Molecular Evolution and Phylogeny of the Atpb– Rbcl Spacer of Chloroplast DNA in the True Mosses

417

Molecular evolution and phylogeny of the atpB– rbcL spacer of chloroplast DNA in the true mosses

Tzen-Yuh Chiang and Barbara A. Schaal

Abstract: The nucleotide variation of a noncoding region between the atpB and rbcL genes of the chloroplast genome was used to estimate the phylogeny of 11 species of true mosses (subclass Bryidae). The A+T rich (82.6%) spacer sequence is conserved with 48% of bases showing no variation between the ingroup and outgroup. Rooted at liverworts, Marchantia and Bazzania, the monophyly of true mosses was supported cladistically and statistically. A nonparametric Wilcoxon Signed-Ranks test Ts statistic for testing the taxonomic congruence showed no significant differences between gene trees and organism trees as well as between parsimony trees and neighbor-joining trees. The reconstructed phylogeny based on the atpB–rbcL spacer sequences indicated the validity of the division of acrocarpous and pleurocarpous mosses. The size of the chloroplast spacer in mosses fits into an evolutionary trend of increasing spacer length from liverworts through ferns to seed plants. According to the relative rate tests, the hypothesis of a molecular clock was supported in all species except for Thuidium, which evolved relatively fast. The evolutionary rate of the chloroplast DNA spacer in mosses was estimated to be (1.12 ± 0.019) × 10–10 nucleotides per site per year, which is close to the nonsynonymous substitution rates of the rbcL gene in the vascular plants. The constrained molecular evolution (total nucleotide substitutions, K ≈ 0.0248) of the chloroplast DNA spacer is consistent with the slow evolution in morphological traits of mosses. Based on the calibrated evolutionary rate, the time of the divergence of true mosses was estimated to have been as early as 220 million years ago. Key words: atpB–rbcL noncoding spacer, chloroplast DNA, gene tree, molecular evolution, molecular clock, mosses, phylogeny. Résumé : La variation nucléotidique dans la région non-codante située entre les gènes atpB et rbcL du génome chloroplastique a été employée afin d’estimer la phylogénie chez onze espèces de mousses véritables (sous-classe des Bryidae). La séquence riche en A+T (82,6 %) de l’espaceur est conservée puisque 48 % des positions ne montrent aucune variation entre les groupes interne et externe. Prenant racine chez les hépatiques Marchantia et Bazzania, le carac- tère monophylétique des mousses véritables était supporté à la fois sur les plans cladistiques et statistiques. Une statistique non-paramétrique Ts de Wilcoxon pour échantillons appariés permettant de tester la conformité taxonomique n’a montré aucune différence significative entre les arbres basés sur les gènes et ceux plus globaux (au niveau de l’organisme) de même qu’entre les arbres dérivés de méthodes d’analyse de parsimonie ou « neighbor-joining ». La phylogénie fondée sur la séquence de l’espaceur atpB–rbcL a validé la justesse de la division des mousses en deux groupes : les acrocarpes et les pleurocarpes. La taille de l’espaceur chloroplastique chez les mousses dénote une ten- dance à l’allongement de l’espaceur à partir des hépatiques, en passant par les fougères, jusqu’aux plantes à graines. Selon les tests de taux relatifs, l’hypothèse d’une horloge moléculaire était supportée chez toutes les espèces à l’exception du Thuidium, lequel a évolué relativement rapidement. Le taux d’évolution de l’espaceur chloroplastique chez les mousses a été estimé à (1,23 ± 0,019) × 10–10 nucléotide par site par année, ce qui est proche des taux de substitution non-synonyme au niveau du gène rbcL chez les plantes. L’évolution moléculaire contrainte (K ≈ 0,0248) chez l’espaceur chloroplastique est conforme avec le taux d’évolution lent des caractères morphologiques chez les mousses. En fonction du taux d’évolution calibré, le moment de la divergence des mousses véritables est estimé s’être produit aussi tôt qu’il y a 220 millions d’années. Mots clés : espaceur non-codant atpB–rbcL, ADN chloroplastique, arbre génique, évolution moléculaire, horloge moléculaire, mousses, phylogénie. [Traduit par la Rédaction] Chiang and Schaal 426

Corresponding Editor: G. Bellemare. Received March 15, 1999. Accepted September 30, 1999. Published on the NRC Research Press website on April 27, 2000. T.-Y. Chiang.1 Department of Biology, Cheng-Kung University, Tainan, Taiwan 700. B.A. Schaal. Department of Biology, Washington University, St. Louis, MO 63130–4899, U.S.A.

1Author to whom all correspondence should be addressed (e-mail: [email protected]).

Introduction the tissue mixture was centrifuged at 14 000 rpm for 15 min at room temperature. The supernatant was transferred to an The chloroplast genome has been extensively used for evo- Eppendorf tube followed by addition of 1.2 mL of absolute etha- lutionary and systematic studies (Palmer 1987; Avise 1994). nol. After overnight incubation at 4°C, DNA was recovered by Compared to the nuclear DNA and animal mitochondrial centrifuging the mixture at 14 000 rpm for 15 min at 4°C. The DNA, chloroplast genes evolve relatively slowly (Clegg et brown to black DNA pellet was rinsed in 70% ethanol and centri- al. 1991; Li 1997). Among molecular markers, the rbcL fuged for 5 min at 10 000 rpm. The DNA pellet was resuspended in 20 µL TE. gene has been widely used for systematics at higher levels The extracted genomic DNA was purified on a low-melting- (e.g. Olmstead et al. 1992; Chase et al. 1993; Qiu et al. point agarose gel to remove secondary compounds and RNA. The 1993; Hasebe et al. 1994; Nickrent and Soltis 1995). High band on the gel containing the DNA of the correct size was cut and levels of homoplasy in cpDNA sequence have been found in transferred into an Eppendorf tube. Equal weights of distilled water some groups (Kim et al. 1992) making the sequence some- were added to the gel block containing the purified DNA. Prior to what less reliable in phylogenetic reconstruction than is usu- use of the DNA for polymerase chain reaction (PCR), the gel was ally assumed. Moreover, since the chloroplast genome is heated in a 65°C water bath for 3 min. uniparentally inherited (Sears 1980), the phylogeny inferred Two universal primers, rbcL-1 (5′-AACACCAGCTTTRAATC- from cpDNA sequences may represent a gene tree rather CAA-3′) and atpB-1 (5′-ACATCKARTACKGGACCAATAA-3′), than an organism tree (Pamilo and Nei 1988; Zurawski and were developed for amplifying and sequencing the rbcL-atpB spac- Clegg 1987). ers (Chiang et al. 1998) from the sequences of Marchantia (Umesono et al. 1988), tobacco (Shinozaki et al. 1986), and rice The noncoding region between rbcL and atpB genes has (Nishizawa and Hirai 1987). The PCR amplification protocol uti- been used in phylogenetic studies (e.g., Savolainen et al. lized two units of Taq polymerase (New England BioLab), the Taq 1994; Ehrendorfer et al. 1994; Natali et al. 1995) and the buffer (500 mM KCl, 100 mM Tris–HCl, pH 9.0, and 1.0% Triton function (such as promoters for rbcL) of this spacer in vas- X–100), 2.5 mM MgCl2, 10 pmol of each primer, and 8 mM dNTP cular plants has been well documented (Orozco et al. 1990; in 100 µL reaction. PCR amplification was carried out in 30 cycles Manen et al. 1994; Mullet et al. 1985; Gruissem and of 94°C denaturing for 45 s, 57°C annealing for 1 min 15 s, and Zurawski 1985). The spacer region is variable in size, with 72°C extension for 1 min 15 s, followed by 72°C extension for differences among major groups of plants (Yoshinaga et al. 10 min and 4°C for storing. PCR products were polyacrylamide- 1992; Chiang et al. 1998). In spite of length differences, the gel-purified and sequenced by the dideoxy-mediated chain- evolution of the atpB–rbcL spacer sequence is constrained termination method (Sanger et al. 1977). The fmol™ DNA Se- quencing System (Promega), which uses Taq polymerase, was used relative to the rbcL gene in angiosperms (Zurawski et al. for sequencing. The detergent NP–40 (10%) was added to assist se- 1984). Until now no work has compared the spacer sequence quencing through G+C rich regions and secondary structure (Wang evolution between mosses and other major groups. et al. 1992). Both strands of DNA were sequenced with about 50- In this study, we investigate the tempo and mode of base overlap. evolution of the atpB–rbcL chloroplast spacer in mosses. Mosses have been described as “primitive” terrestrial plants Data analysis (Lemoigne 1970). Based on the fossil evidence, the pace of evolution in mosses is thought to be slower than in angio- Sequence alignment sperms (Delcourt and Delcourt 1991). However, no evidence Sequences were aligned by multiple alignments without weight- from molecular perspectives has been considered. ing transversions or transitions using the CLUSTAL V Program This study had four goals: (1) to reconstruct the gene tree (Higgins et al. 1992). The fixed gap penalty was 35 and the float- ing penalty was 4. The sequences of Marchantia polymorpha of atpB–rbcL spacer; (2) to investigate the evolutionary mode (Umesono et al. 1988) and Bazzania fauriana were used as out- of this spacer; (3) to test the hypothesis of a molecular groups. clock; and (4) to estimate the time of divergence from a common ancestor of the true mosses. Phylogenetic analyses The cladistic analyses of sequencing data were performed by the Materials and methods maximum parsimony using Phylogenetic Analysis Using Parsi- mony Program (PAUP v. 3.1.1., Swofford 1993) and the neighbor- joining (NJ) method using Molecular Evolutionary Genetics Anal- Plant materials ysis Program (MEGA v. 1.01, Kumar et al. 1993). Parsimony analy- Eleven species representing 11 families of both acrocarpous and ses were conducted using heuristic searches with TBR branch pleurocarpous mosses (subclass Bryidae) were sampled (Table 1). swapping, accelerated transformation (ACCTRAN), an uncon- Most of the plants, except for Rhytidium, Leskea, and Thuidium, strained number of maximum trees, and retention of multiple most were collected from the field in the United States and were air- parsimonious trees (MULPARS). Neighbor-joining analyses were dried without any special field treatment. Voucher specimens are conducted by calculating Kimura’s (1980) 2-parameter distance. deposited in the herbaria of Missouri Botanical Garden (Mo.) and Both strict (Sokal and Rohlf 1981) and 50% majority-rule Academia Sinica, Taipei (HAST). (Margush and McMorris 1981) consensus trees were computed rooted at both Marchantia and Bazzania. DNA extraction and sequencing Ag1 test (Huelsenbeck 1991) of skewed tree-length distributions Leaf tissue from single individuals was frozen in liquid nitrogen was calculated from 10 000 random trees generated by PAUP in or- and ground in Eppendorf tubes with a metal dounce. Genomic der to measure the information content of the data. Critical values µ DNA was extracted from the powdered tissue in 600 L2×CTAB of the g1 test were obtained from Hillis and Huelsenbeck (1992). (cetyltrimethylammonium bromide) buffer (Doyle and Doyle 1987) The fit of character data on phylogenetic hypotheses (Swofford with 0.4% (v/v) β-mercaptoethanol and incubated for1hat65°C. 1991) was evaluated and calculated by the consistency index, CI After adding equal volume of 24:1 chloroform : isoamyl alcohol, (Kluge and Farris 1969) and the retention index, RI (Archie 1989;

Table 1. Materials of true mosses and a liverwort (Bazzania, outgroup) collected for DNA isolation and nucleotide sequencing. Species Family Locality Voucher Size (bp) EMBL No. Hylocomium splendens Hylocomiaceae Smoky Mt., N.C., U.S.A. Chiang 31091 553 AJ249047 Rhytidium ruginosum Rhytidiaceace Sichuan, China Redfearn 35492 557 AJ249046 Pleurozium schreberi Amblystegiaceae Idaho, U.S.A. Chiang s. n. 556 AJ249048 Ptilium crista-castrensis Hypnaccae Smoky Mt., N.C., U.S.A. Chiang s. n. 556 AJ249045 Eurhynchium pulchellum Brachytheciaceae Idaho, U.S.A. Chiang s. n. 547 AJ249041 Thuidium cymbifolium Thuidiaceae Yunnan, China He 32026 546 AJ249042 Leskea gracilescens Leskeaceae Iowa, U.S.A. Allen 10910 523 AJ249044 Antitrichia curtipendula Leucodontaccae Washington, U.S.A. Chiang s. n. 555 AJ249038 Hedwigia ciliata Hedwigiaceae Kentucky, U.S.A. Chiang s. n. 555 AJ249043 Mnium sp. Mniaceae Washington, U.S.A. Chiang s. n. 538 AJ249039 Campylopus sp. Dicranaceae Missouri, U.S.A. Chiang s. n. 519 AJ249040 Bazzania fauriana Lepidoziaceae Tainan, Taiwan Chiang s. n. 470 AJ249037 Note: s. n., no collection number.

Farris 1989). The statistical significance of CI was determined ac- al. 1993; 841, Begonia, Liu et al. 1998; and 900, Quercus, cording to Klassen et al. (1991). The confidence of the clades was Hong et al. 1999), and to gymnosperms (1000, Cunninghamia, tested by bootstrapping (Efron 1982; Felsenstein 1985) with 400 Chiang et al. 1998), appears to be emerging. Insertions and replicates (Hedges 1992) of heuristic searches on the 50% majority (or) deletions (indels) are a common phenomenon in mosses rule trees. The nodes with bootstrap values greater than 0.70 are as well as in grasses (Golenberg et al. 1993). In 357 indel significantly supported with ≥95% probability (Hillis and Bull 1993). events, 289 (73.5%) are single-base indels, 37 (14.4%) are two-base indels, 23 (8.9%) are three- to seven-base indels. Tests of taxonomic congruence and alternative trees Eight (3%) large indels with more than 10 bases were ob- Different analytic methods (in this case, PAUP and MEGA) may re- served, respectively, in the Marchantia (11 bases between sult in different topologies. Moreover, the phylogeny inferred from positions 39 and 49; 23 bases between positions 290 and the chloroplast spacer sequence represents a gene phylogeny and 312; 12 bases between positions 320 and 331), Campylopus may conflict with the organism tree. To test the taxonomic congru- (10 bases between positions 26 and 35; 13 bases between ence between topologies as well as gene trees versus organism positions of 275 and 287; 18 bases between positions 302 trees, a nonparametric Wilcoxon Signed-Ranks Test was employed and 109), Leskea (33 bases between positions 401 and 433), (Templeton 1983; Larson 1994). Two-tailed probabilities were and Thuidium (12 bases between positions 301 and 312) se- used to examine the significance levels (Felsenstein 1985; statisti- quences. cal tables see Rohlf and Sokal 1981). The information on characters favoring each tree with signs of different steps according to the Nucleotides A and T are rich in the chloroplast spacer, assumption of parsimony was obtained from the computer program which is consistent with the nucleotide composition of most MACCLADE (Maddison and Maddison 1992). noncoding spacers and pseudogenes due to low functional constraints (cf. Li 1997). The average A+T content is 82.6%. Relative rate tests Among the taxa, Campylopus has the highest level of A The hypothesis of a molecular clock (Zuckerkandl and Pauling (40.7%), and Rhytidium, Hedwigia, and Thuidium have the 1965) was tested by relative rate tests (Sarich and Wilson 1973; highest levels of T (44.5%). Wu and Li 1985). The total number of nucleotide substitutions (K), The ratios between transitions and transversions obtained which is the number of transitional and transversional substitutions from MEGA ranged from 0.630 (between Campylopus and per site, was calculated from each lineage using Marchantia as the reference species. The data on number and ratio of transversion Antitrichia) to 1.40 (between Thuidium and Antitrichia) in versus transition between taxa was obtained from the MEGA pro- mosses. In total, 1161 transitions and 1132 transversions gram. The null hypothesis of a molecular clock suggests that the were observed (with a ratio of transitions/transversions of number of nucleotide substitutions between two lineages would be 1.03). Biased substitution patterns toward transitions, with the same. Based on the assumption of a normal distribution of nu- deviation from random mutation (with an expected transi- cleotide substitutions (Wu and Li 1985), the hypothesis of a molec- tions/transversions ratio of 0.5), have been found in several ular clock will be rejected with 95% significance, when the fast-evolving genes, such as primate mtDNA control region difference of substitution rates between two lineages is greater than (ratio ≈15.0–15.7, Kocher and Wilson 1991; Vigilant et al. 1.96 times the standard error (sx ). 1991; Tamura and Nei 1993) and nuclear satellite DNA (Wu et al. 1999), both of which are generally subject to very Results and discussion weak selective constraints (cf. Li 1997). However, more like the coding sequences of mtDNA (Brown et al. 1982), the DNA sequences and the mode of evolution atpB–rbcL noncoding spacer of the chloroplast DNA has a The size of the atpB–rbcL spacer is variable among moss much lower ratio between transitions and transversions both in families from 519 to 557 base pairs (average = 549 bp, angiosperms (e.g., 1.5 between barley and maize, Zurawski et Table 1). An evolutionary trend of increasing size of the al. 1984) and mosses (1.03). A lower transition/transversion chloroplast spacer from liverworts (470, Bazzania; 507, ratio indicates a conserved nature of this chloroplast spacer Marchantia), through mosses (549), ferns (ca. 600, Angiopteris, and low evolutionary rates in the plants (discussed below). Yoshinaga et al. 1992), angiosperms (899, grasses, Golenberg et But, unlike the strong functional and selective constraints in

Fig. 1. Variable sites of the aligned atpB–rbcL sequences of true mosses and outgroups (Marchantia and Bazzania).

the mtDNA coding region, the mechanisms causing the low 0.409–0.571. The absence of the biased trend may be closely variation in this noncoding spacer have remained unknown. correlated with the long evolutionary history between mosses In contrast to the bias toward transitions in true mosses, the and liverworts, which allowed the evolutionary changes in transition/transversion ratios between Marchantia and the the chloroplast spacer between the two lineages reached sat- mosses were much closer to random mutation with a range uration.

Fig. 1 (continued).

Phylogenetic reconstruction upon request from the authors. The sequences are conserved, Cladistic analyses were conducted on the aligned se- with 284 bases (48%) having no variation among moss taxa quences of 584 bases (Fig. 1). Aligned sequence is available and liverworts. Nevertheless, the level of synapomorphy,

Fig. 1. (concluded). Variable sites of the aligned atpB–rbcL sequences of true mosses and outgroups (Marchantia and Bazzania).

233 bases out of a total of 584 (39.9%), is high. Within the (Fig. 2). A g1 statistic of –1.275 indicates significant signal variable bases, 77% (233 of 300) were synapomorphies. (P ≤ 0.01) of the data matrix on the phylogenetic hypothesis. A single most parsimonious tree with 394 steps, a CI of The monophyly of the true mosses was significantly sup- 0.871 (P ≤ 0.01), and an RI of 0.571, was recovered by PAUP ported with a bootstrap value of 0.90 (P ≥ 95%). Within true

Fig. 2. The most parsimonious tree rooted at Marchantia and Fig. 3. The K2P tree recovered by MEGA based on the genetic Bazzania reconstructed by PAUP based on the nucleotide variation distance of the chloroplast DNA spacer between atpB and rbcL of the atpB–rbcL spacer of the chloroplast DNA. Numbers at genes in true mosses and outgroups (Marchantia and Bazzania). nodes indicate the bootstrap values of the clades. A, acrocarpous Numbers at nodes indicate the bootstrap values. mosses; P, pleurocarpous mosses.

Apparently, this close relationship between families was not supported by the atpB–rbcL spacer sequence analysis. In contrast, a more traditional classification, which places the Hedwigiaceae in the acrocarpous mosses and closely related to the Grimmiaceae (Hedwig 1801), was supported by the molecular data. Nonetheless, the systematic position of the Hedwigiaceae remains problematic (cf. Mishler and de Luna 1991). The phylogeny inferred from the chloroplast DNA sequences may merely represent gene trees instead of the organism trees. To achieve better understanding of the phylogeny of the Hedwigiaceae, more molecular data as well as ontogenetic data are required.

Tests of alternative trees The most parsimonious tree identified by PAUP is not com- pletely consistent with the organism trees inferred from both mosses, two monophyletic groups, i.e., the acrocarpous mos- morphological data (Rohrer 1985) and the combined data of ses (bootstrap value = 87%) and the pleurocarpous mosses ITS (internal transcribed spacer) of nrDNA and atpB–rbcL (bootstrap value = 77%), were recognized and well sup- spacers (Chiang 1994), which suggests that Hylocomium is ported statistically. The close relationship between Hedwigia related to Rhytidium and that Pleurozium and Antitrichia are and Mnium was revealed by the cladistic analysis with a closely related. Templeton’s test was used to determine the bootstrap value of 74%. Two nodes had bootstrap values character fit to the topology of the chloroplast tree and the greater than 50% (but less than 70%): the clade of Hylo- organism trees. Eleven characters favored the chloroplast comium and Pleurozium (56%), and the clade of Leskea and tree and three characters favored the organism trees. A Ts- Ptilium (63%). statistic of 22.5 obtained from the Wilcoxon Signed-Ranks Neighbor-joining analysis was conducted based on the tests therefore suggested a non-significant difference be- distance matrix. A K2P tree (Fig. 3) was obtained with com- tween the gene tree and organism trees (P ≤ 0.0688). That is, plete resolution, but the topology is not fully congruent with the organism trees are suboptimal to the trees inferred from the tree identified by PAUP. On the neighbor-joining tree, the chloroplast spacer. Hylocomium was closely related to Eurhynchium and Anti- Likewise, a significant difference between K2P tree and trichia instead of Pleurozium. Rhytidium and Thuidium were parsimony tree is not supported by Templeton’s test. Ten the basal taxa of the pleurocarpous mosses instead of clus- characters, of which two have sign of +2 (two steps shorter tering together. than in the alternative tree) and eight have sign of +1, fa- Interestingly, both analytic methods supported the taxo- vored the parsimony tree and six characters, with sign of –1, nomic position of Hedwigia being more closely related to favored the K2P trees. The Ts statistic is equal to 45.0 (P < the acrocarpous mosses instead of the pleurocarpous mosses. 0.10, non-significant). Recent classifications, such as Buck and Vitt (1986), have The taxonomic hypothesis of the Hedwigiaceae belonging placed the Hedwigiaceae close to the family Leucodontaceae, to the pleurocarpous mosses was also tested. The Wilcoxon another member of the pleurocarpous mosses, based on im- Signed-Ranks test showed that 27 characters, with a sign of mersed capsules, autoicous plants, and papillose leaf-cells. +1, favored the chloroplast tree, in which the Hedwigiaceae

is related to acrocarpous mosses, and five characters, with sign of –1, favored the alternative tree. The Ts statistic of 82.5 suggested a significant difference (P < 0.01). That is, the hypothesis of the Hedwigiaceae in the pleurocarpous mosses was rejected by the atpB–rbcL spacer sequence analysis.

Relative rate tests The differences in nucleotide substitutions per site between mosses and Marchantia varied from 0.096 to 0.104 with an average of 0.0985 ± 0.0017 (sx ). In contrast, differences in nucleotide substitutions within mosses were highly = standard error) (below diagonal). as the reference species; where K13 variable from 0.008 to 0.050 (average = 0.0248). Obviously, x s

( the nucleotide substitutions obtained within mosses are less , Thuidium. x s

11 than those between mosses and Marchantia due to the long time of divergence between mosses and liverworts from their

Marchantia common ancestor. It is noteworthy that the evolutionary rate of the atpB– rbcL noncoding spacer in mosses is much slower than in

, Eurhynchium; vascular plants. For example, the difference in nucleotide

10 substitution between maize and barley is 0.0691 (Zurawski et al. 1984), which apparently have much shorter coales- cence time than do mosses. Furthermore, as a noncoding region, the atpB–rbcL spacer should have evolved faster than

, Antitrichia; the rbcL gene due to the lower functional constraints. How- 9 ever, compared to the third position substitution rate of 0.190 for the rbcL gene between barley and maize (Bousquet et al. 1992), the number of substitutions per site

Pleurozium; of the noncoding spacer in mosses is even slower (average =

8, 0.02). Accordingly, the evolutionary rate of atpB–rbcL spacers found here is close to the nonsynonymous rates of the rbcL gene in gymnosperms and ferns (Savard et al. 1994). Among the taxa analyzed, the chloroplast spacer of Thuidium evolved relatively rapidly. When pairwise compar- , Hylocomium;

7 isons of relative rate tests were made using Marchantia as a reference species, most lineages are congruent with the hypothesis of a molecular clock, except for the pairs of Thuidium and Leskea, and Thuidium and Pleurozium (Table 2).

, Rhytidium; The time between the common ancestor of true mosses 6 and other groups was estimated from the molecular clock constructed from the noncoding atpB–rbcL spacer. Savard et al. (1994) suggested that liverworts and seed plants diverged , Ptilium; 5 440 million years ago. According to a cladistic study of the phylogeny of bryophytes and related major groups (Mishler and Churchill 1984), bryophytes appear to be a paraphyletic , Leskea; group, within which mosses were more related to vascular ;4 plants than to liverworts or hornworts. Therefore, 440 million years can be used as the reference for the branching of mosses from a common ancestor. The rate of evolution for

Campylopus the chloroplast spacer was estimated to be (1.12 ± 0.019) × –10 3, 10 substitutions per site per year. The evolution of the true mosses, excluding Thuidium, can thus be traced back to 220 million years ago. Mnium; 2, Conclusions In this study, we investigated the molecular evolution of atpB–rbcL spacer of the chloroplast genome in the true Hedwigia; Differences (×100) in number of nucleotide substitutions per site K (=Kl3 – K23) for chloroplast DNA spacers using mosses. Not only the morphological traits, but also the mo- 0.05 (significance level).

:1, lecular evolution of the cpDNA spacer in mosses was con- Յ

P strained. The sequences of the noncoding region are highly Note * 12345 0.1556 0.5347 0.6008 0.062±0.40 0.369 0.4529 0.413 0.251±0.47 0.864 0.556 0.252±0.42 0.280 0.936 0.305 –0.190±0.42 0.188±0.51 0.137 0.00 0.310 0.144 0.815 –0.380±0.44 0.190±0.46 0.442 1.171 –0.250±0.45 0.126±0.45 0.240 –0.440±0.47 0.250 0.000±0.51 1.420 0.125±0.41 0.717 0.660 0.063±0.46 0.440±0.54 –0.130±0.42 0.342 0.419 –0.630±0.54 0.062±0.43 0.126±0.42 1.090 –0.130±0.54 –0.570±0.50 –0.440±0.31 –0.190±0.43 0.613 1.091 –0.130±0.54 –0.630±0.38 0.064±0.45 1.370 –0.380±0.53 –0.130±0.38 –0.640±0.50 0.791 –0.120±0.54 –0.130±0.31 –0.190±0.31 –0.820±0.57 1.290 –0.380±0.35 1.800 0.314±0.29 –0.130±0.33 1.740 –0.820±0.41 0.315±0.23 0.505±0.39 0.000 0.253±0.32 0.833 0.505±0.28 0.316±0.26 –0.380±0.35 0.505±0.29 0.000±0.24 1.000 0.506±0.32 –0.250±0.30 –0.250±0.39 0.000±0.27 –0.700±0.40 –0.250±0.25 0.000±0.25 –0.70±0.35 0.252±0.27 –0.440±0.31

Table 2. (K23) is the difference of substitutionsTaxa between species 1 (2) and species 3 (reference taxon) (above 1 diagonal) – absolute value of K/ 210 311 0.300 1.140 4 0.142 1.261conserved 0.222 1.442 5 0.394both 2.000* within 6 1.220 1.090 mosses 1.580 and 0.641 7 between 0.000 1.750 mosses 8 0.000 2.000* and liv- 9 0.933 1.190 1.890 10 –0.70±0.35 11

© 2000 NRC Canada Chiang and Schaal 425 erworts. The conserved nature of these sequences suggests Ehrendorfer, F., Manen, J.F., and Natali, A. 1994. cpDNA inter- that this spacer may not be an appropriate marker for phy- genic sequences corroborate restriction site data for reconstruct- logeny at lower levels. Based on the relative rate tests, the ing Rubiaceae phylogeny. Plant Syst. Evol. 190: 245–248. molecular clock ticked at nearly a regular rate, which was Farris, J.S. 1989. The retention index and homoplasy excess. Syst. estimated to be (1.12 ± 0.019) × 10–10 substitutions per site Zool. 38: 406–407. per year in this noncoding region of true mosses. Neverthe- Felsenstein, J. 1985. Confidence limits on phylogenies: An ap- less, since the relative rate test only considers the numbers proach using the bootstrap. Evolution, 39: 783–791. of nucleotide substitutions and not the numbers of indels, Golenberg, E.M., Clegg, M., Durbin, M.L., Doebley, J., and Ma, these tests may be biased when applied to genes with high D.P. 1993. Evolution of a noncoding region of the chloroplast numbers of indels. genome. Mol. Phylogenet. Evol. 2: 52–64. Gruissem, W., and Zurawski, G. 1985. Analysis of promoter regions for the spinach chloroplast rbcL, atpB and psbA genes. Acknowledgements EMBO J. 4: 3375–3383. We thank Dr. Brent D. Mishler for his advice on DNA ex- Hasebe, M., Omori, T., Nakazawa, M., Sano, T., Kato, M., and Iwatsuki, K. 1994. rbcL gene sequences provide evidence for traction. We are indebted to Steve O’Kane and Cheng-Fang the evolutionary lineages of leptosporangiate ferns. Proc. Natl. Chiang for the assistance in DNA sequencing and data anal- Acad. Sci. U.S.A. 91: 5730–5734. ysis. We are grateful to Prof. Peter H. Raven for his encour- Hedges, S.B. 1992. The number of replications needed for accurate agement on this research. estimation of the bootstrap P value in phylogenetic studies. Mol. Biol. Evol. 9: 366–369. References Hedwig, J. 1801. Species Muscorum Frondosorum. J.A. Barth, Leip- zig, Germany. pp. 353. Archie, J.W. 1989. Homoplasy excess ratios: New indices for mea- Higgins, D.G, Bleasby, A.J., and Fuchs, R. 1992. CLUSTAL V:Im- suring levels of homoplasy in phylogenetic systematics and a proved software for multiple sequence alignment. CABIOS, 8: critique of the consistency index. Syst. Zool. 38: 253–269. 189–191. Avise, J.C. 1994. Molecular markers, natural history and evolution. Hillis, D.M., and Bull, J.J. 1993. An empirical test of bootstrap- Chapman & Hall, New York. ping as a method assessing confidence in phylogenetic analysis. Bousquet, J., Strauss, S.H., Doerksen, A.H., and Price, R.A. 1992. Syst. Biol. 42: 182–192. Extensive variation in evolutionary rate of rbcL gene sequences Hillis, D.M., and Huelsenbeck, J.P. 1992. Signal, noise, and reli- among seed plants. Proc. Natl. Acad. Sci. U.S.A. 89: 7844– ability in molecular phylogenetic analyses. J. Hered. 83: 189– 7848. 195. Brown, W.M., Prager E.M., Wang, A., and Wilson, A.C. 1982. Mi- Hong, K.S., Chen, T.Y., and Chiang, T.Y. 1999. Sequence announce- tochondrial DNA sequences of primates: Tempo and mode of ment: Promoter of rbcL gene from isolate H002 of Quercus evolution. J. Mol. Evol. 18:225–239. dodonaeifolius. Plant Mol. Biol. 39: 389. Buck, W.R., and Vitt, D.H. 1986. Suggestions for a new familial Huelsenbeck, J.P. 1991. Tree-length distribution skewness: An in- classification of pleurocarpous mosses. Taxon, 35: 21–60. dicator of phylogenetic information. Syst. Zool. 40: 257–270. Chase, M.W., Soltis, D.E., Olmstead, R.G., Morgan, D., Les, D.H., Mishler, B.D., Duvall, M.R., Price, R.A., Hills, H.G., and Qiu, Kim, K.J., Jansen, R.K., Wallace, R.S., Michaels, H.J., and Palmer, Y.L. 1993. Phylogenies of seed plants: An analysis of nucleotide J.D. 1992. Phylogenetic implication of rbcL sequence variation sequences from the plastid gene rbcL. Ann. Mo. Bot. Gard. 80: in the Asteraceae. Ann. Mo. Bot. Gard. 79: 428–445. 528–580. Kimura, M. 1980. A simple method for estimating evolutionary Chiang, T.Y. 1994. Molecular systematics of the Hylocomiaceae rates of base substitutions through comparative studies of nucle- (Order Hypnales) inferred from DNA sequences of the internal otide sequences. J. Mol. Evol. 16: 111–120. transcribed spacers of ribosomal DNA and the atpB–rbcL non- Klassen, G.J., Mooi, R.D., and Locke, A. 1991. Consistency indi- coding spacers of chloroplast DNA. Ph.D. Dissertation, Depart- ces and random data. Syst. Zool. 40: 446–457. ment of Biology, Washington University, St. Louis. Kluge, A.G., and Farris, J.S. 1969. Quantitative phyletics and the Chiang, T.Y., Schaal, B.A., and Peng, C.I. 1998. Universal primers evolution of anurans. Syst. Zool. 18: 1–32. for amplification and sequencing a noncoding spacer between Kocher, T.D., and Wilson, A.C. 1991. Sequence evolution of mito- the atpB and rbcL genes of chloroplast DNA. Bot. Bull. Acad. chondrial DNA in human and chimpanzees: Control region and Sin. (Taipei) 39: 245–250. a protein-coding region. In Evolution of life: Fossils, molecules, Clegg, M.T., Learn, G.H., and Golenberg, E.M. 1991. Molecular and culture. Edited by S. Osawa and T. Honjo. Springer-Verlag, evolution of chloroplast DNA. In Evolution of the Molecular Tokyo. Level. Edited by R.K. Selander, A.G. Clark, and T.S. Whittam. Kumar, S., Tamura, K., and Nei, M. 1993. MEGA: Molecular Evolu- Sinauer Association Inc. Publishers, Sunderland, Mass. pp. 135– tionary Genetics Analysis, version 1.01. Pennsylvania State Uni- 149. versity, University Park, Pennsylvania. Delcourt, H.R., and Delcourt, P.A. 1991. Quaternary Ecology. A Larson, A. 1994. The comparison of morphological and molecular paleoecological perspective. Chapman and Hall, New York. data in phylogenetic systematics. In Molecular Ecology and Evo- Doyle, J.J., and Doyle, J.L. 1987. A rapid isolation procedure for lution: Approaches and applications. Edited by B. Schierwater, B. small quantities of fresh leaf tissue. Phytochem. Bull. 19: 11– Streit, G.P. Wagner, and R. DeSalle. Birkhauser Verlag, Basel, 15. Switzerland. pp. 371–390. Efron, B. 1982. The jackknife, the bootstrap, and other resampling Lemoigne, Y. 1970. Nouvfles diagnoses du genre Rhynia et de plans. CBMS-NSF Regional Conference Series in Applied 1’espece Rhynia gwynne-vaughnii. Bulletin de la société bota- Mathematics, Monograph 38. Society of Industrial and Applied nique de France, 117: 307–320. Mathematics, Philadelphia. pp. 1–92. Li, W.H. 1997. Molecular evolution. Sinauer, Sunderland, Mass.

Liu, S.L., Chiang, T.Y., and Peng, C.I. 1998. Sequence announce- Savolainen, V., Manen, J.F., Douzery, E., and Spichiger, R. 1994. ment: Promoter of rbcL gene from Begonia chitoensis. Plant Molecular phylogeny of families related to Celastrales based on Mol. Biol. 38: 907. rbcL 5′ flanking sequences. Mol. Phylogenet. Evol. 3: 27–37. Maddison, W.P., and Maddison, D.R. 1992. MACCLADE: Analysis of Sears, B.B. 1980. Elimination of plastids during spermatogenesis phylogeny and character evolution, Version 3. Sinauer Associ- and fertilization in the plant kingdom. Plasmid, 4: 233–255. ates, Sunderland, Mass. Shinozaki, K., Tanaka, M., Wakasugi, T., Hayashida, N., Matsubayashi, Manen, J., Savolainen, V., and Simon, P. 1994. The atpB and rbcL T., Zaita, N., Chunwongse, J., Obokata, J., Yamaguchi-Shinozaki, promoters in plastid DNAs of a wide dicot range. J. Mol. Evol. K., Ohto, C., Torazawa, K., Meng, B.Y., Sugita, M., Deno, H., 38: 577–582. Karnogashira, T., Yamada, K., Kusuda, J., Takaiwa, F., Kato, A., Margush, T., and McMorris, F.R. 1981. Consensus n-trees. Bull. Tohdoh, N., Shimada, H., and Sugiura, M. 1986. The complete nu- Nath. Biol. 43: 239–244. cleotide sequence of tobacco chloroplast genome: Its gene organiza- Mishler, B.D., and Churchill, S.P. 1984. A cladistic approach to the tion and expression. EMBO J. 5: 2043–2049. phylogeny of the “bryophytes.” Brittonia, 36: 406–424. Sokal, R.R., and Rohlf, F.J. 1981. Biometrics. W.H. Freeman, San Mishler, B.D., and de Luna, E. 1991. The use of ontogenetic data Francisco. in phylogenetic analysis of mosses. Adv. Bryol. 4: 121–167. Swofford, D.L. 1991. When are phylogeny estimates from molecular Mullet, J.E., Orozco, E.M., and Chua, N. 1985. Multiple tran- and morphological data incongruent? In Phylogenetic Analysis of scripts for higher plant rbcL and atpB genes and localization of DNA Sequences. Edited by M.M. Miyamoto and J. Cracraft. Ox- the transcription initiation site of the rbcL gene. Plant Mol. Biol. ford University Press. New York, Oxford. pp. 295–333. 4: 39–54. Swofford, D.L. 1993. PAUP: Phylogenetic Analysis Using Parsi- Natali, A., Manen, J., and Ehrendorfer, F. 1995. Phylogeny of the mony, Version 3. 1. 1. Computer program distributed by the Illi- Rubiaceae-Rubioideae, in particular the tribe Rubieae: Evidence nois Natural History Survey, Champaign, Ill. from a non-coding chloroplast DNA sequence. Ann. Mo. Bot. Tamura, K., and Nei, M. 1993. Estimation of the number of nucle- Gard. 82: 428–439. otide substitutions in the control region of mitochondrial DNA Nickrent, D.L., and Soltis, D.E. 1995. A comparison of angio- in humans and chimpanzees. Mol. Biol. Evol. 10: 512–526. sperm phylogenies from nuclear 18S rDNA and rbcL sequences. Templeton, A. 1983. Phylogenetic inference from restriction endo- Ann. Mo. Bot. Gard. 82: 208–234. nuclease cleavage site maps with particular reference to the evolution of humans and the apes. Evolution, 37: 221–244. Nishizawa, Y., and Hirai, A. 1987. Nucleotide sequence and ex- Umesono, K., Inokuchi, H., Shiki, Y., Takeuchi, M., Chang, Z., pression of the gene for the large subunit of rice ribulose 1,5- Fujuzawa, H., Kochi, T., Shirai, H, Ohayama, K., and Ozeki, II. bisphosphate carboxylase. Jpn. J. Genet. 62: 223–229. 1988. Gene organization of the large copy region from rps′12 to Olmstead, R.G., Michaels, H.J., Scott, K.M., and Palmer, J.D. atpB. J. Mol. Biol. 203: 299–331. 1992. Monophyly of the Asteridae and identification of their Vigilant, L., Stoneking, M., Harpending, H., Hawkes, K., and Wil- major lineages inferred from DNA sequences of rbcL. Ann. Mo. son, A.C. 1991. African populations and the evolution of human Bot. Gard. 79: 249–265. mitochondrial DNA. Science, 253: 1503–1507. Orozco, E.M., Chen, L., and Eilers, R.J. 1990. The divergently Wang, B., Fang, Q., Williams, W.V., and Weuber, D.B. 1992. Dou- transcribed rbcL and atpB genes of tobacco plastid DNA are ble-stranded DNA sequencing by linear amplification with Taq separated by nineteen base pairs. Curr. Genet. 17: 65–71. DNA polymerase. BioTec. 13: 527–529. Qiu, Y.L., Chase, M.W., Les, D.H., and Parks, C.R. 1993. Molecu- Wu, C.I., and Li, W.H. 1985. Evidence for higher rates of nucleo- lar phylogenetics of the Magnoliidae: Cladistic analyses of nu- tide substitution in rodents than in man. Proc. Natl. Acad. Sci. cleotide sequences of the plastid gene rbcL. Ann. Mo. Bot. U.S.A. 82: 1741–1745. Gard. 80: 587–606. Wu, W.L., Wang, J.P., Tseng, M.C., and Chiang, T.Y. 1999. Clon- Palmer, J.D. 1987. Chloroplast DNA evolution and biosystematic ing and genetic variability of an Hind III repetitive DNA in uses of chloroplast DNA variation. Am. Nat. 130(Suppl.): S6– Acrossocheilus paradoxus (Cyprinidae). Genome, 42: 780–788. S29. Yoshinaga, K., Kubota, Y., Ishii, T., and Wada, K. 1992. Nucleo- Pamilo, P., and Nei, M. 1988. Relationships between gene trees tide sequence of atpB, rbcL, trnR, dedB and psaL chloroplast and species trees. Mol. Biol. Evol. 5: 568–583. genes from a fern Angiopteris lygodiifolia: A possible emer- Rohlf, V.L., and Sokal, R.R. 1981. Statistical Tables. W.H. Free- gence of Spermatophyta lineage before the separation of man and Co., San Franscisco. Bryophyta and Pteridophyta. Plant Mol. Biol. 18: 79–82. Rohrer, J.R. 1985. A generic revision of the Hylocomiaceae. J. Zuckerkandl, E., and Pauling, L. 1965. Evolutionary divergence Hattori Bot. Lab. 59: 241–278. and convergence in proteins. In Evolving genes and proteins. Sanger, F., Nicklen, S., and Coulson, A.R. 1977. DNA sequencing Edited by V. Bryson and H.J. Vogel. Academic Press, New York. with chain-terminating inhibitors. Proc. Natl. Acad. Sci. U.S.A. pp. 97–166. 74: 5463–5467. Zurawski, G., and Clegg, M.T. 1987. Evolution of higher-plant Sarich, V.M., and Wilson, A.C. 1973. Generation time and genomic chloroplast DNA-encoded genes: Implications for structure- evolution in primates. Science, 179: 1144–1147. function and phylogenetic studies. Ann. Rev. Plant Physiol. 38: Savard, L., Li, P., Strauss, S.H., Chase, M.W., Michaud, M., and 391–418. Bousquet, J. 1994. Chloroplast and nuclear gene sequences indi- Zurawski, G., Clegg, M.T., and Brown, A.D.H. 1984. The nature of nu- cate late Pennsylvanian time for the last common ancestor of ex- cleotide sequence divergence between barley and maize chloroplast tant seed plants. Proc. Natl. Acad. Sci. U.S.A. 91: 5136–5167. DNA. Genetics, 106: 735–749.