MOLECULAR PHYLOGENETICS AND EVOLUTION Molecular Phylogenetics and Evolution 31 (2004) 915–928 www.elsevier.com/locate/ympev Phylogeny of the Polytrichales (Bryophyta) based on simultaneous analysis of molecular and morphological data Jaakko Hyv€ onen,a,* Satu Koskinen,b Gary L. Smith Merrill,c Terry A. Hedderson,d and Soili Stenroose a Division of Systematic Biology, P.O. Box 65, FIN-00014 University of Helsinki, Helsinki, Finland Laboratory of Plant Physiology and Molecular Biology, University of Turku, FIN-20014 Turku, Finland c Department of Botany, Field Museum of Natural History, Chicago, IL 60605-2496, USA Bolus Herbarium, Department of Botany, University of Cape Town, Private Bag, Rondebosch 7701, South Africa e Herbarium, University of Turku, FIN-20014 Turku, Finland b d Received 19 February 2003; revised 10 October 2003 Abstract Phylogenetic analyses of Polytrichales were conducted using morphology and sequence data from the chloroplast genes rbcL and rps4 plus the trnL-F gene region, part of the mitochondrial nad5 and the nuclear-encoded 18S rDNA. Our analyses included 46 species representing all genera of Polytrichales. Phylogenetic trees were constructed with simultaneous parsimony analyses of all sequences plus morphology and separate combinations of sequence data only. Results lend support for recognition of Polytrichales as a monophyletic entity. Oedipodium griffithianum appears as a sister taxon to Polytrichales or as a sister taxon of all mosses excluding Sphagnales and Andreaeles. Within Polytrichales, Alophosia and Atrichopsis, species without the adaxial lamellae (in Atrichopsis present but poorly developed on male gametophyte) otherwise typical of the group are sister to the remaining species followed by a clade including Bartramiopsis and Lyellia, species with adaxial lamellae covering only the central portion of the leaves. Six taxa with an exclusively Southern Hemisphere distribution form a grade between the basal lineages and a clade including genera that are mostly confined to the Northern Hemisphere. Ó 2003 Elsevier Inc. All rights reserved. 1. Introduction The green plants are one of the major clades of eukaryotes (e.g. Lipscomb et al., 1998). Their morphological and chemical diversity and ecological dominance in almost all habitats make them the most important group of organisms in terrestrial ecosystems. Reconstruction of their phylogenetic relationships is important for understanding some of the most significant evolutionary events, such as the original conquest of dry land habitats. Bryophytes are small plants with a haplo-diplontic life cycle, and they probably were among the first plants to gain a hold in terrestrial environments. Three groups can be clearly distinguished: Marchantiophyta, Antho- * Corresponding author. Fax: +358-(0)9-19157788. E-mail address: jaakko.hyvonen@helsinki.fi (J. Hyv€onen). 1055-7903/$ - see front matter Ó 2003 Elsevier Inc. All rights reserved. doi:10.1016/j.ympev.2003.11.003 cerophyta, and Bryophyta (Newton et al., 2000). Traditionally, bryophytes have been recognized as a monophyletic entity but relationships among the three major lineages are still in dispute (Hedderson et al., 1998; Lewis et al., 1997; Nickrent et al., 2000; Renzaglia et al., 2000). While the reason for this may be inadequate sampling, it might well be that we will never resolve this part of organismal history. It would not be surprising that dispersal to practically empty ‘‘dry’’ land habitats caused such an explosive evolutionary diversification that branch lengths of the resulting ‘‘tree’’ are simply too short to recover after several hundred million years and presumably rampant extinction. However, there is no doubt that each of the three major lineages of bryophytes is monophyletic. Bryophyta, the mosses, is the largest of all bryophyte groups. The number of species is estimated at 7000–8000 (Crosby, 1999), and mosses can be found in virtually all terrestrial and in many fresh water habitats. 916 J. Hyv€onen et al. / Molecular Phylogenetics and Evolution 31 (2004) 915–928 They are also extremely important, even dominant, ecologically in habitats such as mires and forests of the boreal zone as well as humid cloud forests of the tropical and subtropical mountains. Within Bryophyta four major lineages can be distinguished: Sphagnopsida, Andreaeaopsida, Polytrichopsida, and Bryopsida (Buck and Goffinet, 2000). In addition, there are genera like Andreaeobryum Murray, Takakia S. Hatt. and Inoue, Buxbaumia Hedw., Oedipodium Schw€ agr., Tetraphis Hedw. and Tetrodontium Schw€ agr. whose affinity to the four major groups is still unclear. Buck and Goffinet (2000) include all but the first two genera within Polytrichopsida in order Tetraphidales but it is still equivocal whether they belong here together with Polytrichales or in Bryopsida. Bryopsida is by far the largest group of mosses with well over 90% of all species. Polytrichopsida, whether it is interpreted to include only Polytrichales or also Tetraphidales, is a much smaller group but still the second largest of the major groups both in species number and ecological variability. Polytrichales are typically pioneer plants of open, sometimes even dry, habitats. Despite the small number of species, the order exhibits great diversity from miniature plants with reduced leaves such as Pogonatum pensilvanicum (Hedw.) P. Beauv. of eastern North America to giants of Australasia and New Zealand like Dawsonia superba Grev. with the best-developed gametophyte of all land plants. The most typical features of the polytrichalean gametophyte are the closely spaced adaxial lamellae on the leaves, forming a pseudoparenchyma, and differentiation of leaves into a distinct blade and sheathing base. The calyptra is typically hairy in many common species of the Northern Hemisphere, enveloping the developing capsules of the sporophyte generation. This has given the whole group its name, although most genera have a practically naked calyptra. Capsules of the Polytrichales normally have a well-developed peristome with at least 16 teeth formed of whole cells. The epiphragm covering the mouth of the capsule is a unique character that distinguishes Polytrichales from all other groups of mosses. Size and shape of the urn vary greatly among genera (Schofield, 1985; Smith, 1971). Nineteen genera are currently accepted in Polytrichales, comprising approximately 200 species (Hyv€ onen et al., 1998). Eopolytrichum antiquum Konopka et al. (1997), the sole species of Eopolytrichum, is known only from late Cretaceous fossils. Many of the remaining genera are monotypic, and all the others, with the exception of Pogonatum (ca. 50 spp.) and Polytrichum, (ca. 30 spp.) are fairly small. Some species, like Polytrichum juniperinum Hedw., have almost cosmopolitan distributions, while there are also narrow endemics, some possibly even threatened by extinction. Ecologically, Polytrichales range from xerophytes like P. piliferum Hedw. to species of peaty, wet, and to some extent flooded habitats like P. commune Hedw. Although their structure appears obviously adapted to dry environments, Polytrichales are largely absent from extremely arid regions, and the group exhibits greatest diversity in areas with humid or moist subtropical and tropical climates. Phylogenetic relationships of Polytrichales are particularly relevant to considerations of the evolutionary history of mosses since the group is probably among the first of the lineages that diverged from the common ancestor of all mosses (Mishler and Churchill, 1984). Earlier cladistic analyses of the group have been done by Hyv€ onen (1989), Forrest (1995), and Hyv€ onen et al. (1998). The first two were based solely on morphology, and the latter on sequences of three genes (the chloroplast-encoded rbcL and rps4 loci and the nuclear-encoded 18S rRNA gene) plus morphology from 22 species. However, sequences were not obtained from all three genes for all species, and large genera like Pogonatum were represented by only a few species. In addition, only morphological characters were available for some of the species. The aim of the current study was to enlarge our matrix significantly with respect to both taxa and characters in order to develop a more robust hypothesis of Polytrichalean phylogeny. 2. Material and methods 2.1. Plant material and data sets Our morphological matrix includes 43 characters, and the treatment mostly follows Hyv€ onen et al. (1998). This data set is based on extensive study of specimens from several herbaria and in most cases also included the DNA voucher specimens. The data matrix and list of characters can be found in Appendices A and B. Our analyses include 46 species, representing all known genera of Polytrichales. With few exceptions, we were able to obtain sequence data for all five loci. Of course, the fossil Eopolytrichum is represented only by morphology. Andreaea rupestris Hedw., Buxbaumia aphylla Hedw., B. piperi Best, Diphyscium foliosum (Hedw.) Mohr, Funaria hygrometrica Hedw., Oedipodium griffithianum (Dicks.) Schwaegr., Sphagnum palustre L., Tetraphis pellucida Hedw., T. geniculata Girg. ex Milde, and Timmia sibirica Lindb. et Arn. were used as outgroup taxa based on previous higher-level analyses (Cox and Hedderson, 1999; Hedderson et al., 1996, 1998). Our data included sequences for the nuclear-encoded 18S rRNA gene, the chloroplast-encoded rbcL, rps4, and trnL-F regions, and a stretch of ca. 700 bp from the 30 terminus of the mitochondrial nad5 gene. We tried to get all sequences from the same voucher specimen but this was not always possible, and matrices were supplemented with available sequences from GenBank. We used composite taxa, i.e., J. Hyv€onen et al. / Molecular Phylogenetics and Evolution 31 (2004) 915–928 combining data of two species, only in two cases: nad5 sequences were not available for Sphagnum palustre and Timmia sibirica, and we used sequences of S. fallax H. Klinggr. and T. bavarica Hessl., respectively, to supplement our matrix. See Table 1 for details of the vouchers and available sequences. 2.2. DNA extraction, amplification, and sequencing Total DNA was extracted from fresh, herbarium (oldest specimen 25 years), or silica-dried specimens. Extractions were made using the two different methods given in Hyv€ onen et al. (1998) and the Dneasy Plant Mini Kit (Qiagen) according to the manufacturerÕs instructions.Template DNA suitable for cycle sequencing was prepared via PCR. Amplification was done using either the DynaZyme DNA Polymerase Kit (Finnzymes Oy) and Personal Minicycler (MJ Research) or the Taq DNA Polymerase Kit (Promega) and DNA Thermal Cycler 480 (Perkin–Elmer) or AmpliTaq Gold DNA polymerase (Perkin–Elmer) and Gene Amp PCR system 9700 (Perkin–Elmer). For nad5 and 18S rDNA we used a program comprising a 95 °C initial denaturation step (12 min) followed by 35 cycles of 95 °C denaturation (30 s), 52 °C annealing (1 min), and 72 °C extension (3 min) with a final annealing step at 72 °C for 7 min. The reaction volume was 50 ll. The program used for rbcL and rps4 included a 97 °C initial denaturation step (12 min) followed by 35 cycles of 97 °C denaturation (30 s), 55 °C annealing (1 min), and 72 °C extension (1 min 30 s) with a final annealing step of 72 °C for 7 min. All reactions were done in 50 ll volumes. See also Hyv€ onen et al. (1998) for two alternative PCR programs. A negative control, including all reaction components except the target DNA, was also used. The PCR products were inspected on agarose gels and product sizes were determined from a DNA sizestandard ladder of 50–20,000 bp (Bio-Rad Laboratories). PCR products were purified with the PCR Purification Kit (QIAquick) according to the manufacturerÕs instructions. Cycle sequencing reactions were prepared using the DNA Sequence Kit (ABI, Perkin–Elmer), with either Dye Terminator or dRhodamine Terminator Cycle Sequencing reactions or the Big Dye Terminator Cycle Sequencing Ready Reaction Kit. Sequences were visualized using ABI 373 or 377 automated sequencers. The PCR primers used for 18S rDNA were NS1 and B, and the internal sequencing primers were ERC, G, H, KRC, and Q (Hedderson et al., 1998). Primers rps5 and trnas were used for both PCR amplification and sequencing of rps4 (Cox and Hedderson, 1999). For rbcL, primers M28 (NM34 of Newton et al., 2000) and M1390r were used for PCR and, along with internal primers M740r and M1010r, for sequencing. For some species we were unable to amplify rbcL in one piece and therefore had to use an additional primer (M636) for PCRs. For the trnL-F gene region we used primers C and F both for 917 the PCR and sequencing reactions (see Cox et al., 2000 for details of these oligonucleotide primers). For nad5, K- and L-primers were used for PCR and only L-primer for sequencing (Steinhauser et al., 1999). The individual sequencing products from different primer reactions were aligned as a composite strand using either programs of the Lasergene package (DNASTAR) or manually with a text-editor using a color-coded font BKGCuclc (M. Sogin, Marine Biological Laboratory, Woods Hole, MA). Portions of the completed sequence for each gene are based on reads in only one direction, and the extent of single-read sequence varies among taxa. Discrepancies between reads were solved manually by inspection of the original electropherograms. In doubtful cases IUPAC ambiguity codes were assigned. 2.3. Data analysis Sequence alignment was performed initially with the program CLUSTAL X (Jeanmougin et al., 1998), and sequences were adjusted manually using a color-coded font BKGCuclc. The chloroplast genes rps4 and rbcL and the mitochondrial nad5 did not pose alignment problems. Similarly, the nuclear 18S rRNA-coding gene was not particularly length-variable over the range of taxa in this study and thus was easily aligned; only positions 194–199 and 1396–1399 were ambiguous, and these 10 nucleotides were removed from the final matrix. However, whether these 10 nt were included or not did not alter the resulting topologies. Some of the sequences generated were of too poor quality to be used so we do not have complete sequences for every taxon included in the analyses. All sequences are deposited in GenBank as indicated in Table 1, and the complete matrix has been submitted to TreeBase (http://www.treebase.org). Our final matrix included 4794 characters, and of these, 988 (21%) were parsimonyinformative. Parsimony-informative characters were distributed among data sets as indicated in Table 2. Difficulties were experienced with aligning the noncoding region at the 30 end of the rps4 sequences, so these positions were excluded from the analyses. Besides this, the trnL-F gene region shows considerable length variation in non-coding regions, and therefore alignment was problematic. There are basically two ways to treat ambiguous alignment of sequences. It has been proposed that all such sequences should be excluded from the analyses, which equals ignoring part of the data. The logic behind this is, however, that there genuinely are sequences for which we cannot find optimal alignment. If we adopt this approach for the current material, we have to ignore most of the trnL-F sequences as well as the intergenic spacer at the 30 end of the rps4 gene, which is typically long for Polytrichales and some other mosses but practically lacking, for example, in Bryales (Goffinet et al., 2001). We made preliminary alignments for the trnL-F matrix with Clustal X (Jeanmougin et al., 1998). Taxon Azores Chile. Reg. de Magallanes Brazil. S~ao Paulo USA Lousiana Mexico. Veracruz Finland. Uusimaa Canada. British Columbia Papua New Guinea Australia. Queensland New Zealand. N Island New Zealand. N Island Argentina. Tierra del Fuego Chile. Region de los Lagos Brazil, Sao Paulo Canada. Ellesmere Island USA Colorado New Zealand, S Island Argentina. Rio Negro Finland. Uusimaa Canada. British Columbia Sweden. Sk
ane Brazil. S~ao Paulo Taiwan. Taichung Canada. British Columbia Finland. Uusimaa Japan. Honshu Taiwan. Taichung Taiwan. Taichung Japan. Honshu Brazil. S~ao Paulo Japan. Honshu Australia. Queensland Finland. Uusimaa Chile. Region de Los Lagos Brazil. Minas Gerais Finland. Etel€a-H€ame Finland. Uusimaa Finland, Varsinais-Suomi Brazil. Minas Gerais Finland. Uusimaa Finland. Uusimaa Finland. Uusimaa Malawi. Mulanji Canada. Ellesmere Island Mexico. Veracruz Accession No. Rumsey 18.3.1997 (RNG) Smith B1407b (AAS) Hyv€onen 6387(H) Hedderson 10393 (RNG) Hyv€onen 6504 (H) Hyv€onen 6170 (H) Hedderson 10044 (RNG) Baker 662 (RNG) Schulman 125 (H) Stenroos 4677 (H) Hyv€onen 6083 (H) Hyv€onen 2557 (H) Kelt 26.5.1986 (H) Ahti 51824(H) Hedderson 6825 (RNG) Weber WWB36612 (H) Hyv€onen 6069 (H) Hyv€onen 5625 (H) Enroth 25.7.1998 (H) Hedderson 10043 (RNG) Hyv€onen 6486 (H) Hyv€onen 6392 (H) Hyv€onen 4008 (H) Hedderson 5803 (H) Hyv€onen 6169 (H) Nishimura 10601(H) Hyv€onen 4087(H) Hyv€onen 4021 (H) Hayashi 7038 (H) Hyv€onen 6393(H) Chishiki 1862 (H) Hyv€onen 6025(H) Hyv€onen 6173 (H) Hyv€onen 5865 (H) Hyv€onen 6276 (H) Hyv€onen 6204 (H) Hyv€onen 6197(H) Hyv€onen 6506 (H) Hyv€onen 6230(H) Hyv€onen 6168 (H) Hyv€onen 6193 (H) Hyv€onen 6205 (H) Wigginton M1397a (H) Hedderson 5938 (RNG) Hedderson 12898 (H) 18S rbcL rps4 trnL nad5 AY126951 AF548459 a AY126952 U18492 AY126953 X85093 AY126954 AF228669 AY126956 AY126955 AF208402 AY126957 AY126958 AY126959 AF208403 AY126960 AF208404 AY126961 AY126962 AY126963 AY126964 AY126965 AY126966 AY126967 AY126968 AY126969 AY126970 AY126971 AY126972 AY126973 AY126974 AY126975 AF208406 AF208407 AY126976 AY126977 X80982 AY126978 AY126979 U18518 AY126980 AY126981 AY126982 AY126983 AY126984 AF208408 AY118233 AY118234 AF231061 AY118235 AY118236 AF208409 AF208410 AY118238 AY118237 AAF208411 AY118239 AY118240 AF208412 AF208413 AY118241 AF208414 AY118242 AY118243 AF208415 AY118244 AY118245 AY118246 AY118247 AY118248 AY118249 AY118250 AY118251 AY118252 AY118253 AY118254 AY118255 AY118256 AY118257 AF261074 AY118258 AY118259 AY118260 AY118261 U87087 AY118262 AY118263 AY118264 AF208416 AY118265 AY137679 AF544997 AF544998 AF544999 AF545000 AF545001 AF545002 AF545003 AF246704 AF545005 AF545004 AF545006 AF545007 AF545008 AF545009 AF545010 AF545011 AF545012 AF545013 AF545014 AF545015 AF545016 AF545017 AF545018 AF545019 AF545020 AF545021 AF545022 AF545023 AF545024 AF545025 AF545026 AF545027 AF545028 AF545029 AF545030 AF545031 AF545032 AF545033 AF545034 AF545035 AF545036 AF545037 AF545038 AF545039 AF545040 AY137713 — — AF208417 AY137680 AY137681 AF208418 AF208419 AY137683 AY137682 AF208420 AY137684 AY137685 AF208421 AF208422 AF208423 AY137686 AY137687 AY137688 AF208424 AY137689 AY137690 AY137691 AF208425 AY137692 AY137693 AY137694 AY137695 AY137696 AY137697 AY137698 AY137699 AF208426 AF208427 AY137700 AY137701 AY137702 AY137703 AY137704 AF208428 AY137705 AY137706 AY137707 AF208429 AY137708 — AY137714 AY137715 AY137716 AY137717 AY137718 AY150372 AY137720 AY137719 AY137721 AY137722 AY137723 AY137724 AY137725 AY137726 AY137727 AY137728 AY137729 AY137730 AY137731 AY137732 AY137733 AY137734 AY137735 AY137736 AY137737 AY137738 AY137739 AY137740 AY137741 AY137742 AY137743 AY137744 AY137745 AY137746 AY137747 AY137748 AY137749 — AY137751 AY137752 AY137753 AY137754 AY137755 J. Hyv€onen et al. / Molecular Phylogenetics and Evolution 31 (2004) 915–928 Alophosia azorica Atrichopsis compressa Atrichum androgynum A. angustatum A. oerstaedianum A. undulatum b Bartramiopsis lescurii Dawsonia papuana D. polytrichoides D. superba Dendroligotrichum dendroides D. squamosum Hebantia rigida Itatiella ulei Lyellia aspera Meiotrichum lyallii Notoligotrichum australe Oligotrichum austro-aligerum O. hercynicum O. parallelum Pogonatum aloides P. campylocarpum P. cirratum P. contortum P. dentatum P. japonicum P. microstomum P. neesii P. nipponicum P. pensilvanicum P. spinulosum P. subulatum P. urnigerum Polytrichadelphus magellanicus P. pseudopolytrichum Polytrichastrum alpinum P. formosum b P. longisetum Polytrichum brachymitrium P. commune b;c P. juniperinum P. piliferum P. subpilosum Psilopilum laevigatum Steereobryon subulirostrum Collection reference 918 Table 1 Voucher numbers for taxa sampled in analysis, followed by GenBank Accession numbers for DNA sequences 919 — — — Accession numbers in italics indicate sequences downloaded from GenBank. See text for further details. a 18S sequences submitted in two pieces, the other No. is AF548460. b 18S sequences not from the same specimen. c rbcL sequences not from the same specimen. d nad5 sequences not from the same specimen. — Z98963 Table 2 Size of the each matrix and percentage of parsimony informative characters plus values for consistency (CI) and retention indeces (RI) as measured for the two equally parsimonious trees found in the analysis of the whole combined matrix — AY137712 AJ224855 — AF231902 AF544996 AF231908 AF023715 — — AF231892 AF306955 AF231896 AF023775 — AF231887 AY118232 U87091 AF275166 — Hotchkill 98-70 Schofield 103022 (DUKE) Pittillo 9764(DUKE) Hedderson 6341(RNG) USA Hawaii Canada. British Columbia USA North Carolina Canada. Ellesmere Island Y11370 AY126950 U18527 AF023678 — AY137709 AY137710 AY137711 Z98959 AF548458 AJ001225 AY137677 AY137678 AF223034 AF023776 AF306968 AY118230 AY118231 AF232692 AF005513 AF246289 Y17603 AY126949 AF230415 X80212 AF228668 AF231905 AF231909 AF544994 AF544995 AF023716 AF246290 Hedderson 10439 (RNG) Anderson 24716(DUKE) Belland 16889 (DUKE) Schofield 81571 (DUKE) Goffinet 4492 (DUKE) Cox 148(RNG) Schofield 98670 (DUKE) Outgroup taxa Andreaea rupestrisc;d A. rothii Buxbaumia aphyllab B. piperi Diphyscium foliosum Funaria hygrometricab; c; d Oedipodium griffithianumb Sphagnum. fallax S. palustreb; d Tetraphis geniculata T. pellucida b Timmia sibirica d T.bavarica Canada. New Foundland USA North Carolina Canada. Nova Scotia Canada. British Columbia USA North Carolina Great Britain. England USA Alaska AJ617675 — L13473 — U18490 — — AJ001227 J. Hyv€onen et al. / Molecular Phylogenetics and Evolution 31 (2004) 915–928 Matrix Characters Informative (%) CI RI 18S rbcL rps4 trnL-F nad5 Morphology 1834 1367 600 261 689 43 173 404 208 66 99 39 0.54 0.40 0.53 0.50 0.62 0.34 0.62 0.44 0.57 0.67 0.66 0.64 9 30 35 25 14 91 This resulted in a matrix of 948 characters. However, none of the individual sequences had this length, and alignment gaps of various length were inserted in all. In order to retain only unambiguously aligned stretches of sequences, we removed nucleotides in positions 1–73, 148–151, 175–202, 240–499, and 627–948 from the matrix. This resulted in a reduced matrix of only 260 nucleotides that were unambiguously aligned, corresponding to parts of the trnL intron and the trnL 30 exon in the trnLF region. In addition to this, we were able to align sequences of Buxbaumia piperi and Tetraphis geniculata only by their 30 end and therefore had to remove the first 499 bp of these sequences. Unfortunately, we were not able to use the same approach for the 30 terminus AT-rich indel of rps4 because even partially unambiguous alignment was not obtained for these sequences. Parsimony analyses were performed using the program NONA (Goloboff, 1994) in conjunction with a Winclada shell (Nixon, 1999) with the following settings: hold * (holding all trees that memory allows, in current settings with Winclada this is 10,000), mult*100 (search replicated 100 times), hold/2 (keeping 2 starting trees for each replication), and using multiple tree-bisection reconnection algorithm (mult*max*). We also performed more extensive analysis of the whole material (hold*, mult*1000, hold/10) but obtained exactly the same result as with the smaller search. In all sequences, gaps were treated as missing data. In the analyses all characters were weighted equally, with no distinction between transitions and transversions, and morphological characters of the sporophyte and gametophyte generation were treated equally as well. All morphological characters were treated as unordered. Traditionally, more ‘‘weight’’ has been given to sporophytic characters in studies of moss phylogeny. However, we preferred to avoid additional assumptions which a priori character weighting necessitates (Kluge, 1997). The following analyses were performed: 1. Simultaneous analyses (Nixon and Carpenter, 1996) including all five sequence matrices plus morphology based on (A) inclusion of all taxa, (B) analysis leaving out the fossil taxon Eopolytrichum antiquum with a high proportion of missing entries. 920 J. Hyv€onen et al. / Molecular Phylogenetics and Evolution 31 (2004) 915–928 2. Analyses based only on sequence data (A) from all five available matrices, (B) leaving out nuclear 18S rDNA sequence. Different methods and indices have been proposed to study ‘‘reliability’’ or ‘‘strength’’ of different phylogenetic hypotheses included in each cladogram. The aim of calculating support values is to estimate how well our hypotheses will hold up when more, possibly conflicting, data is obtained. We calculated jackknife values (Farris et al., 1996) for our material using the parsimony jackknifer xac (Farris, 1997a) which includes branch-swapping. The search was performed with 10,000 replicates. Some authors (e.g., Kluge, 1997) have challenged the use of these metrics altogether, arguing convincingly that comparisons of the equally parsimonious tree(s) with suboptimal topologies or utilizing only part of the available evidence within the cladistic framework is not warranted. Real tests of the current hypotheses will be provided only by further data, i.e., the next added character or taxon. The congruence between different data sets was tested by performing the incongruence length test (ILD) of Mickevich and Farris (1981) with the program xarn (Farris, 1997a,b). This was performed with 1000 repetitions and three rounds of branch-swapping. 3. Results Our simultaneous parsimony analysis of the total matrix with all of the 56 taxa included yielded two equally parsimonious trees of 4419 steps, with a consistency index (CI, Kluge and Farris, 1969) of 0.47 and retention index (RI, Farris, 1989) of 0.54. One of these trees is shown in Fig. 1. The only difference between the two trees is in the position of Pogonatum contortum, which is either sister to P. cirratum or sister to the rest of the same clade. Our material includes a large proportion of missing entries for the fossil Eopolytrichum antiquum but this did not seem to have an effect on results. This fossil species was unambiguously nested within the Polytrichum clade in all trees, and excluding it from the analysis did not change the general topology. Excluding morphological characters resulted only in a single, but very surprising, change: Atrichum angustatum was nested within Polytrichum as the sister of P. commune! Both Polytrichum and Atrichum are very welldefined, distinct genera that can be distinguished unambiguously by their morphology. What is the reason for this spurious grouping? Long-branch attraction has in many cases been given as an explanation when novel and unexpected groupings are encountered (e.g., Buck et al., 2000). However, as pointed out by Siddall and Whiting (1999) the presence of long-branch attraction as an artifact is possible only when two branches are attracted to each other. It should be evident if the topol- ogy is altered by removing one of the taxa from the analysis. When P. commune was removed, A. angustatum was back with other species of Atrichum. Branches leading to these two problematic species are the longest within Atrichum and Polytrichum, respectively. However, there are also longer branches on the overall tree. But when we examine changes in 18S sequences, it is evident that branches leading to A. angustatum and P. commune are exceptionally long (53 and 48 changes, respectively). There are two other branches that come close, one leading to Atrichopsis compressa (46 changes) and the one leading to the clade of Pogonatum neesii plus P. subulatum (47). Reduction of the data set still further by including only plastid sequences in the analysis gave eight equally parsimonious trees (not shown) without an A. angustatum–P. commune grouping. It seems that this unexpected grouping was due to the exceptional 18S sequences downloaded from GenBank. Both were deposited by one of us (TAH) as part of an early study of land plant relationships. They were generated from RNA templates using reverse transcriptase, an enzyme with a high known error rate. Resequencing of 18S rDNA for these two taxa seems to be warranted. The ILD test performed with xarn (Farris, 1997a) revealed all data sets to be highly incongruent with each other. The low a-values (0.001) indicating highly incongruent data sets were obtained for all comparisons between different data sets irrespective of whether they represent the same or different organellar genome or morphology. The only exceptions were comparisons between partial sequences of the mitochondrial nad5 and the chloroplast trnL-F region (0.057) and between the nuclear-encoded 18S and nad5 (0.002). 4. Discussion 4.1. Sequence analyses Our data included nuclear, chloroplast, and mitochondrial DNA sequences from coding as well as noncoding regions. The utility of these regions in resolving lineages of different age probably varies a great deal. In the present context this is a clear advantage because in Polytrichales we likely have a group of great antiquity but possibly also with lineages that have undergone speciation quite recently. Nuclear-encoded ribosomal 18S rDNA sequences have been used extensively to address ‘‘deep’’ (Mishler, 2000) reconstruction problems in many groups of organisms from family to kingdom level. It shows the lowest percentage of informative characters (Table 2) in our material, but remains more or less at the same level (8%) even if we compare Polytrichales only, and provides information even within the most apical clade consisting essentially of Psilopilum, Polytrichum, and Pogonatum (5%). Leaving financial J. Hyv€onen et al. / Molecular Phylogenetics and Evolution 31 (2004) 915–928 Fig. 1. The other one of the two equally parsimonious trees based on simultaneous analysis of the total matrix composed of six datasets. The length of the trees are 4419 steps, with a consistency index (CI) of 0.47 and retention index (RI) of 0.54. Supporting characters given for each node are shown as divided between different sources of data. 921 922 J. Hyv€onen et al. / Molecular Phylogenetics and Evolution 31 (2004) 915–928 matters aside (18S rDNA as compared to other gene regions in this analysis is expensive to sequence because of its length), it seems that sequencing of even fairly ‘‘conservative’’ gene regions is valuable and can provide information also at ‘‘lower’’ taxonomic levels. It has been argued that substitutions at certain sites (such as third codon positions in protein coding genes) become randomized (saturated) and will be phylogenetically uninformative (Swofford et al., 1996) but recent findings (e.g. Yang, 1998) show that these concerns are exaggerated. As pointed out by Chase and Albert (1998) and K€ allersj€ o et al. (1999), based on examples from large rbcL data sets, there is no direct connection between homoplasy and ‘‘value’’ of characters in phylogeny reconstruction; leaving out or down-weighting characters according to their homoplasy would be a mistake. Also in our own rbcL matrix the percentage of informative characters was highest in third positions (66%) but they also showed much homoplasy, with lower CI (0.38) than first (0.40) and second (0.56) positions of the gene. Even assuming (based on a model of molecular evolution) that there are undetected multiple substitutions, one is still dealing with many, possibly conflicting characters. It is extremely unlikely that these ‘‘saturated’’ sites would covary to such an extent that they would obscure the phylogenetic signal in the material. Certainly this is far less of a problem when an analysis, such as ours, is based on multiple genes and other sources of data. When we use simultaneous analysis of all available data, it might still be valuable to explore the data with separate analyses. For example, when we left out nuclear-encoded 18S rDNA sequences the resulting topology included some curious groupings. A lineage including Pogonatum microstomum, P. urnigerum, and Polytrichastrum alpinum is now placed within Pogonatum, albeit as a basal branch. In plants plastid genomes are predominantly inherited from the maternal lineage and therefore one can suspect that this conflict in results is possibly due to an ancient hybridization between two genera in Polytrichales. To answer this question other types of data and analyses are needed and even then getting unambiguous answers might be difficult because of missing data due to extinction (Derda and Wyatt, 1999). Other differences from the results obtained with simultaneous analysis include separation of P. aloides, P. campylocarpum, and P. pensilvanicum from other species of the genus forming a clade together with Alophosia azorica and Atrichopsis compressa. The taxa involved in this unexpected and novel grouping all show exceptionally long-branch lengths for the chloroplast data. Are these novel groupings examples of long-branch attraction? If yes, combining data from different sources might have another benefit for the analyses by leveling out length differences between branches and removing problems associated with exceptionally long branches. When we inspect characters supporting each branch (Fig. 1) one can immediately see that in our material contribution to the total length of branches from different gene regions is not uniform but varies a great deal throughout the tree. As soon as multiple sequence data sets became available for phylogenetic analyses concerns were raised over their congruence, i.e., if they could (or should) be combined or analyzed separately (e.g. Miyamoto and Fitch, 1995 versus Kluge, 1989; Eernisse and Kluge, 1993). The simplest way to test this is to compare resulting topologies (taxonomic congruence). This does not, however, take into account the strength of support for individual hypotheses included in the topologies that are being compared. Farris et al. (1994) devised an elegant way to explore this with the incongruence length (ILD) test of Mickevich and Farris (1981). While we acknowledge the power of the test, we agree with Siddall (1997) who argued that incongruence per se does not warrant ignoring part of the available data. He also showed with a simple example that incongruence can be caused by a very small number (actually only one!) of characters that are in conflict with other sources of data. Leaving out part of the data would be warranted only if we knew a priori which part of our data is unreliable. This is, however, something we cannot know. Which part of the data should we trust? For example, in our material clear incongruence was observed among practically all partitions. Should we perform independent analyses and pool the results to find out their taxonomic congruence? In our opinion this is not a viable alternative. Homoplasy is encountered in all data sets, and we agree with Wheeler et al. (1993) that congruence between characters from different data sets provides us with the best test to sort out homology from homoplasy and level out noise in our data sets. When we examine jackknife support values and length of branches it is noteworthy that internal nodes within Polytrichales have either very low values (or they are lacking altogether) and these branches are extremely short as compared with more terminal branches (Figs. 1 and 2). There might be real biological reasons for these short branches. For example, in the present case short branches are characteristic for the part of tree where genera that represent Gondwanan and Northern Hemisphere elements, respectively, branch off. It is possible that short branches indicate dispersal to new ‘‘empty’’ areas and habitats and subsequent rapid diversification. 4.2. Classification, morphology, and biogeography Our analysis provides support for the monophyly of Polytrichales with Oedipodium griffithianum, formerly included in the Bryales, as a sister taxon. The position of Oedipodium as a member of Polytrichopsida is, however, ambiguous. In the analysis based only on plastid sequences O. griffithianum is in a still more basal position within mosses, being placed between Andreaea and a clade leading to Bryopsida (including Tetraphidales J. Hyv€onen et al. / Molecular Phylogenetics and Evolution 31 (2004) 915–928 sensu Buck and Goffinet, 2000) and Polytrichales. Similar results were also obtained by Newton et al. (2000) in their analysis of phylogenetic relationships among major moss lineages, and our jackknife tree (Fig. 2) shows the same position. However, in analyses performed by Newton et al. (2000) Polytrichales were represented by only two species whereas in our own analysis we included only few species of Bryales. In order to test whether our results were due to unequal sampling of Polytrichales versus Bryales we expanded our matrix with 10 more Fig. 2. The tree obtained from the jackknife analysis with the program xac based on the total matrix composed of six datasets. Jackknife support values are given below the nodes. 923 outgroup taxa. Sequences were downloaded from GenBank and many of these additional outgroup taxa were composites as indicated in Table 3 and morphology was not scored for these additional taxa. In this analysis Oedipodium still remained as sister-taxon of Polytrichales. However, we should aim for still wider sampling, including morphology, to resolve phylogeny of the most basal lineages of mosses. With the currently available programs with extremely powerful algorithms and access to parallel computing we should aim to expand data sets, not reduce them. Due to extinct lineages, our sampling will always be only representative rather than complete and there is no reason to reduce it still further. In order to obtain reliable hypotheses of the deeper nodes within mosses, such genera as Diphyscium, Buxbaumia, and Tetraphis with their allies are in a pivotal position. More detailed sampling of them is needed and such studies have already been undertaken by Magombo (2003). Within Polytrichales our data give clear and strong support for some traditionally distinguished genera (e.g., Atrichum, Dawsonia) while the largest genera, Pogonatum and Polytrichum, appear as paraphyletic or are distinguished as clades that also include other species, respectively. As mentioned above, most of the internal branches are weakly supported. However, the tree presented is still our best hypothesis given the data we currently have. Our results are to a large extent compatible with those obtained in a preliminary analysis based on restricted sampling of both taxa and characters (Hyv€ onen et al., 1998). The traditional division of the order into two families, Polytrichaceae and Dawsoniaceae (e.g. Brotherus, 1925; Crum, 2001), is still strongly contradicted. Dawsonia is monophyletic but firmly included within Polytrichaceae as noted by Smith (1971). Alophosia azorica appears to be sister to all the other Polytrichales but now joined by another taxon lacking the typical adaxial lamellae—Atrichopsis compressa. In this species lamellae are actually present, but poorly developed and found only on the male gametophyte. When we recoded this character for A. compressa as ‘‘lamellae present’’ it did not change the topology, but only added one more step. The latter species was represented in our earlier analysis only by morphology. Both of these species are narrow endemics with very restricted distribution. Alophosia azorica is found only in Macaronesia (Azores and Madeira) while Atrichopsis compressa has been found only in the extremely oceanic areas of the western and southern coasts of southernmost South America. The species of Bartramiopsis, Lyellia, and Dawsonia and the rest of Polytrichales show a grade of more elaborate development of photosynthetic adaxial lamellae from the species without them (Alophosia azorica and Atrichopsis compressa), through species with lamellae only in the central part of the leaf (Bartramiopsis and Lyellia) to species with numerous high lamellae with specialized enlarged apical cells with ornamented outer 924 J. Hyv€onen et al. / Molecular Phylogenetics and Evolution 31 (2004) 915–928 Table 3 Outgroup taxa sampled in the extended analysis, all sequences downloaded from the GenBank Taxon Dicranum scoparium Encalypta rhaptocarpa E. streptocarpa Fontinalis antipyretica Hedwigia ciliata Hookeria acutifolia H. lucens Mnium hornum Rhodobryum giganteum Takakia lepidozioides Takakia sp. Theriotia lorifolia Tortula ruralis Accession No. 18S rbcL rps4 trnL-F nad5 X89874 AF023680 AF231067 AJ275167 AF234158 AF023777 AF234159 AF023717 Z98956 — — — — AF023714 AJ275010 AJ275183 AF231073 AF158170 AF023817 AJ251309 AF023771 AF233587 AJ291556 AJ291570 Z98966 — — — — AF215906 AF023767 AF023737 Z98969 AJ291567 Z98964 AJ291553 AF231904 AF229893 AF023722 — AJ243168 X80985 AF023699 AJ269686 AF226820 AJ275176 AF244565 AJ251316 AF023796 AF023789 AF306950 — — — AF223007 AF023682 AF232698 AJ275169 AF223036 AF023831 — — — — AJ291562 See text for further details. walls. There are, however, also species of Atrichum and Pogonatum that lack lamellae. Our results lend support to the view that the absence of these structures in some species of these two genera is due to reduction. Similar variation can be also be seen in the leaf form. Alophosia azorica shows the differentiation of leaf parts that is typical for Polytrichales: a distinctly widened leaf-base (sheath) with a long and narrow apical blade. It seems that in most cases undifferentiated leaves are due to reduction but whether this applies also to Atrichopsis compressa and Bartramiopsis lescurii is not clear. It is equally parsimonious to assume that the common ancestor of all Polytrichales already had a differentiated sheath as it is to suppose that differentiation has taken place independently in Alophosia, Lyellia, and in the common ancestor of all other Polytrichales. Leaf margins in Polytrichales vary from entire to distinctly toothed. Taxa with sharp, unicellular teeth on the blade margins do not form a monophyletic group and therefore one has to assume that this kind of teeth evolved independently at least three times. Polytrichales also show variation in the thickness of leaf margins, and this seems to be a highly homoplasious character as well. The hairy calyptra, a structure typical of many Polytrichales, is present in three different groups. The calyptra of Alophosia has both uni- and multiseriate hairs, while those with exclusively multiseriate (Dawsonia) or uniseriate hairs (Polytrichum and Pogonatum) seem to have evolved independently. Whether peristomes of all mosses are homologous is also open to debate according to our results. If Oedipodium is a sister taxon to Polytrichales then it is more parsimonious to assume that the peristomes have evolved independently in Bryopsida, Atrichopsis and the rest of the Polytrichales. However, if Oedipodium is in the more basal position as a sister taxon to all mosses excluding Sphagnopsida and Andreaeaopsida then it is equally parsimonious to assume that lack of peristome in Alophosia, Bartramiopsis, and Lyellia is due to reduction. At the moment it seems that none of the morphological characters is a very good indicator of phylogeny, since all of them show considerable homoplasy. However, the level of homoplasy observed does not differ significantly from the values obtained for sequence data. The combined simultaneous analysis provided results that would have been unexpected based solely on morphology as observed in our earlier analysis (Hyv€ onen et al., 1998). Plants such as Polytrichadelphus, Dawsonia, and Polytrichum with large, well-developed gametophytes and leaves with differentiated hinge-tissue, numerous adaxial lamellae and specialized marginal cells appear to be quite unrelated to each other. These elaborate gametophyte structures seem to have evolved independently in these lineages. When we examine the current geographical distributions of the taxa, an interesting pattern is quite obvious (Fig. 3). The most basal clades include species that are today geographically widely separated from each other. Alophosia is a Macaronesian endemic, Atrichopsis is restricted to southernmost South America, Bartramiopsis is confined to equally oceanic climates around the northern Pacific coastline and Lyellia has species in the high arctic and in the Himalayas. However, conclusions of whether the common ancestor of all Polytrichales was found in the north or in the south should not be made based on the pattern illustrated in Fig. 3. As mentioned above, our sampling of the basal moss lineages is still so unbalanced that such conclusions would be unwarranted. While Atrichopsis obviously represents an archaic element in the Southern Hemisphere, most of the genera with contemporary Southern Hemisphere distributions seem to have originated later. All these genera form a grade within Polytrichales, and it is parsimonious to assume that widespread and common Northern Hemisphere genera such as Atrichum, Polytrichum, and Pogonatum represent younger elements of the Northern Hemisphere J. Hyv€onen et al. / Molecular Phylogenetics and Evolution 31 (2004) 915–928 925 Fig. 3. The distribution of the Polytrichales included in the analysis mapped on the strict consensus tree of the two trees obtained from the analysis based on simultaneous analysis of the total matrix composed of six datasets. Species found exclusively in the southern Hemisphere are marked with thick light gray bars and those that are present in both Hemispheres with dark gray bars. Unmarked species are confined to the northern Hemisphere. Polytrichales and originated from their common ancestor with Meiotrichum and some other smaller northern genera. This corresponds to a large extent with the biogeographic scenario presented by Smith (1972) although at least with the current sampling it seems that Atrichum, Polytrichum, and Pogonatum did not originate in the south but instead spread there later. If we follow recommendations and conventions by Wiley (1981) our results would necessitate numerous changes in nomenclature. Four of these would require only adoption of the older names of the taxa, which in some cases, have been continuously widely used. However, at this point we decline to make any formal changes because data are accumulating at an ever-increasing pace 926 J. Hyv€onen et al. / Molecular Phylogenetics and Evolution 31 (2004) 915–928 and we are sure that our results will be challenged shortly with wider sampling of characters and taxa. If short internal branches that are typical for our results at this point still persist after analyses of much larger data matrices, it will be time to reconsider the existing nomenclature in order to make it compatible with phylogenetic relationships and as informative as possible. 11. 12. 13. Acknowledgments 14. We are grateful to Bernard Goffinet, Naoki Nishimura, Daniel H. Norris, and Jon Shaw for providing recently collected specimens for our study and curators of AAS for the specimens of Atrichopsis compressa. We thank Cymon Cox, Bernard Goffinet, Louise Lewis, Brent Mishler, Angela Newton, and Steffen Schaffer for generously providing unpublished primers and sequences. Colleagues that have shared their ideas on mosses and phylogenetic methods are too numerous to list but we want especially to thank two for their insightful ideas: Leif Tibell on simultaneous analysis and Gonzalo Giribet on representative sampling. We thank two anonymous reviewers for their constructive comments. The financial support to JH by Academy of Finland (projects 39482 and 50620) is cordially acknowledged. The support NSF/USDA/DOE (Grant No. 94-37105-0713) to the Green Plant Phylogeny Research Coordination Group (GPPRCG) is also acknowledged. Appendix A Different character states are coded with (0), (1), (2), and (3). These codes do not, however, designate a priori which of the states is plesiomorphic or apomorphic. Unless otherwise stated treatment of these characters follow Hyv€ onen et al. (1998) and this paper should be consulted for more detailed discussion of the characters and their states. 1. Branching: not or sparingly branched (0); dendroid (1); branches in fascicles (2). 2. Sheath type: sheath differentiated, broad (0); sheath not differentiated (1). 3. Hyaline sheath margin: present (0); absent (1). 4. Sheath/leaf base margin: entire (0); ciliate (1); serrate (2). 5. Hinge tissue: present (0); absent (1). 6. Leaf (blade) margin: serrate (0); toothed (1); entire (2). 7. Leaf border: absent (0); Atrichum-type (1). 8. Thickness of leaf margin: unistratose (0); two- or more stratose (1). 9. Adaxial lamellae: present (0); absent (1). 10. Extent of adaxial lamellae: numerous, occupying full width of lamina (0); restricted to median 15. 16. 17. 18. 19. 20. 21. 22. strip (1). Taxa lacking lamellae were scored as ‘‘)’’ (inapplicable) for this and following three characters. Thickness of lamella-free lamina: unistratose (0); bistratose (1). Lamella marginal cells (LMCs, apical cells of lamellae): single (0); geminate (1). Size of LMC lumen: comparable to lower cells of lamellae (0); elongated as seen in side view (1); higher than lower cells (1). Form of LMCs as seen in cross-section: rounded (undifferentiated) (0); ovoid to bottle-shaped (1); flattened (2); retuse (3). LMC cell-walls: undifferentiated (0); incrassate (1); only outer wall incrassate (2); outer wall notched (3). Lamella cuticle: smooth (0); papillose (1). Paraphyses: present (0); absent (1). Calyptra: present (0); absent (1). Calyptra hair: uniseriate (0); multiseriate (1); sparse or none (2). Pseudopodium: absent (0); present (1). Seta: present (0); absent (1). Seta surface: smooth (0); papillose (1). The capsule form was treated with two characters (capsule cross-section and angles) by Hyv€ onen et al. (1998). We now think that it is better to distinguish capsule cross-sectional symmetry from the capsule angles and therefore include here three distinct characters: symmetry, angles, and angle form. 23. Capsule cross sectional symmetry: symmetrical (0); dorsiventral (1); bilaterally compressed (2). 24. Angles of capsule: none (0); two (1); 4–6 (2); numerous (6–8) or practically absent (3). 25. Capsule angle form: blunt (0); sharp, knife-edged (1); ribbed (2). 26. Capsule dehiscence: longitudinal slits (0); operculum (1). 27. Exothecium: smooth (0); mamillose (1); papillose (2). 28. Exothecial pitting: none (0); thin-spots (1); pitted (2). 29. Apophysis: tapering (0); contracted (1); discoid (2). 30. Stomata: present (0); absent (1). 31. Stomata type: superficial (0); cryptopore (1). 32. Stomata extent: restricted to base (0); dispersed (1). 33. Peristome: present (0); absent (1). 34. Peristome type: polytrichoid (0); dawsonioid (1); tetraphid (2); arthrodont (3). 35. Tooth structure: simple (0); compound, sinus broad (1); compound, sinus narrow (2). 36. Tooth number: 32 (0); 64 (1); 4 (2); 16 (3). 37. Peristome pigmentation: pale (0); intensively colored (1). 38. Epiphragm type: discoid (0); absent (1); stopper (2); cylindric (rod) (3). 39. Capsule rim disc: narrow (0); broad (disc) (1). J. Hyv€onen et al. / Molecular Phylogenetics and Evolution 31 (2004) 915–928 40. Spore sac: overarching columella (0); cylindrical (1). 41. Spore origin: from endothecium (0); from exothecium (1). 927 42. Spore surface: papillose (0); echinulate (1); Bartramiopsis-type (2); Oedipodium-type (3). 43. Brood-bodies: absent (0); present (1). Appendix B Different character states are coded with (0), (1), (2), and (3). Polymorphism (0,1) marked with M, and (0,2) with R. (?) denote unknown information and (–) inapplicable characters. 928 J. Hyv€onen et al. / Molecular Phylogenetics and Evolution 31 (2004) 915–928 References Brotherus, V.F., 1925. Musci (Laubmoose) 2. H€alfte. In: Engler, A. (Ed.), Die nat€ urlichen Pflanzenfamilien. 2. Aufl. 11. Engelmann. Leipzig. Buck, W.R., Goffinet, B., 2000. Morphology and classification of mosses. In: Shaw, A.J., Goffinet, B. (Eds.), Bryophyte Biology. Cambridge University Press, Cambridge, pp. 71–123. Buck, W.R., Goffinet, B., Shaw, A.J., 2000. Novel relationships in pleurocarpous mosses as revealed by cpDNA sequences. Bryologist 103, 774–789. Chase, M.W., Albert, V.A., 1998. A perspective on the contribution of plastid rbcL DNA sequences to angiosperm phylogenetics. In: Soltis, D.E., Soltis, P.E., Doyle, J.J. (Eds.), Molecular Systematics of Plants II. Kluwer Academic Publishers, Boston, MA, pp. 488–507. Cox, C., Hedderson, T.J., 1999. Phylogenetic relationships among the ciliate arthrodontous mosses: evidence from nuclear and chloroplast gene sequences. Plant Syst. Evol. 215, 119–139. Cox, C.J., Goffinet, B., Newton, A.E., Shaw, A.J., Hedderson, T.A.J., 2000. Phylogenetic relationships among the diplolepideous-alternate mosses (Bryidae) inferred frrom nuclear and chloroplast DNA sequences. Bryologist 103, 224–241. Crosby, M.R., 1999. An overview of the species of mosses of the world. Lecture in the XVIth International Botanical Congress, St. Louis, MO. Crum, H., 2001. Structural diversity of bryophytes. The University of Michigan Herbarium, Ann Arbor, MI. Derda, G.S., Wyatt, R., 1999. Isozyme evidence regarding the origins of three allopolyploid species of Polytrichastrum (Polytrichaceae, Bryophyta). Plant Syst. Evol. 220, 37–53. Eernisse, D.J., Kluge, A.G., 1993. Taxonomic congruence versus total evidence, and amniote phylogeny inferred from fossils, molecules, and morphology. Mol. Biol. Evol. 10, 1170–1195. Farris, J.S., 1989. The retention index and rescaled consistency index. Cladistics 5, 417–419. Farris, J.S., 1997a. xac. Computer program and documentation distributed by the author. Farris, J.S., 1997b. xarn. Computer program and documentation distributed by the author. Farris, J.S., K€ allersj€ o, M., Kluge, A.G., Bult, C., 1994. Testing significance of incongruence. Cladistics 10, 315–320. Farris, J.S., Albert, V.A., K€allersj€ o, M., Lipscomb, D., Kluge, A.G., 1996. Parsimony jackknifing outperforms neighbor-joining. Cladistics 12, 99–124. Forrest, L.L., 1995. A phylogenetic analysis of Polytrichaceae (Musci). M.Sc. thesis, Department of Botany, University of Reading. Goffinet, B., Cox, C.J., Shaw, A.J., Hedderson, T.A.J., 2001. The Bryophyta (Mosses): systematic and evolutionary inferences from an rps4 gene (cpDNA) phylogeny. Ann. Bot. 87, 191–208. Goloboff, P.A., 1994. NONA: A tree searching program. Program and documentation, Available from ftp.unt.edu.ar/pub/parsimony. Hedderson, T.A., Chapman, R.L., Rootes, W.L., 1996. Phylogenetic relationships of bryophytes inferred fron small-subunit rRNA gene sequences. Plant Syst. Evol. 200, 213–224. Hedderson, T.A., Chapman, R.L., Cox, C.J., 1998. The origins and diversification of land plants: new evidence from molecules. In: Ashton, N.W., Bates, J.W., Duckett, J.G. (Eds.), Bryology for the Twenty-first Century. Maney, Leeds, pp. 65–77. Hyv€ onen, J., 1989. A synopsis of genus Pogonatum (Polytrichaceae, Musci). Acta Bot. Fennica 138, 1–87. Hyv€ onen, J., Hedderson, T.A., Smith Merril, G.L., Gibbings, J.G., Koskinen, S., 1998. On phylogeny of the Polytrichales. Bryologist 101, 489–504. Jeanmougin, F., Thompson, J.D., Gouy, M., Higgins, D.G., Gibson, T.J., 1998. Multiple sequence alignment with Clustal X. Trends Biochem. Sci 23, 403–405. Kluge, A.G., 1989. A concern for evidence and a phylogenetic hypothesis of relationships among Epicrates (Boidae, Serpentes). Syst. Zool. 38, 7–25. Kluge, A.G., 1997. Testability and the refutation and corroboration of cladistic hypotheses. Cladistics 13, 81–96. Kluge, A.G., Farris, J.S., 1969. Quantitative phyletics and the evolution of anurans. Syst. Zool. 18, 1–32. Konopka, A.S., Herendeen, P.S., Smith Merrill, G.L., Crane, P.R., 1997. Sporophytes and gametophytes of Polytrichaceae from the Campanian (Late Cretaceous) of Georgia, USA. Int. J. Plant Sci. 158, 489–499. K€allersj€ o, M., Albert, V.A., Farris, J.S., 1999. Homoplasy increases phylogenetic structure. Cladistics 15, 91–94. Lewis, L.A., Mishler, B.D., Vilgalys, R., 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. Lipscomb, D.L., Farris, J.S., K€allersj€ o, M., Tehler, A., 1998. Support, ribosomal sequences and the phylogeny of the eukaryotes. Cladistics 14, 303–338. Magombo, Z., 2003. Inferences on evolution and phylogeny of peristomate mosses based on chloroplast genes. Syst. Bot. 28, 24–38. Mickevich, M.F., Farris, J.S., 1981. The implications of congruence in Menidia. Syst. Zool. 30, 351–369. Mishler, B.D., 2000. Deep phylogenetic relationships among ‘‘plants’’ and their implications for classification. Taxon 49, 661–683. Mishler, B.D., Churchill, S.P., 1984. A cladistic approach to the phylogeny of the ‘‘bryophytes’’. Brittonia 36, 406–424. Miyamoto, M.M., Fitch, W.M., 1995. Testing species phylogenies and phylogenetic methods with congruence. Syst. Biol. 44, 64–76. Newton, A., Cox, C., Duckett, J., Wheeler, J.A., Goffinet, B., Hedderson, T.A.J., Mishler, B.D., 2000. Evolution of the major moss lineages: phyologenetic analyses based on multiple gene sequences and morphology. Bryologist 103, 187–211. Nickrent, D.L., Parkinson, C.L., Palmer, J.D., Duff, R.J., 2000. Multigene phylogeny of land plants: hornworts are basal and mosses are sister to liverworts. Mol. Biol. Evol. 17, 1885–1895. Nixon, K.C., 1999. Winclada (beta) ver. 0.9. Published by the author, Ithaca, NY. [Available from http://www.cladistics.com]. Nixon, K.C., Carpenter, J.M., 1996. On simultaneous analysis. Cladistics 12, 221–242. Renzaglia, K.S., Duff, R.J., Nickrent, D.N., Garbary, D.J., 2000. Vegetative and reproductive innovations of early land plants: implications for a unified phylogeny. Philos. T. R. Soc. B 355, 769–793. Schofield, W.B., 1985. Introduction to Bryology. Macmillan, New York. Siddall, M.E., 1997. Prior agreement: arbitration or arbitrary? Syst. Biol. 46, 765–769. Siddall, M.E., Whiting, M.F., 1999. Long-branch abstractions. Cladistics 15, 9–24. Smith, G.L., 1971. A conspectus of the genera of Polytrichaceae. Mem. N. Y. Bot. Gard. 21, 1–83. Smith, G.L., 1972. Continental drift and the distribution of Polytrichaceae. J. Hattori Bot. Lab. 35, 41–49. Steinhauser, S., Beckert, S., Capesius, I., Malek, O., Knoop, V., 1999. Plant mitochondrial RNA editing—extreme in hornworts and dividing the liverworts? J. Mol. Evol. 48, 303–312. Swofford, D.L., Olsen, G.L., Waddell, P.J., Hillis, D.M., 1996. Phylogenetic inference. In: Hillis, D.M., Morowitz, C., Mable, B.K. (Eds.), Molecular Systematics. Sinauer, Sunderland. MA, pp. 407–514. Wheeler, W.C., Cartwright, P., Hayashi, C.Y., 1993. Arthropod phylogeny: a combined approach. Cladistics 9, 1–39. Wiley, E.O., 1981. Phylogenetics: the theory and practice of phylogenetic systematics. Wiley, New York. Yang, Z., 1998. On the best evolutionary rate for phylogenetic analysis. Syst. Biol. 47, 125–133.