the ML tree found in the local rearrangement search (fig. 1). The estimated log-likelihood under the JTT model (Jones, Taylor, and Thornton 1992), adjusted with the empirical amino acid frequency (JTT-F model) of each site under the 105 topologies, was calculated using ProtML (Adachi and Hasegawa 1996). The log-likelihood under the JTT model (Jones, Taylor, and Thornton 1992) was calculated for each gene to take account of rate differences among genes. The likelihood under a discrete gamma model of rate heterogeneity among sites (Yang 1994) was calculated with PAML version 3.13d (Yang 1997). The P values for the bootstrap and approximately unbiased (AU) test (Shimodaira 2002) were calculated using the estimated log-likelihood with CONSEL (Shimodaira and Hasegawa 2001). Maximum parsimony analyses were performed with PAUP* version 4.0b10 (Swofford 1998). The most parsimonious trees were recovered with a heuristic search strategy employing tree bisection and reconnection (TBR) with 1,000 random addition sequence replications. Bootstrap tests were performed with 1,000 replications with 10 replications of random addition sequence in each run. Analyses to overcome compositional heterogeneity with paralinear/LogDet distances (Lake 1994; Lockhart et al. 1994) followed Nishiyama and Kato (1999).
Table 1 Accession Numbers and Codon Usage for Leucine in Chloroplast Genomes
FIG. 1.— Phylogenetic relationship of representative lineages in land plants inferred from a maximum-likelihood analysis using 8,979 amino acid residues for 51 genes from 20 chloroplast genomes. Local bootstrap probabilities are shown on the branches. The horizontal branch lengths are proportional to the estimated number of amino acid substitutions per site. This tree is rooted using all the taxa other than embryophytic land plants. Higher taxonomic categories and common names are indicated.
Results
Heterogenetity in base composition at all codon positions is recognized as a problem for phylogenetic analyses of chloroplast genes (Steel, Lockhart, and Penny, 1993; Nishiyama and Kato 1999). We observed such compositional differences across entire chloroplast genomes; GC composition ranged from 12.9% in the third codon of Marchantia polymorpha to 38.3% in Adiantum capillus-veneris. We also found differences in codon usage for the first codon positions of leucine (table 1). Fifty-five percent of the leucine codons start with U in Nephroselmis olivacea compared to 26% in Chaetosphaeridium globosum. Most conventional methods of phylogenetic analysis are statistically consistent (i.e., reach the true topology with infinite data) only when the sequences evolved under a stationary Markov process (Gu and Li 1996). The large differences in the base composition indicate that the evolutionary process is not stationary. Therefore, conventional methods may give positively misleading results with high bootstrap probabilities (Lockhart et al. 1994). To overcome this problem, we applied three approaches: (1) we eliminated codon usage differences by analyzing amino acids sequences; (2) we removed the sites exhibiting compositional heterogeneity, i.e., all the leucine codons and third codon positions; and (3) we used LogDet/paralinear distances (Lake 1994; Lockhart et al. 1994; Steel 1994; Gu and Li 1996), which are consistent under nonstationary processes.
Analysis of Deduced Amino Acid Sequence Data
A local rearrangement search starting with the NJ tree (Saitou and Nei 1987) under the JTT-F model (Jones, Taylor, and Thornton 1992; Adachi and Hasegawa 1996) found an ML tree (ln likelihood = –89478.09; fig. 1). The three bryophytes, Marchantia polymorpha, Physcomitrella patens, and Anthoceros angustus, formed a monophyletic group with high local bootstrap probability (99%). Previously well-recognized monophyletic groups, the eudicots, monocots, angiosperms, seed plants, and land plants (reviewed in Judd et al. 2002, pp. 153–183), each formed a monophyletic group with high local bootstrap probability. Vascular plants formed a monophyletic group, though the bootstrap value was low (55%). This contrasts previous studies with high bootstrap support for vascular plants (Hedderson, Chapman, and Cox 1998; Nickrent et al. 2000). To test whether other possible trees have a significantly less likelihood than the ML tree, the bootstrap test and AU test (Shimodaira 2002) were used for all 105 possible topologies among seed plants, monilophytes, hornworts, liverworts, mosses, and outgroups. We constrained the topology within seed plants and within outgroup taxa. In the bootstrap test, topology 1 (fig. 2), which had the same topology as the ML tree, had 52.8% bootstrap support. Topology 2 (fig 2), which had the second highest likelihood (ln = –89480.91), occupied 44.3% of the bootstrap replicates. Bryophytes were also monophyletic in topology 2, but the bryophyte clade was nested within vascular plants, and monilophytes were basal among land plants. Other topologies were significantly less likely. The AU test showed that all topologies except topology 2 were significantly less likely than topology 1 at the 5% level (table 2; table S1 in the online Supplementary Material).
FIG. 2.— The best three alternative topologies found in the analyses of 105 trees. S, seed plants; F, monilophytes (Adiantum capillus-veneris plus Psilotum nudum); H, a hornwort (Anthoceros angustus); L, a liverwort (Marchantia polymorpha); M, a moss (Physcomitrella patens); O, outgroup taxa.
Table 2 Statistical Test for the 105 Topologies Under Various Models of Rate Heterogeneity
To consider rate heterogeneity among genes and sites, we performed separate gene analyses, in which each gene is assumed to have evolved independently ("independent gene" in table 2). This was performed under the JTT model using ProtML. We also performed a discrete gamma approximation with eight categories ("gamma 8 category" in table 2) using PAML (Yang 1997). The log-likelihood of every tree was calculated. These analyses also supported topology 1 as the ML tree (table 2). We also obtained the same topology when we used the genomic sequence of Anthoceros angustus to deduce the amino acid sequence (data not shown), indicating that RNA editing does not cause serious problems for these phylogenetic analyses. We also tested 20 data sets, each one omitting one taxon from the matrix. Although the monophyly of vascular plants changed when an outgroup taxon or Adiantum capillus-veneris was removed, the bryophyte monophyly was robust against taxon removal (data not shown).
Analysis of Nucleotide Sequence Data
We found a single most parsimonious tree (14,487 steps) from a TBR search based on nucleotide sequence data omitting third codon positions. The liverwort Marchantia polymorpha was resolved as the earliest diverging lineage in land plants, but bootstrap support for the phylogenetic relationships among bryophytes was low (fig. 3). When all the leucine codons and third codon positions were excluded, Marchantia polymorpha and Physcomitrella patens formed a monophyletic group (fig. 4). Phylogenetic relationships between this moss + liverwort clade and other taxa did not have high bootstrap support.
FIG. 3.— The most parsimonious tree using nucleotides of the first and second codon positions. 4,228 of 17,958 sites were parsimony-informative. Tree length = 14,487. Consistency index excluding uninformative characters = 0.52. Retention index = 0.62. Bootstrap probabilities are indicated on branches. Branch lengths are proportional to the number of substitutions on the branches by ACCTRAN optimization.
FIG. 4.— The most parsimonious tree excluding the nucleotide sites coding leucine and third codon positions. 2,731 of 14,860 sites were parsimony-informative. Tree length = 8,998. Consistency index excluding uninformative characters = 0.63. Retention index = 0.65. Bootstrap probabilities are indicated on branches. Branch lengths are proportional to the number of substitutions on the branches by ACCTRAN optimization.
Because LogDet/paralinear distances assume a uniform rate among sites (Lake 1994; Lockhart et al. 1994; Steel 1994; Gu and Li 1996), we used only the fourfold degenerate sites, of which there were 2,695 across taxa. An NJ tree was calculated using paralinear/LogDet distances (fig. 5). Again, the bootstrap values for the basal branches were low.
FIG. 5.— Neighbor-Joining tree using LogDet distance based on 2,695 fourfold degenerate sites. The number on each branch indicates a bootstrap probability (%) estimated by 10,000 replicates. Branch lengths are proportional to the LogDet distance.
Discussion
The monophyly of extant bryophytes was inferred by our analyses using translated amino acid sequence data of currently available chloroplast genomes. Analyses using nucleotide sequences did not resolve phylogenetic relationships of the three bryophytes with statistical confidence. Amino acid and nucleotide data each have their own merits and drawbacks. Amino acid sequences have frequently been used to infer deep phylogeny because they avoid problems with saturation of silent substitution and differential GC content (Russo, Takezaki, and Nei 1996; Nishiyama and Kato 1999; Simmons, Ochoterena, and Freudenstein 2002 and references therein). But, Simmons, Ochoterena, and Freudenstein (2002) criticized the use of amino acid translation, showing that nucleotide data outperform deduced amino acid data in a phylogenetic analysis of angiosperms, which diverged during the last 200 Myr (Judd et al. 2002, pp. 175–176). However, analyses using nucleotide sequences may not be robust when nucleotide compositions are apparently different among taxa, because phylogenetic analyses usually assume a stationary Markov process (Lockhart et al. 1994). In an attempt to remove characters with different nucleotide compositions, we excluded third codon positions (fig. 3) or the nucleotide sites coding leucine and third codon positions (fig. 4). We also compensated for differences in nucleotide composition by LogDet transformation (fig. 5). Neither the removal of characters nor the LogDet transformation resolved the relationships of extant bryophytes with high statistical support, whereas resolution with high statistical support was achieved with amino acid translation. These results might be explained if a uniform rate of change does not describe evolution of the fourfold degenerate sites. These results may also suggest that nucleotide composition bias can increase in proportion to divergence time among taxa. Nucleotide sequences including the third codon position data are more likely to be useful for inferring phylogeny among taxa that diverged relatively recently, such as angiosperms, which diverged within 200 Myr (Judd et al. 2002, pp. 175–176). If compositional heterogeneity is not extreme and the spatial pattern of substitution is simple, it would follow that nucleotide data will outperform deduced amino acid data because the former has more phylogenetic information than the latter. However, when sequences become highly diverged, the spatial substitution pattern may become more complex and problematic for distance-based correction formulae (Steel, Huson, and Lockhart 2000). The main lineages of land plants probably diverged as many as 470 MYA (Kenrick and Crane 1997a, 1997b; Kenrick 2000).
Phylogenetic analyses using DNA sequences suggest that extant bryophytes are paraphyletic (Lewis, Mishler, and Vilgalys 1997; Duff and Nickrent 1999; Nishiyama and Kato 1999; Nickrent et al. 2000). However, only the MP-3Ti analysis by Nickrent et al. (2000) provided evidence of strong statistical support. Moreover, monophyly of bryophytes was not rejected in the Kishino-Hasegawa (1989) test by Nickrent et al. (2000). The sister relationship of mosses and liverworts received relatively strong support in recent analyses using multiple genes (Nishiyama and Kato 1999; Nickrent et al. 2000). Marchantia appears on a basally diverging branch in land plants with high bootstrap probability in a combined analysis of atpB, rbcL, nad5, and the small subunit rRNA gene (Karol et al. 2001). However, that study used DNA sequences of chloroplast genes without attempting to compensate for the nucleotide composition bias, which means that the bootstrap probabilities may not be a good estimator of the reliability of branches and tree topologies (Lockhart et al. 1994; Swofford et al. 2001; Felsenstein 2003, pp. 272–274). Therefore, the extant bryophyte monophyletic hypothesis inferred in this study stands as the best one statistically supported by molecular sequence data.
Phylogenetic studies of land plant relationships using morphological characters are also incongruent. Most studies suggest that extant bryophytes are paraphyletic (Mishler and Churchill 1984, 1985; Renzaglia et al. 2000), whereas monophyly of extant bryophytes and a sister relationship of mosses and liverworts were inferred using characters of spermatogenesis (Garbary, Renzaglia, and Duckett 1993; Renzaglia et al. 2000; Renzaglia and Garbary 2001). Analyses of spermatogenesis characters also supported monilophytes involving ferns, psilotophytes (whisk ferns), and equisetophytes (horsetails), which are supported by DNA sequence data (Nickrent et al. 2000; Pryer et al. 2001). If our inference of monophyletic bryophytes reflects the true relationships, this would suggest that spermatogenesis characters have retained more phylogenetic signal than other morphological characters.
While our hypothesis was supported with high statistical confidence with presently available data, the bryophyte monophyly hypothesis should be tested with more data and further analysis. Current analyses include only one sample per group from mosses, liverworts, and hornworts and lack a major lineage of land plants, lycophytes. Some major lineages of bryophytes and lycophytes should be included based on previous phylogenetic analyses (Lewis, Mishler, and Vilgalys 1997; Garbary and Renzaglia 1998; Hedderson, Chapman, and Cox 1998; Nickrent et al. 2000; Renzaglia et al. 2000). Sphagnum, Takakia, Andreaea, and Andraeobryum from mosses and Treubia, Haplomitrium, Blasia, one simple thalloid, and one leafy liverwort from liverworts are candidate representatives. In addition to one homosporous and one heterosporous lycophyte, equisetophytes should be targeted for vascular plants. Furthermore, data from Charales may be a better outgroup to land plants than Coleochaetales (Turmel et al. 2002). Mitochondrial and nuclear genome data are also candidate sources of phylogenetic information that could provide an independent test of results based on chloroplast genome data.
The fossil record of plants older than rhyniophytes includes cuticles (Edwards, Duckett, and Richardson 1995) and spore tetrads (Gray 1993). These fossils have been considered to be related to extant liverworts (Edwards, Duckett, and Richardson 1995; Taylor 1995). Our results are not in line with this interpretation because we infer that liverworts do not descend from a basal node in bryophytes (fig. 2). Our results, however, are consistent with a "bryophyte-like ancestor" hypothesis (Graham and Gray 2001), whereby the above fossils represent a common ancestor of the three extant groups of bryophytes.
Supplementary Material
The complete table for statistical tests among 105 topologies is available online as a supplementary table. The nucleotide data matrix is available online in nexus format.
Acknowledgements
We thank two anonymous reviewers, Brent Mishler, Hiroyuki Akiyama, Hironori Deguchi, and members of the Hasebe laboratory for discussion and comments on the manuscript. Work on Anthoceros angustus has been published under the name of A. formosae. Thanks also to Carol Rowe for help sequencing the Adiantum chloroplast genome. The computations were partly done in the Computer Lab of the National Institute for Basic Biology. The research was supported by National Science Foundation grant #DEB-0228432 to P.G.W. and grants by MEXT and JSPS. Adiantum capillus-veneris, Anthoceros angustus, Physcomitrella patens, and Psilotum nudum were sequenced by P.G.W., R.B.S, and M.H.; M.K. and K.Y.; M.S. and C.S.; and T.Y. and K.Y, respectively. Correspondence in general should be addressed to M.H. (e-mail: mhasebe@nibb.ac.jp). Correspondence and requests for materials for chloroplast DNAs should be addressed to P.G.W. (Adiantum, e-mail: wolf@biology.usu.edu), K.Y. (Anthoceros, e-mail: sckyosi@ipc.shizuoka.ac.jp), M.S. (Physcomitrella, e-mail: sugita@gene.nagoya-u.ac.jp), and K.Y. (Psilotum, e-mail: yamada@sci.toyama-u.ac.jp).
References
Adachi, J., and M. Hasegawa. 1996. Molphy Version 2.3: programs for molecular phylogenetics based on maximum likelihood. The Institute of Statistical Mathematics, Tokyo, Japan.
Chapman, R. L., and D. A. Waters. 2002. Green algae and land plants—an answer at last? J. Phycol. 38:237–240.
Cummings, M. P., S. P. Otto, and J. Wakeley. 1995. Sampling properties of DNA sequence data in phylogenetic analysis. Mol. Biol. Evol. 12:814–822.
Duff, R. J., and D. L. Nickrent. 1999. Phylogenetic relationships of land plants using mitochondrial small-subunit rDNA sequences. Am. J. Bot. 86:372.
Edwards, D., J. G. Duckett, and J. B. Richardson. 1995. Hepatic characters in the earliest land plants. Nature 374:635–636.
Felsenstein, J. 2003. Inferring phylogenies. Sinauer Associates, Sunderland, Mass.
Garbary, D. J., and K. S. Renzaglia. 1998. Bryophyte phylogeny and the evolution of land plants: evidence from development and ultrastructure. Pp. 45–63 in J. W. Bates, N. W. Ashton, and J. G. Duckett, eds. Bryology for the twenty-first century. Maney Publishing and British Bryological Society, Leeds, United Kingdom.
Garbary, D. J., K. S. Renzaglia, and J. G. Duckett. 1993. The phylogeny of land plants—a cladistic analysis based on male gametogenesis. Plant Syst. Evol. 188:237–269.
Graham, L. E., and J. Gray. 2001. The origin, morphology, and ecophysiology of early embryophytes: neontological and paleontological perspectives. Pp. 140–158 in P. G. Gensel and D. Edwards, eds. Plants invade the land: evolutionary and environmental perspectives. Columbia University Press, New York.
Graham, L. K., and L. W. Wilcox. 2000. The origin of alternation of generations in land plants: a focus on matrotrophy and hexose transport. Phil. Trans. R. Soc. Lond. B 355:757–766.
Gray, J. 1993. Major Paleozoic land plant evolutionary bio-events. Palaeogeogr. Palaeoclimatol. Palaeoecol. 104:153–169.
Gu, X., and W. H. Li. 1996. Bias-corrected paralinear and LogDet distances and tests of molecular clocks and phylogenies under nonstationary nucleotide frequencies. Mol. Biol. Evol. 13:1375–1383.
Hedderson, T. A., R. Chapman, and C. J. Cox. 1998. Bryophytes and the origins and diversification of land plants: new evidence from molecules. Pp. 65–77 in J. W. Bates, N. W. Ashton, and J. G. Duckett, eds. Bryology for the twenty-first century. Maney Publishing and British Bryological Society, Leeds, United Kingdom.
Jones, D. T., W. R. Taylor, and J. M. Thornton. 1992. The rapid generation of mutation data matrices from protein sequences. Comput. Appl. Biosci. 8:275–282.
Judd, W. S., C. S. Campbell, E. A. Kellogg, P. F. Stevens, and M. J. Donoghue. 2002. Plant systematics: a phylogenetic approach. Sinauer Associates, Sunderland, Mass.
Karol, K. G., R. M. McCourt, M. T. Cimino, and C. F. Delwiche. 2001. The closest living relatives of land plants. Science 294:2351–2353.
Katoh, K., K. Misawa, K. Kuma, and T. Miyata. 2002. MAFFT: a novel method for rapid multiple sequence alignment based on fast Fourier transform. Nucleic Acids Res. 30:3059–3066.
Kenrick, P. 2000. The relationships of vascular plants. Phil. Trans. R. Soc. Lond. B 355:847–855.
Kenrick, P., and P. R. Crane. 1997a. The origin and early evolution of plants on land. Nature 389:33–39.
———. 1997b. The origin and early diversification of land plants: a cladistic study. Smithsonian Institution Press, Washington, D.C.
Kishino, H., and M. Hasegawa. 1989. Evaluation of the maximum likelihood estimate of the evolutionary tree topologies from DNA sequence data, and the branching order in Hominoidea. J. Mol. Evol. 29:170–179.
Kugita, M., A. Kaneko, Y. Yamamoto, Y. Takeya, T. Matsumoto, and K. Yoshinaga. 2003. The complete nucleotide sequence of the hornwort (Anthoceros formosae) chloroplast genome: insight into the earliest land plants. Nucleic Acids Res. 31:716–721.
Lake, J. A. 1994. Reconstructing evolutionary trees from DNA and protein sequences: paralinear distances. Proc. Natl. Acad. Sci. USA 91:1455–1459.
Lewis, L. A., B. D. Mishler, and R. Vilgalys. 1997. Phylogenetic relationships of the liverworts (Hepaticae), a basal embryophyte lineage, inferred from nucleotide sequence data of the chloroplast gene rbcL. Mol. Phylogenet. Evol. 7:377–393.
Lockhart, P. J., M. A. Steel, M. D. Hendy, and D. Penny. 1994. Recovering evolutionary trees under a more realistic model of sequence evolution. Mol. Biol. Evol. 11:605–612.
Mishler, B. D., and S. P. Churchill. 1984. A cladistic approach to the phylogeny of the "bryophytes." Brittonia 36:406–424.
———. 1985. Transition to a land flora: phylogenetic relationships of the green algae and bryophytes. Cladistics 1:305–328.
Nickrent, D. L., C. L. Parkinson, J. D. Palmer, and R. J. Duff. 2000. Multigene phylogeny of land plants with special reference to bryophytes and the earliest land plants. Mol. Biol. Evol. 17:1885–1895.
Nishiyama, T., and M. Kato. 1999. Molecular phylogenetic analysis among bryophytes and tracheophytes based on combined data of plastid coded genes and the 18S rRNA gene. Mol. Biol. Evol. 16:1027–1036.
Pryer, K. M., H. Schneider, A. R. Smith, R. Cranfill, P. G. Wolf, J. S. Hunt, and S. D. Sipes. 2001. Horsetails and ferns are a monophyletic group and the closest living relatives to seed plants. Nature 409:618–622.
Qiu, Y. L., Y. Cho, J. C. Cox, and J. D. Palmer. 1998. The gain of three mitochondrial introns identifies liverworts as the earliest land plants. Nature 394:671–674.
Qiu, Y. L., J. Lee, F. Bernasconi-Quadroni, D. E. Soltis, P. S. Soltis, M. Zanis, E. A. Zimmer, Z. Chen, V. Savolainen, and M. W. Chase. 1999. The earliest angiosperms: evidence from mitochondrial, plastid and nuclear genomes. Nature 402:404–407.
Renzaglia, K. S., R. J. T. Duff, D. L. Nickrent, and D. J. Garbary. 2000. Vegetative and reproductive innovations of early land plants: implications for a unified phylogeny. Phil. Trans. R. Soc. Lond. B 355:769–793.
Renzaglia, K. S., and D. J. Garbary. 2001. Motile gametes of land plants: diversity, development, and evolution. Crit. Rev. Plant Sci. 20:107–123.
Russo, C. A. M., N. Takezaki, and M. Nei. 1996. Efficiencies of different genes and different tree-building methods in recovering a known vertebrate phylogeny. Mol. Biol. Evol. 13:525–536.
Saitou, N., and M. Nei. 1987. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4:406–425.
Shimodaira, H. 2002. An approximately unbiased test of phylogenetic tree selection. Syst. Biol. 51:492–508.
Shimodaira, H., and M. Hasegawa. 2001. CONSEL: for assessing the confidence of phylogenetic tree selection. Bioinformatics 17:1246–1247.
Simmons, M. P., H. Ochoterena, and J. V. Freudenstein. 2002. Amini acid vs. nucleotide charactors: challenging preconceived notions. Mol. Phylogenet. Evol. 24:78–90.
Steel, M. 1994. Recovering a tree from the leaf volourations it generates under a Markov model. Appl. Math. Lett. 7:19–23.
Steel, M. A., D. Huson, and P. J. Lockhart. 2000. Invariable site models and their use in phylogeny reconstruction. Syst. Biol. 49:225–232.
Steel, M. A., P. J. Lockhart, and D. Penny. 1993. Confidence in evolutionary trees from biological sequence data. Nature 364:440–442.
Sugiura, C., Y. Kobayashi, S. Aoki, C. Sugita, and M. Sugita. 2003. Complete chloroplast DNA sequence of the moss Physcomitrella patens: evidence for the loss and relocation of rpoA from the chloroplast to the nucleus. Nucleic Acids Res. 31:5324–5331.
Swofford, D. L. 1998. PAUP*: phylogenetic analysis using parsimony (*and other methods). Version 4. Sinauer Associates, Sunderland, Mass.
Swofford, D. L., P. J. Waddell, J. P. Huelsenbeck, P. G. Foster, P. O. Lewis, and J. S. Rogers. 2001. Bias in phylogenetic estimation and its relevance to the choice between parsimony and likelihood methods. Syst. Biol. 50:525–539.
Taylor, W. A. 1995. Spores in earliest land plants. Nature 373:391–392.
Turmel, M., M. Ehara, C. Otis, and C. Lemieux. 2002. Phylogenetic relationships among streptophytes as inferred from chloroplast small and large subunit rRNA gene sequences. J. Phycol. 38:364–375.
Wellman, C. H., P. L. Osterloff, and U. Mohiuddin. 2003. Fragments of the earliest land plants. Nature 425:282–285.
Wolf, P. G., C. A. Rowe, R. B. Sinclair, and M. Hasebe. 2003. Complete nucleotide sequence of the chloroplast genome from the leptosporangiate fern Adiantum capillus-veneris L. DNA Res. 10:59–65.
Yang, Z. 1994. Maximum likelihood phylogenetic estimation from DNA sequences with variable rates over sites: approximate methods. J. Mol. Evol. 39:306–314.
———. 1997. PAML: a program package for phylogenetic analysis by maximum likelihood. Comput. Appl. Biosci. 13:555–556.(Tomoaki Nishiyama*, Paul )