Taxonomy and phylogeny in the earliest diverging pleurocarps: square holes and bifurcating pegs

Dietmar Quandt

Taxonomy and phylogeny in the earliest diverging pleurocarps: square holes and bifurcating pegs Author(s): Neil E. Bell, Dietmar Quandt, Terry J. O'Brien, and Angela E. Newton Source: The Bryologist, 110(3):533-560. Published By: The American Bryological and Lichenological Society, Inc. DOI: http://dx.doi.org/10.1639/0007-2745(2007)110[533:TAPITE]2.0.CO;2 URL: http://www.bioone.org/doi/ full/10.1639/0007-2745%282007%29110%5B533%3ATAPITE%5D2.0.CO %3B2 BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the biological, ecological, and environmental sciences. BioOne provides a sustainable online platform for over 170 journals and books published by nonprofit societies, associations, museums, institutions, and presses. Your use of this PDF, the BioOne Web site, and all posted and associated content indicates your acceptance of BioOne’s Terms of Use, available at www.bioone.org/page/ terms_of_use. Usage of BioOne content is strictly limited to personal, educational, and non-commercial use. Commercial inquiries or rights and permissions requests should be directed to the individual publisher as copyright holder. BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions, research libraries, and research funders in the common goal of maximizing access to critical research. Taxonomy and phylogeny in the earliest diverging pleurocarps: square holes and bifurcating pegs NEIL E. BELL Botanical Museum, P.O. Box 7, 00014 University of Helsinki, Finland e-mail: neil.bell@helsinki.fi DIETMAR QUANDT Technische Universität Dresden, Institut für Botanik, Plant Phylogenomics and Phylogenetics Group, 01062 Dresden, Germany e-mail: dietmar.quandt@tu-dresden.de TERRY J. O’BRIEN Department of Biological Sciences, Rowan University, Glassboro, NJ 08028, U.S.A. e-mail: obrien@rowan.edu ANGELA E. NEWTON Department of Botany, The Natural History Museum, Cromwell Road, London SW7 5BD, England, U.K. e-mail: a.newton@nhm.ac.uk ABSTRACT. The extant members of the earliest diverging pleurocarpous moss lineages comprise few species but span a wide range of structural and molecular diversity, most of it restricted to temperate and high-altitude tropical forests in the Southern Hemisphere. We present the most comprehensive molecular phylogenetic study of these lineages to date, based on parsimony and Bayesian analyses of four regions from the chloroplast and mitochondrial genomes. In addition to corroborating the findings of parsimony methods in this and previous studies, the results of heterogeneous Bayesian analyses provide strong support for sub-topologies that are also consistently found under parsimony, but are rarely well supported. Careful model specification and investigation of potential sources of error increase confidence in the Bayesian results, which provide the basis for a substantially revised classification reflecting the best currently available hypothesis of evolutionary history. The genera previously classified in the Rhizogoniaceae, together with Orthodontium, Orthodontopsis, Aulacomnium and Calomnion, are recognized in three families, the Orthodontiaceae, Rhizogoniaceae and Aulacomniaceae, and three monofamilial orders, the Orthodontiales ord. nov., Rhizogoniales and Aulacomniales ord. nov. Many of the species previously recognized in Hypnodendron are placed in Sciadocladus, Touwiodendron gen. nov., Dendro-hypnum or Mniodendron. These genera, with the exception of Sciadocladus, are placed in the Hypnodendraceae together with Spiridens, Franciella, Cyrtopus and Bescherellia. Braithwaitea is excluded from the Hypnodendraceae and recognized in the monogeneric Braithwaiteaceae fam. nov., while Sciadocladus is placed with Pterobryella and THE BRYOLOGIST 110(3), pp. 533–560 Copyright Ó2007 by the American Bryological and Lichenological Society, Inc. 0007-2745/07/$2.95/0 534 the bryologist 110(3): 2007 Cyrtopodendron in the Pterobryellaceae. The Hypnodendraceae, Pterobryellaceae, Braithwaiteaceae and Racopilaceae are recognized in the order Hypnodendrales comb. et stat. nov. We discuss the advantages and limitations of ranked classification systems and propose the abandonment of intercalated Linnaean ranks between order and class levels in Bryopsida. Three node-based informal names, the Pleurocarpids, the Core Pleurocarps and the Homocostate pleurocarps, are defined to represent evolutionarily significant clades within the pleurocarpous group. KEYWORDS. Dendro-hypnum, Hypnodendraceae, Hypnodendron, molecular phylogeny, nodebased, pleurocarps, pleurocarpy, posterior probability, Rhizogoniaceae, taxonomy, Touwiodendron. ^ The pleurocarps are a highly diverse and speciose group of mosses derived from within the diplolepidous-alternate clade, the major entity that also includes several large and familiar acrocarpous families such as the Bryaceae, Mniaceae, Orthotrichaceae and Bartramiaceae (Cox et al. 2000, 2004; Cox & Hedderson 1999; Goffinet & Buck 2004; Newton et al. 2000) and for which the eponymous peristome type is synapomorphic. An apical pleurocarpous clade, comprising the orders Ptychomniales, Hookeriales and Hypnales, includes 98% of pleurocarpous species. It is currently recognized as subclass Hypnidae (Goffinet & Buck 2004) and now widely acknowledged to be monophyletic (Bell & Newton 2004; Buck et al. 2005, 2000a, b; De Luna et al. 1999, 2000; Goffinet & Buck 2004; Goffinet et al. 2001; O’Brien 2007; Quandt et al. 2007; Tsubota et al. 2002). However, the relationships of the earlier diverging pleurocarpous lineages to the Hypnidae were ambiguous until very recently and consequently are inadequately represented in current taxonomy. While there has been much progress in phylogenetic reconstruction and character interpretation (Bell & Newton 2004, 2005, 2007; Newton 2007; Newton et al. 2007; O’Brien 2007), the results sit uneasily with current classification schemes (e.g., Goffinet & Buck 2004) due to marked asymmetry in extant species diversity at several nodes as well as a number of parallelisms and reversals in linked suites of morphological characters. This taxonomic ambiguity continues to contribute to the relative obscurity of these taxa, as does their almost complete restriction to ^ ^ the Southern Hemisphere and the widely held perception that they are not ‘‘true’’ pleurocarps due to their determinate, usually orthotropous growth form (Meusel 1935; Newton 2007). Early results from molecular studies using rbcL only with limited taxon sampling (e.g., Withey 1996a) appeared to corroborate such views by suggesting that these lineages were not directly related to the majority of the pleurocarps and that their pleurocarpous (or pseudo-pleurocarpous) traits were convergent. This hypothesis is refuted by recent studies (Bell & Newton 2004, 2005, 2007; Newton 2007; Newton et al. 2007; O’Brien 2007), all of which recognize a single large clade of predominantly pleurocarpous taxa. The earlier pleurocarp lineages lack many of the most distinctive features of the group as a whole and in aspects of habit, leaf morphology and growth form more closely resemble acrocarps, suggesting that pleurocarpy did not evolve simultaneously with the associated secondary traits that are thought to be largely responsible for its success. Under this scenario, which is consistent with the significantly earlier date estimated by Newton et al. (2007) for the origin of pleurocarpy (194–161 mya) compared with that for the diversification of the Hypnidae (165–131 mya), pleurocarpy sensu stricto (Newton & De Luna 1999) was an existing preädaptation in the common ancestor of the Hypnidae. Hence we must look to the earlier diverging lineages to reconstruct the evolution of the trait. The condition of pleurocarpy is the production of female gametangia on extremely short branches that Bell et al.: Phylogeny of earliest diverging pleurocarps are almost entirely specialized for this purpose, have perichaetial leaves and do not produce normal vegetative leaves, at least prior to fertilization (Bell & Newton 2007; Newton 2007; Newton & De Luna 1999). This adaptation can be thought of as providing a spatially well-delimited gametophytic structure at the modular level of organization (see Mishler & De Luna 1991, for a description of the ontogenetic hierarchy in mosses) that is highly specialized for fertilization and support of the sporophyte generation. It is either a synapomorphy for the clade that includes Hymenodon and all subsequently derived pleurocarpous lineages or is independently derived in Hymenodon and in the other pleurocarps (Bell & Newton 2007). Isolated reversals to acrocarpy occur in the rhizogoniaceous taxa. While there are rare occurrences of at least superficially similar conditions in other distantly related groups outside of the diplolepidous-alternate mosses (e.g., Pleurochaete, see Newton 2007 for discussion), these are clearly independently derived. In non-pleurocarpous mosses nearly all modular-level structures have a significant vegetative role, and hence fertilization and sporophyte support functions must be performed by what are, in effect, normal stems and branches. As production of gametangia terminates module growth, this constrains the form, branching and orientation of the entire gametophyte. Freedom from this constraint is hypothesized to have allowed pleurocarpous mosses to diversify into a range of novel forms and to exploit many new niches (e.g., Vitt 1984), although these appear to be different in the earliest and in the more recently derived lineages. The groups within the rhizogoniaceous grade (Bell & Newton 2004; O’Brien 2007) comprise predominantly orthotropous or short-stemmed complanate-tufted, sparsely branched plants growing on soil, tree bases, dead wood or as small epiphytes on tree ferns, while the hypnodendroid pleurocarps (Bell & Newton 2004) are mostly very robust plants of stable forest habitats, either umbellate-dendroid colonists of the forest floor or large epiphytes with a pinnate-dendroid or sparsely branched architecture (Bell & Newton 2005). Both the rhizogonioid and hypnodendroid mosses nearly always have clearly determinate primary modules that branch from the base and strong costae that are well differentiated in 535 transverse section, in contrast to the Hypnidae in which growth of the primary module is usually nonor indistinctly determinate associated with a plagiotropous habit, and the complexity of the costa is much reduced (Hedenäs 1994; Newton 2007; Newton & De Luna 1999). There is a marked geographical pattern in the phylogeny of the early diverging pleurocarpous mosses. All the major lineages in both the rhizogoniaceous grade and the hypnodendroid pleurocarps sensu Bell and Newton (2004), as well as the Ptychomniales, now strongly supported as the earliest diverging lineage within the Hypnidae (Buck et al. 2005), are distributed almost exclusively in southern South America, Australasia, Malesia, South Africa and the Pacific. A few outlier taxa occur in southern and eastern continental Asia and Japan, tropical and North America and the Caribbean. Generic diversity in all these groups is highest in the Nothofagus zones of eastern Australia, New Zealand, New Caledonia and southwestern South America, i.e., the extant ecosystems that are closest in their combined floristic and climatic affinities to the moist high latitude forests of late Cretaceous Gondwana (Hill 2001; McLoughlin 2001). There is tentative evidence, however, that the distribution of the rhizogoniaceous mosses may have been more extensive in the past. Frahm (2001, 2004) described two fossils of Rhizogonium from Eocene Baltic amber (which from the descriptions and photographs seem closer to Pyrrhobryum), while the acrocarpous Aulacomnium, now shown to be derived from within a clade of pleurocarpous rhizogoniaceous taxa (Bell & Newton 2004, 2007; O’Brien 2007), is most diverse in the Northern Hemisphere. Bell and York (2007) have described Vetiplanaxis pyrrhobryoides, a plant showing some similarities to extant members of Pyrrhobryum and Rhizogonium, from Burmese amber (Middle Cretaceous). The taxonomic history of what were previously considered the ‘‘bryalean’’ or ‘‘eubryalean’’ pleurocarpous taxa is briefly outlined in Bell and Newton (2004). Recent treatments (Buck & Goffinet 2000; Goffinet & Buck 2004) have recognized the Hypnales, Hookeriales and Ptychomniales as subclass Hypnidae and treated the earlier diverging lineages within a single order Rhizogoniales, as the monoördinal 536 the bryologist 110(3): 2007 superorder Rhizogonianae in subclass Bryidae (Goffinet & Buck 2004). Phylogenetic analyses including significant taxon sampling, however (Bell & Newton 2004, 2005; O’Brien 2007; Quandt et al. 2007) support a clade comprising the Hypnodendraceae, Racopilaceae, Pterobryellaceae, Cyrtopodaceae and Spiridentaceae (the hypnodendroid pleurocarps of Bell & Newton 2004) as the sister group of the Hypnidae. The other non-Hypnidaean pleurocarps (all currently members of the Rhizogoniaceae), in addition to the Aulacomniaceae, Calomniaceae and Orthodontiaceae, are placed in at least three clades that are the earliest diverging lineages within a larger monophyletic assemblage from which the Hypnidae-hypnodendroid clade is derived. The current taxonomy is thus outmoded in several respects: 1) the Rhizogoniales/ Rhizogonianae are paraphyletic; 2) the hypnodendroid pleurocarps, the Hypnidae-hypnodendroid clade and the pleurocarp clade sensu lato all lack recognition within the existing nomenclature; 3) the placement of the early diverging pleurocarpous lineages in the predominantly acrocarpous (and paraphyletic) subclass Bryidae is at odds with their pleurocarpous status and their close relationships to the Hypnidae; and 4) generic and higher level taxonomy within both the hypnodendroid group and the rhizogoniaceous grade is in need of extensive revision. In this study we bring together new and existing data to present the most comprehensive phylogenetic analysis of the early diverging pleurocarpous lineages to date. A major objective is to provide a sound phylogenetic hypothesis on which to base a rationalization of the higher-level taxonomy of these organisms, a difficult path between avoiding artificial (i.e., clearly paraphyletic) groups on the one hand and disruptive and unpopular nomenclatural change on the other. We discuss the limitations of an extended Linnaean taxonomic hierarchy for representing the totality of pleurocarp phylogeny and suggest that an informal non-ranked approach to the naming of groups between the ordinal and class levels might better serve the needs of both systematists and other users of classifications at this time. We combine data from previous analyses (including those of Bell & Newton 2004, 2005 and O’Brien 2007) and generate a number of new sequences in order to compile a comprehensive matrix from the mitochondrial nad5 region and the chloroplast rbcL, rps4 and trnL regions, with dense taxon sampling across all the early diverging pleurocarpous lineages. For introns and non-transcribed spacers we employ a strategy informed by known mutational mechanisms in non-coding DNA to guide primary homology assessment (i.e., alignment; see Kelchner 2000). This approach uses knowledge of molecular processes and structure to avoid unrealistic hypotheses of homology and to minimize subjectivity in the alignment process, and can be contrasted with other strategies that implicitly or explicitly assume random and independent evolution of individual sites in non-coding regions. Indels in the alignment are thus considered to be informed hypotheses of primary homology to be tested by congruence or by conformity to models of evolution. They are coded separately using the simple indel coding method of Simmons and Ochoterena (2000). MATERIALS AND METHODS Sampling. We maximized sampling from both the rhizogoniaceous grade and the hypnodendroid pleurocarps by combining data from our previous studies with newly generated sequences as well as others that have become available on GenBank (Table 1). The study of Bell and Newton (2004) sampled extensively from the rhizogonioid grade, as did that of O’Brien (2007). Bell and Newton (2005) focused on the hypnodendroid taxa and employed similar character sampling to Bell and Newton (2004), with the addition of the chloroplast region trnL, also used by O’Brien (2007). Sampling from the hypnodendroid group was expanded to include all nine sections of Hypnodendron (five of which are monospecific) with the addition of H. milnei. The sole South American species, H. microstictum, was added and sampling from Spiridens, Hypnodendron sect. Comosa and H. sect. Phoenicobryum was increased. The monospecific Cyrtopodendron was included. In the rhizogonioid grade we added Orthodontopsis, expanded sampling from Rhizogonium to include South American material of R. novae-hollandiae, and included Hymenodontopsis stresemannii (not included in Bell & Newton 2004). Significantly, the different but overlapping character sampling of Bell and Newton (2004) and Bell et al.: Phylogeny of earliest diverging pleurocarps 537 Table 1. Taxa included in the phylogenetic analyses with GenBank accession numbers. Voucher specimen information for sequences generated by the authors is provided in the GenBank records or in the studies in which they were first published. Asterisks (*) indicate taxa included in the resampled dataset used to explore the influence of non-uniform clade priors on Bayesian posterior probabilities. Taxon Acidodontium sprucei (Mitt.) A. Jaeger Amblyodon dealbatus (Hedw.) Bruch & Schimp. Anacolia webbii (Mont.) Schimp. Anomobryum julaceum (P. Gaertn., B. May. & Scherb.) Schimp. *Aulacomnium androgynum (Hedw.) Schwägr. *Bartramia halleriana Hedw. Bartramia stricta Brid. Bescherellia cryphaeoides (Müll. Hal.) M. Fleisch. Bescherellia elegantissima Duby Brachymenium nepalense Hook. Braithwaitea sulcata (Hook.) A. Jaeger Bryum alpinum With. Bryum argenteum Hedw. *Calomnion brownseyi Vitt & H. A. Mill. Calomnion complanatum (Hook.f. & Wilson) Lindb. Cinclidium stygium Sw. *Cryptopodium bartramioides (Hook.) Brid. Cyrtomnium hymenophyllum (Bruch & Schimp.) Holmen *Cyrtopodendron vieillardii (Müll. Hal.) M. Fleisch. Cyrtopus setosus (Hedw.) Hook.f. Dicranum scoparium Hedw. *Funaria hygrometrica Hedw. Garovaglia elegans (Dozy & Molk.) Bosch & Sande Lac. Glyphothecium sciuroides (Hook.) Hampe *Goniobryum subbasilare (Hook.) Lindb. *Hampeella alaris (Dixon & Sainsbury) Sainsbury Hedwigia ciliata (Hedw.) P. Beauv. Hildebrandtiella endotrichelloides Müll. Hal. Hookeria lucens (Hedw.) Sm. *Hymenodon pilifer Hook.f. & Wilson *Hymenodon sphaerothecius Besch. *Hymenodontopsis stresemannii Herzog Hypnodendron arcuatum (Hedw.) Lindb. ex Mitt. Hypnodendron auricomum Broth. & Geh. Hypnodendron camptotheca (Duby ex Besch. ex Paris) Touw Hypnodendron comatulum (Geh. ex Broth. & Watts) Touw Hypnodendron comatum (Müll. Hal.) Touw Hypnodendron comosum (Labill.) Mitt. Hypnodendron dendroides (Brid.) Touw Hypnodendron diversifolium Broth. & Geh. Hypnodendron fuscomucronatum (Müll. Hal.) A. Jaeger *Hypnodendron kerrii (Mitt.) Paris Hypnodendron menziesii (Hook.) Paris Hypnodendron microstictum Mitt. ex A. Jaeger Hypnodendron milnei Mitt. Hypnodendron reinwardtii (Schwägr.) A. Jaeger Hypnodendron spininervium (Hook.) A. Jaeger nad5 rps4 rbcL trnL AY908357 AY908377 AY908949 AY908353 AJ291564 Z98961 AY312870 AY608580–1 AY608578–9 AY908354 AY524524 AY631210 AY908945 AY631211 AY631212 AY908361 AY631213 AY908363 AY908768 AY608582–3 Z98956 Z98959 AY631145 AY631216 AY631217 AY524519 Z98966 AY908702 EF107537 AY631218 AY631219 AY908782 AY524540 EF562512 AY524531 AY524523 EF562513 AY524521 AY524537 AY524535 — AY524525 AY524522 AY908771 EF562514 EF562515 AY524536 AY078316 AF223062 AF4910292 AF023786 AF023811 EF107539 AF023799 AY524473 AY524472 AY078338 AY524469 AF023783 AY078318 AY631140 AY631141 AF023791 AY631142 AF023792 AY908010 AY524479 AF231277 AJ845203 AY631181 AY631147 AY631148 AY524463 AF478289 AY306925 AJ251316 AY631149 AY631150 AY142998 AY524484 AY524487 AY524476 AY524468 AY857779 AY524466 AY524482 AY524480 AY524488 AY524470 AY524467 AY908008 EF562529 EF562530 AY524481 AY163012 AJ275181 AF4351332 AJ275172 AY631174 AF231090 AY312926 AY524445 AY524444 EF107529 AY524441 AY631176 AY163024 AY631177 AY631178 EF107530 AF231084 EF107531 — AY524451 AF231067 AF005513 AY631215 AY631183 AY631184 AY524435 AF478234 EF107535 AY631185 AY631186 AY631187 AY853997 AY524456 AY524459 AY524448 AY524440 — AY524438 AY524454 AY524452 AY524460 AY524442 AY524439 — EF562526 EF562527 AY524453 AY078290 AY501425 EF107540 AF023739 AY857795 EF107539 AF023756 AY524501 AY524500 AY078311 AY524497 AF023738 AY078291 EF562516 AY857811 AF023763 AY857802 AF023764 — AY524507 AF234159 AF023716 DQ194218 DQ194221 AF023753 AY524491 AF478336 AY306759 AF215906 AY857804 EF562517 AY857803 AY524512 AY524515 AY524504 AY524496 AY857814 AY524494 AY524510 AY524508 AY524516 AY524498 AY524495 AY02899 EF562518 EF562519 AY524509 538 the bryologist 110(3): 2007 Table 1. Continued. Taxon Hypnodendron subspininervium (Müll. Hal.) A. Jaeger *Hypnodendron vitiense Mitt. Hypnum cupressiforme Hedw. *Leptostomum inclinans R. Br. *Leptotheca boliviana Herzog *Leptotheca gaudichaudii Schwägr. Leucodon sciuroides (Hedw.) Schwägr. Leucolepis acanthoneura (Schwägr.) Lindb. *Lopidium concinnum (Hook.) Wilson Macrocoma tenuis (Hook. & Grev.) Vitt *Mesochaete taxiforme (Hampe) Watts & Whitel. *Mesochaete undulata Lindb. Mielichhoferia elongata (Hoppe & Hornsch.) Nees & Hornsch. Mnium hornum Hedw. Myurium hochstetteri (Schimp.) Kindb. *Orthodontium lineare Schwägr. *Orthodontopsis bardunovii Ignatov & B. C. Tan *Phyllodrepanium falcifolium (Schwägr.) Crosby Physcomitrella patens (Hedw.) Bruch & Schimp. Plagiobryum zierii (Hedw.) Lindb. Plagiomnium cuspidatum (Hedw.) T. J. Kop. Pleurozium schreberi (Brid.) Mitt. Pohlia nutans (Hedw.) Lindb. Powellia involutifolia Mitt. Pterobryella praenitens Müll. Hal. Pterobryon densum Hornsch. Ptychomnion cygnisetum (Müll. Hal.) Kindb. *Pyrrhobryum bifarium (Hook.) Manuel *Pyrrhobryum dozyanum (Sande Lac.) Manuel *Pyrrhobryum medium (Besch.) Manuel *Pyrrhobryum mnioides (Hook.) Manuel *Pyrrhobryum novae-caledoniae (Besch.) Manuel *Pyrrhobryum paramattense (Müll. Hal.) Manuel *Pyrrhobryum spiniforme (Hedw.) Mitt. *Pyrrhobryum vallis-gratiae (Müll. Hal.) Manuel Racopilum cuspidigerum (Schwägr.) Ångstr. Racopilum spectabile Reinw. & Hornsch. *Rhacocarpus purpurascens (Brid.) Paris *Rhizogonium distichum (Sw.) Brid. *Rhizogonium graeffeanum (Müll. Hal.) A. Jaeger *Rhizogonium novae-hollandiae (Brid.) Brid. (Australia) *Rhizogonium novae-hollandiae (Brid.) Brid. (Chile) *Rhizogonium pennatum Hook.f. & Wilson Rhizomnium gracile T. J. Kop. Rhodobryum giganteum (Schwägr.) Paris Rhytidiadelphus triquetrus (Hedw.) Warnst. Rutenbergia madagassa Geh. & Hampe Spiridens camusii Thér. Spiridens reinwardtii Nees Spiridens vieillardii Schimp. nad5 rps4 rbcL trnL AY524538 AY524526 AY908444 AY908359 AY908779 AY631220 AY908716 AY908366 AY631221 AY908940 AY524518 AY631222 AY312878 AJ291567 AY908439 AJ291566 AY908780 AY908374 Z98960 AY908355 AY908365 AY908642 AJ291565 AY524520 AY524539 AY908693 DQ200901 AY631224 AY631225 AY631226 AY631227 AY631228 AY631229 AY524541 AY631230 AY524532 AY524533 Z98967 AY524517 AY631231 AY631232 EF562511 AY631233 AY908362 AY312886 Z98971 AY524542 AY524530 AY908772 AY524529 AY532393 AY524471 DQ467883 AY907974 AF023816 AY631151 AY908186 AY857789 AY631153 AY618361 AY524462 AY631154 AF023793 AF023796 AY908180 AF023800 AY908178 AF143074 NC005087 AY078335 AY907978 AY908281 AY907983 AY524465 AY524483 AF143013 DQ186846 AY631159 AY631160 AY631162 AY631163 AY631164 AY631165 AY524485 AY631167 AY524477 AY524478 AY857792 AY524461 AY631168 AY631169 EF562531 AY631170 AY907976 AF023789 AY524464 AY524486 AY524475 AY908009 AY524474 AY532391 AY524443 DQ467878 AJ2751783 AY631188 AY631189 DQ467876 AJ275177 AY631190 AF005524 AY524434 AY631191 AF232693 AF226820 DQ645999 AJ275174 — EF061306 NC005087 AY163147 EF107533 AB024664 AY631193 AY524437 AY524455 AF158175 DQ196095 AY631195 AY631196 AY631198 AY631199 AY631200 AY631201 AY524457 AY631202 AY524449 AY524450 AJ275171 AY524433 AY631203 AY631204 EF562528 AY631205 EF107532 AJ275176 AY524436 AY524458 AY524447 AF231087 AY524446 AY532394 AY524499 AF397812 AY078287 AF023749 AF548759 AF397786 AY857821 AY306780 AY636030 AY524490 AY857798 AF023766 AF023767 AF509542 AF023768 EF562520 AF161167 NC005087 AY078308 AF497138 AF152384 DQ108957 AY524493 AY524511 AF397838 DQ194226 AY857805 EF562521 EF562522 AY143062 EF562523 AY857807 AY524513 AF023754 AY524505 AY524506 AF023724 AY524489 AY857800 AF023752 EF562524 AY857801 AY747691 AF023737 AY524492 AY524514 AY524503 AF023748 AY524502 Bell et al.: Phylogeny of earliest diverging pleurocarps 539 Table 1. Continued. Taxon Tetraplodon mnioides (Hedw.) Bruch & Schimp. Thamnobryum alopecurum (Hedw.) Gangulee Trachyloma planifolium (Hedw.) Brid. *Ulota crispa (Hedw.) Brid. nad5 rps4 rbcL trnL AY908376 AJ291571 AY908445 AJ291568 AY499644 AF023834 AY908606 AF306972 AY312937 AY532392 AY631207 AY631208 AY501416 AF023769 EF562525 AY636026 O’Brien (2007) allowed us to combine data from the mitochondrial nad5 gene and the chloroplast trnL region, both highly informative for these lineages. The nad5 group I intron in particular provides much phylogenetic information for the hypnodendroid pleurocarps in the form of both substitutions and large-scale deletions and insertions (Bell & Newton 2005). We included 16 exemplars from the Hypnidae with a bias towards early diverging taxa, 17 Bryalean species, three Bartramiales, and two exemplars each from the Hedwigiales, Orthotrichales and Splachnales. Outgroups were Funaria hygrometrica, Dicranum scoparium and Physcomitrella patens. All genomic regions were obtained for all exemplars, other than Hypnodendron fuscomucronatum, for which nad5 was missing, Cyrtopodendron vieillardii, for which rbcL and trnL were not available, and three other taxa for which rbcL alone was not obtained (Table 1). DNA extraction, PCR amplification and sequencing. Isolation of DNA was carried out using commercially available DNA isolation kits (MachereyNagel or Qiagen). PCR amplifications were performed in 50 ll reactions containing 1.5 U Taq DNA polymerase, 1 mM dNTP mix (each 0.25 mM), 13 buffer, 1.5–2.5 mM MgCl2 and 12.5 pmol of each amplification primer. The mitochondrial nad5 region was amplified using the temperature profiles, primers and amplification strategy described in Bell and Newton (2004), whereas the plastid regions were amplified using the approach of Quandt et al. (2007). PCR products were purified using the Nucleospint PCR Purification Kit (Macherey-Nagel). For cycle sequencing we either used the DTCS QuickStart Reaction Kit (Beckman Coulter) or the ABI Prisme Big Dye Terminator Cycle Sequencing Ready Reaction Kit (Perkin Elmer), applying standard protocols for all reactions. Extension products were either run on a Beckman Coulter CEQ 8000 or an ABI PRISM 310 or 377 automated sequencer (Perkin Elmer), respective- ly. In some cases the cleaned PCR products were sequenced by Macrogen Inc., South Korea (www. macrogen.com). Sequences were edited manually with PhyDEt v0.992 (Müller et al. 2005). The alignments and trees are available from the authors upon request. Alignment and phylogenetic analyses. Based on the criteria laid out in Kelchner (2000) and Quandt and Stech (2005), DNA sequences were manually aligned using PhyDEt v0.992. Following the approach in Quandt et al. (2003) and Quandt and Stech (2004, 2005), the data matrix was screened for inversions using secondary structure models calculated with RNA structure (Mathews et al. 2004). Detected inversions were positionally separated in the alignment. As discussed in Quandt et al. (2003) and Quandt and Stech (2004), presence or absence of detected inversions was not coded for the phylogenetic analyses. Phylogenetic reconstructions using maximum parsimony were performed using PAUP 4.0b10 (Swofford 2002) in combination with PRAP (Müller 2004). The latter program generates command files for PAUP* that allow parsimony ratchet searches as designed by Nixon (1999). In the present study, ten random addition cycles of 200 ratchet iterations each were used. Each iteration comprised two rounds of TBR branch swapping, one on a randomly reweighted data set (25% of the positions), and the other on the original matrix saving one shortest tree. Since each random addition cycle rapidly converged to the same tree score, cycles were not extended to more than 200 iterations, nor were further cycles added. Shortest trees collected from the different tree islands were used to compute a strict consensus tree. Furthermore, the data set was analyzed employing a simple indel coding (SIC) strategy (Simmons & Ochoterena 2000) as implemented in SeqState (Müller 2005). SeqState generates a ready-to-use nexus file containing the sequence alignment with an automat- 540 the bryologist 110(3): 2007 ically generated indel matrix appended. This file can be analyzed either directly in PAUP or MrBayes. Alternatively, for ratchet and decay (ratchet) analyses it can be submitted to PRAP. Heuristic bootstrap searches were performed with 1000 replicates and 10 random addition cycles per bootstrap replicate on both data matrices (with and without indel coding). Decay values as further measurement of support for the individual clades were obtained employing a ratchet search (command file generated by PRAP) in PAUP with the same options in effect as in the ratchet. Heterogeneous Bayesian analysis was performed using MrBayes 3.1.2 (Huelsenbeck & Ronquist 2001; Ronquist & Huelsenbeck 2003). Partitioning data for the testing and application of different models in Bayesian analysis is a compromise between allowing each potentially discrete compartment to be modelled separately (thus increasing the fit of the combined model to the data) and over-parameterization, whereby individual compartments do not contain sufficient data to allow parameters to be estimated accurately. We adopted an approach in which data from each genomic region was modelled separately as a single compartment unless it contained sizeable subregions likely to be evolving under very different constraints, for example coding and non-coding DNA. However, as many of the nad5 sequences had missing data at the 3 0 end of the coding region, we did not separate this gene into coding and non-coding (intron) compartments. Additionally, we combined the non-coding rps4-trnS intergenic spacer, sequenced along with the rps4 gene, with the trnL region, also predominantly non-coding and containing a spacer region. We considered that this relatively short spacer, probably not containing sufficient data to be effectively modelled separately, was more likely to be accurately modelled in combination with other noncoding chloroplast DNA than with the protein-coding part of the rps4 region. Hence nucleotide partitions were as follows : nad5, rps4 coding, rbcL, and trnL þ rps4 non-coding. The hierarchical likelihood ratio test and the AIC criterion as implemented in MrModeltest 2.2 (Nylander 2004) were used to select the best fitting models for each compartment. In the analysis including indels, the restriction site (binary) model was used for the indel compartment (coding ¼ variable). Compartments were unlinked to allow parameters to vary independently. Final analyses were conducted through the online interface of the Computational Biology Service Unit (CBSU) of Cornell University (http://cbsuapps.tc. cornell.edu/mrbayes.aspx). For each analysis, three independent runs using the default prior settings, each with five chains (temp parameter ¼ 0.085), were run simultaneously for 2 3 106 generations with trees sampled every 100 generations. A majority rule consensus tree was constructed from the sampled trees after testing for stationary and discarding the trees from the burn-in phase. To test whether non-uniform clade priors could be artificially inflating posterior probabilities for nodes in the rhizogoniaceous grade (Picket & Randle 2005, see discussion), we further analyzed a resampled matrix (excluding indels) that included all taxa within the grade but excluded most exemplars from the Hypnidae, the hypnodendroid pleurocarps and the Bryales. The resampled matrix of 37 taxa is indicated by asterisks in Table 1. The analysis was conducted with the same settings as for the main analyses. Consensus topologies and support values from the different methodological approaches were compiled and drawn using TreeGraph (Müller & Müller 2004). RESULTS Alignment and model specification. As reported previously we observed a hairpin-associated inversion in front of the trnF gene that was present in 50% of the taxa (Quandt & Stech 2005; Quandt et al. 2004, 2007). Due to its homoplastic nature it was excluded from subsequent phylogenetic analyses. In addition an ambiguous poly-A stretch, located in the P8 stem loop structure of the trnL intron, was excluded from the analyses, as mononucleotide stretches are problematic in indel-coding approaches (Provan et al. 2001). After exclusion of terminal regions with missing data, the trnL-F inversion and the P8 poly-A stretch, the alignment resulted in a matrix of 5213 nucleotide characters and 357 indels. Total numbers of characters and numbers of parsimony informative characters respectively for each region were as follows: nad5: 2159/358, rps4: 768/246, rbcL: 1322/315, trnL: 964/ 157. Numbers of characters in the compartments used Bell et al.: Phylogeny of earliest diverging pleurocarps for the Bayesian analyses were as follows: nad5: 2159, rps4 coding: 605, rbcL: 1322, trnL þ rps4 non-coding: 1127. Both the likelihood ratio test and the AIC criterion as implemented within MrModeltest agreed in specifying the GTR þC þ I model as the best fit for the data for all the four nucleotide data compartments individually. Parsimony. Parsimony analysis of the matrix with indels excluded resulted in 16 most parsimonious trees (MPTs) of length 5093, consistency index (CI) ¼ 0.40, rescaled consistency index (RC) ¼ 0.28, retention index (RI) ¼ 0.68. With indels included the analysis produced 74 MPTs of length 5570, CI ¼ 0.43, RC ¼ 0.30, RI ¼ 0.69. The strict consensus of the trees from the analysis including indels is considerably less resolved due to the greater number of optimal topologies, although it is otherwise entirely consistent with the consensus of the 16 trees found in the analysis excluding indels. In particular, the nodes supporting the relationships of the major clades in the rhizogoniaceous grade are not found in all the MPTs from the analysis including indels. Figure 1 shows a strict consensus of the trees found in the non-indel analysis. Bootstrap support (BS) and Bremer support (BR) values from both analyses are included. Analysis of the indel matrix alone produced 39 MPTs (CI ¼ 0.57, RC ¼ 0.63, RI ¼ 0.80). The only clades recovered in the strict consensus (not shown) were species groups (mainly pairs) below the genus level and a small number at the generic or small family level. This is consistent with the indels providing signal at a low phylogenetic level that may help to resolve species relationships, but which is likely to be saturated (and hence a potential source of error within a parsimony framework) at higher phylogenetic levels. The differences between the results of the main analyses excluding and including indels (Fig. 1) can readily be explained in this context. As our focus here is on higher level relationships, we subsequently describe the parsimony results based on the analysis excluding indels. The pleurocarpous clade sensu lato, i.e., including the common ancestor of all the previously Bryalean lineages that contain pleurocarps, is resolved as monophyletic and well supported by a BS of 91% and BR of 5. The Bartramiales together with the 541 Hedwigiales appear as the sister group of the pleurocarp clade, although this hypothesis receives no bootstrap support (BS , 50%, BR ¼ 1). Among the Bryalean exemplars the Mniaceae and Phyllodrepanium falcifolium form a clade (BS ¼ 100%, BR ¼ 17) sister to Leptostomataceae þ Bryaceae (BS ¼ 93%, BR ¼8; for the sister relationship BS ¼ 76%, BR ¼3). Within the pleurocarp clade the Hypnidae and the hypnodendroid pleurocarps are reciprocally monophyletic to the exclusion of the rhizogonioid taxa, which occur in three clades and represent a grade. Mesochaete, Aulacomnium and Hymenodontopsis, together with Pyrrhobryum vallis-gratiae, P. bifarium and P. mnioides, form a poorly supported clade (BS ¼ 65%, BR ¼ 1) that is sister to a combined hypnodendroid pleurocarp þ Hypnidae clade, again with poor support (BS ¼ 54%, BR ¼ 1). The monophyly of the latter two groups combined is moderately supported (BS ¼ 72%, BR ¼ 4). Rhizogonium, Goniobryum, Calomnion, Cryptopodium, Pyrrhobryum sect. Pyrrhobryum and P. dozyanum (referred to as the ‘‘core rhizogonioids’’ by Bell & Newton 2005 and O’Brien 2007) form a wellsupported clade (BS ¼ 92%, BR ¼ 5) that is sister to this grouping, although without bootstrap support, while a Hymenodon, Leptotheca, Orthodontium and Orthodontopsis clade (BS ¼ 52%, BR ¼ 1) is sister to this larger clade in turn. Orthodontopsis is sister to Orthodontium lineare (BS ¼ 100%, BR ¼ 20), while Leptotheca boliviana is well supported (BS ¼ 95%, BR ¼ 6) as sister to Hymenodon. The relations of the (Hymenodon þ Leptotheca boliviana) clade, (Orthodontium þ Orthodontopsis) and L. gaudichaudii vary between MPTs. The topology in the core rhizogonioids places Pyrrhobryum dozyanum as sister to Pyrrhobryum sect. Pyrrhobryum, although without support, while Cryptopodium and Calomnion form a clade (BS ¼ 98%, BR ¼ 8) and all these taxa together are moderately supported as monophyletic (BS ¼ 90%, BR ¼ 4). A second well-supported clade (BS ¼ 98%, BR ¼ 7) consists of Rhizogonium and Goniobryum. In the third major rhizogonioid grouping Aulacomnium is moderately well supported (BS ¼ 85%, BR ¼ 8) as sister to Mesochaete, this clade being sister to the majority of Pyrrhobryum sect. Bifariella, out of which Hymenodontopsis is derived (forming a clade with Pyrrho- 542 the bryologist 110(3): 2007 Bell et al.: Phylogeny of earliest diverging pleurocarps bryum bifarium and P. mnioides, BS ¼ 85%, BR ¼ 8). Pyrrhobryum vallis-gratiae is the earliest diverging lineage in this latter clade, although there is no bootstrap support for its membership of the group. In the hypnodendroid pleurocarp clade there is strong support (BS ¼ 100%, BR ¼ 14) for the grouping of Cyrtopodendron with Pterobryella and for a clade comprising these taxa together with Hypnodendron sect. Sciadocladus (BS ¼ 100%, BR ¼ 15). Other relationships between the early diverging taxa lack convincing support, however. Braithwaitea is sister to the Racopilaceae (BS ¼ 67%, BR ¼ 2), and there is an unsupported clade grouping all these taxa to the exclusion of the other hypnodendroid pleurocarps. The latter occur in a very well-supported clade (BS ¼ 100%, BR ¼ 29) that includes all sections of Hypnodendron other than Sciadocladus in addition to Spiridens, Bescherellia and Cyrtopus. Hypnodendron arcuatum, the sole member of sect. Lindbergiodendron, and the exemplars from H. sect. Hypnodendron form a well-supported clade with Spiridens (BS ¼ 97%, BR ¼ 5). Hypnodendron arcuatum itself is fairly well supported (BS ¼ 84%, BR ¼ 2) as sister to the sole South American Hypnodendron species, H. microstictum, and thus appears derived from within H. sect. Hypnodendron. Nearly all the remaining sections of Hypnodendron occur in a single large and well-supported group (BS ¼ 95%, BR ¼ 5) that is sister to (H. sect. Hypnodendron þ H. arcuatum þ Spiridens) (BS ¼ 67%, BR ¼ 2). Sister to this combined clade is a grouping (BS ¼ 73%, BR ¼ 2) of the Cyrtopodaceae with Hypnodendron diversifolium (the sole member of H. sect. Tristichophyllum). Within the large clade first mentioned are two well-supported groups, one corresponding to H. sect. Comosa (BS ¼ 100%, BR ¼ 10) and the other including H. sects. Phoenicobryum, Mniodendropsis, Leiocarpos and Pseudomniodendron (BS ¼ 88%, BR ¼ 4). There is strong support for the grouping of Phoenicobryum (here recognized within the genus Dendro-hypnum) with Mniodendropsis (BS 543 ¼ 100%, BR ¼ 9) and Leiocarpos with Pseudomniodendron BS ¼ (100%, BR ¼ 12). Bayesian. Both of the Bayesian analyses (i.e., including and excluding indel coding) reached stationarity after approximately 30% of the 2 3 106 generations had been completed as determined by a visual comparison of the distribution of log likelihood values in the output files of the four runs. Runs were congruent after 2 3 106 generations, with the average standard deviation of split frequencies , 0.01. For the analysis excluding indel coding, the first 7000 of the 20000 sampled trees in each of the three runs were discarded, and for the analysis including indel-coding 8000 trees were discarded. The potential scale reduction factor (PSRF) calculated based on the retained trees deviated from 1.0 by less than 0.03 for all the parameters in both analyses. The topology of the most probable trees was identical for each, while the majority consensus trees differed only in resolution amongst some of the Bryalean taxa and in the Ptychomniales. Posterior probabilities (PP) for individual nodes within the rhizogoniaceous grade and in the hypnodendroid pleurocarps were identical for the majority of clades and differed by less than 10% for all. Henceforth we exclusively refer to the results of the analysis including indel-coding. The majority consensus of all sampled trees including clade probabilities in is shown in Fig. 2. The results show strong support (PP ¼ 100%) for the exemplars of the Orthotrichales as sister group to the pleurocarp clade (as opposed to the unsupported hypothesis of the (Bartramiales þ Hedwigiales) as sister to the pleurocarps under parsimony). Within the pleurocarp clade sensu lato the general topology is the same as that recovered under parsimony, although there are differences in the proposed relationships of a small number of individual exemplar taxa and small clades that lacked well-supported resolution under parsimony. In addition, posterior probability scores are relatively much higher (i.e., relative to other nodes in the same topology) for some major groupings that Figure 1. Strict consensus of 16 MPTs obtained from parsimony analysis of the matrix with indels excluded. Numbers above branches are bootstrap and Bremer support values respectively. The strict consensus of the 74 MPTs obtained from the analysis of the matrix including indels is wholly compatible with this topology but is less resolved; numbers below branches are the corresponding support values. New combinations made in this study are in bold. 544 the bryologist 110(3): 2007 Bell et al.: Phylogeny of earliest diverging pleurocarps had relatively weak bootstrap support under parsimony. The three rhizogonioid subclades as well as the larger clades represented by the ‘‘backbone’’ nodes in the rhizogoniaceous grade all have high posterior probabilities, specifically 100% for the grouping of the (Aulacomnium þ Mesochaete þ Hymenodontopsis þ Pyrrhobryum sect. Bifariella) clade with the combined (Hypnidae þ hypnodendroid pleurocarp) clade and also for the grouping of the (Hymenodon þ Leptotheca þ Orthodontium þ Orthodontopsis) clade with the other pleurocarps. The grouping of the core rhizogonioids with its sister group has a posterior probability of 97%. The three major clades in the rhizogoniaceous grade, from the earliest diverging to the latest diverging are supported by posterior probabilities of 100%, 100% and 94%, respectively. Leptotheca gaudichaudii is sister to the combined (Leptotheca boliviana þ Hymenodon) clade, although with a fairly low posterior probability (82%). Within the core rhizogonioids, the precise relationship of Pyrrhobryum dozyanum is highly uncertain, with its position in the most probable tree (not shown) differing from its position in the consensus tree (i.e., in the majority of trees sampled, Fig. 2). In the most probable tree it is sister to a monophyletic (Calomnion þ Cryptopodium þ Pyrrhobryum sect. Pyrrhobryum) clade, although this sub-topology is only present in 30% of trees sampled (PP ¼ 30%). In the majority of trees (56%) Pyrrhobryum dozyanum is sister to Pyrrhobryum sect. Pyrrhobryum, as in the parsimony topology, while in 15% it forms a clade with Calomnion and Cryptopodium. Within the latest diverging rhizogonioid clade the grouping of Hymenodontopsis with (Pyrrhobryum bifarium þ P. mnioides) and Aulacomnium with Mesochaete both have posterior probabilities of 100%. Pyrrhobryum vallis-gratiae is well supported (PP ¼ 99%) as a member of the (Pyrrhobryum sect. Bifariella þ Hymenodontopsis) clade. The hypnodendroid pleurocarps and the Hypnidae are both supported by posterior probabilities of 545 100%, as is the monophyly of the two clades combined. Hypothesized relationships amongst the early diverging lineages in the hypnodendroid pleurocarps are the same as under parsimony, although the grouping of Braithwaitea sulcata with the Racopilaceae is highly probable (PP ¼ 97%). A significant difference from the parsimony results within the larger clade is the position of the group comprising the exemplars from the Cyrtopodaceae as sister to the combined (Hypnodendron sect. Hypnodendron þ H. arcuatum þ Spiridens) clade (PP ¼ 94%) rather than to Hypnodendron diversifolium. The latter is unresolved in the consensus tree (Fig. 2) in relation to the two major clades in the large apical group. In 31% of topologies (PP ¼ 31%) it is sister to the (Hypnodendron sect. Hypnodendron þ H. arcuatum þ Spiridens þ Cyrtopodaceae) clade (this topology is also found in the single most probable tree, not shown), while in 29% (PP ¼ 29%) it is sister to the other large clade. In 36% (PP ¼ 36%) it is sister to these two groups combined. Other nodes within the hypnodendroid pleurocarp clade correspond to those found under parsimony and have comparable levels of relative support. The analysis of the resampled matrix excluding most taxa in the Bryales, hypnodendroid pleurocarps and the Hypnidae reached stationarity before 10% of the 2 3 106 generations had been completed. All statistics (see above) indicated convergence of the parallel runs and accurate sampling from the posterior probability distribution. The topology was entirely congruent with that of the main analyses for the nodes of interest. Clade posterior probabilities were 100% for all ‘‘backbone’’ nodes in the rhizogoniaceous grade, while the three rhizogonioid clades were supported (in order of divergence) by PP scores of 100%, 100% and 98% respectively. DISCUSSION Traditional classification schemes are based on overall morphological similarity or on intuition of synapomorphy, with a tendency for the former to Figure 2. Majority consensus of trees sampled after stationarity in the Bayesian analysis of the matrix including indels, with posterior probabilities for individual clades. A phylogram for the pleurocarp clade is shown on the left. New combinations made in this study are in bold, with the proposed higher-level taxonomy indicated on the right. Informal node-based names for major clades are indicated on the corresponding nodes. Where posterior probability values from the analysis excluding indels differ by more than 5%, these are indicated on the branches after the values from the analysis including indels. 546 the bryologist 110(3): 2007 prevail unless contrary evidence from the latter is strong, for example as judged by the perceived unlikelihood of a particular shared feature having arisen on multiple occasions. However, when rates of morphological change on a phylogeny are highly heterogeneous, as will be the case when stabilizing selection and directional selection occur on adjacent branches, overall morphological similarity may be a poor guide to evolutionary history. In ancient lineages with long branches, extinction may have amplified this problem by erasing morphologically intermediate terminal nodes. In both the rhizogoniaceous grade and the hypnodendroid pleurocarps there are several examples of highly distinctive taxa being derived from larger groups in which plesiomorphic suites of characters predominate. This, combined with apparent parallelisms and reversals in linked suites of characters on a fine phylogenetic scale, has largely confounded elucidation of relationships in the past. The current study closely corroborates the results of other recent molecular phylogenetic analyses of the early diverging pleurocarps while increasing confidence in critical nodes and expanding taxon sampling. Specifically, Bayesian analyses using complex heterogeneous models converge strongly on topologies that under parsimony have always been highly stable, but for which convincing support is difficult to obtain. Within the pleurocarp clade the relationships of the major groups are identical to those found by Bell and Newton (2004, 2005), and are also consistent with the results of O’Brien (2007) other than for the relative positions of the core rhizogonioids and the (Hymenodon þ Leptotheca þOrthodontium þ Orthodontopsis) clade. Congruence with previous studies extends also to lower-level relationships, although these are generally more informative due to increased taxon sampling. For example, sampling of all species from Pyrrhobryum sect. Bifariella (as in Bell & Newton 2004), combined with the inclusion of Hymenodontopsis stresemannii (as in O’Brien 2007), better clarifies the status of these taxa. While three of the four species placed by Manuel (1980) in Pyrrhobryum sect. Bifariella (P. mnioides, P. bifarium and P. vallisgratiae) form a clade together with Hymenodontopsis, the fourth member, P. dozyanum, is more closely related to P. sect. Pyrrhobryum, although its precise position remains ambiguous. Similarly, increased sampling from Hypnodendron allows confident circumscription of generic-level monophyletic groups within the Hypnodendraceae. The results provide a sufficient basis for a revised classification that reflects probable evolutionary relationships. Differences in topology and support under parsimony and Bayesian methods. Topological differences in the phylogenetic hypotheses proposed under parsimony and Bayesian methods are minor. There are few instances where a relationship proposed under one method is clearly contradicted by the other and none where conflicting topologies receive high support under both methods. The position of the Orthotrichales as sister to the pleurocarp clade is well supported in the Bayesian analyses, for example, but the alternative topology under parsimony is unsupported. As in Quandt et al. (2007), separate analyses of the mitochondrial and plastid sequence data revealed that the mitochondrial data support the (Orthotrichales þ pleurocarp) relationship regardless of the analysis method used. This topology is also found under Bayesian analysis of the chloroplast data, whereas under parsimony the (Bartramiales þ Hedwigiales) are favored as sister group to the pleurocarps s.l., but again without support. Another notable difference is the position of the Cyrtopodaceae (sister to Hypnodendron diversifolium under parsimony and to Spiridens þHypnodendron sect. Hypnodendron þ H. arcuatum under Bayesian). In addition, the precise position of Pyrrhobryum dozyanum varies between the optimal topologies found under each method, although in the Bayesian analysis its position in the majority of trees sampled corresponds with the parsimony topology. From a systematic perspective it is more significant that substantially different support measures are obtained for some critical nodes, particularly those within the rhizogoniaceous grade (Figs. 1, 2). While posterior probabilities for the majority of these clades (both the ‘‘backbone’’ nodes and the three principal groupings) are mostly 100%, parsimony bootstrap support for the same nodes (with the exception of the core rhizogonioids) is low or entirely lacking. Interpretation of these differences is problematic, as the impracticality of performing likelihood bootstraps on the matrix means that any differences due to the specification of the model are conflated Bell et al.: Phylogeny of earliest diverging pleurocarps with differences due to the functioning of the support measures themselves. There is a substantial literature devoted to the interpretation of posterior probabilities and their relationship to bootstrap and jackknife percentages (e.g., Cummings et al. 2003; Erixon et al. 2003; Huelsenbeck & Rannala 2004; Pickett & Randle 2005; Simmons et al. 2004). While there is a broad consensus that Bayesian methods tend to overestimate support (Cummings et al. 2003; Erixon et al. 2003; Simmons et al. 2004), the circumstances in which this may occur are complex (also, other methods may underestimate support, e.g., Simmons et al. 2004). In particular, posterior probabilities are sensitive to under-parameterization of the model, while correspondingly there is evidence that if a correct (i.e., sufficiently complex) model is specified these problems can be avoided (Erixon et al. 2003; Huelsenbeck & Rannala 2004). Huelsenbeck and Rannala (2004) advocated the use of the most complex models available and maintained that the posterior probability of a tree is the probability that the tree is correct, assuming that the model is correct. In our analyses we were careful to specify a heterogeneous model in which compartmentalization allowed for potential differences in mode of evolution between genomes, gene regions and coding vs. non-coding regions as much as possible while avoiding overly small compartments or compartments with extensive missing data for which parameters would be difficult to estimate. It is significant that for every compartment both the likelihood ratio test and the AIC criterion as implemented in MrModeltest agreed in specifying the GTR þC þI model (the most complex tested for) as the best fit to the data. Pickett and Randle (2005) illustrated a potential problem inherent in the standard practice of specifying uniform topological priors in Bayesian analysis. Specifying all topologies as equally probable in terms of Bayesian priors does not equate to specifying that the probabilities of all clades are equal. The authors demonstrated that this may influence posterior probabilities, increasing relative support for the smallest and largest clades in an analysis. This bias becomes more pronounced as the number of taxa increases. For the nodes in the rhizogoniaceous grade with high posterior probabilities and low or absent 547 support under parsimony, we considered whether this phenomenon could be a factor. In the case of the three ‘‘backbone’’ nodes this seems less likely, as at least two of them represent clades of approximately medium size relative to the total number of taxa in the analysis (72 out of 101, 66/101 and 51/101). However, the critical (Orthodontium þ Orthodontopsis þ Hymenodon þ Leptotheca) and (Aulacomnium þ Mesochaete þ Pyrrhobryum sect. Bifariella þ P. bifarium) clades are both relatively small in relation to the size of the matrix. In our analysis of the resampled matrix in which we excluded most exemplars from the Hypnidae, the hypnodendroid pleurocarps and the Bryales, these clades have a size that is much larger in relation to the total number of taxa in the matrix, and any positive influence of non-uniform clade priors on posterior probabilities should be much less pronounced. The results show that this phenomenon is not likely to be a significant factor for the support measures obtained in the main analysis. Another approach to investigating the differences between Bayesian posterior probabilities and parsimony bootstrap values for individual clades is to examine what the bootstrap replicates themselves can tell us about ‘‘hidden’’ signal in the dataset relating specifically to the taxa of interest. Examination of the bipartition table produced by PAUP* allows identification of non-optimal sub-topologies recovered in some bootstrap replicates that conflict with the monophyly of the group in question and thus partially contribute towards reducing the bootstrap percentage for that node. We applied this technique to the (Orthodontium þ Orthodontopsis þ Hymenodon þ Leptotheca) clade, supported by a Bayesian posterior probability of 100% but a parsimony bootstrap of only 52% in the analysis excluding indels. Table 2 lists all bipartitions occurring in 5% or more of bootstrap replicates that specifically contradict the monophyly of the six exemplars in this clade. It can be seen that the strongest signal in the matrix that is incompatible with the monophyly of this group is a tendency for Hymenodon, together with one or both Leptotheca species, to group with other clades (either the core rhizogonioids or the large group that includes all the other pleurocarps) to the exclusion of Orthodontium and Orthodontopsis. In particular, in 14.9% of bootstrap replicates all Hymenodon and Leptotheca 548 the bryologist 110(3): 2007 Table 2. Bipartitions found in 5% or more of parsimony bootstrap replicates that are incompatible with the monophyly of (Hymenodon pilifer, H. sphaerothecius, Leptotheca boliviana, L. gaudichaudii, Orthodontium lineare, Orthodontopsis bardunovii). ‘‘Other pleurocarps’’ ¼ core rhizogonioids þ (Aulacomnium, Mesochaete, Hymenodontopsis, Pyrrhobryum bifarium, P. medium, P. vallis-gratiae) þ hypnodendroid pleurocarps þ Hypnidae. Freq. % Bipartition (smallest partition) Hymenodon pilifer, H. sphaerothecius, Hymenodon pilifer, H. sphaerothecius, Hymenodon pilifer, H. sphaerothecius, Hymenodon pilifer, H. sphaerothecius, Hymenodon pilifer, H. sphaerothecius, þ other pleurocarps. Leptotheca Leptotheca Leptotheca Leptotheca Leptotheca boliviana, Leptotheca gaudichaudii þ core rhizogonioids. boliviana, Leptotheca gaudichaudii þ other pleurocarps. boliviana þ core rhizogonioids. boliviana þ other pleurocarps. boliviana, Orthodontium lineare, Orthodontopsis bardunovii species group with the core rhizogonioids to the exclusion of Orthodontium and Orthodontopsis. This is consistent with the results of constraining the nonmonophyly of the clade under investigation, which results in a single MPT in which Hymenodon and Leptotheca form a sister group to the core rhizogonioids. There does not appear to be any significant hidden signal that groups Orthodontium or Orthodontopsis with other taxa. It is plausible that saturation of individual sites in both the stem lineage of the core rhizogonioids and in a (Hymenodon þ Leptotheca) clade (see Felsenstein 1978) creates hidden artifactual signal that reduces bootstrap support while being insufficiently powerful to overwhelm phylogenetic signal and distort the topology itself. If so, then the considerably better support for this clade under Bayesian methods probably reflects the ability of the model to account for this factor rather than any differences between the bootstrap and posterior probability metrics themselves. Taxonomy: conceptual framework. We are broadly convinced of the merits of a monophyletic Linnaean nomenclature (Stevens 2006) for supraspecific taxa, as opposed to non-hierarchic formal naming of monophyletic groups (e.g., the PhyloCode, PC, Cantino and de Queiroz 2004) or the recognition of paraphyletic taxa within the Linnaean framework (e.g., Hörandl 2006; Mayr & Bock 2002). Although enforcing strict monophyly of species may be impractical due to the prolonged nature of the speciation process itself and associated problems of lineage sorting and incomplete coalescence, higherlevel taxa should correspond to lineages that have been isolated for longer than the timescales on which these processes operate and should be monophyletic if 14.9 10.6 10.3 8.7 6.6 they are to represent historical entities. We also believe that a flagged hierarchy as opposed to a rank hierarchy (sensu Stevens 2002, 2006) is the only feasible way to implement such a system. In a flagged hierarchy, ranks convey information about the position of a group relative to larger groups of which it is a part or smaller groups that are included within it; there is no necessary implication of biological or historical equivalence between taxa of the same rank. The current classification of mosses (Goffinet & Buck 2004) appears to be a Linnaean flagged hierarchy, although it recognizes a number of paraphyletic taxa within class Bryopsida, the arthrodontous mosses. Subclass Hypnidae is derived from a paraphyletic subclass Bryidae, and superorder Rhizogonianae from a paraphyletic superorder Bryanae. The Rhizogonianae and the Rhizogoniales are also paraphyletic. Difficulties arise from two sources. Firstly, hypotheses of relationships between the major clades of diplolepidous alternate mosses sensu lato Goffinet and Buck (2004) have changed rapidly in the last few years, and as some important nodes still lack resolution or support, further changes are likely to occur in the future. Secondly, even with the use of two intercalated levels in the Linnaean hierarchy (subclass and superorder), there are not enough ranks to provide names for all the evolutionarily significant entities within Bryopsida if monophyly is to be maintained. This leads to a partially arbitrary (and hence unstable) naming of some groups at the expense of others, the adoption of more inclusive but paraphyletic taxa, or both. Considering only the pleurocarp clade sensu lato, there are three significant levels of monophyletic hierarchy above the order: 1) the Hypnidae, i.e., the Ptychomniales, Hookeriales Bell et al.: Phylogeny of earliest diverging pleurocarps and Hypnales; 2) the combined hypnodendroid pleurocarp þ Hypnidae clade; and 3) the pleurocarp clade itself. Currently only the first of these is a taxon, and to recognize all three with formal Linnaean ranks would require either lumping all the Hypnidae into a single order Hypnales or else the recognition of the pleurocarp clade sensu lato as a class in its own right (or the adoption of yet another intercalated rank). The higher-level taxonomy of mosses has been subject to several revisions during the last few years and any comprehensive general system we propose is unlikely to be stable. Hence we suggest that, at least while hypotheses of phylogenetic relationships are still in flux and debates about the future of nomenclature continue, informal names rather than intercalated Linnaean ranks are the best way to refer to evolutionarily significant clades between the ordinal and class levels within the Bryophyta. These have the advantage of greater stability of usage, as when applied to well-circumscribed groups they are not subject to change deriving simply from reassignment of rank. This is similar to the approach taken by the Angiosperm Phylogeny Group (APG 1998, 2003) whose supra-ordinal informal names (e.g., eudicots, asterids) must be some of the more successful communication tools in the recent history of systematics. Kuntner and Agnarsson (2006), in their recently proposed model for a ‘‘combination approach’’ to formal nomenclature, suggested the maintenance of classical supergeneric ranks, although based on phylogenetic definitions rather than type taxa, and the abandonment of intermediate ranks. In the current study, borrowing from both of these approaches while conforming to the ICBN, we will recognize monophyletic taxa at the class, order, family and generic levels, but avoid intercalated higher ranks. In order to reference other evolutionarily significant clades between order and class, we will propose informal names using node-based phylogenetic definitions (see PC article 9.4.1). We stress that we are strongly opposed to the adoption of the PhyloCode as a replacement for the ICBN. Informal node-based names, however, may be able to act as nomenclatural ‘‘buffer-zones’’ between stable taxa at higher and lower levels. Provided that care is taken to ensure that groups chosen as taxa are monophyletic as far as can be determined, then all clades may be 549 referred to as taxa or as small sets of taxa (i.e., short lists) when precision is required, while in other contexts any clade that is not a taxon may be named informally, but with a reduced risk of ambiguity. This seems to us a functional compromise between the differing requirements that phylogenetic systematists and other end users (e.g., ecologists or conservationists) have of nomenclature. Taxonomy: implementation. The hypnodendroid pleurocarps require recognition at the ordinal level on the grounds of both distinctiveness and their phylogenetic position as sister to the Hypnidae. In addition, the results of this analysis corroborate those of Bell and Newton (2005) in strongly supporting a clade comprising all exemplars of Hypnodendron, with the exception of H. sect. Sciadocladus, together with the Spiridentaceae and the Cyrtopodaceae. The latter taxa are robust, sparsely branched epiphytes that are probably derived from more or less dendroid ancestors (Bell & Newton 2005). Hypnodendron sect. Sciadocladus, treated as a genus in its own right prior to Touw’s (1971) revision, is closely related to Pterobryella, with which it shares a plesiomorphic exothecial structure absent from the larger clade. In both these genera a single outer layer of the exothecium is clearly differentiated, being composed of large, thick-walled cells, while the other members of Hypnodendron sensu Touw (1971), as well as the Spiridentaceae and Cyrtopodaceae, have several more or less equally thickened cell layers (Bell & Newton 2005). This character was first described for Sciadocladus by Norris and Koponen (1996). Most species of Pterobryella, a small genus of very robust stipitate plants, are pinnate epiphytes with narrowly triangular leaves. The affinities to Sciadocladus are most apparent when comparing the Lord Howe Island endemic Pterobryella praenitens to Hypnodendron menziesii. In their usual forms both of these plants have a multiple reiterating tiered structure with new stipes originating from the frond section of the previous primary module. Although this structure is also found in Hypnodendron sect. Comosa, both Pterobryella and Sciadocladus additionally have elongate, naked stipes with very fragile leaves, which in conjunction with the reiterating structure confers a distinctive shared architecture. Pterobryella praenitens has palmate fronds, and although Touw (1971) 550 the bryologist 110(3): 2007 described Hypnodendron menziesii as having ‘‘umbellate, occasionally palmate or subumbellate’’ fronds, palmate specimens seem not to be uncommon. The monospecific Cyrtopodendron is closely related to Pterobryella. The significant support under Bayesian methods for the grouping of Braithwaitea and the Racopilaceae is intriguing, although it is clear that both of these taxa are highly distinct entities, both in relationship to each other and to the other Hypnodendrales. This is amply supported by morphological characters as well as by the molecular data. The Racopilaceae are prostate, creeping plants quite unlike the other members of the order and Braithwaitea, while sharing the robust, stipitate form of many of the Hypnodendrales, is distinguished by a reduced peristome, a distinct stipe anatomy and pinnate branching that is most developed distally rather than proximally, amongst other features (Touw 1971). We propose that Braithwaitea be newly recognized at the family level and that Sciadocladus and Cyrtopodendron be included within an expanded Pterobryellaceae. The remaining Hypnodendron species, together with Spiridens, Franciella, Cyrtopus and Bescherellia, are recognized within the Hypnodendraceae. Franciella is clearly very close to Spiridens and Withey (1996b) proposed synonymy of the two genera. Her work suggests that Franciella may be sister to Spiridens, and as its diagnostic elongate seta may be a plesiomorphic character indicative of its shared ancestry with dendroid taxa; we retain it at the generic rank. There is clearly more than one entity at the generic level within the emended Hypnodendraceae in addition to the taxa currently classified in the Cyrtopodaceae and Spiridentaceae. A number of morphological characters agree with the molecular phylogeny in supporting the distinctness of Hypnodendron diversifolium, including its unique tristichous arrangement of dimorphous branch leaves, the morphology of the stipe leaves and its highly conspicuous foliose pseudoparaphyllia. The tristichous branch morphology, best observed in hydrated specimens and usually very distinctive, is occasionally obscure according to Norris and Koponen (1996), who preferred the sharply defined, strongly reflexed and downward-pointing acumen of the stipe leaves as a diagnostic character. The combination of this feature with the large, foliose pseudoparaphyllia is immediately recognizable. Sections Hypnodendron and Lindbergiodendron, in agreement with the molecular phylogeny, are separated from the clade containing most of the sections of Hypnodendron by their complanate branches with anisomorphic leaves and opercula that are bluntly rather than sharply rostrate (Touw 1971). Complanate branches, sometimes with asymmetric leaves, also occur in sects. Leiocarpos and Phoenicobryum, but adjacent leaves are usually only very slightly anisomorphic, whereas in sects. Hypnodendron and Lindbergiodendron at least some branches usually have lateral and dorsal leaves that differ noticeably in size and shape, while lacking the tristichous arrangement of H. diversifolium. Other strong tendencies also distinguish these two groups. Members of sect. Hypnodendron nearly all have closely appressed stipe leaves (although in sect. Lindbergiodendron, i.e., H. arcuatum, they are widely spreading), while species in the larger clade nearly always have strongly spreading to squarrose-recurved leaves (the exceptions being Hypnodendron auricomum, H. opacum and H. fuscomucronatum). The grouping of Hypnodendron milnei with H. sect. Phoenicobryum is consistent with several features shared between these taxa, H. milnei differing principally in its palmate rather than pinnate fronds while retaining elements of leaf morphology and general aspect that are intermediate between Hypnodendron sects. Comosa and Phoenicobryum. Hypnodendron auricomum (together with the closely related H. opacum) and H. fuscomucronatum are relatively isolated taxa with morphological affinities to several other groups, although from our results are clearly most closely related to the (Hypnodendron sect. Phoenicobryum þ H. milnei) clade. Hypnodendron sect. Comosa (previously Mniodendron) is immediately recognizable by the combination of strongly reflexed stipe leaves with dense tomentum on the stipe, and generally also by the form of both the stipe and the branch leaves. We restrict Hypnodendron to the species classified by Touw (1971) in sects. Hypnodendron and Lindbergiodendron and recognize sects. Phoenicobryum, Mniodendropsis, Pseudomniodendron and Leiocarpos in Bell et al.: Phylogeny of earliest diverging pleurocarps the genus Dendro-hypnum Hampe (1872). The genus Mniodendron (Hypnodendron sect. Comosa) is reı̈nstated. Dendro-hypnum is an exclusively Malesian and western Pacific genus, while Mniodendron is an Australasian genus that includes one widespread (and apparently relatively recently derived) Malesian species, Mniodendron dendroides. Touw (1971) commented on the attractiveness of Hypnodendron diversifolium in his revision, and in recognition of his outstanding contribution to the taxonomy of the Hypnodendraceae we name this highly distinctive taxon in his honor at the generic level. It is a tribute to the quality and accuracy of Touw’s revision that molecular analyses support the monophyly of all his sections with the exception of sect. Hypnodendron. In addition, we suspect that all his species are natural groups. Touw’s work remains the definitive reference source for these entities, and the nomenclatural changes necessitated by the paraphyly of Hypnodendron sensu lato should in no way detract from this. Within the rhizogoniaceous grade, the pectinate nature of relationships between the major clades, as well as the wider taxonomic context, pose challenges for monophyletic nomenclature at the ordinal and family levels. The sister group relationship of Hypnodendrales to the Hypnidae precludes recognition of the rhizogonioid and hypnodendroid taxa in the same order and also the recognition of the rhizogonioids as a single monophyletic taxon. The only possible solution within the Linnaean framework is to recognize each of the monophyletic entities within the rhizogoniaceous grade at the ordinal level. In fact, this is not greatly at odds with the magnitude of morphological (particularly large-scale structural) variation found between these taxa (e.g., Bell & Newton 2007). Unfortunately, considerable morphological variation at the generic level, combined with frequent reversals and convergences, means that monophyletic orders and families are not easily defined by traditional morphological characters. However, three clades are consistently recovered and increasingly well supported in all taxon-dense phylogenetic analyses of molecular data (e.g., Bell & Newton 2004, 2005; O’Brien 2007; as well as by this study), and we recognize these here as Rhizogoniales, Aulacomniales and Orthodontiales. The Orthodontiales contain Orthodontium, Or- 551 thodontopsis, Hymenodon and Leptotheca in the single family Orthodontiaceae and can be supported morphologically by reduction of the peristome, more pronounced in some taxa than in others and taking different (but possibly functionally equivalent) directions in the more derived taxa. Although peristome reduction is not regarded as a reliable taxonomic character in the Hypnidae due to high levels of homoplasy (e.g., Buck 1991; Buck & Vitt 1986; Hedenäs 1998, 2001; Huttunen et al. 2004), it is rather rare in the rhizogonioid and hypnodendroid taxa, among which even many strongly epiphytic taxa have fully-developed ‘‘bryalean’’ peristomes (e.g., Hypnodendron sect. Phoenicobryum, Spiridens, Cryptopodium). The double peristomes of specimens of Leptotheca gaudichaudii and Orthodontium lineare we have examined are both characterized by moderate reduction of both endostome and exostome. Endostome processes are highly narrowed to filamentous throughout nearly all their length, abruptly widening to a very low basal membrane (Fig. 3). The bases of the endostome segments are prominently exposed due to reduction of the exostome, the teeth being relatively narrow and widely separated with trabeculae weakly developed or absent. Reduction is less pronounced in Leptotheca gaudichaudii than in Orthodontium lineare and the outer exostome is more ornamented in the former, but the pattern is otherwise identical. This type of peristome is generally characteristic for Orthodontium, although there is a trend towards very delicate endostomes with higher basal membranes and rudimentary segments in Orthodontium pallens, Orthodontium inflatum and Orthodontium infractum (Meijer 1952). Sporophytes are unknown in Leptotheca boliviana. In Hymenodon, the exostome is reduced further and the endostome is lost entirely (Shaw & Anderson 1986), while in Orthodontopsis it is the exostome that is lost and the endostome that is further reduced (Ignatov & Tan 1992). These conditions likely represent alternative, although possibly functionally convergent, trends towards further reduction from a plesiomorphic state retained in Orthodontium and Leptotheca. Further evidence for a close relationship between Leptotheca and Hymenodon is provided by the uniseriate, filamentous axillary propagules characteristic of Leptotheca that were found by Karttunen and 552 the bryologist 110(3): 2007 Figure 4. Proximal cell of axillary propagule of Leptotheca gaudichaudii (Child 433, BM); close-up showing dehiscence structure. Figure 3. Scanning electron micrographs of the basal part of the peristome in Orthodontialean taxa. A. Leptotheca gaudichaudii (Child 628, BM) B. Orthodontium lineare (Child s.n., BM). Bäck (1988) also to occur in specimens of Hymenodon aeruginosus from the West Indies. On this basis they segregated the Cuban, Jamaican and Dominican populations as H. reggaeus. Scanning electron microscopic examination of the propagules of Leptotheca gaudichaudii and Hymenodon reggaeus (Bell unpublished) shows them to have a near-identical structure unlike that of uniseriate propagules found in other moss species (Ulota phylantha, Pilotrichum bipinnatum, Orthotrichum lyellii). They appear to have adaptations towards active dispersal in dry conditions, including specialized dehiscence cells (Fig. 4) and an ability to rapidly twist and contract in response to dehydration (Bell unpublished). Our results strongly suggest that Leptotheca is paraphyletic and that Leptotheca boliviana is sister to Hymenodon. As the relations of Leptotheca gaudichaudii remain uncertain, however (its position in the Bayesian analysis is relatively poorly supported) we prefer to retain the current generic taxonomy until further data are available. Orthodontialian taxa are relatively slender, strongly nerved and commonly found as epiphytes or on rotting wood. Leaf form, areolation and manner of gametangial production vary dramatically between genera. Hymenodon is unique in the group in being pleurocarpous, and may have developed this trait independently. Bell and Newton (2007) discussed this possibility together with the alternative hypothesis of a single origin of the trait, although in the light of the current study the latter seems less likely. Orthodontopsis has the unusual feature of having archegonia scattered along the stem as well as in terminal perigonia (Ignatov & Tan 1992), a highly interesting trait that requires further study in the context of the evolution of pleurocarpy. The majority of the species traditionally classified in the Rhizogoniaceae are here treated in an emended order Rhizogoniales, containing the single family Rhizogoniaceae. Most of these taxa are tufted pleurocarps in which nearly all branching is from the extreme base of primary modules (Bell & Newton 2007). Perichaetial modules are also produced basally, and frequently have subperichaetial innovations giving rise to further fertile modules. However, the two major clades of species that have this morphology (Rhizogonium and Pyrrhobryum sect. Pyrrhobryum) are each closely related to a single species with an architecture closer to the more derived pleurocarps (Fig. 2). Pyrrhobryum dozyanum produces perichaetial modules non-basally and has been classified with Bell et al.: Phylogeny of earliest diverging pleurocarps P. bifarium, P. mnioides and vallis-gratiae in P. sect. Bifariella, while Goniobryum subbasilare, although having basal vegetative branching and subperichaetial innovations, has perichaetial modules that are not fully basal. Additionally, the clade comprising Cryptopodium and Calomnion represents a reversal to acrocarpy. Bell and Newton (2007) hypothesized that basal branching of pleurocarpous perichaetial modules is plesiomorphic, with subperichaetial innovations arising as a response to this in Rhizogonium and Pyrrhobryum sect. Pyrrhobryum. The third order we recognize from the rhizogoniaceous grade is the Aulacomniales, again comprising a single family, the Aulacomniaceae. The wellsupported sister group relationship of the mostly Northern Hemisphere acrocarpous Aulacomnium to the Australian endemic Mesochaete (see also Bell & Newton 2004; O’Brien 2007) is initially surprising, until Mesochaete and the most basal member of Aulacomnium, A. heterostichum, are compared (O’Brien 2007). Morphological traits shared by these taxa include sulcate capsules, deciduous apical leaves, undulate, oblong-ovate and asymmetrical leaves with coarsely-toothed margins, and smooth leaf cells. Mesochaete and Aulacomnium heterostichum also have comparable habitat preferences, typically occurring on shaded, mesic mineral soils. The remaining species of Aulacomnium have diverged from Mesochaete and A. heterostichum both morphologically and in terms of habitat preference. The other members of the order are Pyrrhobryum vallis-gratiae, P. bifarium, P. mnioides and Hymenodontopsis stresemannii. The Pyrrhobryum species are three of the four taxa placed by Manuel (1980) in Pyrrhobryum sect. Bifariella. They have distally located perichaetial modules that do not appear to have subperichaetial innovations, produce vegetative branches both basally and distally, and usually have large portions of their stems clothed with tomentum. Otherwise, leaf morphology is similar to the basally branching members of Pyrrhobryum sect. Pyrrhobryum. Hymenodontopsis stresemannii is a Malesian endemic that appears to be derived from within this group of Pyrrhobryum species (Fig. 2). It is a rather variable plant and robust forms closely resemble Pyrrhobryum, while it has a very similar leaf morphology (albeit in a more delicate form), having a more or less bistratose border with 553 geminate teeth and a dorsally toothed costa. Additionally its architecture, although superficially resembling Hymenodon, is in fact structurally much closer to the distally branching Pyrrhobryum species. If the dense felt of rhizoids covering the lower stems of Hymenodontopsis stresemannii is dissected, it is apparent that vegetative branching is initiated from positions fairly high up on primary modules and that perichaetial modules, while in fairly basal positions, are not truly basal (see diagram and discussion in Bell & Newton 2007). Although Hymenodontopsis stresemannii is a very distinct plant defined by a reduced single peristome (an endostome; Shaw & Anderson 1986), it is ultimately simply a highly derived member of the same clade as Pyrrhobryum bifarium, P. mnioides and P. vallis-gratiae, and retains the underlying morphology of these taxa. Thus we expand the concept of Hymenodontopsis to encompass these three Pyrrhobryum species. As described above, we adopt informal nodebased names to refer to evolutionarily significant monophyletic groupings of orders in the pleurocarpous mosses and do not recognize intercalated Linnaean ranks. These replace other informal terms we have used in the current and previous studies to refer to the same or similar groups, e.g., ‘‘pleurocarps sensu lato,’’ ‘‘apical pleurocarps,’’ etc. Although the manner in which these names are defined equates to one of the options allowed under the provisional PhyloCode, we do not subscribe to the ambition of the PhyloCode to replace the basic Linnaean rank hierarchy and believe that such names should informally supplement the latter. The Homocostate Pleurocarps equate to the apical pleurocarpous clade that includes the Ptychomniales, Hookeriales and Hypnales. Although this group has often been viewed as the ‘‘true’’ pleurocarpous group or even ‘‘the pleurocarps,’’ it represents a relatively recent, albeit massively speciose, radiation of pleurocarpous taxa (Newton et al. 2007; Shaw et al. 2003). The homogeneous costa (Hedenäs 1994) is the only reliable feature distinguishing these plants from the austral Hypnodendrales, and we propose the name Core Pleurocarps to refer to the clade that includes both of these reciprocally monophyletic groups. Finally, the Pleurocarpids are the pleurocarp clade sensu lato, i.e., the clade that includes the Core 554 the bryologist 110(3): 2007 Pleurocarps in addition to the three rhizogonioid orders. Future research. The phylogenetic hypothesis presented here should be tested on an ongoing basis by the addition of further data and the employment of new analytical techniques. In the Orthodontiales, further sampling from Orthodontium and Hymenodon may help to firmly resolve the position of Leptotheca gaudichaudii and the status of Leptotheca, as may inclusion of more exemplars of the disjunct but apparently genetically uniform L. gaudichaudii itself (McDaniel & Shaw 2003). The precise position of Pyrrhobryum dozyanum remains ambiguous. Given its morphological affinities to the distally branching pleurocarps in the Aulacomniales, this issue is critical to understanding the early development of pleurocarpy, particularly whether the basally branching condition is plesiomorphic or locally apomorphic (see Bell & Newton 2007). Again, inclusion of more exemplars may provide a solution. Within the Hypnodendraceae the remaining species of Dendro-hypnum and Hypnodendron in particular should be sampled, not least in order to test the monophyly of the genera. We would, however, be surprised if these entities are not natural. The dramatic derivation of robust, simply branched taxa (Spiridens, Franciella, Bescherellia and Cyrtopus) from within a predominantly dendroid family provides an ideal subject for a detailed study of the adaptive significance of these growth forms in the context of austral forest ecosystems, particularly as a strikingly parallel (and plausibly contemporaneous) development appears to have taken place in the family Hypopterygiaceae in the Homocostate Pleurocarps. Finally, major questions concerning the geographical, temporal and adaptive origins of the first pleurocarps remain largely mysterious, although in the light of a well-resolved phylogeny perhaps more intriguingly than abominably so. Interdisciplinary studies involving ecology, evolutionary development, historical biogeography and paleobotany in conjunction with phylogeny offer fertile avenues for research. CLASSIFICATION AND NOMENCLATURE Scheme. We here present the supra-generic taxonomy and definition of informal node-based names. PLEUROCARPIDS (informal node-based name) The least inclusive clade containing Hylocomium splendens and Hymenodon pilifer. Etymology: ‘‘inclined to pleurocarpy.’’ Order ORTHODONTIALES (Broth.) N. E. Bell, A. E. Newton & D. Quandt, comb. et stat. nov.; Bryaceae subfam. Orthodontioideae Broth., Nat. Pflanzenfam., ed. 2, 10: 347. 1924. ORTHODONTIACEAE (Broth.) Goffinet TYPE: Orthodontium Wilson Leptotheca Schwägr., Hymenodon Hook.f. & Wilson, Orthodontium Wilson, Orthodontopsis Ignatov & B. C. Tan ORDER RHIZOGONIALES (M. Fleisch) Goffinet & W. R. Buck RHIZOGONIACEAE Broth. TYPE: Rhizogonium Brid. Rhizogonium Brid., Goniobryum Lindb., Pyrrhobryum Mitt., Calomnion Hook.f. & Wilson, Cryptopodium Brid. ORDER AULACOMNIALES (Schimp.) N. E. Bell, A. E. Newton & D. Quandt, comb. et stat. nov.; Aulacomniaceae Schimp., Synop. Musc. Europ. 411. 1860. AULACOMNIACEAE Schimp. TYPE: Aulacomnium Schwägr. Aulacomnium Schwägr., Mesochaete Lindb., Hymenodontopsis Herzog CORE PLEUROCARPS (informal node-based name) The least inclusive clade containing Hylocomium splendens and Spiridens reinwardtii. ORDER HYPNODENDRALES (Broth.) N. E. Bell, A. E. Newton & D. Quandt, comb. et stat. nov.; Hypnodendraceae Broth., Nat. Pflanzenfam. I(3): 1166. 1909. BRAITHWAITEACEAE N. E. Bell, A. E. Newton & D. Quandt, fam. nov. Hypnodendraceis et Pterobryellaceis similis sed foliis ramulinis ad apicem obtusis cymbiformibus, costa forti in mucronem parvum excurrenti, peristomio reducto, trabeculis exostomii vestigialibus, endostomio segmentis linearibus et membrana basali humilissima instructo, differt. TYPUS: Braithwaitea Lindb. Braithwaitea Lindb. Bell et al.: Phylogeny of earliest diverging pleurocarps RACOPILACEAE Kindb. TYPE: Racopilum P. Beauv. Racopilum P. Beauv., Powellia Mitt. PTEROBRYELLACEAE (Broth.) W. R. Buck & Vitt TYPE: Pterobryella (Müll. Hal.) A. Jaeger Pterobryella (Müll. Hal.) A. Jaeger, Sciadocladus Lindb. ex Kindb., Cyrtopodendron M. Fleisch. HYPNODENDRACEAE Broth. TYPE: Hypnodendron (Müll. Hal.) Lindb. ex Mitt. Hypnodendron (Müll. Hal.) Lindb. ex Mitt., Spiridens Nees, Franciella Thér., Cyrtopus (Brid.) Hook.f., Bescherellia Duby, Touwiodendron N. E. Bell, A. E. Newton & D. Quandt, Dendro-hypnum Hampe, Mniodendron Lindb. ex Dozy & Molk. HOMOCOSTATE PLEUROCARPS (informal nodebased name) The least inclusive clade containing Hylocomium splendens and Ptychomnion aciculare. Etymology: referring to the synapomorphy of the costa being homogeneous in transverse section (Hedenäs 1994). PTYCHOMNIALES W. R. Buck, C. Cox, A. J. Shaw & Goffinet HOOKERIALES M. Fleisch. HYPNALES (M. Fleisch) W. R. Buck & Vitt Nomenclatural changes at genus and species levels. Included here are new and proposed combinations and descriptions of new taxa. All species delimitations are unchanged from Touw (1971), other than for the new combination Dendro-hypnum opacum, which follows the recognition of Hypnodendron opacum by Akiyama (1989). Hymenodontopsis bifaria (Hook.) N. E. Bell, A. E. Newton & D. Quandt, comb. nov.; Hypnum bifarium Hook., Musci Exot. 1: 57 1818; Pyrrhobryum bifarium (Hook.) Manuel, Cryptog. Bryol. Lichénol. 1: 70. 1980. Hymenodontopsis mnioides (Hook.) N. E. Bell, A. E. Newton & D. Quandt, comb. nov.; Hypnum mnioides Hook., Musci Exot. 1: 77 1818; Pyrrhobryum mnioides (Hook.) Manuel, Cryptog. Bryol. Lichénol. 1: 70. 1980. Hymenodontopsis vallis-gratiae (Hampe ex Müll. Hal.) N. E. Bell, A. E. Newton & D. Quandt, comb. nov.; Mnium vallis-gratiae Hampe ex Müll. Hal., 555 Bot. Zeitung (Berlin) 17: 205. 1859; Pyrrhobryum vallis-gratiae (Hampe ex Müll. Hal.) Manuel, Cryptog. Bryol. Lichénol. 1: 70. 1980. Mniodendron Lindb. ex Kindb., Bot. Centralb. 77: 393. 1909. TYPE (lectotype designated by Touw 1971): Hypnum divaricatum Reinw. & Hornsch. Mniodendron camptotheca Duby ex Besch. ex Paris; Hypnodendron camptotheca (Duby ex Besch. ex Paris) Touw Mniodendron colensoi (Hook.f. & Wilson) Besch.; Isothecium colensoi Hook.f. & Wilson; Hypnodendron colensoi (Hook.f. & Wilson) Mitt. Mniodendron comatulum Geh. ex Broth. & Watts; Hypnodendron comatulum (Geh. ex Broth. & Watts) Touw Mniodendron comatum (Müll. Hal.) Lindb. ex Paris; Hypnum comatum Müll. Hal.; Hypnodendron comatum (Müll. Hal.) Mitt. ex Touw Mniodendron comosum (Labill.) Lindb. ex A. Jaeger; Hypnum comosum Labill.; Hypnodendron comosum (Labill.) Mitt. Mniodendron comosum var. sieberi (Müll. Hal.) N. E. Bell, A. E. Newton & D. Quandt comb. nov.; Hypnum sieberi Müll. Hal., Syn. Musc. Frond. 2: 504. 1851; Hypnodendron comosum var. sieberi (Müll. Hal.) Touw, Blumea 19: 319. 1971. Mniodendron dendroides (Brid.) Wijk & Margad.; Bryum dendroides Brid.; Hypnodendron dendroides (Brid.) Touw Mniodendron tahiticum Besch. ex Besch.; Hypnodendron tahiticum (Besch. ex Besch.) Touw Dendro-hypnum Hampe, Nuovo Giorn. Bot. Ital. 4: 289. 1872. TYPE: Dendro-hypnum beccarii Hampe. NOTE: Touw (1971) considered Dendro-hypnum to be invalid for a number of reasons, but nevertheless accepted the sole species as validly published. This was, however, incorrect. The name meets all the requirements of valid publication as a genericospecific description. Although the name predates Hypnodendron, with the segregation of this group at the generic level, Hypnodendron is no longer threatened. Dendro-hypnum auricomum (Broth. & Geh.) N. E. Bell, A. E. Newton & D. Quandt, comb. nov.; Hypnodendron auricomum Broth. & Geh., Oefver. Förh. Finska Vetensk.-Soc. 40: 190. 1898. Dendro-hypnum auricomum ssp. celebensis (Dix- 556 the bryologist 110(3): 2007 on) N. E. Bell, A. E. Newton & D. Quandt, comb. nov.; Sciadocladus celebensis Dixon, Ann. Bryol. 7: 28. 1934; Hypnodendron auricomum ssp. celebensis (Dixon) Touw, Blumea 19: 259. 1971. Dendro-hypnum auricomum ssp. sumatranum (Baumgartner) N. E. Bell, A. E. Newton & D. Quandt, comb. nov.; Hypnodendron sumatranum Baumgartner, Ann. Naturhist. Mus. Wien 59: 84. 1953; Hypnodendron auricomum ssp. sumatranum (Baumgartner) H. Akiyama, Bryologist 92: 200. 1989. Dendro-hypnum beccarii Hampe; Hypnodendron beccarii (Hampe) A. Jaeger. Dendro-hypnum brevipes (Broth) N. E. Bell, A. E. Newton & D. Quandt, comb. nov.; Hypnodendron brevipes Broth., Oefvers. Förh. Finska Vetensk.-Soc. 40: 189. 1898. Dendro-hypnum flagelliferum (Broth. & Watts) N. E. Bell, A. E. Newton & D. Quandt, comb. nov.; Hypnodendron flagelliferum Broth. & Watts, J. & Proc. Roy. Soc. New South Wales 49: 156. 1915. Dendro-hypnum fuscomucronatum (Müll. Hal.) N. E. Bell, A. E. Newton & D. Quandt, comb. nov.; Hypnum fusco-mucronatum Müll. Hal., Bot. Zeitung (Berlin) 20: 393. 1862; Hypnodendron fuscomucronatum (Müll. Hal.) A. Jaeger, Ber. Thätigk. St. Gallischen Naturwiss. Ges. 1877–78: 360 (Gen. Sp. Musc. 2: 1380). 1880. Dendro-hypnum fuscomucronatum ssp. chalmersii (Mitt.) N. E. Bell, A. E. Newton & D. Quandt, comb. nov.; Hypnodendron chalmersii Mitt., Proc. Linn. Soc. New South Wales 7: 103. 1882; Hypnodendron fuscomucronatum ssp. chalmersii (Mitt.) Touw, Blumea 19: 337. 1971. Dendro-hypnum milnei (Mitt.) N. E. Bell, A. E. Newton & D. Quandt, comb. nov.; Hypnodendron milnei Mitt., Fl. Vit. 401. 1873. Dendro-hypnum milnei ssp. parvum (Müll. Hal.) N. E. Bell, A. E. Newton & D. Quandt, comb. nov.; Mniodendron parvum Müll. Hal., Hedwigia 41: 133. 1902; Hypnodendron milnei ssp. parvum (Müll. Hal.) Touw, Blumea 19: 301. 1971. Dendro-hypnum milnei ssp. korthalsii (Bosch & Sande Lac. ex Paris) N. E. Bell, A. E. Newton & D. Quandt, comb. nov.; Mniodendron korthalsii Bosch & Sande Lac. ex Paris, Index Bryol., ed. 2, 3: 263. 1905; Hypnodendron milnei ssp. korthalsii (Bosch & Sande Lac. ex Paris) Touw, Blumea 19: 302. 1971. Dendro-hypnum opacum (M. Fleisch.) N. E. Bell, A. E. Newton & D. Quandt, comb. nov.; Hypnodendron opacum M. Fleisch., Musci Fl. Buitenzorg 4: 1610. 1923; Hypnodendron auricomum ssp. opacum (M. Fleisch.) Touw, Blumea 19: 259. 1971. Dendro-hypnum reinwardtii (Schwägr.) N. E. Bell, A. E. Newton & D. Quandt, comb. nov.; Hypnum reinwardtii Schwägr., Sp. Musc. Frond., Suppl. 3, 1(1): pl. 223, figs. 1, 3–6, 10–16. 1827; Hypnodendron reinwardtii (Schwägr.) Lindb. ex A. Jaeger, Ber. Thätigk. St. Gallischen Naturwiss. Ges. 1877–78: 358 (Gen. Sp. Musc. 2: 1378). 1880. Dendro-hypnum reinwardtii ssp. caducifolium (Herzog) N. E. Bell, A. E. Newton & D. Quandt, comb. nov.; Hypnodendron caducifolium Herzog, Hedwigia 61: 292. 1919; Hypnodendron reinwardtii ssp. caducifolium (Herzog) Touw, Blumea 19: 245. 1971. Dendro-hypnum subspininervium (Müll. Hal.) N. E. Bell, A. E. Newton & D. Quandt, comb. nov.; Hypnum subspininervium Müll. Hal., Bot. Zeitung (Berlin) 15: 782. 1857; Hypnodendron subspininervium (Müll. Hal.) A. Jaeger, Ber. Thätigk. St. Gallischen Naturwiss. Ges. 1877–78: 359 (Gen. Sp. Musc. 2: 1379). 1880. Dendro-hypnum subspininervium ssp. arborescens (Mitt.) N. E. Bell, A. E. Newton & D. Quandt, comb. nov.; Trachyloma arborescens Mitt., J. Proc. Linn. Soc., Bot. Suppl. 1: 91. 1859; Hypnodendron subspininervium ssp. arborescens (Mitt.) Touw, Blumea 19: 237. 1971. Sciadocladus Lindb. ex Kindb., Bot. Centralbl. 77: 393. 1909. TYPE (designated here): Sciadocladus kerrii (Mitt.) A. Jaeger ex Broth., Nat. Pflanzenfam. I(3): 1168. 1909. Sciadocladus kerrii (Mitt.) A. Jaeger ex Broth.; Trachyloma kerrii Mitt.; Hypnodendron kerrii (Mitt.) Paris Sciadocladus menziesii (Hook.) Lindb. ex Broth.; Hypnum menziesii Hook.; Hypnodendron menziesii (Hook.) Paris Sciadocladus menziesii ssp. splendidum (Besch.) N. E. Bell, A. E. Newton & D. Quandt, comb. nov.; Hypnodendron splendidum Besch., Ann. Sci. Nat. Bot., V, 18: 245. 1873; Hypnodendron menziesii ssp. splendidum (Besch.) Touw, Blumea 19: 265. 1971. Touwiodendron N. E. Bell, A. E. Newton & D. Quandt, gen. nov. Bell et al.: Phylogeny of earliest diverging pleurocarps Planta dendroidea frondibus umbellatis vel palmatis. Rami complanati. Stipes tomento in caespibus dispersus plerumque dispositi. Pseudoparaphyllia foliosa conspicua. Folia stipium squarrosissima acumine longo reflexo. Folia ramorum tristicha eis lateralibus grandibus effusis eis dorsalibus minoris. TYPE: Touwiodendron diversifolium (Broth. & Geh.) N. E. Bell, A. E. Newton & D. Quandt Touwiodendron diversifolium (Broth. & Geh.) N. E. Bell, A. E. Newton & D. Quandt, comb. nov.; Hypnodendron diversifolium Broth. & Geh., Oefvers. Förh. Finska Vetensk.-Soc. 40: 191. 1898. KEY TO THE FAMILIES AND GENERA OF THE HYPNODENDRALES This key employs easily and universally observable characters where possible in order to allow quick identification. In particular, features of the exothecium in transverse section that are diagnostic for the Pterobryellaceae are avoided in favor of simpler characters at the expense of an increased number of couplets. Some characters are derived (although not uncritically) from existing works, including those of Touw (1971), Sastre-De Jesús (1987), Koponen and Norris (1986) and Withey (1996b). Note that in some dendroid taxa stipe leaves are delicate and often decayed, although largely intact leaves can usually be found towards the top of the stipe. 1. Plants mat-forming, small to medium sized, growth indeterminate; stems prostrate and creeping ............ RACOPILACEAE 1. Plants tufted, determinate, usually robust; erect axes projecting from substrate .........................................................2 2. Plants pinnate-dendroid; apex of branch leaves obtuse, cymbiform (concave and shaped like the prow of a boat); costa projecting in small mucro ...................................... BRAITHWAITEACEAE (Braithwaitea) 2. Plants variously formed, if pinnate-dendroid then apex of branch leaves not as above ........................................3 3. Plants dendroid (stipitate, with regularly pinnate, palmate or umbellate frond)........................................................................4 3. Plants not or only vaguely dendroid; stems simple or sparsely and irregularly branched .............................. HYPNODENDRACEAE 4. Stipe leaves spreading-squarrose ....................................5 4. Stipe leaves appressed to weakly spreading ..................6 5. Stipe leaves strongly toothed ....................... HYPNODENDRACEAE 5. Stipe leaves weakly serrulate to crenulate or with a few teeth near base of acumen ...................................... PTEROBRYELLACEAE 6. Branch leaves very narrowly triangular or narrowly ovate and long-acuminate, stipe tomentose at base only; capsules ovate to globular.................... PTEROBRYELLACEAE 557 6. Branch leaves ovate-acute, not long-acuminate, stipe tomentose or otherwise, or if triangular then broadly so and stipe more or less tomentose; capsules not ovate or globular ................................................. HYPNODENDRACEAE RACOPILACEAE A. Plants small, lateral and dorsal leaves usually poorly or only slightly differentiated; seta generally , 2 cm............ Powellia A. Plants mostly medium sized, lateral and dorsal leaves usually clearly differentiated; seta generally . 2 cm ......... Racopilum PTEROBRYELLACEAE A. Stipe leaves squarrose-recurved ..........................Sciadiocladus A. Stipe leaves appressed ............................................................. B B. Laminal cells narrow, elongate; branch leaves 6 straight in both moist and dry states or acumina becoming slightly flexuose or falcate when dry ................................................................... Pterobryella B. Laminal cells rectangular to irregularly quadrate; branch leaves straight when moist, curled inwards or falcate when dry ..................................... Cyrtopodendron HYPNODENDRACEAE A. Plants not or only very weakly dendroid; stems simple or sparsely and irregularly branched .......................................... B A. Plants dendroid; frond regularly pinnate, palmate or umbellate ...................................................................................E B. Cells of lower leaf lamina distinctly differentiated, elongate near costa and 6 isodiametric towards margin ............................................................................. C B. Lower laminal cells various but not differentiated as above................................................................................D C. Seta elongate (. 1 cm); endostome absent ......... Bescherellia C. Seta reduced (, 1 cm); Both exostome and endostome present .......................................................................... Cyrtopus D. Seta elongate (. 15mm, usually þ/ 40mm), costa with single row of guide cells..........................Franciella D. Seta very short (, 5mm), costa with double row of guide cells. ..........................................................Spiridens E. Plants of Australia, New Zealand or South America............F E. Plants of Southeast Asia or the western Pacific ...................G F. Stipes entirely tomentose ............................Mniodendron F. Stipes tomentose at base only ................. Hypnodendron G. Plants with branch leaves tristichous, differentiated into larger, spreading lateral leaves and smaller dorsal leaves; stipe leaves spreading to squarrose with long, strongly reflexed acumens; stipes with conspicuous foliose pseudoparaphyllia ............................................................................ Touwiodendron G. Plants various but not as above; branch leaves differentiated in size or isomorphous but not tristichous; pseudoparaphyllia not conspicuous, appressed or else small and obscured by tomentum...........................................................H 558 the bryologist 110(3): 2007 H. Stipes entirely tomentose; stipe leaves widely spreading to squarrose.................................................Mniodendron H. Stipes tomentose at base only, or if tomentose above then stipe leaves 6 appressed ....................................... I I. Stipe leaves strongly appressed to only slightly spreading, not entirely covering stipe; stipes tomentose at base only .............................................................................. Hypnodendron I. Stipe leaves strongly spreading to squarrose, or if 6 appressed then stipes either more or less tomentose or else completely covered with leaves ...........................................Dendro-hypnum ACKNOWLEDGMENTS The first author’s work at the Botanical Museum of the University of Helsinki was funded by the Academy of Finland under the project ‘‘Polytrichales: towards a modern phylogenetic monograph and the development of a model of sporophyte evolution’’ (No. 108629). The second author acknowledges support by the Deutsche Forschungsgemeinschaft (DFG Qu 153-1), and the Deutscher Akademischer Austauschdienst (DAAD). Some of the molecular data was generated during the first author’s Ph.D. work at the Natural History Museum in London, funded by the Department of Botany, and a small number of newly generated DNA sequences made use of material collected in southern Patagonia, Chile (2005), in conjunction with Universidad de Magallanes and the Omora Foundation as part of a project funded by the U.K. Darwin Initiative. Bayesian analyses were carried out using the resources of the Computational Biology Service Unit from Cornell University which is partially funded by Microsoft Corporation. We thank Jaakko Hyvönen and Bill Buck for support and help with nomenclature, Katherine Challis and Norman Robson for help with Latin, and the reviewers, Lars Hedenäs and Ray Tangney, for their helpful comments and for agreeing to examine the manuscript promptly and at short notice. LITERATURE CITED Akiyama, H. 1989. A reevaluation of Hypnodendron opacum (Hypnodendraceae, Musci). The Bryologist 92: 198–201. APG (The Angiosperm Phylogeny Group). 1998. An ordinal classification for the families of flowering plants. Annals of the Missouri Botanical Garden 85: 531–553. ——— II. 2003. An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants: APG II. Botanical Journal of the Linnaean Society 141: 399–436. Bell, N. E. & A. E. Newton. 2004. Systematic studies of nonhypnanaean pleurocarps: establishing a phylogenetic framework for investigating the origins of pleurocarpy. In B. Goffinet, V. C. Hollowell & R. E. Magill (eds.), Molecular systematics of bryophytes. Monographs in Systematic Botany from the Missouri Botanical Garden 98: 290–319. ——— & ———. 2005. The paraphyly of Hypnodendron and the phylogeny of related non-hypnanaean pleurocarpous mosses inferred from chloroplast and mitochondrial sequence data. Systematic Botany 30: 34–51. ——— & ———. 2007. Pleurocarpy in the rhizogoniaceous grade. In A. E. Newton & R. Tangney (eds.), Pleurocarpous mosses: systematics and evolution. Systematics Association Special Volume Series 71: 41–64. CRC Press, Boca Raton. ——— & P. V. York. 2007. Vetiplanaxis pyrrhobryoides, a new fossil moss genus and species from Middle Cretaceous Burmese amber. The Bryologist 110: in press. Buck, W. R. 1991. The basis for familial classification of pleurocarpous mosses. Advances in Bryology 4: 169–185. ———, C. J. Cox, A. J. Shaw & B. Goffinet. 2005. Ordinal relationships of pleurocarpous mosses, with special emphasis on the Hookeriales. Systematics and Biodiversity 2: 121–145. ——— & B. Goffinet. 2000. Morphology and classification of mosses. Pages 71–123. In A. J. Shaw & B. Goffinet (eds.), Bryophyte Biology. Cambridge: Cambridge University Press, Cambridge. ———, ——— & A. J. Shaw. 2000a. Testing morphological concepts of orders of pleurocarpous mosses (Bryophyta) using phylogenetic reconstructions based on trnL-trnF and rps4 sequences. Molecular Phylogenetics and Evolution 16: 180–198. ———, ——— & ———. 2000b. Novel relationships in pleurocarpous mosses as revealed by cpDNA sequqnces. The Bryologist 103: 774–789. ——— & D. H. Vitt. 1986. Suggestions for a new familial classification of pleurocarpous mosses. Taxon 35: 21–60. Cantino, P. D. & K. de Queiroz. 2004. PhyloCode: A phylogenetic code of biological nomenclature. http://www. ohiou.edu/phylocode. Cox, C. J., B. Goffinet, A. E. Newton, A. J. Shaw & T. A. J. Hedderson. 2000. Phylogenetic relationships among the diplolepideous-alternate mosses (Bryidae) inferred from nuclear and chloroplast DNA sequences. The Bryologist 103: 224–241. ———, ———, A. J. Shaw & S. B. Boles. 2004. Phylogenetic relationships among the mosses based on heterogeneous Bayesian analysis of multiple genes from multiple genomic compartments. Systematic Botany 29: 234–250. ——— & T. A. J. Hedderson. 1999. Phylogenetic relationships among the ciliate arthrodontous mosses: evidence from chloroplast and nuclear DNA sequences. Plant Systematics and Evolution 215: 119–139. Cummings, M. P., S. A. Handley, D. S. Myers, D. L. Reed, A. Rokas & K. Winka. 2003. Comparing bootstrap and posterior probability values in the four taxon case. Systematic Biology 52: 477–487. De Luna, E., W. R. Buck, H. Akiyama, T. Arikawa, H. Tsubota, D. González, A. E. Newton & A. J. Shaw. 2000. Ordinal phylogeny within the hypnobryalean pleurocarpous mosses inferred from cladistic analyses of three chloroplast DNA sequence data sets: trnL-F, rps4, and rbcL. The Bryologist 103: 242–256. ———, A. E. Newton, A. Withey, D. González & B. Mishler. Bell et al.: Phylogeny of earliest diverging pleurocarps 1999. The transition to pleurocarpy: a phylogenetic analysis of the main diplolepidous lineages based on rbcL sequqnces and morphology. The Bryologist 102: 634–650. Erixon, P., B. Svennblad, T. Britton & B. Oxelman. 2003. Reliability of Bayesian posterior probabilities and bootstrap frequencies in phylogenetics. Systematic Biology 52: 665– 673. Felsenstein, J. 1978. Cases in which parsimony or compatibility methods will be positively misleading. Systematic Zoology 27: 401–410. Frahm, J.-P. 2001. Neue Laubmoosfunde aus sächsischem und baltischem Bernstein. Nova Hedwigia 72: 271–281. ———. 2004. A new contribution to the moss flora of Baltic and Saxon amber. Review of Palaeobotany and Palynology 129: 81–101. Goffinet, B. & W. R. Buck. 2004. Systematics of the Bryophyta (mosses): from molecules to a revised classification. In B. Goffinet, V. C. Hollowell & R. E. Magill (eds.), Molecular systematics of bryophytes. Monographs in Systematic Botany from the Missouri Botanical Garden 98: 205–239. ———, C. J. Cox, J. A. Shaw & T. A. J. Hedderson. 2001. The Bryophyta (mosses): systematic and evolutionary inferences from an rps4 gene (cpDNA) phylogeny. Annals of Botany 87: 191–208. Hampe, E. 1872. Musci frondosi in insulis Ceylon et Borneo a D.r Od. Beccari lecti. Nuovo Giornale Botanico Italiano 4: 273–291. Hedenäs, L. 1994. The basal pleurocarpous diplolepidous mosses—A cladistic approach. The Bryologist 97: 225–243. ———. 1998. Cladistic studies on pleurocarpous mosses: research needs, and use of results. Pages 125–141. In J. W. Bates, N. W. Ashton & J. G. Duckett (eds.), Bryology for the Twenty-First Century. Maney Publishing and the British Bryological Society, Leeds. ———. 2001. Environmental factors potentially affecting character states in pleurocarpous mosses. The Bryologist 104: 72–91. Hill, R. S. 2001. Biogeography, evolution and palaeoecology of Nothofagus (Nothofagaceae): the contribution of the fossil record. Australian Journal of Botany 49: 321–332. Hörandl, E. 2006. Paraphyletic versus monophyletic taxa— evolutionary versus cladistic classifications. Taxon 55: 564– 570. Huelsenbeck, J. P. & B. Rannala. 2004. Frequentist properties of Bayesian posterior probabilities of phylogenetic trees under simple and complex substitution models. Systematic Biology 53: 904–913. ——— & F. Ronquist. 2001. MRBAYES: Bayesian Inference of phylogeny. Bioinformatics 17: 754–755. Huttunen, S., M. S. Ignatov, K. Müller & D. Quandt. 2004. Phylogeny and evolution of epiphytism in the three moss families Meteoriaceae, Brachytheciaceae and Lembophyllaceae. In B. Goffinet, V. C. Hollowell & R. E. Magill (eds.), Molecular systematics of bryophytes. Monographs in 559 Systematic Botany from the Missouri Botanical Garden 98: 328–361. Ignatov, M. S. & B. C. Tan. 1992. Orthodontopsis, a new genus of Bryaceae (Musci) from southern Siberia, USSR. Journal of the Hattori Botanical Laboratory 71: 165–173. Karttunen, K. & S. Bäck. 1988. Taxonomy of Hymenodon (Musci, Rhizogoniaceae). Annales Botanici Fennici 25: 89– 95. Kelchner, S. A. 2000. The evolution of non-coding chloroplast DNA and its application in plant systematics. Annals of the Missouri Botanical Garden 87: 482–498. Koponen, T. & D. H. Norris. 1986. Bryophyte flora of the Huon Peninsula, Papua, New Guinea. XVII. Grimmiaceae, Racopilaceae and Hedwigiaceae. Acta Botanica Fennica 133: 81– 106. Kuntner, M. & I. Agnarsson. 2006. Are Linnean and phylogenetic nomenclatural systems combinable? Systematic Biology 55: 774–784. Manuel, M. G. 1980. Miscellanea bryologica II. Classification of Rhizogonium Brid., Penzigiella hookeri Gangulee, and some nomina nuda. Cryptogamie Bryologie, Lichenologie 1: 67– 72. Mathews, D.H., M. D. Disney, J. L. Childs, S. J. Schroeder, M. Zuker & D. H. Turner. 2004. Incorporating chemical modification constraints into a dynamic programming algorithm for prediction of RNA secondary structure. Proceedings of the National Academy of Sciences of the United States of America 101: 7287–7292. Mayr, E. & W. J. Bock. 2002. Classifications and other ordering systems. Journal of Zoological Systematics and Evolutionary Research 40: 169–194. McDaniel, S. F. & A. J. Shaw. 2003. Phylogeographic structure and cryptic speciation in the trans-Antarctic moss Pyrrhobryum mnioides. Evolution 57: 205–215. McLoughlin, S. 2001. The breakup history of Gondwana and its impact on pre-Cenozoic floristic provincialism. Australian Journal of Botany 49: 271–300. Meijer, W. 1952. The genus Orthodontium. Acta Botanica Neerlandica 1: 3–80. Meusel, H. 1935. Wuchsformen und Wuchstypen der europäischen Laubmoose. Nova Acta Leopoldina (Neue Folge) 3: 123–277. Mishler, B. D & E. De Luna. 1991. The use of ontogenetic data in phylogenetic analyses of mosses. Advances in Bryology 4: 121–167. Müller, J. & K. Müller. 2004. TreeGraph: automated drawing of complex tree figures using an extensible tree description format. Molecular Ecology Notes 4: 786–788. Müller, K. 2004. PRAP-computation of Bremer support for large data sets. Molecular Phylogenetics and Evolution 31: 780– 782. ———. 2005. SeqState - primer design and sequence statistics for phylogenetic DNA data sets. Applied Bioinformatics 4: 65–69. 560 the bryologist 110(3): 2007 ———, D. Quandt, J. Müller & C. Neinhuis. 2005. PhyDe v0.92: Phylogenetic Data Editor. http://www.phyde.de. Newton, A. E. 2007. Branching architecture in pleurocarpous mosses. In A. E. Newton & R. Tangney (eds.), Pleurocarpous mosses: systematics and evolution. Systematics Association Special Volume Series 71: 287–307. CRC Press, Boca Raton. ——— & E. De Luna. 1999. A survey of morphological characters for phylogenetic study of the transition to pleurocarpy. The Bryologist 102: 651–682. ———, J. G. Duckett, J. A. Wheeler, B. Goffinet, T. A. J. Hedderson & B. D. Mishler. 2000. Evolution of the major moss lineages: phylogenetic analyses based on multiple gene sequences and morphology. The Bryologist 103: 187–211. ———, N. Wikström, N. E. Bell, L. L. Forrest & M. S. Ignatov. 2007. Dating the diversification of the pleurocarpous mosses. In A. E. Newton & R. Tangney (eds.), Pleurocarpous mosses: systematics and evolution. Systematics Association Special Volume Series 71: 337–366. CRC Press, Boca Raton. Nixon, K. C. 1999. The parsimony ratchet, a new method for rapid parsimony analysis. Cladistics 15: 407–411. Norris, D. H. & T. Koponen. 1996. Bryophyte flora of the Huon Peninsula, Papua New Guinea. LVIII. Hypnodendron (Hypnodendraceae, Musci). Acta Botanica Fennica 156: 23– 48. Nylander, J. A. A. 2004. MrModeltest v2. Program distributed by the author. Evolutionary Biology Centre, Uppsala University. O’Brien, T. J. 2007. The phylogenetic distribution of pleurocarpous mosses: evidence from cpDNA sequences. In A. E. Newton & R. Tangney (eds.), Pleurocarpous mosses: systematics and evolution. Systematics Association Special Volume Series 71: 19–40. CRC Press, Boca Raton. Pickett, K. M. & C. P. Randle. 2005. Strange Bayes indeed: uniform topological priors imply non-uniform clade priors. Molecular Phylogenetics and Evolution 34: 203–211. Provan, J., W. Powell & P. M. Hollingsworth. 2001. Chloroplast microsatellites: new tools for studies in plant ecology and evolution. Trends in Ecology and Evolution 16: 142–147. Quandt, D., N. E. Bell & M. Stech. 2007. Unravelling the knot: the Pulchrinodaceae fam. nov. (Bryales). Nova Hedwigia Beiheft 131: 21–39 ———, S. Huttunen, H. Streimann, J.-P. Frahm & W. Frey. 2004. Molecular phylogenetics of the Meteoriaceae s. str.: focusing on the genera Meteorium and Papillaria. Molecular Phylogenetics and Evolution 32: 435–461. ———, K. Müller & S. Huttunen. 2003. Characterisation of the chloroplast DNA psbT-H region and the influence of dyad symmetrical elements on phylogenetic reconstructions. Plant Biology 5: 400–410. ——— & M. Stech. 2004. Molecular evolution and phylogenetic utility of the chloroplast trnT-trnF region in bryophytes. Plant Biology 6: 545–554. ——— & ———. 2005. Molecular evolution and secondary structure of the chloroplast trnL intron in bryophytes. Molecular Phylogenetics and Evolution 36: 429–443. Ronquist, F. & J. P. Huelsenbeck. 2003. MRBAYES 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19: 1572–1574. Sastre-De Jesús, I. 1987. Revision of the Cyrtopodaceae and transfer of Cyrtopodendron to the Pterobryaceae. Memoirs of The New York Botanical Garden 45: 709–721. Shaw, J. & L. E. Anderson. 1986. Morphology and homology of the peristome teeth in Hymenodon and Hymenodontopsis (Rhizogoniaceae: Musci). Systematic Botany 11: 446–454. ———, C. J. Cox, B. Goffinet, W. R. Buck & S. B. Boles. 2003. Phylogenetic evidence of a rapid radiation of pleurocarpous mosses (Bryophyta). Evolution 57: 2226–2241. Simmons, M. P. & H. Ochoterena. 2000. Gaps as characters in sequence-based phylogenetic Analyses. Systematic Biology 49: 369–381. ———, K. M. Pickett & M. Miya. 2004. How meaningful are Bayesian support values? Molecular Biology and Evolution 21: 188–199. Stevens, P. F. 2002. Why do we name organisms? Some reminders from the past. Taxon 51: 11–26. ———. 2006. An end to all things?—plants and their names. Australian Systematic Botany 19: 115–133. Swofford, D. L. 2002. PAUP*: Phylogenetic analysis using parsimony (*and other methods), version 4.0b10. Sunderland, MA. Touw, A. 1971. A taxonomic revision of the Hypnodendraceae (Musci). Blumea 19: 211–354 Tsubota, H., T. Arikawa, H. Akiyama, E. De Luna, D. González, M. Higuchi & H. Deguchi. 2002. Molecular phylogeny of hypnobryalean mosses as inferred from a large-scale dataset of chloroplast rbcL, with special reference to the Hypnaceae and possibly related families. Hikobia 13: 645–665. Vitt, D. H. 1984. Classification of the Bryopsida. Pages 696–759. In R. M. Schuster (ed.), New Manual of Bryology, Vol. 2. The Hattori Botanical Laboratory, Nichinan. Withey, A. 1996a. Phylogenetic studies of the Spiridentaceae (Musci): observations of three morphological characters associated with pleurocarpy. Anales del Instituto de Biologia, Serie Botánica 67: 5–14. ———. 1996b. Systematic Studies of the Spiridentaceae (Musci). Ph.D. dissertation, Duke University, Durham, NC. ms. received April 3, 2007; accepted April 19, 2007.

RELATED PAPERS

RELATED TOPICS

Log In

Taxonomy and phylogeny in the earliest diverging pleurocarps: square holes and bifurcating pegs

Taxonomy and phylogeny in the earliest diverging pleurocarps: square holes and bifurcating pegs

Related Papers

RELATED PAPERS

RELATED TOPICS