Phylogeny of the Brachytheciaceae (Bryophyta) based on morphology and sequence level data

Sanna Huttunen; Michael S. Ignatov

Cladistics Cladistics 20 (2004) 151–183 www.blackwell-synergy.com Phylogeny of the Brachytheciaceae (Bryophyta) based on morphology and sequence level data Sanna Huttunen* and Michael S. Ignatov *Botanical Museum and Division of Systematic Biology, PO Box 7 and 65, FIN-00014 University of Helsinki, Finland Main Botanical Garden of Russian Academy of Sciences, Botanicheskaya 4, Moscow 127276, Russia Accepted 3 March 2004 Abstract Brachytheciaceae is often considered a taxonomically difficult group of mosses. For example, morphological variation has led to difficulty in generic delimitation. We used DNA sequence data (chloroplast psbT-H and trnL-F and nuclear ITS2) together with morphology (63 characters) to examine the relationships within this family. The combined unaligned length of the DNA sequences used in the phylogenetic analyses varied between 1277 and 1343 bp. For phylogeny reconstruction we performed direct optimization, as implemented in POY. Analyses were performed with three different gap costs and the morphological data partition was weighted both: (1) equal to gap cost, and (2) with a weight of one. The utility of sensitivity analysis has recently been cast into doubt; hence in this study it was performed only to explore the effects of weighting on homology statements and topologies and to enable more detailed comparisons between earlier studies utilizing the direct optimization method. The wide sequence length variation of non-coding ITS2 sequences resulted in character optimizations (i.e., ‘‘alignments’’) of very different lengths when various gap costs were applied. Despite this variation, the topologies of equally parsimonious trees remained fairly stable. The inclusion of several outgroups, instead of only one, was observed to increase the congruence between data sets and to slightly increase the resolution. An inversion event in the 9 bp loop region in the chloroplast psbT-N spacer in mosses has been postulated to include only uninformative variation, thus possibly negatively impacting the phylogeny reconstruction. Despite this inversion, its variation within Brachytheciaceae was clearly congruent with information from other sources, but inclusion of these 9 bp in the analysis had only a minor effect on the phylogenetic results. In the most parsimonious topology, which was obtained with equal weighting of all data, Meteoriaceae and Brachytheciaceae were resolved as monophyletic sister groups, which had recently been suggested based on a few shared morphological characters. Our study revealed some new generic relationships within the Brachytheciaceae, which are discussed in light of the morphological characters traditionally used for generic delimitation. The Willi Hennig Society 2004. Brachytheciaceae includes about 570 species (Crosby et al., 1999), and constitutes one of the largest families of mosses. These species belong to the pleurocarpous mosses, presumably the youngest group of mosses, and the family Brachytheciaceae is also possibly still rapidly evolving (McAdam, 1982). Observations of low sequence variation among pleurocarpous mosses compared with other bryophyte groups support this idea (Vanderpoorten et al., 2002), although even rough age estimates are very difficult to give due to the extremely scattered fossil data. Among the pleurocarpous mosses, Brachytheciaceae has been suggested to be phylogenetically close to Amblystegiaceae, Leskeaceae, and Thu The Willi Hennig Society 2004 idiaceae (Robinson, 1962) or to Hylocomiaceae and Ctenidiaceae (Hedenäs, 1989). According to the latest phylogenetic studies of pleurocarpous mosses, Brachytheciaceae belongs to the order Hypnales (Buck and Goffinet, 2000), and based on morphological and molecular data it is closely related to Meteoriaceae (Buck, 1998; Ignatov, 1999; Buck et al., 2000a; Quandt et al., 2004b; Huttunen et al., 2004). Phylogenetic analyses by Buck et al. (2000a) of the chloroplast trnL-F region and rps4 gene, and Tsubota et al.’s (2002) work on the rbcL gene suggest another family, Plagiotheciaceae (in the s. str., including only Plagiothecium Schimp.), is very closely related to Brachytheciaceae. 152 S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 Despite the many difficulties encountered in the delimitation of Brachytheciaceae (for details see Ignatov and Huttunen, 2002), the family is now considered to be one of the best-defined pleurocarp families (Hedenäs, 1989, 1992; Ignatov, 1998; Ignatov, 1999). Characters separating it from other pleurocarpous mosses mostly include those that have been neglected by previous authors, such as the shape of the stomatal pore, perichaetial leaf patterns, and the arrangement of pseudoparaphyllia around the young branch primordia. Generic delimitation, however, remains very unclear. The latest worldwide revision of Brachytheciaceae was undertaken by Brotherus (1925), who borrowed the concept of the three largest genera of the family from Bruch et al. (1851–1855). The genera Brachythecium Schimp., Rhynchostegiella (Schimp.) Limpr., Rhynchostegium Schimp., and Eurhynchium Schimp. are assumed to be non-monophyletic (Hedenäs, 1992; Ignatov et al., 1999; many others). This conclusion was, however, only based on morphological data, and their monophyly has not been tested from a phylogenetic perspective. Only one attempt was made to change the classification of Brachythecium, in which Robinson (1962) split it into two parts of almost equal size, Brachythecium s. str. and Chamberlainia Grout. This idea did not receive wide acceptance and Robinson (1987) himself later denied it. At the same time almost all revisions of Brachythecium accepted 5–8 sections introduced by Kindberg (1897), then used by Brotherus (1905–1909) and most subsequent authors, with relatively minor corrections and alternations (Takaki, 1955a,b, 1956; Robinson, 1962; McFarland, 1994; Ignatov, 1998). Robinson (1987) also attempted to redefine Rhynchostegium by placing tropical species in the genus Steerecleus H. Robinson, a view not accepted in recent studies (McFarland, 1994; Buck, 1998; Ignatov et al., 1999; Hedenäs, 2002). Many authors in the early 20th century followed Warnstorf (1905), who raised the rank of Oxyrrhynchium (Schimp.) Warnst. (as a group of species centered around Eurhynchium hians (Hedw.) Sande Lac.) to generic level. However, the generic status of Oxyrrhynchium was not accepted in the latter half of the 20th century. The aim of the present study was to determine whether molecular data can provide further resolution and help in clarifying the separation of Brachytheciaceae from Meteoriaceae, as well as providing evidence for the generic classification of Brachytheciaceae. Only a limited attempt has been made to analyze Brachytheciaceae phylogenetically before this study (Stech and Frahm, 1999), and the current classification of the family is mainly based on a rather intuitive view of the relationships among the taxa. We believe that our study provides a new, or at least a better-founded hypothesis of phylogenetic relationships, which may lead to a more natural classification of the family. Due to the considerable morphological variation, and presumably their homoplastic evolution, cladistic analyses based only on morphological data have been suspected of providing only limited information on phylogenetic relationships. We use nuclear and chloroplast molecular data, in addition to morphological evidence. Genomic regions sequenced for this study include the nuclear Internal Transcripted Spacer 2 (ITS2), the chloroplast psbT-H gene complex, including psbT, psbN and psbH genes coding small photosystem II proteins, and the non-coding spacers between them, and the trnL-F region, consisting of partial ribosome coding tRNA-Leu gene and the non-coding spacer preceding it. Since the majority of sequence level data in this study originated from non-coding genomic regions, we use direct optimization alignment (Wheeler, 1996) as implemented in POY (Gladstein and Wheeler, 2001). While the phylogenetic information from insertion and deletion events (‘‘gaps’’) have traditionally been neglected by treating them as missing data, the direct optimization method utilizes this variation in the same way as information from all other types of substitution events. Here we attempt to evaluate the effect of indel events on the resulting hypothesis of relationships within the study group. The value of information from gaps and how they should be treated in phylogenetic analyses has been the focus of several papers (for example, Giribet and Wheeler, 1999; Simmons and Ochoterena, 2000; Bapteste and Philippe, 2002; Kawakita et al., 2003). In our study, we attempt to determine if this kind of exploration affects the hypotheses of phylogenetic relationships. We also examine the effect of differential treatment of short inversion regions on the hypothesis of relationships. Similar inversions are expected to be common in plant chloroplast genomes (Kelchner and Wendel, 1996; Graham et al., 2000), and recent studies (Kelchner and Wendel, 1996; Quandt et al., 2003) have suggested that these regions may affect phylogeny reconstruction by increasing homoplasy within the data and adding extra weight for the inversion event in analyses. Although we agree that differential weighting does not lead to the most parsimonious results (see Frost et al., 2001; Grant and Kluge, 2003), we performed the sensitivity analyses (sensu Wheeler, 1995) in order to be able to compare the results favored by this method to those obtained from the equal weighting of all characters. We also aimed to explore the contribution of different data partitions on the hypothesis of relationships when differential weighting of gaps and morphology are used in direct optimization analyses. Materials and methods Selection of taxa We included in our study representatives of all genera of Brachytheciaceae for which we had recently collected specimens available (Table 1). In those cases S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 153 Table 1 List of specimens used for sequencing. All specimens are held in the Finnish Museum of Natural History (H) if not indicated otherwise. Of all specimens in the Main Botanical Garden of Russian Academy of Sciences, Moscow (MHA), duplicates are also available in H Species trnL-F psbT-H ITS2 Voucher specimen for sequences AY044072 AF417395 AF395634 AF397780 AF417386 AF403619 AF397802 AF417401 AF403628 Ecuador, Loja Prov., D. H. Norris 92175 & M. Bolivar, December 18, 1997 China, Hunan Prov., T. Koponen, S. Huttunen & P.-C. Rao 52753, October 9, 1997 USA, California, D. H. Norris 82503, February 21, 1994 AF397864 AF417442 AF403654 USA, North Carolina, L. E. Anderson 24257, March 3, 1983 AF397875 AF397828 AF417444 AF417352 AF403646 AF403582 Finland, Regio aboënsis, J. Pykälä 6775, September 21, 1990 China, Hunan Prov., T. Koponen, S. Huttunen & P.-C. Rao 51827, October 8, 1997 Brachythecium buchananii (Hook.) A. Jaeger Brachythecium collinum (Müll. Hal.) Schimp. Brachythecium complanatum Broth. AF397792 AF417348 AF403595 AY184776 AY184757 AY166440 AY184781 AY184762 AY166444 Brachythecium erythrorrhizon Schimp. ssp. asiaticum Ignatov Brachythecium falcatulum (Broth.) Paris Brachythecium frigidum (Müll. Hal.) Besch. Brachythecium geheebii Milde AY184783 AY184763 AY166446 AF397774 AF417468 AF403662 AF397874 AF397776 AF417456 AF417467 AF403638 AF403660 Brachythecium geheebii Milde – AY184770 AF503533 Brachythecium glaerosum (Spruce) Schimp. Brachythecium glaciale Schimp. Brachythecium hylotapetum B. L. Higinb. & N. L. Higinb. Brachythecium koponenii Ignatov AF397865 AF417457 AF403656 – AF397871 AY184766 AF417445 AY166448 AF403653 AF397872 AF417446 AF403657 AY184782 AF397806 – AF417394 AY166445 AF403584 AY184786 AY184767 AF503534 Brachythecium mildeanum (Schimp.) Schimp. Brachythecium plumosum (Hedw.) Schimp AY184777 AY184758 AY166441 AF397814 AF417359 AF403586 Brachythecium populeum (Hedw.) Schimp. AF397873 AF417458 AF403640 Brachythecium oedipodium (Mitt.) A. Jaeger Brachythecium ornellanum (Molendo) Venturi & Bott. Brachythecium reflexum (Starke) Schimp. Brachythecium rivulare Schimp. AF397784 AF417388 AF403649 China, Hunan Prov., T. Koponen, S. Huttunen, S. Piippo & P.-C. Rao 53972, September 30., 1998 Russia, Volgograd Prov., M. Ignatov, August 7, 1999 (MHA) Russian Far East, Upper Bureya River, M. Ignatov 97–172, August 28, 1997 (MHA) Russian Far East, Upper Bureya River, l M. Ignatov 97–1120, August 25, 1997 (MHA) Russia, Altai Mts., M. Ignatov #0 ⁄ 1680, June 26, 1989 (MHA) USA, California, Duell, April 23., 1981 Russia, Bashkiria, E. Ignatova #15 ⁄ 68, September 13, 1990 (MHA) Poland, Sudetes Mts., R. J. Wojcicki, August 26, 1986 (Ochyra & Bednarek-Ochyra, Musci Poloniae exs. 983) Ukraina, Lvov Prov., M. Ignatov, September 13, 1991 (MHA) Russia, Caucasus, E. Ignatova, August 8, 1986 (MHA) Canada, British Colombia, R. Love & J. Rogers B27688, June 15, 1981 Papua New Guinea, Morobe Prov., T. Koponen 32122, July 6, 1981 USA, Oklahoma, B. Allen 6208, April 23, 1988 Papua New Guinea, Morobe Prov., T. Koponen 28748, May 20, 1981 Russia, Kola Peninsula, Khibiny Mts., M. Ignatov, July 31, 1998 (MHA) Russia, Central European part, Kursk Prov., M. Ignatov, May 20, 1999 (MHA) China, Hunan Prov., T. Koponen, S. Huttunen, S. Piippo & P.-C. Rao 56777, October 14, 1998 Finland, Regio aboënsis, J. Pykälä 7029, September 30, 1990 Finland, Tavastia australis, S. Huttunen 1439, April 14, 2000 AF397785 AF417354 AF403593 Russia, Altai Mts, M. Ignatov 16 ⁄ 44, July 18, 1991 (MHA) AF397858 AY044068 AF417459 AF417471 AF403655 AF395627 Brachythecium rivulare Schimp. Brachythecium ruderale (Brid.) W. R. Buck Brachythecium rutabulum (Hedw.) Schimp. Brachythecium salebrosum (F. Weber & D. Mohr) Schimp. Brachythecium salebrosum (F. Weber & D. Mohr) Schimp. Brachythecium subalbicans Broth., nom illeg. AF397866 AY184784 AF397867 AY184780 AF417460 AY184764 AF417447 AY184761 AF403651 AF503535 AF403644 AY16643 Finland, Savonia australis, S. Huttunen 1195, May 1997 China, Hunan Prov., T. Koponen, S. Huttunen, S. Piippo & P.-C. Rao 48671, October 4, 1998 Finland, Ostrobottnia kajanensis, A. Parnela, May 19, 1996 Haiti, W. R. Buck 4826, March 23, 1982 Finland, Nylandia, S. Huttunen 1415, April 16, 2000 Russia, Kursk Prov, M. Ignatov, August 15, 1996 (MHA) AF397857 AF417448 AF403648 Finland, Tavastia australis, S. Laaka 1835, October 18, 1987 AF397788 AF417373 AF403583 China, Hunan Prov., T. Koponen, S. Huttunen & P.-C. Rao 49112, October 1, 1997 Brachytheciaceae Aerolindigia capillacea (Hornsch.) M. Menzel Aerobryum speciosum (Dozy & Molk.) Dozy & Molk. Bestia longipes (Sull. & Lesq.) Broth. Brachythecium acuminatum (Hedw.) Austin Brachythecium albicans (Hedw.) Schimp. Brachythecium buchananii (Hook.) A. Jaeger Brachythecium laetum (Brid.) Schimp Brachythecium lamprocarpum (Müll. Hal.) A.Jaeger Brachythecium latifolium Kindb. 154 S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 Table 1 Continued Species trnL-F Brachythecium turgidum (Hartm.) Kindb. Brachythecium udum I. Hagen. AY184785 AY184765 AY166447 psbT-H ITS2 Russia, Altai Mts, M. Ignatov 31 ⁄ 291, August 1, 1992 (MHA) AY184778 AY184759 AF503536 Brachythecium velutinum (Hedw.) Schimp. Bryhnia novae-angliae (Sull. & Lesq.) Grout Bryhnia novae-angliae (Sull. & Lesq.) Grout Bryhnia novae-angliae (Sull. & Lesq.) Grout Bryoandersonia illecebra (Hedw.) H. Robinson Cirriphyllum cirrosum (Schwägr.) Grout Cirriphyllum cirrosum (Schwägr.) Grout AF397832 AF417440 AF403667 Russia, Siberia, Yakutiya, M. Ignatov 00–843, August 19, 2000 (MHA) Finland, Regio aboënsis, M. Kiirikki, August 23, 1988 AF397843 AF417412 AF403665 AF397826 AF417405 AF403588 China, Hunan Prov., T. Koponen, S. Huttunen & P.-C. Rao 52900, October 11, 1997 Russia, Kostroma Prov., S. Popov 26, May 11, 1999 (MHA) AY184779 AY184760 AY166442 Russia, Altai Mts., M. Ignatov 34 ⁄ 227 (MHA) AF397819 AF417365 AF403626 USA, Arkansas, Bowers 22214, May 24, 1999 AF397856 AF417449 AF403641 AY184793 AY184775 AY447036 China, Xinjiang Prov., B. C. Tan 93–846, July 24, 1993 Russia, Yakutia, M. Ignatov 00–26, August 17, 2000 (MHA) Russia, Caucasus, M. Ignatov, August 23, 1999 (MHA) Cirriphyllum crassinervium (Taylor) AF397868 AF417433 AF403668 Loeske & M. Fleisch. Cirriphyllum flotowianum (Sendtn.) Ochyra AF397869 AF417461 AF403636 Cirriphyllum piliferum (Hedw.) Grout Cirriphyllum tommasinii (Boulay) Grout Clasmatodon parvulus (Hampe) Sull. Eurhynchiadelphus eustegium (Besch.) Ignatov & Huttunen Eurhynchium angustirete (Broth.) T. J. Kop. Eurhynchium hians (Hedw.) Sande Lac. Poland, Krakow-Czestochowa Upland, H. & S. Bednarek, June 8, 1985 AF397799 AF417403 AF403608 Finland, Regio aboënsis, T. Koponen & S. Huttunen 1324, May 24, 1999 AF397870 AF417464 AF403639* Poland, Western Carpathians, R. Ochyra, August 26, 1982 AF397813 AF417430 AF403614 USA, Missouri, P. L. Redfearn, Jr. & B. Allen, April 23, 1992 AF397790 AF417378 AF403602 China, Jilin Prov., T. Koponen 36592, September 20, 1981 AF397825 AF417436 AF403621 Russia, Moscow Prov., M. Ignatov, July 3, 1998 (MHA) AF397815 AF417372 AF403603 China, Hunan Prov., T. Koponen, S. Huttunen, S. Piippo & P.-C. Rao 53740, September 28, 1998 Finland, Regio aboënsis, T. Koponen & S. Huttunen 1321, May 24., 1999 Georgia, Caucasus, M. Ignatov, August 8, 1987 (MHA) China, Hunan Prov., T. Koponen, S. Huttunen & P.-C. Rao 51775, October 8, 1997 Russia, Caucasus, Makridin, August 13., 1998 (MHA) Papua New Guinea, Morobe Prov., T. Koponen 34413, July 27, 1981 Eurhynchium pulchellum (Hedw.) Jenn. AY044069 AF417384 AF395635 Eurhynchium pumilum (Wilson) Schimp. Eurhynchium savatieri Besch. AY184790 AY184772 – AF397859 AF417453 AF403659 Eurhynchium striatum (Hedw.) Schimp. Eurhynchium vagans (A. Jaeger) E. B. Bartram var. bergmaniae (E. B. Bartram) Ignatov Eurhynchium vagans (A. Jaeger) E. B. Bartram Homalotheciella subcapillata (Hedw.) Broth. Homalothecium aureum (Spruce) H. Robinson Homalothecium fulgescens (Mitt.) Lawton Homalothecium laevisetum Sande Lac. AY184788 AY184769 AF503538 AF397862 AF417450 AF403652 Homalothecium H. Robinson Homalothecium H. Robinson Homalothecium A. Jaeger Homalothecium Schimp. Homalothecium Voucher specimen for sequences lutescens (Hedw.) AF397827 AF417379 AF403642 megaptilium (Sull.) AF397842 AF417431 AF403597 China, Hunan Prov., T. Koponen, S. Huttunen, S. Piippo & P.-C. Rao 49717, September 28, 1998 USA, Missouri, B. Allen & P.L. Redfearn, Jr., March 29, 1995 (NY) Turkmenia, I.B. Sirotina & A. Abramova, November 22, 1983 (MHA) Canada, British Colombia, H. Streiman 43136, August 20, 1989 China, Hunan Prov., T. Koponen, S. Huttunen, S. Piippo & P.-C. Rao 54066, July 30, 1998 Czech Republic, Central Bohemia, J. Vana & J. Enroth, Apri.30, 1999 USA, Oregon, A. Newton 5206, May 30, 1992 (BM) nuttallii (Wilson) AF397798 AF417399 AF403589 USA, California, Arcata, M. Ignatov, August 5, 1989 philippeanum (Spruce) AF397863 AF417451 AF403647 Georgia, Caucasus, V. Vasak B-90123, July 31, 1982 sericeum (Hedw.) Schimp. AF397805 AF417349 AF403587 Finland, Regio aboënsis, T. Koponen & S. Huttunen 1322, May 24, 1999 Finland, Regio aboënsis, T. Koponen & S. Huttunen 1320, May 24, 1999 Isothecium alopecuroides (Dubois) Isov. AF397878 AF417465 AF403645 AF397860 AF417462 AF403658 AF397876 AF417463 AF403650 AF397877 AF417466 AF403637 AF397820 AF417351 AF403616 AY044065 AF417353 AF395636 S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 Table 1 Continued Species trnL-F psbT-H ITS2 Kindbergia praelonga (Hedw.) Ochyra AF397804 AF417385 AF403591 Kindbergia praelonga (Hedw.) Ochyra ‘‘serricuspis’’ Kindbergia praelonga (Hedw.) Ochyra AF397818 AF417381 Lindigia debilis (Mitt.) A. Jaeger AF397803 AF417376 Meteoridium remotifolium (Müll. Hal.) Manuel Myuroclada maximowiczii (G. G. Borszcz.) Steere & W. B. Schofield Palamocladium euchloron (Müll. Hal.) Wijk & Margad. Palamocladium leskeoides (Hook.) E. Britton Platyhypnidium austrinum (Hook. & Wilson) M. Fleisch. Platyhypnidium patulifolium (Cardot & Thériot) Broth. Platyhypnidium riparioides (Hedw.) Dixon AF397833 AF417418 China, Hunan Prov., T. Koponen, S. Huttunen, S. Piippo & P.-C. Rao 50467, October 7, 1998 AF403590 China, Hunan Prov., T. Koponen, S. Huttunen, S. Piippo & P.-C. Rao 53561, September 29, 1998 AY184794 Colombia, Urrao, S. P. Churschill, I. Sastre-de Jesus & M. Escoba Acosta 13418, July 1, 1985 AF403617 Ecuador, Provincia de Napo, D. H. Norris 90677 & Eduardo Barahona M., December 5, 1997 AF403601 Mexico, A. Newton 4399 (BM) AF397779 AF417392 AF403625 China, Hunan Prov., T. Koponen, S. Huttunen, S. Piippo & P.-C. Rao 53505, September 29, 1998 AF397851 AF417416 AF403623 Russia, Caucasus, M. Ignatov, August 23, 1999 (MHA) AF397791 AF417387 AF403596 AF397850 AF417443 – AF397797 AF417470 AF403663 AF397854 AF417423 AF403600 Rhynchostegiella macilenta (Renauld & Cardot) Cardot Rhynchostegiella tenella (Dicks.) Limpr. Rhynchostegiella teneriffae (Mont.) Dirkse & Bouman Rhynchostegium confertum (Dicks.) Schimp. Rhynchostegium murale (Hedw.) Schimp. AF397781 AF417370 AF403570 AY044070 AF417400 AF397783 AF417368 AF395633 AF403569 AF397837 AF417413 AF403622 AF397817 AF417454 – Rhynchostegium murale (Hedw.) Schimp. Rhynchostegium murale (Hedw.) Schimp. AF397816 AF397775 AF417355 AF417424 AF403624 AF403661 Rhynchostegium pallidifolium (Mitt.) A. Jaeger Rhynchostegium psilopodium Ignatov & Huttunen Rhynchostegium rotundifolium (Brid.) Schimp. Rhynchostegium serrulatum (Hedw.) A. Jaeger Rozea subjulacea Besch. AF397807 AF417371 AF403618 AF397861 AF417452 AF403643 AF397809 AF417396 AF403611 China, Hunan Prov., T. Koponen, S. Huttunen, S. Piippo & P.-C. Rao 53870, October 1, 1998 Russia, Ural Mts., E. Baisheva, July 9, 1993 (MHA) Russia, Perm Prov., A. G. Bezgodov #222, June 30, 1999 (MHA) China, Hunan Prov., T. Koponen, S. Huttunen, & P.-C. Rao 51301, October 3, 1997 China, Hunan Prov., T. Koponen, S. Huttunen & P.-C. Rao 51803, October 8, 1997 Russia, Caucasus, V. Onipchenco, 31. August 1999 (MHA) AF397829 AF417380 AF403620 USA, New Jersey, B. C. Tan, September 12, 1992 AF397796 AF417366 AF403578 Rozea andrieuxii (Müll. Hal.) Besch. AF397839 AF417409 AF403579 Scleropodium obtusifolium (Mitt.) Kindb. Scorpiurium circinatum (Brid.) M. Fleisch. & Loeske Scorpiurium deflexifolium (Solms.) M. Fleisch. & Loeske Squamidium brasiliense (Hornsch.) Broth. AF397793 AF397834 AF417434 AF417410 AF403615 AF403598 Venezuela, Estado Mérida, Dana Griffin & Manuel Lopez No. PV-904, February 19, 1985 (NY) Dominican Republic, Prov. San Juan, W. R. Buck 14205, April 28, 1982 (NY) USA, California, M. Ignatov & D. H. Norris, August 13, 1989 Georgia, Caucasus, M. Ignatov, August 8, 1987 (MHA) AF397844 AF417407 AF403599 AY044063 AF417393 AF395637 Unclejackia longisetula (E. B. Bartram) Ignatov, T. J. Kop. & D. H. Norris Zelometeorium patulum (Hedw.) Manuel AF397794 AF417357 AF395643 AF397787 AF417362 – Plasteurhynchium striatulum (Spruce) M. Fleisch. Pseudoscleropodium purum (Hedw.) M. Fleisch. Rhynchostegiella brachypodia M. Fleisch. AY184795 – Voucher specimen for sequences China, Hunan Prov., T. Koponen, S. Huttunen & P.-C. Rao 52596, October 10, 1997 AY184791 AY184773 AY166449 Australia, Victoria, H. Streimann 49544 China, Hunan Prov., T. Koponen, S. Huttunen, S. Piippo & P.-C. Rao 53920, October 1, 1998 AF397847 AF417411 – China, Hunan Prov., T. Koponen, S. Huttunen, & P.-C. Rao 51843, October 9, 1997 AY184792 AY184774 AY166450 Georgia, Caucasus, M. Ignatov, August 1, 1987 (MHA) Czech Republic, Central Bohemia, J. Vana & J. Enroth, April 30, 1999 Papua New Guinea, Morobe Prov., T. Koponen 33007, July 11, 1981 Portugal, Madeira, L. Hedenäs, May 13, 1993 (B4503) (S) Georgia, Caucasus, M. Ignatov, October 2, 1997 (MHA) Portugal, Madeira, S. Fontinha, L. Hedenäs, M. Nobrega, June 10, 1991 (B9121; S) Georgia, Caucasus, M. Ignatov, August 27, 1987 (MHA) Portugal, Madeira, L. Hedenäs & I. Bisang, November 21, 1999 (B22455; S) Tanzania, Arusha District, T. Pócs, R. Ochyra & H. Bednarek-Ochyra with C. N. Lema 88161 ⁄ R, June 29, 1988 Papua New Guinea, Morobe Prov., T. Koponen 32496, July 8, 1981 Ecuador, Loja Prov., D. H. Norris 91902, M. Bolivar & M. Aguirre, December 16, 1997 155 156 S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 Table 1 Continued Species Meteoriaceae Aerobryidium filamentosum (Hook.) M. Fleisch. Barbella flagellifera (Cardot) Nog. Chrysocladium retrorsum (Mitt.) M. Fleisch. Duthiella wallichii (Mitt.) Müll. Hal. Floribundaria floribunda (Dozy & Molk.) M. Fleisch. Meteorium polytrichum Dozy & Molk. Pseudospiridentopsis horrida (Cardot) M. Fleisch. Trachycladiella aurea (Mitt.) M. Menzel Trachypus bicolor Reinw. & Hornsch. Lembophyllaceae Dolichomitriopsis diversiformis (Mitt.) Nog. Lembophyllum clandestinum (Hook. & Wilson) Lindb. Amblystegiaceae Amblystegium serpens (Hedw.) Schimp. Donrichardsia macroneuron (Grout) H. A. Crum & L. E. Anderson Campyliaceae Campylium stellatum (Hedw.) C. E. O. Jensen Hylocomiaceae Hylocomium splendens (Hedw.) Schimp. Meteoriella soluta (Mitt.) S. Okamura Rhytidiadelphus triquetrus (Hedw.) Warnst. Hypnaceae Ctenidium molluscum (Hedw.) Mitt. Hypnum cupressiforme Hedw. Leskeaceae Leskea polycarpa Hedw. Okamuraea brachydiction (Cardot) Nog. Leucodontaceae Leucodon sciuroides (Hedw.) Schwägr. Myriniaceae Helicodontium capillare (Hedw.) A. Jaeger trnL-F psbT-H ITS2 Voucher specimen for sequences AF397789 AF417347 AF403613 China, Yunnan Prov., T. Koponen 37908, October 25, 1981 AY044071 AF417360 AF395630 China, Hunan Prov., T. Koponen, S. Huttunen, S. Piippo & P.-C. Rao 56995, October 13, 1998 AY044061 AF417390 AF395626 China, Hunan Prov., T. Koponen, S. Huttunen, S. Piippo & P.-C. Rao 55572, October 6, 1998 AF397782 AF417375 AF403612 China, Hunan Prov., T. Koponen, S. Huttunen, S. Piippo & P.-C. Rao 54168, October 1. 1997 AF397800 AF417428 AF403580 Papua New Guinea, West Sepik Prov., T. Koponen 35719, August 7, 1981 AY044073 AF417350 AF395629 Australia, Queensland, H. Streimann 57477, October 22, 1995 AF397801 AF417383 AF403571 China, Hunan Prov., T. Koponen, S. Huttunen, S. Piippo & P.-C. Rao 55834, October 11, 1998 AF397778 AF417397 AF403573 China, Hunan Prov., T. Koponen, S. Huttunen, S. Piippo & P.-C. Rao 55508, October 6, 1998 AY044060 AF417369 AF395624 China, Hunan Prov., T. Koponen, S. Huttunen, S. Piippo & P.-C. Rao 50721, October 2, 1997 AF397777 AF417469 AF403664 Russian Far East, Kunashir Island,V. A. Nedoluzhko, July 18, 1990 (MHA) AF397823 AF417374 AF403630 New Zealand, South Island, D. H. Vitt 29644, December 22, 1981 AF397836 AF417420 AF403633 Finland, Nylandia, S. Huttunen 1416, April 16, 2000 AY009848 – AF167350 USA, Redfearn 27208 (Sequences from GenBank, Vanderpoorten et al., 2002) AF397821 AF417358 AF403609 Finland, Regio aboënsis, J. Pykälä 7033, October 30, 1990 AF397840 AF417408 AF403610 Finland, Tavastia australis, S. Huttunen 1441, April 14, 2000 AF397808 AF417389 AF403606 China, Hunan Prov., T. Koponen, S. Huttunen & P.-C. Rao 49610, October 1, 1997 AF397811 AF417425 AF403631 Finland, Tavastia australis, S. Huttunen 1440, April 14, 2000 – AF417414 AF403632 Finland, Regio aboënsis, J. Pykälä 8706 & J. Nurmi, August 22, 1991 AF397812 AF417361 AF403607 Finland, Nylandia, S. Huttunen 1438, November 1999 AF397810 AF417367 AF403604 Finland, Regio aboënsis, J. Pykälä 10872, September 15, 1992 AY184789 AY184771 AF503537 China, Hunan Prov., T. Koponen, S. Huttunen, & P.-C. Rao 48969, September 26, 1997 AF397786 AF417398 AF403634 Russia, Karelian Republic, S. Huttunen & H. Wahlberg 819, June 27, 1997 AF397855 AF417435 – Colombia, Departamento del Valle, S. Churchill, A. E. Franco & N. Hollaender, May 4, 1990 Plagiotheciaceae Plagiothecium denticulatum (Hedw.) Schimp. AF397845 AF417419 AF403635 Finland, Karelia australis, S. Huttunen 1398, May 23, 2000 *ITS3 sequence instead of primers ITS3 and ITS4 primers 5.SR and LC4-R were used for PCR and sequencing. when a genus exhibited more or less contrasting infrageneric units (either recognized in the taxonomic literature or observed by us in previous studies of the family), we tried to include representatives of each those entities. Type species of the genera were included whenever possible, in order to make subsequent taxonomic conclusions better founded. Several samples were included for morphologically highly variable species: Brachythecium salebrosum (Web. & Mohr.) Schimp., Bryhnia novae-angliae (Sull. & Lesq.) Grout, Eurhynchium vagans (A. Jaeger) E. B. Bartram, Kindbergia praelonga (Hedw.) Ochyra, Rhynchostegium murale (Hedw.) Schimp. These samples were usually taken from different geographic regions. Outgroup species were S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 selected based on earlier analyses of larger data sets (Huttunen et al., 2004). Based on this study, Meteoriaceae is sister to Brachytheciaceae. Both of these taxa are nested within a monophyletic Hypnales (Buck et al., 2000a). In addition to Meteoriaceae species, some outgroup species were included, representing other presumed closely related Hypnales families, such as Amblystegiaceae, Hylocomiaceae, Hypnaceae, Lembophyllaceae, Plagiotheciaceae, and Leucodontaceae (Table 1). None of the earlier phylogenetic analyses suggested a close relationship between Brachytheciaceae and Leskeaceae, and Leskea polycarpa Hedw. was used as an outgroup in direct optimization analyses. Morphological data The morphological data set included 63 characters (48 gametophytic and 15 sporophytic). It was mainly compiled using the specimens employed in DNA analysis (Table 1), and supplemented with a study of herbarium material available in the Finnish Museum of Natural History, Helsinki (H), particularly for the sporophytic characters. For these characters, data were supplemented with specimens in areas that are closely situated geographically to the sequenced specimen. If additional material was not available, the sporophytic characters were left unknown. We found most of the available morphological characters to vary considerably on the generic level, and in some cases even on the species level, and hence we doubted the information content of the morphological data in phylogenetic analyses of Brachytheciaceae. For morphologically variable species (Brachythecium salebrosum, Bryhnia novae-angliae, Eurhynchium vagans, Kindbergia praelonga, Rhynchostegium murale), we coded gametophytic character states only according to the specimens used for molecular study. Variability in some characters in a few specimens made character coding complicated, and we could not therefore avoid including some polymorphic characters in the data set. Due to difficulties in including data on these characters in POY analyses, all polymorphisms were coded as ‘‘?’’. The quantitative characters used in this data matrix were mostly for separating extreme cases, such as species with very narrow leaves (see Appendix 1, character 20) or very long cells (Appendix 1, character 35). Molecular data We used the information from three, mainly noncoding, DNA sequence regions for the phylogeny reconstruction: chloroplasts psbT-H and trnL-F, and nuclear ITS2. These were sequenced using primers C and F (Taberlet et al., 1991) for the trnL-F region, universal primers ITS3 and ITS4 (White et al., 1990) for the ITS2 region, and primers psbT and psbH (Hong 157 et al., 1995; Shaw, 2000) for the psbT-H region. Sequenced psbT-H region consisted of partial psbT (102 bp) and psbH (108 bp), genes and the whole psbN gene (132 bp) between them, which are all coding for the small proteins of photosystem II. Between these genes there are two non-coding spacers, psbT-N and psbN-H, whose lengths vary among Bryopsida between 67–95 bp and 88–105 bp (Quandt et al., 2003). In the trnL-F region, length variation is concentrated in trnL intron in mosses, which is situated within the ribosome coding tRNA-Leu gene. Compared to some other groups of land plants, in mosses the intron is relatively short. For example, Stech et al. (2003) reported a mean length of 308 bp (SD 59.4) for a data set including 24 mosses from different groups of Bryopsida. Some indel events also occur in the non-coding trnL-F spacer, whose length in mosses is 82 bp (SD 41.3) according to Stech et al. (2003). In addition to these two non-coding fragments, our analyses included the 50 bp long end of the tRNALeu gene between them. The nuclear Internal Transcripted Spacer 2 (ITS2) is situated between the ribosome coding the 5,8S and 26S genes. For example, in the data set including 66 mosses from the subclass Bryidae, where Brachytheciaceae also belongs, the length of the ITS2 region varied between 251 and 360 bp (mean 282.8, SD 25.41; Quandt et al., 2004b). DNA extraction, PCR and sequencing The isolation of DNA was mainly carried out according to the CTAB-technique described in Doyle and Doyle (1990), or in some cases using a Nucleospin Plant DNA Extraction Kit (Machery-Nagel), following the manufacturer’s instructions. Most of the specimens used for the DNA extraction were herbarium specimens, except for some species commonly found in southern Finland. Polymerase chain reactions and cycle sequencing reactions were performed using PTC-100 and PTC200 Thermocyclers (MJ Research). The 25-lL reactions contained 1.5 U Taq DNA polymerase (Amplitaq Gold, Applied Biosystems), 1 · buffer, 2.5 mmol MgCl2, 0.3 mmol dNTPs-mix and 0.4 lmol of each primer. The PCR products were purified using a GFX PCR purification kit (Amersham Pharmacia). For the cycle sequencing, 8-lL reactions were used, containing 2.2 lL of Big Dye Terminator Cycle Sequencing premix (Applied Biosystems), 1.3 lL of 10 lmol primer, 3.8 lL H2O and 0.7 lL purified PCR product. The sequencing products were purified by eluting through columns filled with 450 lL 5% Sephadex solution (Amersham Pharmacia). An ABI PRISM 377 automated sequencer was used to analyse the sequencing products. All PCR products were sequenced using two primers. All sequences were submitted to GenBank, and a list of their accession numbers and voucher specimens is given in Table 1. 158 S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 Phylogenetic analyses Phylogenetic analyses were performed for two data sets, first for a wider selection of taxa (124 terminals) to test for the monophyly of the family. Due to the wide sequence length variation within this large data set, the effects of possible differences in character optimization obtained with POY were tested with a smaller data set (99 terminals) that included only species of Brachytheciaceae and Floribundaria floribunda (Dozy & Molk.) M. Fleisch. (Meteoriaceae) as the outgroup. Simmons and Freudenstein (2003) showed how the increasing genetic distance within data causes errors in alignment if programs such as ClustalX (Thompson et al., 1997) are used, but how this possibly affects direct optimization has not been explored. The ITS2 region has often been stated to be too variable at higher taxonomic levels, and in phylogenetic studies on pleurocarpous mosses based on manual alignments some sequence fractions are often left out due to ambiguity (e.g., Quandt et al., 2000, 2004a,b). Direct optimization, as implemented in POY (Gladstein and Wheeler, 2001), was used. Alignment and phylogenetic analysis are performed simultaneously in direct optimization and several different hypotheses of character transformation were tested during the analyses by optimizing characters. This implies that the resulting set of most parsimonious trees may consist of several equally most parsimonious trees that are each based on different character optimizations, unlike the situation in all other phylogenetic methods. As a result of this, the definitions for homology also differ from those of traditional methods (Giribet et al., 2002). The only homology statement made prior to phylogeny reconstruction is that the submitted sets of sequences are assumed to represent the same genomic region in all taxa. In contrast to static alignment methods, the most parsimonious hypothesis of character transformation between sequences is obtained as a result of analyses performed in POY, and those results are presented in the form of an implied alignment. Hence, if ordinary static alignments are regarded as primary homology statements sensu De Pinna (1991), implied alignments obtained from POY describe true synapomorphies and can thus be considered secondary homology statements (sensu De Pinna). In static alignments, the nucleotides that are expected to be homologous are placed in the same column, and running cladistic analyses tests this hypothesis of homology. POY implied alignments, however, result from the analyses, and the synapomorphies that they present between studied sequences have already been tested in the preceding direct optimization analyses. Phylogenetic analyses were performed using a parallel version of POY with eight processors of the Silicon Graphics Origin 2000 ⁄ 128 cluster in the Center of Scientific Computing, Espoo, Finland. Prior to POY analysis, the sequences were aligned using ClustalV (Higgins et al., 1991) and then split using Winclada (Nixon, 1999a) into shorter fragments within conserved regions to save computation time and memory (Gladstein and Wheeler, 2001). The command line used in all POY analyses is given in Appendix 2. Leskea polycarpa Hedw. was used as an outgroup for larger data sets and Floribundaria floribunda for smaller sets. Implied alignments reported by POY were always saved and checked afterwards to avoid obvious errors in homology. We found the hypothesis of homology in POY character optimizations highly improbable in two coding regions, and hence an extra 6 bp CGCGCG repeat was added prior to analyses at the beginning of the psbT-N and trnL-F spacers (see Quandt et al., 2004b, for more information). We are aware that the strict consensus trees of POY topologies may include branches that are only supported by one character optimization. Unfortunately, checking parsimonious trees against original data matrices and removing ambiguously supported branches may be a fairly complicated task if the direct optimization analysis results in more than one topology and character optimization. Since the parsimonious trees obtained from analysis may be based on different character optimizations, removing these nodes would demand the study of all POY topologies together with the character optimization they are connected with, and this should be done using some other program capable of handling substitution matrices, such as PHAST (Goloboff, 1996). To compare the effects of information from gaps on topology, POY character optimizations were used as static alignments and analyzed using the program Nona (Goloboff, 1994) within the Winclada (Nixon, 1999a) shell, and unlike in POY, the gaps were treated as missing data. In general, the use of POY character optimizations as static alignments suffers from fundamental weaknesses, and this method should not be used as a standard method for phylogenetic analysis. Despite the previous extensive discussion on the importance of information from gaps (for example Giribet and Wheeler, 1999; Simmons and Ochoterena, 2000; Bapteste and Philippe, 2002; Kawakita et al., 2003), we still wanted to see if (and to what extent) they affected the topology in our study. One method for doing this is to simply compare the topology obtained from POY and Nona. For Nona, a multi-ratchet option with three sequential ratchet runs was used (Nixon, 1999b). Each replicate included 200 iterations, and 20 trees were held in memory during the iterations. During the ratcheting, 20% of the characters were resampled. Topological differences were studied by comparing these Nona topologies with those from the POY. Since each POY run most often resulted in more than one character optimization, they all were analyzed in Nona to deter- S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 mine if the parsimonious trees based on these analyses were of the same length despite the differences in treatment of the gaps. Only phylogenies with the shortest tree lengths were examined with ILD metrics and included in comparisons of tree topology between the results of POY and Nona analyses, but for comparisons of parsimony informative (pi) characters in data sets, suboptimal character optimizations were also included. For the most parsimonious Nona trees, Bremer support (Bremer, 1994) for clades was calculated in Nona by saving suboptimal trees in a series of up to 15 steps longer than that of the most parsimonious topology. During the search, up to 10 000 trees were saved from 300 replications and in each replication 20 different starting trees were used. Jackknifing was performed with Nona within the Winclada shell. In jackknifing 1000 replications, including 10 searches, were performed, and in each replication 20 starting trees were used. These trees were submitted for additional tree bisection-reconnection (TBR) branch swapping (do max*). Quandt et al. (2003) showed that the 9 bp short loop region in the psbT-N spacer region is inverted in some mosses, and that this inversion appears to have occurred independently several times in different lineages. The authors stated that the original and inverted reverse complement forms cannot be regarded as homologous positions, and inclusion of this region in the original form in phylogenetic analyses could negatively affect phylogeny reconstruction. They further concluded that inclusion of this region in the original form will also lead to weighting of the inversion event. We tested the effect of including this region on phylogeny reconstruction by analyzing data in three different ways using the 124-terminal data set. First we included the 9 bp region and treated it as we did all other sequences (Table 2). Despite the inversion event in some taxa, the loop region appeared as a homologous character in POY character optimiza- Table 2 Base composition of 9-bp inverted loop region in psbT-N spacer in brachytheciacean species and alternative coding method tested in analyses. Alternative coding includes absence ⁄ presence (0 ⁄ 1) coding for inversion event in loop region and reverse complement sequence of original 9 bp in loop region in those terminals with A-rich type of inversion (Rhynchostegielloideae type) Common type Brachythecium type Sciurohypnum type Rhynchostegielloideae type or Original Reverse complement or original sequence +0 ⁄ 1 coding for inversion event TTTATTTTA CTCATTTTA TTCATTTTA CAAAATAAA TTTATTTTA 0 CTCATTTTA 0 TTCATTTTA 0 TTTATTTTG 1 TAAAATAAA TTTATTTTA 1 159 tions. In another analysis the loop positions were excluded, and in the final analysis the A-rich forms of the loop region were transformed into reverse complement sequences and included. The information on the presence ⁄ absence of the inversion event was included in analyses as a separate character matrix. The last method removed the effect of the additional weighting of the inversion event, but still could not avoid possible erroneous expectations of the positional homology when both the uninverted and inverted loop regions are placed in POY character optimizations as homologous positions. Sensitivity analyses and weighting of morphological data Sensitivity analysis (sensu Wheeler, 1995) has become somewhat a standard way of exploring the data in direct optimization analyses. Sensitivity analysis aims to explore the effect of the different substitution costs on the stability of the obtained hypothesis of relationships and on congruence among data partitions. These results identify the most stable tree hypotheses, presenting the groups shared by all analyses with differing gap costs, or selecting the optimal topology using congruence as an optimality criterion (see Giribet, 2003). This method has recently been criticized (Frost et al., 2001; Grant and Kluge, 2003). The increased congruence identified with the ILD or RILD metrics does not make the less parsimonious results more acceptable (see Grant and Kluge, 2003), and in this paper we consider the equal weighting of all data to be the best fitting result. Due to the common practice of applying sensitivity analyses in direct optimization analyses, we explored the impact of weighting gaps and morphology on the output of POY, and the contribution of data partitions in differentially weighted analyses. Three different gap costs (1, 2 and 4) and two weighting methods for morphological data were used. This also provides for the possibility of comparing this research to other studies utilizing the direct optimization method and sensitivity analysis. For all weighting methods, data were analyzed as combined, and additionally the morphological data, chloroplast (psbT-H and trnL-F) and nuclear (ITS2) regions were analyzed separately. These separate analyses were used when evaluating the impact of different data partitions on the topology obtained from combined analyses and when calculating the ILD metric. The congruence among data sets in the different analyses was estimated using ILD metrics (Mickevich and Farris, 1981; Farris et al., 1995). ILD utilizes tree lengths of combined and separately analyzed data sets (ILD ¼ (tree length of combined data set – S tree lengths of individual data sets) ⁄ tree length of combined data set). Morphological data were either always weighted as one or as the same weight as determined for indel events. In direct optimization analyses (Wheeler, 1996) of 160 S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 combined molecular and morphological data, the latter have often been given the same weight as the indel events (for example, Giribet et al., 2002; Schulmeister et al., 2002). This practice follows Wheeler and Hayashi’s (1998) study of chelicerates, in which they found that an equal weighting of morphology and indel events gave the most congruent results. The method is also said to avoid the possible overwhelming effects of the molecular data on the usually much smaller morphological matrix. The effects of two different weighting methods on morphological data were tested with ILD metric (Mickevich and Farris, 1981; Farris et al., 1995). The method in which the morphology was given a weight of one appeared to lead to lower ILD values and was therefore used in all analyses testing the effect of inversion in the psbT-N region. Results Sequence variation and information content of data partitions The ITS2 region showed the largest sequence length variation (Table 3). The length of the sequenced ITS2 fragments varied in the 124 terminal data set between 340 and 415 bp, and in the 99 terminal data set between 352 and 406 bp. In the psbT-H region, all the length variation was concentrated in the psbT-N spacer, where the length varied between 66 and 94 bp in the large data set and between 79 and 90 bp in the smaller one. In the trnL-F region, lengths of the non-coding sequences trnL intron and trnL-F spacer varied, while the ribosomal coding trnL and trnF regions showed no length variation. The trnL intron of 256–288 bp constitutes the majority of the total length of the trnL-F region. In the trnL-F spacer the 17 bp and 9 bp long insertions at the beginning of the spacer in two species, Trachypus bicolor and Brachythecium frigidum, constitute the majority of the length variation in this region. If POY implied alignments from analyses with a gap cost equal to one are studied as static alignments, and gaps treated as missing data, the number of pi base positions in both chloroplast regions together was about the same as in the nuclear ITS2 region (Tables 3 and 4). In this implied alignment, 17.8% of positions in the 863 bp long ITS2 were pi, while both in the 558 bp long psbT-H and in the 552 bp long trnL-F proportions of the pi positions were only 13.8%. However, if the proportions of pi positions in each non-coding and coding region in trnL-F and psbT-H are studied alone, the psbT-N spacer (24.4% of positions pi) and trnL intron (21.1%) are the most informative regions. If the 9 bp inversion in the psbT-N is excluded, the proportion of pi positions decreases from 24.4 to 21.1%. The high information content of the ITS2 region clearly affected the resolution when we compared tree topologies based on separate analyses of the nuclear ITS2 and chloroplast data. Within most of the genera in the Brachytheciaceae, the sequence variation of trnL-F and psbT-H was extremely low, and most of the information for resolving relationships inside the genera came from the ITS2 region. Genus- and species-level relationships were best resolved in the strict consensus trees based on ITS2, but at the family level Brachytheciaceae was not resolved as monophyletic. Separate analyses of the nuclear ITS2, chloroplast psbT-H and trnL-F, and morphological data resulted in clearly different topologies (not shown). Most of the generic groupings in ITS2 trees closely resembled those obtained from analyses of combined data. Some problematic placements of species in combined analyses also appeared to originate from ITS2 data. The genus Palamocladium Müll. Hal. was monophyletic in all analyses based on chloroplast data, while ITS2 analyses resolve P. leskeoides (Hook.) E. Britton as sister to the Chinese Bryhnia novae-angliae. In the latter analyses, P. euchloron (Müll. Hal.) Wijk & Margad. was placed close to Plasteurhynchium striatulum Table 3 Sequence length variation on chloroplastic psbT-H and trnL-F, and nuclear ITS2 regions in 124- and 99-terminal data set, and number of parsimony informative (pi) characters when gaps are treated as missing data. In genomic regions marked with asterisk (*) number of pi characters varies between different POY implied alignments, depending on gap costs and data sets used in POY analyses; see Table 4 psbT-H trnL-F psbT coding psbN psbH coding region psbT-N coding psbN-H region trnL (partial) spacer region spacer (partial) intron Length variation 102 —124-terminal data set (bp) Number of pi characters 13 Length variation—99-terminal 102 data set (bp) Number of pi characters 102 trnL trnF coding coding trnL-F region ITS2 region spacer (partial) 66–94 132 97 108 256–288 50 51–80 40 340–415 * 79–90 14 132 12 97 9 109 * 0 256–281 50 * 51–80 1 40 * 352–406 10 9 8 * * * 0 0 * 161 S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 Table 4 Number of characters and parsimony informative positions (pi, gaps treated as missing data) in POY character optimizations for genomic regions with variable sequence length. Analyses were performed separately for 124 terminal data set (98 Brachytheciaceae taxa and 26 outgroup species from other pleurocarpous moss families) and 99 terminal data set (98 Brachytheciaceae taxa and Floribundaria floribunda, Meteoriaceae, as outgroup) Number of taxa Ti ⁄ Tv ⁄ gap psbT-N loop 1:1 pi 1:1 pi 1:1 pi 1:1 pi 1:1 pi 1:1 pi 1:1 pi 1:1 pi 1:1 pi 1:1 pi 1:1 pi 1:1 pi :1 included 99 :2 included 99 :4 included 99 :1 included 124 :1 excluded 124 :1 rev.compl. + 0 ⁄ 1 coding 124 :2 included 124 :2 excluded 124 :2 rev.compl. + 0 ⁄ 1 coding 124 :4 included 124 :4 excluded 124 :4 rev.compl. + 0 ⁄ 1 coding 124 (Spruce) M. Fleisch and Eurhynchium. Consensus trees based on analyses of chloroplast trnL-F and psbT-H regions were much less resolved than topologies based on ITS2, especially with regard to the relationships between genera of Brachytheciaceae. The analyses of morphological data alone resulted in a topology with only poor resolution (not shown), and Brachytheciaceae was not monophyletic. Despite this, morphology appeared to be important, especially in cases where taxa within one species or genus showed wide sequence variation. For example, Rhynchostegium murale, Kindbergia praelonga, and Palamocladium always appeared to be polyphyletic in analyses including only molecular data (trnL-F, psbT-H and ITS2) but they were monophyletic in at least the most parsimonious analyses, including morphology (Fig. 1). In some cases sequence variation did not follow the taxonomic species concept, which suggests that the delimitation of some species is in need of re-evaluation. In Bryhnia novaeangliae, even the inclusion of morphological data in analyses did not resolve this species as a monophyletic entity (Fig. 1). Sequences of Bryhnia novae-angliae from southern China were much closer to those of Brachythecium frigidum (Müll. Hal.) Besch., while two other Bryhnia novae-angliae specimens (from European and Asian boreal regions of Russia) shared several mutations in both trnL-F and ITS2. This clearly explains the position of these terminals in cladograms. In all other ITS2 trnL intron trnL-F spacer psbT-N spacer 594 or 596 99 or 101 540–543 108 or 110 524 132 863 154 864–868 161–163 864 151 798 186 764 or 772 185 or 191 797 187 695–697 235 700 233–234 704 236–237 301 30 301 30 291 32 353 52 353 52 353 52 335 52 341 52 335 52 329 53 330 54–55 330 55 94 12 94 12 94 12 109 23 109 23 109 23 109 23 109 23 109 23 109 23 109 23 109 23 104 or 105 25 103 25 102 25 119 29 109 23 109 28 116 or 117 29 107 24 116 28 116 29 108 23 117 28 analyses the Chinese specimen was resolved to the same clade with Brachythecium frigidum, while both Northern Eurasiatic Bryhnias always formed a clade (Figs 1, 2 and 3). Effect of inversion in psbT-N spacer loop region Topologies obtained using direct optimization were in some details affected by the inversion in the psbT-N loop, and it had a minor effect on the congruence between data sets (Table 5). In the most parsimonious analyses with a gap cost equal to one, the inclusion of the loop region in original form resulted in the highest congruence between data partitions. Despite the short length, 6 out of a total of 9 bp in the loop are pi when this region is included in analyses in the original form. Even in reverse-complement form (A-rich inversions reverse-complemented in T-rich form, see Table 2) five pi positions remained. These positions appear to be very informative in some clades within Brachytheciaceae. Species with rarer A-rich forms of inversion are concentrated in Rhynchostegielloideae and the Squamidium clade, which includes Squamidium brasiliense (Hornsch.) Broth., Meteoridium remotifolium (Müll. Hal.) Manuel, and Zelometeorium patulum (Hedw.) Manuel. The latter clade appears in some analyses as sister to Rhynchostegielloideae (Figs 1a, 1c, 2 and 3, Table 6). When the loop region is excluded, the position 162 S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 Table 5 Lengths of the most parsimonious trees for analyses of chloroplastic (trnL-F + psbT-H), nuclear (ITS2), combined molecular (trnL-F + psbT-H + ITS2), and combined molecular + morphological data from 124 terminals with different gap costs and analysis methods, and congruence among data partitions (ILD) in these analyses. Nona analyses are based on character optimizations obtained from direct optimization, but gaps are treated as missing data Gap cost ⁄ Weight psbT-N Morphology ILD No. of spacer treatment for Chloroplast Nuclear Molecular & molecular ILD for for all terminals loop region of gaps morphology DNA DNA Morphology combined combined molecular data POY 1 : 1 : 1 1:1:2 1:1:4 99 99 99 POY 1 : 1 : 1 124 1:1:1 124 1:1:1 124 1 1 1 1 : : : : 1 1 1 1 : : : : 2 2 2 2 124 124 124 124 1 1 1 1 : : : : 1 1 1 1 : : : : 4 4 4 4 124 124 124 124 Nona 1 : 1 : 1 124 1:1:1 124 1:1:1 124 1:1:2 124 1:1:2 124 1:1:2 124 1:1:4 124 1:1:4 124 1:1:4 124 included included included 1 2 4 1 1 1 536 671 893 767 1078 1592 525 525 525 included excluded rev.compl. + 0⁄1 coding included included excluded rev.compl. + 0⁄1 coding included included excluded rev.compl. + 0⁄1 coding 1 1 1 1 1 1 879 847 866 1392 1392 1392 799 799 799 2541 3515 3500 3507 0.106 0.126 0.132 0.128 2 2 2 2 1 2 1 1 1102 1102 1067 1090 2156 2156 2156 2156 799 1598 799 799 3508 4486 5415 4428 4458 0.071 0.096 0.103 0.092 0.093 4 4 4 4 1 4 1 1 1466 1466 1404 1413 3195 3195 3195 3195 799 3196 799 799 5035 6076 8836 5992 6038 0.074 0.101 0.111 0.099 0.105 1 636 765 799 2471 0.096 1 606 772 799 2402 0.094 1 617 759 799 2405 0.096 1 643 951 799 2669 0.103 1 615 958 799 2637 0.100 1 623 954 799 2643 0.101 1 644 1268 799 3016 0.101 1 627 1259 799 2963 0.094 1 636 1265 799 2990 0.097 included missing data excluded missing data rev.compl. missing + 0⁄1 data coding included missing data excluded missing data rev.compl missing + 0⁄1 data coding included missing data excluded missing data rev.compl. missing + 0⁄1 data coding 2130 2608 3422 0.142 0.128 0.120 Fig. 1. Strict consensus trees from direct optimization analyses of combined morphological and molecular data including 124 terminals with a gap cost of 1: (a) strict consensus of six most parsimonious trees (L ¼ 5512), 9-bp inversion in psbT-N loop region included in original form (see Table 2); (b) strict consensus of three most parsimonious trees (L ¼ 3500), psbT-N loop region excluded from analyses, and (c) strict consensus of A-rich type of 9-bp loop region reverse-complemented and information on presence ⁄ absence of inversion event included in analyses, strict consensus of four trees (L ¼ 3507). Branches leading to taxa with A-rich type of psbT-N loop region presented with gray color (this region is lacking in Donrichardsia macroneuron). The character evolution of two characters, seta surface structure (marked with square, 0 ¼ setae smooth, 1 ¼ setae rough) and operculum shape (triangle, 0 ¼ operculum rostrate, 1 ¼ operculum conic) is presented in (a). For character optimization of these characters the unambiguous option in Winclada (Nixon, 1999a) was used. S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 of the Squamidium clade switches from Rhynchostegielloideae to basal Brachytheciaceae (Fig. 1b), but otherwise treatment of the inversion region does not have a major effect on the topology of Brachytheciaceae. All 163 taxa in these clades have an A-rich form of the psbT-N loop, although for one species (Donrichardsia macroneuron), the structure of the loop region is unknown due to a lack of the psbT-H sequence. Outside Rhynchos- 164 S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 Fig. 1. Continued tegielloideae and Squamidium only two species of Brachytheciaceae with A-type are found, both in the Rhynchostegioideae clade (Fig. 1). Base compositions in the more common T-rich forms also contain some phylogenetic information. The TTCATTTTA form is found only in the Sciuro- S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 165 Fig. 1. Continued hypnum clade, while occurrences of the CTCATTTTA form are restricted to all other groups in the Homalothecioideae and Brachythecioideae clades. Analyses with 99-terminal data set POY analyses with a smaller data set (99 terminals) resulted in topologies that were almost identical to those 166 S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 Fig. 2. Strict consensus of 20 most parsimonious trees (L ¼ 2130) from direct optimization analyses of combined morphological and molecular data with gap costs of one. Only 99 terminals are included in analyses and 9-bp loop region of psbT-N spacer in its original form. obtained with the larger data set. The number of characters and pi positions in the character optimizations, however, clearly differed (Tables 3 and 4). Decreasing sequence length, especially in the ITS2 region, resulted in character optimizations with fewer indel events and the length of the character optimizations decreased by almost 300 bp (Table 4). The inclusion of a wider selection of outgroups improved S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 167 Fig. 3. Strict consensus of 629 equally parsimonious trees (L ¼ 2471) from Nona analyses of one of six POY character optimizations with a gap cost of one. Original direct optimization and results from Nona are based on combined analyses of morphological and molecular data and 9-bp loop region on psbT-N spacer in original form. Support values for clades are indicated below (jackknife value) and above (Bremer support) each node. Unlike the original direct optimization analyses, gaps in all Nona analyses are treated as missing data. 168 Table 6 Status of groups within Brachytheciaceae under different analysis methods and parameters; black ¼ monophyletic, gray ¼ paraphyletic, white ¼ polyphyletic, patterned ¼ unresolved S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 congruence among the data sets. The ILD value from analyses of small data sets was higher with all gap costs compared with those based on larger data sets (Table 5). Effect of information on indel events and weighting of gaps and morphological data Weighting of morphological data as one, despite the gap cost, led to the most congruent results according to ILD (Table 5). If in analyses with gap costs two or four, and morphology with the same weight as the gaps, the topologies of strict consensus trees, however, were closer to those of equal weighting (POY 1 : 1 : 1 in Tables 5, i.e., the most parsimonious results) (Table 6). In this case, Palamocladium and Kindbergia praelonga appeared to be monophyletic, also with high gap costs. Based on these findings it seemed clear that the monophyly of these taxa were only due to the phylogenetic signal from morphology. Topologies from all combined analyses differed only in some details, and most of the groupings were fairly stable, regardless of differences in gap costs. A summary of the major topological differences is shown in Table 6. Of all combined analyses, only that with a gap cost of four, and an unaltered psbT-N loop region, resolved the Brachytheciaceae as non-monophyletic, due only to the placement of one species, Platyhypnidium riparioides (Hedw.) Dixon, which in this analysis clustered within Meteoriaceae (Table 6). ILD favored analyses with a gap cost of two and the psbT-N loop region excluded (Table 5). Although the most parsimonious results (a gap cost of one) are the least congruent according to ILD, their topology included several details that were more congruent with the traditional classification of Brachytheciaceae than any of the POY topologies with higher gap costs (Table 6). This analysis resolved Palamocladium and Kindbergia praelonga as a monophyletic clade, while Bryhnia novaeangliae always remained polyphyletic. While in analyses with a gap cost of one, the ILD supported the inclusion of an unaltered inversion region on the psbT-N spacer, whereas all other gap costs employed favored its exclusion from the analyses (Table 5). Increasing gap costs led to more compact implied alignments, which, especially in the ITS2 region, increased the number of pi positions if implied alignments were studied as static alignments (Table 4). The lengths of the psbT-H and trnL-F sequences in the implied alignments were quite insensitive to differences in gap costs: the psbT-H region was 556–559 bp and trnL-F 528–552 bp long. The number of pi positions also varied, mainly in the ITS2 region (Tables 3 and 4), with only minor variations in the trnL intron and the psbT-N spacer. The topologies based on combined analyses of all data also strongly reflected the phylogenetic signal from the ITS2 region when higher gap 169 costs were applied. In those analyses, topologies were closer to those obtained from analyses of only ITS2 data. For example, ITS2 never resolved Rhynchostegioideae, Palamocladium, and Kindbergia praelonga as monophyletic, or with higher gap costs the Squamidium clade as sister to Rhynchostegielloideae (Table 6). Since our Nona analyses were based on POY character optimizations in which gaps were treated as missing data, differences between a Nona topology (Fig. 3) and a topology from POY with a gap cost one (Fig. 1a) resulted from a different treatment of gaps. When character optimizations obtained with higher gap costs were analyzed with Nona, some of the topological differences occurred due to weighting of the gap information in the POY analysis. We used abbreviations for Nona analyses based on different POY character optimizations in the text and tables. For example, Nona 1 : 1 : 1 means that Nona analyses based on POY character optimization were obtained with equal weighting for transitions and transversions, and a gap cost of one, and in Nona 1 : 1 : 2 analyses character optimization was obtained from POY analyses with equal weighting for transitions and transversions, and a gap cost of two. Topologies based on the Nona analyses of POY character optimizations constantly differed in some details from those obtained from the original POY analyses (see Table 6). The differences resulting from gap treatment were, however, surprisingly small, considering the large difference in numbers of pi characters. For example, when gaps were treated as missing data, POY character optimizations of ITS2 using a gap cost of one had 154 pi characters in the 124 terminal data set (Table 4), but when gaps were included in analyses as a fifth character state, the number with the same optimizations was 406. Nona resolved the Rhynchostegioideae clade (including Rhynchostegium, Palamocladium, Scorpiurium Schimp., Aerobryum speciosum Dozy & Molk., most Eurhynchium species and Platyhypnidium riparioides) most often as monophyletic in the basal position compared with other Brachytheciaceae (Fig. 3), which never occurred in POY analyses. Bryoandersonia illecebra was also almost always placed within the Rhynchostegioideae clade in Nona analyses, while POY resolved it most often in Rhynchostegielloideae. Based on ILD metrics, POY analyses gave less congruent results than analyses of POY character optimizations with Nona (Table 5). Both Nona 1 : 1 : 1 and Nona 1 : 1 : 4 analyses excluding the psbT-N loop region gave the most congruent results of all Nona analyses (Table 5). Nona 1 : 1 : 1 analyses also resulted the hypothesis of relationships that was very similar to the current classification of the Brachytheciaceae (Table 6). The shortest trees were obtained when POY character optimizations with a gap cost of one were used as a static alignment in Nona (Nona 1 : 1 : 1, 170 S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 Table 5, Fig. 3). Although the topologies from all Nona analyses varied mainly in minor details, Nona 1 : 1 : 2 gave results that were closer to the current classification than any of the other POY or Nona analyses (Table 6). Because of problems with character independence in this method of analysis, we, however, did not regard this or any other Nona topologies as the most parsimonious result of our research. In POY, the homology statements in molecular data partitions may be affected by information from other data partitions, and thus in the implied alignments the independence of information within each partition is not guaranteed anymore (for details see Discussion). Indel events appeared to have an especially strong influence on the topology in POY analyses with high indel costs, because in addition to utilizing the information in them, they are also weighted according to gap cost. This can be estimated when POY topologies with higher gap costs (two or four) and Nona topologies based on character optimizations obtained using these costs are compared with each other (Table 6). Three specimens of Kindbergia praelonga never formed a monophyletic group in POY analyses with gap costs two or four and morphology with a weight one, although all Nona resolved these specimens as a monophyletic group basal to Brachythecioideae. Two Palamocladium species were also included and in Nona analyses this genus was always monophyletic and grouped within Rhynchostegioideae, but were polyphyletic in all other analyses. The position of Squamidium also varied; in POY analyses with gap costs two and four it appeared in the basal position within Homalothecium, and in the Nona analyses it was most often resolved as a sister group of Rhynchostegielloideae or all other species of the entire Brachytheciaceae clade (Nona 1 : 1 : 2). Topologies from all Nona analyses varied much less than the original topologies from POY, and this was found for all weighting schemes used (Table 6). When gaps were treated as missing data in Nona, the POY character optimizations obtained from one POY run did not give trees of the same length. Only the lengths of the shortest trees are shown in Table 5. However, for example, in the case of Nona 1 : 1 : 4 analyses (psbT-H loop included in original form and morphology with a weight of 1), of the total of the obtained eight character optimizations only two gave the parsimonious result with a length of 3016 steps, while the remaining six optimizations resulted in trees with 3017 (one optimization), 3018 (one optimization), 3019 (three optimizations), or 3022 steps (one optimization). Delimitation of Brachytheciaceae and relationships within the family The most parsimonious result (Fig. 1a), as well as all other combined unweighted analyses (Figs 1b,c, 2 and 3), supported a monophyletic Brachytheciaceae + Meteoriaceae (B+M) clade, leaving Isothecium alopecuroides (Dubois) Isov., Rozea subjulacea Besch., R. andrieuxii (Müll. Hal.) Besch., Lindigia debilis (Mitt.) A. Jaeger and Meteoriella soluta (Mitt.) S. Okamura outside that group. The branching order within the outgroup families varied widely in different analyses. Our sampling outside B+M is, however, so restricted that we do not want to draw any conclusions about their phylogeny even if the results of different analyses had been congruent. The limit between Meteoriaceae and Brachytheciaceae is always sharp, showing that they clearly constitute two independent lineages. Species that were transferred recently to Brachytheciaceae from Meteoriaceae (Aerolindigia capillacea (Hornsch.) M. Menzel, Aerobryum speciosum, Meteoridium remotifolium, Squamidium brasiliense, Zelometeorium patulum), Myriniaceae (Clasmatodon parvulus (Hampe) Sull., Helicodontium capillare (Hedw.) A. Jaeger), and Symphyodontaceae (Unclejackia longisetula (E. B. Bartram) Ignatov, T. J. Kop. & D. H. Norris) were always nested within Brachytheciaceae. On the other hand, Duthiella Müll. Hal., which was thought to belong to Brachytheciaceae by Noguchi (1991), was always in Meteoriaceae, lending support to the results by Buck and Goffinet (2000). Four distinct groups can be seen within the Brachytheciaceae (Figs 1 and 3): (1) Rhynchostegioideae includes Rhynchostegium, Eurhynchium striatum (Hedw.) Schimp. and E. angustirete (Broth.) T. J. Kop., Palamocladium, Pseudoscleropodium purum (Hedw.) M. Fleisch., Aerobryum speciosum, Plasteurhynchium striatulum, Platyhypnidium riparoides and Scorpiurium, (2) Rhynchostegielloideae includes Rhynchostegiella, Eurhynchium hians and related species, Cirriphyllum piliferum (Hedw.) Grout, and many small epiphytic taxa from America, (3) Homalothecioideae includes Homalothecium Schimp., Brachythecium subgen. Velutinum Loeske, Eurhynchium pulchellum (Hedw.) Jenn., and (4) Brachythecioideae includes most species of Brachythecium and Bryhnia Kaur. The relationships within Brachytheciaceae and the monophyly of some groups, for example the Rhynchostegioideae clade, varied depending on the treatment of gaps (POY versus Nona), the number of outgroup taxa included in the analyses, and the inverted psbT-N loop region. While the two other major clades, Brachythecioideae + Homalothecioideae and Rhynchostegielloideae, were constant in almost all analyses, only Rhynchostegioideae appeared as a paraphyletic cluster basal to Brachythecioideae in the POY analyses with higher gap costs. In the most parsimonious analyses with a gap cost of one, however, and in all Nona analyses, Rhynchostegioideae was always resolved as monophyletic (Table 6). Rhynchostegioideae clade. This taxon includes all species of Rhynchostegium, Eurhynchium s. str. (E. striatum and E. angustirete), Plasteurhynchium, S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 Palamocladium, Platyhypnidium riparioides, Scorpiurium, Pseudoscleropodium (Limpr.) M. Fleisch., Aerobryum Dozy & Molk., and in Nona analyses also Bryoandersonia H. Robinson. This clade was found to be monophyletic in the most parsimonious results (Fig. 1a), and in all Nona analyses in which gaps were included as missing data (Fig. 3). In Nona analyses it is sister to the other Brachytheciaceae (Fig. 3), while the most parsimonious analyses resolved it as sister to Rhynchostegielloideae (Fig. 1a). Eurhynchium striatum (the type species of the genus) and E. angustirete formed the monophyletic Eurhynchium within Rhynchostegioideae. Plasteurhynchium (Eurhynchium) striatulum is also a close relative of Eurhynchium, but our analysis supports its placement in a genus of its own. Other species traditionally placed in Eurhynchium were found to be only distantly related. Some of these have already been assigned to Oxyrrhynchium, including E. hians, E. savatieri Schimp., E. vagans, and E. pumilum (Wilson) Schimp., and Kindbergia, with K. (Eurhynchium) praelongum (Ignatov and Huttunen, 2002; Ignatov et al., 2004). Platyhypnidium riparioides was nested within Rhynchostegium in the most parsimonious resolved topology. Our results suggest that Platyhypnidium species with rough seta (P. austrinum and P. patulifolium) are not related to P. riparioides. Palamocladium appeared to be monophyletic in the most parsimonious topology (Fig. 1a), where it appeared close to Eurhynchium. Direct optimization of only ITS2 resolved Palamocladium euchloron in the same position as in the combined analyses, while P. leskeoides appeared in a rather surprising position within Brachythecium, as a sister group of the Chinese Bryhnia novae-angliae. The weighting of gaps and the stronger phylogenetic signal from ITS2 was thus the most likely reason for nonmonophyly of the genus in analyses with higher gap costs (Table 6). Scorpiurium is resolved as sister to Rhynchostegioideae in the most parsimonious topology (Fig. 1a). Pseudoscleropodium, Plasteurhynchium, and Aerobryum occur within the same clade with Eurhynchium s. str. Bryoandersonia illecebra was found within this clade only in the Nona analyses (Fig. 3) and analyses of 99 terminal data set (Fig. 2), while direct optimization analyses of 124-terminal data set resolved it within Rhynchostegielloideae (Fig. 1). Rhynchostegielloideae clade. This clade was always clearly delimited. The composition of the group remains the same in all analyses, except for the Squamidium clade, which includes the three pendent epiphytic species Zelometeorium patulum, Meteoridium remotifolium and Squamidium brasiliense (Table 6). In the most parsimonious POY analyses it appears as sister to Rhynchostegielloideae (Fig. 1a). However, if the 9 bp inversion on the psbT-N spacer is excluded, the Squamidium clade is 171 resolved as sister to all other Brachytheciaceae (Fig. 1b). The following groups are always included in Rhynchostegielloideae: 1. Rhynchostegiella s. str. forms a monophyletic cluster including three Mediterranean-Macaronesian species: R. tenella (Dicks.) Limpr., R. teneriffae (Mont.) Dirske & Bouman, and R. macilenta (Renauld & Cardot) Cardot. 2. The tropical Malesian Rhynchostegiella brachypodia M. Fleisch. and tropical South American-African Aerolindigia (Rhynchostegiella) capillacea form a clade and are fairly distant from Rhynchostegiella s. str. They are always sister to the three epiphytic American species, Homalotheciella subcapillacea (Hedw.) Broth., Clasmatodon parvulus, and Helicodontium capillare. The group of these five species, which in the figures (Figs 1–3) is labeled as the Aerolindigia clade, is present in all analyses. 3. The genus Cirriphyllum Grout, as traditionally delimited (Grout, 1931), appeared to be polyphyletic, but a smaller monophyletic Cirriphyllum clade could be found including the type species of genus, C. piliferum and C. crassinervum (Taylor) Loeske & M. Fleisch. (Cirriphyllum clade in Figs 1–3). Brachythecium koponenii Ignatov was found within this group in most parsimonious analyses. Unlike other taxa earlier included in Cirriphyllum, this monophyletic Cirriphyllum (or Cirriphyllum + Okamuraea) subclade appeared sister to Rhynchostegiella. Our results support the positioning of Okamuraea within Brachytheciaceae. This is in agreement with the recent data by Tsubota et al. (2002) and contradicts its position in Leskeaceae, where it had been placed in most of the recent publications (for example, Noguchi, 1991; Crosby et al., 1999). 4. Oxyrrhynchium (Eurhynchium) hians and several related species are the only members of the traditional Eurhynchium that are included in Rhynchostegielloideae. They form a rather stable group in all analyses. Direct optimization most often resolved Oxyrrhynchium as a monophyletic genus, excluding Oxyrrhynchium (Eurhynchium) pumilum and the Chinese variety of O. vagans (Figs 1 and 2, Table 6), although in Nona analyses all Oxyrrhynchium species resolved as a monophyletic group. Interestingly, the rare aquatic Chinese species, known as Platyhypnidium patulifolium and the aquatic North American endemic species Donrichardsia macroneuron (Donrichardsia clade in Figs 1–3) were always found adjacent to or even within this Oxyrrhynchium group. Donrichardsia H.A. Crum & L.E. Anderson was described as a monospecific genus from America, and it had earlier been included in Amblystegiaceae. The position of D. macroneuron within Brachytheciaceae was also found in the molecular phylogenetic study of Vanderpoorten et al. (2002), who, however, thought that it was close to P. riparioides. Another interesting fact is that P. austri- 172 S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 num, an aquatic Australian species, was also always found within Rhynchostegielloideae, but most often in the basal position of the clade including Cirriphyllum, Okamuraea, and Brachythecium koponenii (Figs 1 and 2). Brachythecioideae + Homalothecioideae clade. This clade usually consists of two rather distinct subclades, one centered around Homalothecium and the other around Brachythecium (Figs 1–3). Bryhnia, Kindbergia, Eurhynchiadelphus eustegium (Besch.) Ignatov, Huttunen & T.J. Kop. are, in most parsimonious topologies, resolved in a basal position in Brachythecioideae. Myuroclada maximowiczii is resolved as a sister group of the Brachythecioideae + Homalothecioideae clade. Homalothecioideae clade. Our results do not support the division of Homalothecium into smaller entities, such as Camptothecium Schimp. (including in our analyses H. aureum (Spruce) H. Robinson, H. lutescens (Hedw.) H. Robinson, and H. nuttalii (Wilson) A. Jaeger) and Trachybryum (Broth.) W. B. Schofield (H. megaptilium (Sull.) H. Robinson), because they do not form clearly independent lineages. The closest relatives of Homalothecium appear to be Brachythecium subgen. Velutinum (the Brachytheciastrum clade), represented in our analysis by three species: Brachythecium velutinum (Hedw.) Schimp., B. falcatulum (Broth.) Paris, and B. collinum (Schleich.) Schimp., with Eurhynchium pulchellum as the next closest relative. Brachythecioideae clade. The relationships within this clade are the least supported within Brachytheciaceae (compare Figs 1a–c and 3). The only group that is always very homogeneous is one of the small Brachythecium species centered around B. reflexum (Starke) Schimp. and B. plumosum (Hedw.) Schimp., which are segregated by Ignatov and Huttunen (2002) as a genus of their own, Sciurohypnum. The other species of Brachythecium are found centered around B. rivulare Schimp. (section Rutabula Broth.) and B. salebrosum (Chamberlainia). However, these groupings are not very clear in our analyses, and our results suggest that it is better to treat Brachythecium in a broad sense. B. rivulare and B. rutabulum (Hedw.) Schimp. are always grouped together (B. rivulare clade in Figs 1–3), but otherwise the closest sister group of this clade differed in various analyses. A peculiar endemic New Guinean genus, Unclejackia (E.B. Bartram) Ignatov, T.J. Kop. & D.H. Norris, is always found within Brachythecium close to B. lamprocarpum (Müll. Hal.) A. Jaeger and B. buchananii (Hook.) A. Jaeger. Bryhnia novae-angliae is resolved within a sister group of Brachythecium, but the three specimens do not form a monophyletic group. The Bryhnia clade consists of two specimens from European Russia and Siberia, which are morphologically closer to the type of B. scabrida (Lindb.) Kaur. (the type species of the genus). The specimen from southern China, however, was never in that clade, but always appeared as sister to Brachythecium frigidum (Müll. Hal.) Besch. The branching order within Brachythecium appeared to be sensitive to changes in the treatment of the psbT-N loop region (Fig. 1). The position of Kindbergia also appeared to vary (Table 6). Most of the species traditionally included in Cirriphyllum were found within the Brachythecioideae: C. flotovianum (Sendnt.) Ochyra was resolved within Sciurohypnum and C. cirrosum (Schwägr.) Grout and C. tommasinii (Boulay) Grout within Brachythecium (Figs 1–3). Discussion Morphological circumscription of Brachytheciaceae and its relationships to other families Our results support the monophyly of Brachytheciaceae and Meteoriaceae, following Ignatov (1999). Ignatov’s idea of the close relationships of these two families was based on the unique structure of the pseudoparaphyllia, which is similar in these families. The young developing branch primordia in the B+M clade are covered by three leaf-like structures, the pseudoparaphyllia, that are always situated regularly at a 120 angle to each other (see Ignatov, 1999, for details). The close relationship of the B+M group was also supported by Buck et al. (2000a,b) and Tsubota et al. (2002). However, both of these studies included only very small numbers of species from these families. In contrast to the current delimitation of the family (Buck and Goffinet, 2000), Isothecium Brid. and Rozea Besch. were not included in Brachytheciaceae (Hedenäs, 1992; Ignatov, 1999). This is also supported by the absence of typical brachytheciacean pseudoparaphyllia. In addition, Isothecium was resolved as a close relative of Lembophyllaceae (Quandt et al., 2004a). Our analyses suggest, however, that few exceptions occur in pseudoparaphyllia patterns. Lindigia debilis, Neobarbela comes (Griff.) Nog., and Meteoriella soluta have all been reported to have brachytheciacean pseudoparaphyllia patterns, but not all were found within the monophyletic B+M clade. Although, according to its morphology, L. debilis appears similar to Aerolindigia capillacea, DNA sequence data suggest they are not closely related. Neobarbella comes was not included in our analyses, but Quandt et al. (2004a) showed that it belongs to Lembophyllaceae. Meteoriella soluta was resolved in our analyses in Hylocomiaceae, which is consistent with the current classification of this species (Buck and Goffinet, 2000). In studies of the pseudoparaphyllia of this species we found some formerly neglected features that appear to link it with the other Hylocomiaceae and not Meteoriaceae. Trachypus bicolor Reinw. & Hornsch. is also certainly a S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 member of Meteoriaceae, although its pseudoparaphyllia differ from the B+M type. These exceptions are not discussed here, but will be examined in detail in future studies. Despite the fact that Meteoriaceae and Brachytheciaceae are probably close relatives, the limit between the two families has remained unclear (Buck, 1994; Ignatov, 1999). Several morphological characters have been suggested for distinguishing Meteoriaceae from Brachytheciaceae, such as the presence of papillose and short leaf laminal cells, strongly modified peristome, pendent growth, short seta and hairy calyptra. Due to the wide variation in these characters in Meteoriaceae, it was hard to find a character combination for taxonomic delimitation. Our results within the B+M clade suggest that a single character alone can be used to delimit these families. All Meteoriaceae have papillose laminal cells, while in Brachytheciaceae they are smooth. Surprisingly, this very simple character appeared to be crucial to family level classification, although at the species and genus level, papillosity appears to be in some cases very unreliable (Quandt et al., 2004b). Thus, delimitation of the Brachytheciaceae is becoming easier because all species, with the few exceptions mentioned above, can be separated from close relatives within the pleurocarpous mosses using two characters: the typical brachytheciacean pseudoparaphyllia and non-papillose cells. Our analyses do not support Plagiothecium as a close relative of Brachytheciaceae, as was suggested by Buck et al. (2000a,b) and Tsubota et al. (2002). This may also be the result of an uneven selection of outgroup species and a higher sequence variation within the outgroups, as compared with the ingroup, which may lead to unnatural groupings in the outgroups (Barriel and Tassy, 1998). The branching order of the outgroups was fairly variable in our analyses. New insights on relationships of the main genera of the Brachytheciaceae As was often suspected (Hedenäs, 1992; Ignatov et al., 1999; and many others), some of the largest genera in Brachytheciaceae are unnatural and their classification should be corrected. Formerly, only those attempts to clarify classification that have led to the formation of small, mono- or oligotypic genera were successful. The few attempts to rearrange the large genera of Brachytheciaceae, by dividing them into a few, but still fairly large entities, were all unsuccessful (see details in Ignatov and Huttunen, 2002). The most important factor leading to this situation was the difficulty in finding morphological characters to justify these new groups. In light of our results, it appears that practically all the morphological characters used in generic level classification in Brachytheciaceae have 173 evolved independently several times in many lineages and thus cannot serve by themselves as diagnostic characters. Phylogenetic analysis including molecular data will clearly change the traditional classification, and some new placements were quite unexpected. However, we found all the stable new placements acceptable and, after a re-evaluation of the morphological characters, well founded. Brachythecium, with 100–150 species, has always been the largest genus in the family. It has been stated to be an artificial assemblage by many authors (see Hedenäs, 1992; Ignatov et al., 1999), but at the same time division into several sections by Kindberg (1897) has been widely accepted. Robinson’s (1962) attempt to split the genus into two, Brachythecium (species centered around B. rivulare) and Chamberlainia (species around B. salebrosum) was, however, not accepted. In Chamberlainia, Robinson combined the sections Salebrosa Broth. and Velutina De Not., leaving in Brachythecium the sections Rutabula Broth., Reflexa Broth., and Cirriphyllopsis (Broth.) Takaki (B. plumosum–B. populeum group). Our analysis showed that these sectional limits are stable, while their relationships differed from Robinson’s view (1962). Section Velutina is clearly distinct from other sections, and it is closer to Homalothecium than to any Brachythecium species. Two other sections, Cirriphyllopsis and Reflexa, form a very clearly delimited group. Based on our results, the latter group, as well as sect. Velutina, should be excluded from Brachythecium and placed in genera of their own (Ignatov and Huttunen, 2002). Sequence variation within Brachythecium is very low, which is probably one reason for some of the minor differences which occur in the branching order in different analyses. The basal taxa centered around B. rivulare have rough seta (Fig. 1a), and ecologically they are meso- to hygrophytic, while the apical part of the clade centered around B. salebrosum includes more xeric species with smooth seta (Fig. 1a). These two groups were in most of the analyses somewhat poorly delimited. Based on the current results, there is not enough support for segregating the latter part into Chamberlainia, as Robinson (1962) suggested. Another problem with Brachythecium s. str. is that it also includes the southern Chinese specimen of Bryhnia novae-angliae. It is clear that analyses which include more specimens are needed in order to resolve the relationships of this species. The apical part of Brachythecium includes temperate to tropical species with a slightly reduced sporophyte. These species often have a strongly reduced annulus and capsule inclined to suberect, as exemplified by B. buchananii, B. ruderale (Brid.) W. R. Buck, B. lamprocarpum, B. laetum (Brid.) Schimp., and B. acuminatum (Hedw.) Austin. This group also includes a peculiar species, Unclejackia longisetula, which was originally described as a distinct genus (Ignatov et al., 1999). To redefine the genus level delimitation for the Brachythecium clade, 174 S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 either Unclejackia should be transferred to Brachythecium, or Brachythecium should be divided into several smaller genera. The latter choice would result in a situation with distinct phenotypes of Brachythecium salebrosum being assigned to different genera (Figs 1 and 2). Although this would enable the retention of Unclejackia as its own genus, it does not appear to be a reasonable solution to the problem. Inclusion of Unclejackia in Brachythecium, on the other hand, makes the morphological characterization of Brachythecium extremely broad. U. longisetula is a large, golden-lustrous moss with strongly concave leaves, very flexuose leaf acumen, very short costa (0.1–0.2 of leaf length), and strongly modified peristome. Unclejackia has most likely been derived from an ancestor resembling species in a group of tropical Brachythecium. It has probably evolved rapidly into a morphologically distinct species occurring in a geographically very limited area. U. longisetula is found only in the alpine ‘‘savannas’’ of New Guinea, where it has been collected only on trunks of tree ferns. Unclejackia was recently transferred to Brachytheciaceae (Ignatov et al., 1999), and this placement was immediately criticized. For example, Tan (2000) noted that Unclejackia would be better placed in Chaetomitrium Dozy & Molk. (Symphyodontaceae). Rhynchostegium, with 30–70 species is the second largest genus in the family. It was also considered an unnatural segregate, but in our analysis it was resolved as a very homogeneous and stable monophyletic clade. Our analysis included species showing wide morphological variation: tropical ‘‘typical Steerecleus’’ (R. pallidifolium (Mitt.) A. Jaeger and R. serrulatum (Hedw.) A. Jaeger) with pale color and large complanately arranged leaves, R. murale with imbricate leaves and deep green color, R. confertum which is usually deep green and often with costa ending in a spine, R. psilopodium Ignatov & Huttunen with pure green color and narrow leaves (previously placed in Rhynchostegiella), and R. rotundifolium (Scop.) Schimp. with exceptionally lax areolation of leaves and dark green to brownish color. The position of Platyhypnidium riparioides remains ambiguous; our analyses resolve it as a close relative of Rhynchostegium (Figs 1–3), in which it had formerly been placed (Takaki, 1956; Hedenäs, 1992, etc.). However, a lack of the most informative ITS2 sequence data in this species may affect its position. Morphologically, it shares several leaf characters with Rhynchostegium. Eurhynchium, unlike Rhynchostegium, appeared to be even more heterogeneous than expected. Species placed in this genus were found in all major clades, and it should be segregated into several smaller units and some of the species placed in other genera. Eurhynchium is the third largest genus of the family. Ignatov and Huttunen (2002) retained only two species in Eurhynchium, while the remainder were transferred to Plasteurhynchium, Oxyrrhynchium, Eurhynchiastrum Ignatov & Huttunen, Eurhynchiadelphus, and Kindbergia. Of the species occasionally included in this genus, Rhynchostegium (Eurhynchium) serrulatum was confirmed as belonging to Rhynchostegium, Platyhypnidium (Eurhynchium) riparioides, Eurhynchium muelleri (A. Jaeger) E.B. Bartram, Rhynchostegium aquaticum A. Jaeger to Platyhypnidium, and E. macroneuron (Grout) H.A. Crum to Donrichardsia. Eurhynchium (Platyhypnidium) austrinum (Hook. & Wilson) A. Jaeger was resolved as sister to Rhynchostegielloideae. Our results support the recognition of Oxyrrhynchium for E. hians and its close relatives. The two most important reasons for the long misunderstanding of this group are probably the enormous morphological variation in some of the widespread species of Oxyrrhynchium and the occurrence of many similarly appearing and highly variable species. For example, the E. hians group shows a wide morphological variation that overlaps not only with other Oxyrrhynchium, but also with species in other brachytheciaceaen genera. Finding morphological differences delimiting it from other Oxyrrhynchium as well as from other genera is a real problem. Most species of Cirriphyllum were originally placed in Eurhynchium and have frequently been transferred from one genus to another. Two species of Cirriphyllum, C. piliferum and C. crassinervum, formed a monophyletic clade, while other species were mainly resolved within Brachythecium. The recently described Malesian B. koponenii, a species with ambiguous affinity, is better included in Cirriphyllum. Rhynchostegiella appeared to be polyphyletic, and Ignatov and Huttunen (2002) suggested that it be split into three groups of more or less equal size. These groups have no, or only partial, geographic overlap; unlike Eurhynchium, all occur within only one of four major clades within Brachytheciaceae, namely Rhynchostegielloideae. Only four representatives belonging to two of the three groups given above were included in this analysis, but since two small genera related to Rhynchostegiella, Aerolindigia M. Menzel and Helicodontium (Mitt.) A. Jaeger, were also included in phylogenetic analyses, we have suggested a new classification for the entire genus (see Ignatov and Huttunen, 2002). Homalothecium. A clear tendency towards a more reduced peristome structure can be observed within Homalothecium. All species in the basal part of the clade have a complete peristome and thus have formerly been segregated in a genus of their own, Camptothecium. Two lineages with reduced peristomes can be found at the crown of the clade, the Eurasian H. philippeanum (Spruce) Schimp. and H. sericeum (Hedw.) Schimp., and the North American H. nuttalii. Our results do not support the identification of H. megaptilum as a distinct genus, Trachybryum. While Homalothecium was always monophyletic in analyses of combined data, the trees S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 based only on chloroplast data resolved H. laevisetum Sande Lac. within Rhynchostegium, adjacent to Palamocladium leskeoides. H. laevisetum is the most peculiar member of the genus, with characters resembling some of those seen in Palamocladium. Two rather small genera appeared to be polyphyletic. Three species of Platyhypnidium included in our analyses, P. riparioides, P. austrinum, and P. patulifolium never formed a monophyletic group and they did not even appear to be closely related. While the latter two species were always resolved within Rhynchostegielloideae, P. riparioides appeared in Rhynchostegioideae, usually close to, or among, Rhynchostegium. Of three specimens of Bryhnia novae-angliae, two from Europe and Siberia that were also morphologically most similar to each other, were found to be closely related. At the same time, a specimen from central China was always resolved as sister to Brachythecium frigidum, instead of sister to the other two Bryhnia specimens. The Chinese specimen also differs in morphology from the northern specimens, but improvements in species delimitation are hampered by the inclusion of the East Asiatic Bryhnia novae-angliae specimens, which show all the transitional morphotypes from the Chinese to the Siberian. Evaluation of morphological characters in the Brachytheciaceae Almost all morphological characters of Brachytheciaceae show homoplasy to a large extent. Despite this, two synapomorphies characterizing Brachytheciaceae, non-papillose leaf cells and the presence of a unique pseudoparaphyllia pattern, can also serve as diagnostic characters for separating brachytheciaceaen species from other pleurocarpous mosses. On the genus level, few characters appear to be unique for certain genera within Brachytheciaceae, such as a reduced exostome in Clasmatodon parvulus, immersed to emergent capsules in Squamidium (Müll. Hal.) Broth. and Zelometeorium Manuel, and the presence of paraphyllia in Kindbergia. In practice, most of the clades found in phylogenetic analyses cannot be delimited using currently known morphological characters. Based on our analyses, we can now reconsider the value of those characters that were thought to be especially important in earlier studies of the Brachytheciaceae. Peristome reduction was thought to be very important for classifying the Brachytheciaceae up to the late 19th century. For example, Schimper (1876) did not include Homalothecium in this family due to its reduced peristome. Later, however, peristome reduction was found to be a common phenomenon in many lineages of pleurocarpous mosses. In our analyses, lineages with peristome reduction occur in all major clades and many subclades. In Rhynchostegioideae, Aerobryum and Pal- 175 amocladium are plants showing a clearly reduced peristome structure, and of those taxa not included in the present analysis, but expected to belong to this group, Eriodon also exhibits this character state. Most of the genera in Rhynchostegielloideae have more or less reduced peristomes, with the exception of Oxyrrhynchium, Cirriphyllum, Platyhypnidium austrinum and P. patulifolium. In Homalothecioideae, Homalothecium species have reduced peristomes. The pattern of peristome reductions in various lineages within this genus varies somewhat and is also different from that of Unclejackia longisetula and Brachythecium acuminatum, representatives of the Brachythecioideae with reduced peristomes. Interestingly, some species with relatively ‘‘complete’’, ‘‘hypnoid’’ peristomes display hygrocastique hygroscopic movement (for example, Palamocladium, Homalothecium lutescens, Rhynchostegiella tenella), whereas some tropical taxa with rather modified peristomes with heavily incrassate teeth covered with branched papillae display xerocastique movement (Malesian Rhynchostegiella, referred by Ignatov and Huttunen (2002) to Remyella Müll. Hal.). In the latter case, peristome modification is associated with a number of other characters, which might often have evolved together, such as straight perichaetial leaves and erect capsule. Hence, the presence of modification appears to be a more important character than its hygroscopic movement pattern. Surface structure of the seta appears to be one of the most stable characters, showing no homoplasy in some clades (Fig. 1a). This character is frequently used in keys of Brachythecium, as a character distinguishing sections. The setae are invariably smooth in all species of Rhynchostegioideae (without exception!), but rough in most species of Rhynchostegielloideae. The latter group, however, has some taxa with smooth setae, such as Rhynchostegiella tenella, Eurhynchium vagans, Meteoridium remotifolium, and Squamidium brasiliense. Most of the basal species in Brachythecioideae have rough setae, while most of the apical groups, with some exceptions, have smooth setae. Operculum shape was considered to be one of the most important characters in Brachytheciaceae over the last 150 years. Our analysis shows that this character is relatively stable in some clades, but not in all. Species in Rhynchostegielloideae invariably have a rostrate operculum, as do Rhynchostegioideae (except Pseudoscleropodium; Fig. 1a). The operculum shape in Brachythecioideae is more variable than was previously thought. Two species traditionally placed in Cirriphyllum or Eurhynchium, C. tommasinii and C. flotowianum, appeared in Brachythecioideae. The shape of the operculum is rather variable in basal species of Brachythecioideae and in Homalothecioideae. Thus, to use this as a key character for separating Eurhynchium and Brach- 176 S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 ythecium has led to unnatural classifications, i.e., the misplacement of some species. Phyllodioicy is a rather rare feature in the pleurocarpous mosses. In Brachytheciaceae it has evolved several times independently in three major clades: (1) in Rhynchostegioideae (Aerobryum, Eurhynchium s. str., Palamocladium), (2) in Homalothecioideae (many Homalothecium species and Eurhynchium pulchellum), and (3) Brachythecioideae (Eurhynchium eustegium and Brachythecium auriculatum A. Jaeger). Annulus structure is difficult to evaluate if only herbarium material is available. Hence, it is safe to discuss only the most contrasting character state, that of total absence of specialized rows of annulus cells, which remain after operculum removal at the mouth of the urn. The annulus is absent in the most derived members of the Brachythecium clade, which mainly occur in tropical and temperate areas. Usually these species tend to have only a slightly curved capsule instead of a strongly curved, typical brachythecian one. Most have smooth setae (B. buchananii, B. ruderale, etc.) with only one exception, Unclejackia. Species centered around B. salebrosum commonly have an annulus consisting of only one cell row, which is rather firmly attached and difficult to take apart. This can be assumed to present a form that is transitional to the 2–3-rowed annulus occurring in most species of the family. Shape of the pseudoparaphyllia is a stable character in Rhynchostegioideae, which always has a triangular pseudoparaphyllia. There are several lineages within Rhynchostegielloideae, and especially Brachythecioideae, which show a transition to a long and narrow pseudoparaphyllia (Aerolindigia, Remyella, Rhynchostegiella, Homalothecium, p. p. Kindbergia, the Brachythecium buchananii group and part of the B. salebrosum group, and Unclejackia). In most cases, the shape of the pseudoparaphyllia correlates with epiphytic growth, although Kindbergia is an exception. Some other typical epiphytes, however, have short triangular pseudoparaphyllia (cf. the Squamidium group). Costa ending in a spine appears to lack value as a character at the genus level and above, although it is frequently used in generic keys. In both Rhynchostegium (for example, R. confertum (Dicks.) Schimp.) and Brachythecium (for example, some populations of B. salebrosum) are some specimens with a fairly strong tooth at the end of the costa, whereas Eurhynchium hians, which often has a clear spine, may have a smooth end of the costa in shaded habitats. Size of alar cells in basal corners of leaves and their cell wall thickness appear to be good characters at the species level, whereas their use at higher levels is doubtful. Hairy calyptras are never present in Rhynchostegioideae, but some or all species of Squamidium, Zelometeorium, Okamuraea, Brachythecium (B. complanatum Broth., B. coreanum Cardot, B. auriculatum), and Homalothecium have sparsely hairy calyptras. This character has also been reported in Bryhnia trichomitria Dixon & Thériot, but its position in this genus should be confirmed. It, however, appears to be a close relative of the basal species of the Brachythecium clade, especially B. auriculatum. Although axillary hairs may have a rather variable structure, some peculiarities in shape and color were found to be useful for classification. Species of Rhynchostegium often have long, multicellular axillary hairs with pale basal cells and upper cells which are more intensively colored. In some species axillary hairs were found to be verrucose, but this is probably not a stable character, even at the species level. Axillary hairs in Rhynchostegiella are usually short, consisting of 2–3 cells of the upper cell as in Oxyrrhynchium vesiculateinflated. This pattern can be seen in some, although not in all, axillary hairs within a single specimen. A similar pattern was observed in some Rhynchostegium specimens, as in the Chinese R. inclinatum (Mitt.) A. Jaeger, which however, was not included in our analysis. Axillary hairs in Homalothecioideae are often composed of several relatively short cells, while in Brachythecium s. str. they are usually with 2–3 cells, the upper being long and more or less distinctly tapering at the distal part. The latter character state is less clear in the group of species centered around B. reflexum and B. populeum (Hedw.) Schimp. in the Sciurohypnum clade. Color of plants is in most groups of Brachytheciaceae quite indistinctly green, turning with age to yellow- or brownish-green. Two peculiar groups, however, can be found within the family. In the ‘‘Chamberlainia’’ group (Brachythecium albicans (Hedw.) Schimp., B. buchananii, B. ruderale, and Cirriphyllum cirrosum), the plants never become brownish with age, but instead turn very pale stramineous. Another peculiar group is Rhynchostegioideae, with plants turning silvery white when they become older. This pattern is especially noticeable in tropical Rhynchostegium species, but some silvery white aspect is often also present in such taxa as Eurhynchium angustirete, Aerobryum speciosum, and sometimes in Pseudoscleropodium purum. In some tropical species the older parts of plants easily become black. This character is widespread in Meteoriaceae, while in Brachytheciaceae it is found in two groups, which our analysis suggests are unrelated: the Squamidium clade (including Squamidium, Zelometeorium, and Meteoridium) and Aerobryum. Prorate leaf laminal cells are found in Bryhnia, Eurhynchiadelphus eustegium and Kindbergia, all of which were found to be closely related. Plasteurhynchium striatulum has similar distinctly prorate cells but is not closely related to other species showing this character. Cells are treated as prorate only when the prorae S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 are very distinct and tooth-like, while a small proration can be traced in a powerful light microscope in many other species. Longitudinal leaf plication is characteristic for Homalothecioideae, Brachythecioideae and Rhynchostegioideae, including, for example Palamocladium and Eurhynchium, but this character is also very rarely found in Rhynchostegielloideae (Cirriphyllum, Okamuraea). Methodological discussion The combined analyses of molecular and morphological characters revealed a well-resolved phylogeny of Brachytheciaceae. As expected, morphology alone did not include enough information to resolve the relationships of closely related species and genera, which may be the result of the presumably young age of the family. Most of the clades found in our analyses were very close to some groupings in earlier classifications of Brachytheciaeceae (Brotherus, 1925; Robinson, 1962; Hedenäs, 1992; Ignatov, 1998). It appears that, although finding reliable diagnostic characters for delimiting these groups has been difficult, there is still something in the general appearance of species that has enabled their classification together. In addition to resolving the phylogenetic relationships in Brachytheciaceae, our aim was to determine whether the number of outgroups and the phylogenetic information in indel events, or in short inversion region in psbT-N loop, affect the topology of phylogenetic hypotheses. We showed that: (1) although the indel events in our study include a large amount of phylogenetic information, the topologies obtained with and without these data are only conflicting in two details, the position of Bryoandersonia illecebra and Rhynchostegioideae (Figs 1a and 3); (2) in conflict with results of Quandt et al. (2003), the inversion occurring in the 9 bp loop region in the psbT-N spacer appeared highly informative in Brachytheciaceae; and (3) the large numbers of outgroups and resulting increased sequence variation did not disturb phylogeny reconstruction in direct optimization, but improved the congruence among data partitions, as measured with the ILD. In POY, the optimization process is affected by both parameters used, and interactions between included data sets. Sometimes even fairly small changes in the original data set, such as the exclusion of the 9 bp in the psbT-N spacer in this study, may lead to different character optimization and topology (Fig. 1a–c). Although the treatment of gaps as a fifth character state remarkably increased the information available for phylogeny reconstruction (see Results), this had only a minor effect on topology (Figs 1a and 3). One reason for this may be that, due to the high number of indel events in ITS2, it is possible to optimize this region so that very little conflict results with phylogenetic signals from other data sets. 177 It is rather surprising that analyses based on direct optimization were, according to ILD metrics, always less congruent than results based on the same POY implied alignments analyzed with Nona. The POY topologies obtained were always based on the most parsimonious optimization of all available characters, but when the character optimizations from POY are analyzed in Nona as static alignments, and the gaps treated as missing data, they may actually appear to be suboptimal (see also Wahlberg and Zimmerman, 2000). Recently, character optimizations from POY have often been used like ordinary alignments in phylogenetic studies (Wahlberg and Zimmerman, 2000; Cognato and Vogler, 2001; Belshaw and Quicke, 2002; Quandt et al., 2004b). Despite the possible suboptimality resulting from different treatments of gaps (of the previously mentioned studies, only Cognato and Vogler (2001) used the same gap cost for succeeding phylogeny reconstruction), POY character optimizations are still usually better (in terms of tree length) than static alignments obtained using other methods. Independence of characters in gap-containing regions will, however, be lost, due to the interactions between data partitions during direct optimization. Hence, the use of character optimizations as static alignments in phylogenetic analyses violates the basic assumptions of cladistic analyses and is not recommended as a method for solving sequence alignment problems. The different treatment of indel events, as well as the lack of weighting during analyses also explains the more stable topologies obtained from Nona with different POY character optimizations. The higher gap cost in POY also means a higher weighting of indel events in analyses, while in Nona, information included in the gaps is totally ignored. Due to this the effect of gappy ITS2 is clearly stronger in POY analyses with higher gap costs. Topologies obtained from POY with gap costs of two and four share more features with the topology based only on ITS2, while trees with a gap cost of one more closely resemble Nona topologies (Table 6). Analyses with large numbers of outgroup taxa improved the congruence among data sets. Despite this, topologies from POY analyses remained fairly similar to those obtained from analyses with one outgroup (Floribundaria floribunda). It might be expected that if the sequence variation is high, exclusion of the most variable outgroup species could make character optimization easier and hence improve the results. For example, ITS2 has often been avoided in analyses in many phylogenetic studies of mosses, because this region has been considered as impossible to align. However, wider sequence variation within the large data set (Table 3) increased the congruence between different data sets (Table 5), the lower sequence variation decreased the resolution in the analysis of 99-terminal data set, and cladograms from both analyses differed 178 S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 only in minor details (Figs 1a and 3). The advantages of a wide sequence variation for phylogeny reconstruction have been observed previously (Källersjö et al., 1999; Broughton et al., 2000), and recent phylogenetic studies have reported that adding taxa in analyses increased their resolution (Pollock et al., 2002; Rydin and Källersjö, 2002; Stenroos et al., 2002). The effect of morphological data on our analyses appeared rather small if the topologies based on only molecular data and all data are compared. Some genera (Palamocladium) and species (Kindbergia praelonga, Rhynchostegium murale, Brachythecium buchananii), however, were monophyletic only if the morphological data were included. In Wheeler and Hayashi’s (1998) study, where they used direct optimization and tested different weighting schemes for morphological data, weighting morphology equally with indel cost gave consistently more congruent results in light of the ILD compared with those in which morphology alone was given a weight one. Equal weighting of indel events and morphology is becoming a common method in combined POY analyses (Giribet et al., 2002; Schulmeister et al., 2002), and this is often defended by highlighting the higher value of morphological data compared with sequence level data. Weighting morphological data does not, however, necessarily lead to a higher congruence between data partitions in all analyses. Thus, the differential weighting of morphological data is only based on rather subjective views of the value of that data partition. In our case, for example, the weighting method where morphology always attained a weight of one gave the most congruent results, according to the ILD metric. Although the short inversion of the psbT-N spacer loop was only expected to increase homoplasy within the data set (Quandt et al., 2003), it appeared to be very informative within Brachytheciaceae (Fig. 1, Table 2). Similar short inversions are expected to be common in non-coding chloroplast regions (Kelchner and Wendel, 1996; Graham et al., 2000), although there are few documented examples of this. These regions are said to be prone to frequent changes, which has raised the question of their phylogenetic informativeness and how they should be treated in phylogenetic analyses (Kelchner and Wendel, 1996; Quandt et al., 2003). In our study, the exclusion of a psbT-N loop in analyses with gap costs of two or four increased the congruence between data sets in POY analyses and some groupings in parsimonious trees changed. In most parsimonious analyses, however, any altering of the loop region decreased congruence. Thus, in our analyses, we can see no reason to ignore the information in the psbT-N loop region. It clearly contains some phylogenetic information in our study groups, and even if ILD favored the exclusion of the loop positions from the analyses, we believe that unnecessary exclusions of data should always be avoided. Our results support Wenzel and Siddal’s (1999) view that the globally homoplasious characters may be locally informative, and hence their value in any specific study cannot be known without testing. The base composition within the loop region seems to include phylogenetically important information and, unlike in the Quandt et al. (2003) study, its inclusion in the analyses did not have major negative effects on them. Our final aim was to revise the classification of Brachytheciaceae, and for this purpose we need to select one topology to use as the basis of the new arrangement. Currently, three different methods can be used to select the topology for taxonomic revisions in studies utilizing the direct optimization. First, based on sensitivity analyses, we can regard the most congruent topology according to ILD as the best solution. A second choice is to name only those groups that are stable despite the variation in substitution costs in the direct optimization analyses. The third possibility is to avoid all a priori weightings of data partitions, including the differential weighting of gaps, and to regard the topology showing the smallest number of evolutionary steps as the optimal topology. Of these, the first and third are based on different views of the concept of parsimony in phylogenetic analyses. While the third represents the traditional concept of parsimony in phylogenetic analyses, in the first one taxonomic congruence is accepted as an additional criterion. As discussed by Schulmeister et al. (2002), the first method—the use of ILD as a measure of optimality—is based on the idea that congruence among data partitions can be used as an additional criterion for selecting the most parsimonious results. According to this method, the parameter set resulting in the least conflict between data partitions, and giving the most congruent result, is regarded as the most parsimonious solution. In our case, ILD supported the direct optimization analyses and the topology that was obtained with a gap cost of two and morphology with a weight of one. We find it hard to accept the differential weighting in this analysis, as it results in a topology which differs from the most parsimonious hypothesis of relationships (see Grant and Kluge, 2003). In addition, some details in topology from this analysis, such as the polyphyly of Palamocladium and Kindbergia praelonga, seem surprising in light of the morphological variation within these taxa. Giribet (2003) suggested the use of stability of clades under different substitution costs as a criterion for accepting them as taxonomic units. As with the first method, this is also based on exploring the effect of differential weightings of data partitions on direct optimization analyses. In this case, the instability of clades under different substitution costs will be detected using the sensitivity analyses, and only those clades that appear most stable with different parameters are accep- S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 ted. They can be visualized, for example, by constructing the stability trees that sum up the results from different analyses (see for example Schulmeister et al., 2002) or using the graphic sensitivity plots, ‘‘Navajo rugs’’ (see Wheeler, 1995). Stability is thus used as a measure of support. In our study this would lead to taxonomic conclusions fairly similar to those of the first method. For example, the polyphyly Palamocladium and Kindbergia praelonga, and the paraphyly of Rhynchostegielloideae are always supported in all other gap costs except one (Table 6), and thus the use of stability (sensu Giribet, 2003) would lead discarding them as taxonomic units. Although the use of sensitivity analysis (sensu Wheeler, 1995) for testing the effect of different parameters on topology and congruence between different data sets is becoming almost a standard method for evaluating results, this method has also been criticized. For example, Frost et al. (2001) and Grant and Kluge (2003) stated that this type of search for additional criteria is unnecessary and unrelated to parsimony. According to them, the only truly parsimonious result would be that obtained using equal weighting (i.e., all indel and substitution events and morphology with the same weight) and to avoid testing several indel and substitution parameters. Thus, the optimal most parsimonious topology to use as the basis for classifications is the shortest tree from unweighted analyses of all available data, the third method of those listed above, and including also the data from 9 bp inversion region in psbT-N spacer. This alternative is based on the idea that equal weighting in analyses minimizes the number of additional assumptions because any differential weighting schemes including the weighting of indel events can be regarded as adding unnecessary ad hoc hypotheses leading to unparsimonious results. The weighting of gap positions increases the number of assumed evolutionary events and hence equal weighting provides the most parsimonious explanation for the data. Based on our own observations with the present data, POY analyses with lower gap costs usually lead to ‘‘looser’’ character optimizations allowing, at least in most variable sequence regions such as ITS2, more autapomorphic positions. This reduces the number of pi positions within the data set (Table 4, see also Caterino and Vogler, 2002; Quandt et al., 2004b). Thus, we regard the topology obtained with equal weighting as the only truly parsimonious hypothesis of relationships in Brachytheciaceae (Fig. 1a). In addition, this topology also avoids all of the most doubtful groupings and maintains the need for minimum nomenclatural changes. Although the stability of the nomenclature cannot be used as a desirable goal of phylogenetic systematics (Dominguez and Wheeler, 1997), it is easy for us to prefer the choice that is both a parsimonious explanation of the data as well as the 179 most congruent with the nomenclature and traditional classification of Brachytheciaceae (see also Wiley, 1981). Acknowledgments We thank our colleagues for providing specimens of Helicodontium capillare, Meteoridium remotifolium, Homalothecium megaptilium (Angela Newton, BM); Scorpiurium deflexifolium (Solms.) M. Fleisch. & Loeske, Rhynchostegiella teneriffae, R. macilenta (Lars Hedenäs, S); Lindigia debilis, Homalotheciella subcapillata, Rozea subjulacea, R. andrieuxii (William R. Buck, NY). Sequences of South American Kindbergia praelonga were kindly provided us by Dr Dietmar Quandt. We want to thank Taran Grant, Arnold Kluge and Gonzalo Giribet for giving access to their unpublished manuscripts. We are thankful to three referees, Drs Vesa Selonen, Jaakko Hyvönen, and all our colleagues in the Finnish Museum of Natural History for valuable comments on manuscript. This work was financially supported by the Academy of Finland (to Jaakko Hyvönen’s Bryosphere project and to Timo Koponen’s Biodiversity of SE Asiatic Bryoflora project), Finnish Cultural Foundation, Center for International Mobility (CIMO), SYS Resource researcher exchange program in the Natural History Museum, London, the Program ‘‘Biodiversity’’ of Russian Academy of Sciences, and Russian Foundation for Basic Researches. References Bapteste, E., Philippe, H., 2002. The potential value of indels as phylogenetic markers: position of Trichomonads as a case study. Mol. Biol. Evol. 19, 972–977. Barriel, V., Tassy, P., 1998. Rooting with multiple outgroups: consensus versus parsimony. Cladistics 14, 193–200. Belshaw, R., Quicke, L.J., 2002. Robustness of ancestral state estimates: evolution of life history strategy in ichneumonoid parasitoids. Syst. Biol. 51, 450–477. Bremer, K., 1994. Branch support and tree stability. Cladistics 10, 295– 304. Brotherus, V.F., 1905–09. Musci. (Laubmoose). II. Pleurocarpi. In: Engler, A. Prantl, K. (Eds.), die Natürlichen Pflanzenfamilien. W. Engelmann, Leipzig, pp. 701–1173. Brotherus, V.F., 1925. Musci. In: Engler, A., Prantl, K. (Eds.), die Natürlichen Pflanzenfamilien, 2nd edn. W. Engelmann, Leipzig, pp. 1–522. Broughton, R.E., Stanley, S.E., Durrett, R.T., 2000. Quantification of homoplasy for nucleotide transitions and transversions and reexamination of assumptions in weighted phylogenetic analysis. Syst. Biol. 49, 617–627. Bruch, P.H., Schimper, W.P.H., Guembel, T.H., 1851–1855. Bryologia Europea seu genera muscorum Europaeorum monographice illustrate 5–6. Sumptius Liberiae E. Schweizerbart, Stuttgart. Buck, W.R., 1994. A new attempt at understanding the Meteoriaceae. J. Hattori Bot. Lab. 75, 51–72. Buck, W.R., 1998. Pleurocarpous mosses of the West Indies. Mem. NY Bot. Garden 82, 1–400. 180 S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 Buck, W.R., Goffinet, B., 2000. Morphology and classification of mosses. In: Shaw, A.J., Goffinet, B. (Eds.), Bryophyte Biology. Cambridge University Press, Cambridge, UK, pp. 71–123. Buck, W.R., Goffinet, B., Shaw, A.J., 2000a. Testing morphological concepts of orders of pleurocarpous mosses (Bryophyta) using phylogenetic reconstructions based on trnL-trnF and rps4 sequences. Mol. Phyl. Evol. 16, 180–198. Buck, W.R., Goffinet, B., Shaw, A.J., 2000b. Novel relationships in pleurocarpous mosses as revealed by cpDNA sequences. Bryologist 103, 774–789. Caterino, M.S., Vogler, A.P., 2002. The phylogeny of the Histeroidea (Coleoptera: Staphyliniformia). Cladistics 18, 394–415. Cognato, A.I., Vogler, A.P., 2001. Exploring data interaction and nucleotide alignment in a multiple gene analysis of Ips (Coleoptera: Scolytinae). Syst. Biol. 50, 758–780. Crosby, M.R., Magill, R.E., Allen, B., He, S., 1999. A Checklist of the Mosses. Missouri Botanical Garden, St. Louis. De Pinna, M.C.C., 1991. Concepts and tests of homology in the cladistics. Cladistics 7, 367–394. Dominguez, E., Wheeler, Q.D., 1997. Taxonomic stability is ignorance. Cladistics 13, 367–372. Doyle, J.J., Doyle, J.L., 1990. Isolation of plant DNA from fresh tissue. Focus 12, 13–15. Farris, J.S., Källersjö, M., Kluge, A.G., 1995. Testing significance of incongruence. Cladistics 10, 315–319. Frost, D.R., Rodrigues, M.T., Grant, T., Titus, T.A., 2001. Phylogenetics of the lizard genus Tropidurus (Squamata: Tropiduridae: Tropidurinae): direct optimization, descriptive efficiency, and sensitivity analysis of congruence between molecular data and morphology. Mol. Phyl. Evol. 21, 352–371. Giribet, G., 2003. Stability in phylogenetic formulations, and its relationship to nodal support. Syst. Biol. 52, 554–564. Giribet, G., Edgecombe, G.E., Wheeler, W.C., Babbitt, C., 2002. Phylogeny and systematic position of Opiliones: a combined analysis of chelicerate relationships using morphological and molecular data. Cladistics 18, 5–70. Giribet, G., Wheeler, W.C., 1999. On gaps. Mol. Phyl. Evol. 13, 132–143. Gladstein, D., Wheeler, W., 2001. POY documentation and command summary. Available at ftp://ftp.amnh.org/pub/molecular/poy. Goloboff, P.A., 1994. NONA: A Tree Searching Program. Program and documentation, published by the author, Tucumán, Argentina. Goloboff, P.A., 1996. PHAST, Version 1.1. Program and documentation, published by the author, Tucumán, Argentina. Graham, S.W., Reeves, P.A., Burns, A.C.E., Olmstead, R.G., 2000. Microstructural changes in noncoding chloroplast DNA: interpretation, evolution, and utility of indels and inversions in basal angiosperm phylogenetic inference. Int. J. Plant Sci. 161, S83–S96. Grant, T., Kluge, A.G., 2003. Data exploration in phylogenetic inference: scientific, heuristic, or neither. Cladistics 19, 379–418. Grout, A.J., 1931. Moss Flora of North America North of Mexico, Vol. 3(2). Published by author, New York. Hedenäs, L., 1989. Some neglected character distribution patterns among the pleurocarpous mosses. Bryologist 92, 157–163. Hedenäs, L., 1992. Flora of Madeiran pleurocarpous mosses (Isobryales, Hypnobryales, Hookerales). Bryophyt. Bibl. 44, 1–165. Hedenäs, L., 2002. An overview of the family Brachytheciaceae (Bryophyta) in Australia. J. Hattori Bot. Lab. 92, 51–90. Higgins, D.G., Bleasby, A.J., Fuchs, R., 1991. CLUSTAL V: improved software for multiple sequence alignment. CABIOS 8, 189–191. Hong, L., Stevenson, J.K., Roth, W.B., Hallick, R.B., 1995. Euglena gracilis chloroplast psbB, psbT, psbH and psbN gene cluster: regulation of psbB-psbT pre-mRNA processing. Mol. Gen. Genet. 247, 180–188. Huttunen, S., Ignatov, M.S., Müller, K., Quandt, D., 2004. Phylogeny and evolution of epiphytism in the three moss families Mete- oriaceae, Brachytheciaceae and Lembophyllaceae. Monogr. Syst. Bot., in press. Ignatov, M.S., 1998. Bryophyte flora of Altai Mountains. VIII. Brachytheciaceae. Arctoa 7, 85–152. Ignatov, M.S., 1999. On pseudoparaphyllia in Brachytheciaceae and Meteoriaceae (Musci.). Acta Bot. Fennica 165, 73–84. Ignatov, M.S., Huttunen, S., 2002. Brachytheciaceae (Bryophyta) – family of sibling genera. Arctoa 11, 245–296. Ignatov, M.S., Huttunen, S., Koponen, T., 2004. Bryophyte flora of the Hunan Province, China. 5. Brachytheciaceae (Musci.), with an overview of Eurhynchiadelphus and Rhynchostegiella in southeast Asia. Acta Bot. Fennica, in press. Ignatov, M.S., Koponen, T., Norris, D.H., 1999. Bryophyte flora of the Huon Peninsula, Papua New Guinea. LXII. Brachytheciaceae (Musci.), excluding Homalothecium and Palamocladium. Acta Bot. Fennica 165, 23–72. Källersjö, M., Albert, V.A., Farris, J.S., 1999. Homoplasy increases phylogenetic structure. Cladistics 15, 91–93. Kawakita, A., Sota, T., Ascher, J.S., Ito, M., Tanaka, H., Kato, M., 2003. Evolution and phylogenetic utility of alignment gaps within intron sequences of three nuclear genes in bumble bees (Bombus). Mol. Biol. Evol. 20, 87–92. Kelchner, S.A., Wendel, J.F., 1996. Hairpins create minute inversions in non-coding regions of chloroplast DNA. Curr. Genet. 30, 259– 262. Kindberg, N.C., 1897. Species of European and Northamerican Bryinae (Mosses). Part 1. Pleurocarpous. Linkoeping Lithografiska Aktiebolag, Linkoeping. McAdam, S.V., 1982. Chromosome evolution in family Brachytheciaceae. J. Bryol. 12, 233–258. McFarland, K.D., 1994. Rhynchostegium. Mem. NY Bot. Gard. 69, 932–941. Mickevich, M.F., Farris, J.S., 1981. The implications of congruence in Menidia. Syst. Zool. 30, 351–370. Nixon, K.C., 1999a. Winclada (BETA), Version 0.9.9. Available at http://www.cladistics.com/about_winc.html. Nixon, K.C., 1999b. The parsimony ratchet, a new method for rapid parsimony analysis. Cladistics 15, 407–414. Noguchi, A., 1991. Illustrated Moss Flora of Japan, Vol. 4. Hattori Botanical Laboratory, Nichinan. Pollock, D.D., Zwickl, D.J., McGuire, J.A., Hillis, D.M., 2002. Increased taxon sampling is advantageous for phylogenetic inference. Syst. Biol. 51, 664–671. Quandt, D., Huttunen, S., Streimann, H., Frey, W., Frahm, J.-P., 2004b. Molecular phylogenetics of the Meteoriaceae s. str. focusing on the genera Meteorium and Papillaria. Mol. Phyl. Evol., in press. Quandt, D., Huttunen, S., Tagney, R., Stech, M., 2004a. A generic revision of Lembophyllaceae based on molecular data. Syst. Bot., in press. Quandt, D., Müller, K., Huttunen, S., 2003. Characterisation of the chloroplast DNA psbT-H region and the influence of dyad symmetrical elements on phylogenetic reconstructions. Plant Biol. 5, 400–410. Quandt, D., Tangney, R.S., Frahm, J.-P., Frey, W., 2000. A molecular contribution for understanding the Lembophyllaceae (Bryopsida) based on noncoding chloroplast regions (cpDNA) and ITS2 (nrDNA) sequence data. J. Hattori Bot. Lab. 89, 71–92. Robinson, H., 1962. Generic revisions of North American Brachytheciaceae. Bryologist 65, 73–146. Robinson, H., 1987. Notes on generic concepts in the Brachytheciaceae and the new genus Steerecleus. Mem. NY Bot. Gard. 45, 678–681. Rydin, C., Källersjö, M., 2002. Taxon sampling and seed plant phylogeny. Cladistics 18, 485–513. S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 Schimper, W.P., 1876. Synopsis Muscorum Europaeorum, praemissa introductione de elementis bryologicis tractante. Ed. 2. Sumptibus Librariae E. Schweizerbart (E. Koch), Stuttgart. Schulmeister, S., Wheeler, W.C., Carpenter, J.M., 2002. Simultaneous analysis of the basal lineages of Hymenoptera (Insecta) using sensitivity analyses. Cladistics 18, 455–484. Shaw, A.J., 2000. Phylogeny of the Sphagnopsida based on chloroplast and nuclear DNA sequences. Bryologist 103, 277–306. Simmons, M.P., Freudenstein, J.V., 2003. The effects of increasing genetic distance on alignment of, and tree construction from, rDNA internal transcribed spacer sequences. Mol. Phyl. Evol. 26, 444–451. Simmons, M.P., Ochoterena, H., 2000. Gaps as characters in sequencebased phylogenetic analyses. Syst. Biol. 49, 369–381. Stech, M., Frahm, J.-P., 1999. Systematics of species of Eurhynchium, Rhynchostegiella and Rhynchostegium (Brachytheciaceae, Bryopsida) based on molecular data. Bryobrothera 5, 203–211. Stech, M., Quandt, D., Frey, W., 2003. Molecular circumscription of the hornworts (Anthocerophyta) based on the chloroplast DNA trnL-trnF region. J. Plant Res. 116, 389–398. Stenroos, S., Hyvönen, J., Myllys, L., Thell, A., Ahti, T., 2002. Phylogeny of the genus Cladonia s.lat. (Cladiniaceae, Ascomycetes) inferred from molecular, morphological, and chemical data. Cladistics 18, 237–278. Taberlet, P., Gielly, L., Pautou, G., Bouvet, J., 1991. Universal primers for amplification of three non-coding regions of chloroplast DNA. Plant Mol. Biol. Rep. 17, 1105–1109. Takaki, N., 1955a. Researches on the Brachytheciaceae of Japan and its adjacent areas I. J. Hattori Bot. Lab. 14, 1–28. Takaki, N., 1955b. Researches on the Brachytheciaceae of Japan and its adjacent areas II. J. Hattori Bot. Lab. 15, 1–69. Takaki, N., 1956. Researches on the Brachytheciaceae of Japan and its adjacent areas III. J. Hattori Bot. Lab. 16, 1–71. Tan, B.C., 2000. Additions of the moss floras of Mt. Wilhelm nature reserve and Mt. Gahavisuka Provincial Park, Papua New Guinea. J. Hattori Bot. Lab. 89, 173–196. Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F., Higgins, D.G., 1997. The ClustalX Windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucl. Acids Res. 24, 4876–4882. Tsubota, H., Arikawa, T., Akiyama, H., de Luna, E., Gonzalez, D., Higuchi, M., Deguchi, H., 2002. Moleculart phylogeny of hypnobryalean mosses as inferred from a large-scale data set of chlororplast rbcL, with apecial reference the Hypnaceae and possibly related families. Hikobia 13, 645–665. Vanderpoorten, A., Hedenäs, L., Cox, C.J., Shaw, A.J., 2002. Circumscription, classification, and taxonomy of the Amblystegiaceae (Bryopsida) inferred from nuclear and chloroplast DNA sequence data and morphology. Taxon 51, 115–122. Wahlberg, N., Zimmerman, M., 2000. Pattern of phylogenetic relationships among members of the tripe Melitaeini (Lepidoptera: Nymphalidae) inferred from mitochondrial DNA sequences. Cladistics 16, 347–363. Warnstorf, C., 1905. Kryptogamenflora der Mark Brandenburg. Laubmoose. 2. Bontraeger, Berlin. Wenzel, J.W., Siddall, M.W., 1999. Noise. Cladistics 15, 51–64. Wheeler, W.C., 1995. Sequence alignment, parameter sensitivity, and the phylogenetic analysis of molecular data. Syst. Biol. 44, 321–331. Wheeler, W., 1996. Optimization alignment: The end of ultiple sequence alignment in phylogenetics? Cladistics 12, 1–9. Wheeler, W.C., Hayashi, C.Y., 1998. The phylogeny of the extant chelicerate orders. Cladistics 14, 173–192. White, T.J., Bruns, T., Lee, S., Taylor, J., 1990. Amplification and direct sequencing of fungal ribosomal ribosomal RNA genes for phylogenetics. In: Innis, M., Gelfand, D., Sninsky, J., White, T. 181 (Eds.), PCR Protocols: a Guide to Methods and Applications. Academic Press, San Diego, pp. 315–322. Wiley, E., 1981. Phylogenetics. The Theory and Practise of Phylogenetic Systematics. John Wiley and Sons, New York. Appendix 1 Morphological characters 0. Color of plant: 0—light green, easily turning to pale stramineous (e.g. Brachythecium albicans); 1—pure green to dark green or brownish-green (for example, the grayish green of Helicodontium); 2—light green easily turning to whitish argent (many tropical species of Rhynchostegium). 1. Black color in older parts of plants: 0—absent; 1—present. 2. Growth: 0—stem prostrate, never ascending, sympodial branches arching (Brachythecium reflexum, Eurhynchium angustirete); 1—stem ascending in apical parts (Homalothecium ssp., Brachythecium albicans, B. glareosum, etc.); 2—stem pendent (Aerolindigia, Zelometeorium, Meteoridium). 3. Foliage: 0—loose, stem visible almost throughout except shoot apices, 1—moderately dense, stem only partially visible, 2—dense, stem not seen in well-developed young shoots. 4. Branch foliage: 0—terete; 1—more or less complanate. In freely branched species this character was evaluated for secondary axes. 5. Branching: 0—irregularly pinnate; 1—regularly pinnate with rather densely arranged branches of the same length (Kindbergia, Pseudoscleropodium, some Homalothecium, also outgroup taxa). 6. Branches: 0—not curved, 1—regularly curved (cf. Scorpiurium spp.). Somewhat flexuose and occasionally curved branches are coded as 0. 7. Stipelike pattern: 0—absent; 1—present. This character means, that the proximal part of the sympodial branch has remote foliage with much smaller and strongly modified, usually deltoid leaves that are often reflexed. The pattern is quite clear when observed at a certain distance. It is not expressed in all specimens, but if it was observed in at least some, it was coded as 1. 8. Central strand: 0—present; 1—absent. Among ingroup taxa only Rhynchostegiella and Homalotheciella display the total absence of the central strand. This character was coded as 1 in cases when at least two transversal sections of well developed shoot showed no traces of a central strand, although in some other sections a 1–3-celled strand could be discerned (Zelometeorium, Rhynchostegiella teneriffae, etc.). 9. Stem cortex: 0— 0–4 cell layers thick; 1 to > 5 cell layers thick. Those cases, in which a thick cortex was observed in at least some transverse cross-sections were coded as 1. 182 S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 10. Color of upper cells of axillary hair: 0—hyaline; 1—brownish. If brownish axillary hairs were found only occasionally, this character was still coded as 0. 11. Width of upper cell of axillary hair: 0—essentially not wider than the cell below; 1—wider and inflated compared with cell below (Oxyrrhynchium spp.). Species were coded as 1 even if only at least some of the axillary hairs had this pattern. 12. Axillary hairs: 0— 2–4-celled; 1—with 5 or more cells, at least in some well-developed axillary hairs. Note that not only the upper cells in the axillary hairs were counted, but also the basal cells, which sometimes differ in shape and coloration from apical cells. 13. Pseudoparaphyllia of Brachytheciaceae ⁄ Meteoriaceae type: 0—absent; 1—present. Brachytheciaceae and Meteoriaceae each branch primordia is covered by three leaf-like pseudoparaphyllia and they are forming unique regular pattern (see Ignatov, 1999). 14. Pseudoparaphyllia of Brachytheciaceae ⁄ Meteoriaceae type: 0—acute to obtuse or truncate; 1—longtriangular or acuminate at least at some branch initials (excluding upper 2 mm, where branch initials were regarded as immature). If pseudoparaphyllia are not of B ⁄ M type, the character state was coded as inapplicable. 15. Leaves: 0—straight; 1—falcate or homomallous. Even slightly homomallous cases, as in Platyhypnidium, were coded as 1. 16. Stem leaves: 0—erecto patent to patent (Brachythecium rutabulum); 1—erect to erect-appressed (Homalothecium spp.); 2—imbricate (B. cirrosum); 3—reflexed to squarrose (B. reflexum, Zelometeorium patulum). 17. Stem leaf length: 0— 1.1–2.3 mm; 1 to > 2.3 mm; 2 to <1.1 mm. The largest leaves were measured, and hence species was coded as 1 if at least some leaves were longer than 2.3 mm and 2 if no leaves were longer than 1.1 mm). 18. Stem leaf length to width ratio: 0 to > 2 : 1; 1 to < 2 : 1. Species was coded as 1 if no less than half of the leaves from the optimally developed stems in their apical part (5 mm) showed a ratio of 2 : 1 or if they were still shorter. 19. Leaves: 0— > 0.4 mm wide; 1 to < 0.4 mm wide. The largest leaves were evaluated, and species was coded as 1, if no leaves were found wider than 0.4 mm. 20. Stem leaf shape: 0—acuminate with acumen no less than 1 ⁄ 5 of leaf length; 1—leaves obtuse, acute, apiculate, or very shortly acuminate with acumen less than 1 ⁄ 5 of leaf length; 2—abruptly piliferous from rather rounded part; 3—attenuate. 21. Acumen: 0—not distinctly flexuose; 1—distinctly flexuose (Squamidium, Meteoridium, Zelometeorium, etc.). 22. Stem leaves: 0—plane; 1—with concave basal part; 2—entire leaf concave; 3—channeled from the base. 23. Stem leaves: 0—not twisted when dry in mid-leaf; 1—twisted in mid-leaf (Platyhypnidium, Rhynchostegium rotundifolium). 24. Stem leaves: 0—without distinct longitudinal plicae; 1—with distinct longitudinally plicae (Homalothecium spp., some Brachythecium spp.). The character is observed under stereo-microscope when leaves are still attached on plant, as in slide more plicae appear due to pressing of concave leaves. 25. Stem leaf base: 0—not clasping; 1—clasping. Species was coded as 1 if the majority of leaves had a clasping base. Taxa with U-shaped base, e.g. Aerolindigia, with indistinct aspect of clasping in a few leaves were coded as 0. 26. Stem leaf: 0—not longly and broadly decurrent, or decurrences lacking; 1—longly and broadly decurrent. These differences are usually explained in literature as leaves decurrent ⁄ non-decurrent. In fact, almost all pleurocarpous leaves have some decurrencies, which however rarely remain with leaf after it is detached. Decurrencies remain with leaf usually when they (triangulars extending out of line of insertion) are no less than four cells wide and four cells long, at least in some leaves. The latter case is that is coded here as 1. 27. Stem leaf base: 0—not auriculate; 1—auriculate. 28. Margin of stem leaves: 0—serrulate to serrate; 1—slightly serrulate to subentire; 2—serrate, at least some teeth backward reflexed. 29. Margin of stem leaves toothed: 0—not to the base; 1—almost to the base. 30. Costa in stem leaves: 0—single, including rarely forking case of Platyhypnidium; 1—costa absent or double. 31. Costa length in stem leaves: 0— 0.3–0.8 · total leaf length; 1 to < 0.3; 2 to > 0.8. If costa double or absent, character was coded as inapplicable. Maximal values were used, and hence if in at least two leaves from optimally developed stem costa is longer than 0.8· of total leaf length, character was coded as 2 and 1 if we were not able to find two leaves from one stem in which costa was longer than 0.3· leaf length. 32. Costa width in stem leaves: 0— < 1/4 of leaf width; 1— > 1/4 of leaf width. Some species in the Brachytheciaceae, e.g. Donrichardsia macroneuron, have exceptionally wide costae. 33. Costa in stem leaves ending: 0—without spine; 1—with spine. 34. Laminal cell length in stem leaves: 0—always < 110 lm; 1—at least sometimes > 110 lm. 35. Laminal cell proportion in stem leaves: 0—at least 50% of cells are > 10 : 1; 1—4 to 10 : 1; 2—at least 50% of cells are < 4 : 1. This character was measured in mid-leaf cells, excluding paracostal area. 36. Laminal cell in stem leaves: 0—thin-walled, lumen to wall rate > 3 in more than half of the cells; 1—thickwalled, no less than half of the cells in mid-leaf have lumen to wall rate < 3. This was measured with a Leitz S. Huttunen & M. S. Ignatov / Cladistics 20 (2004) 151–183 Duallux microscope at 40· magnification from a water slide. Error margin in measurements was ± 0.5 lm. 37. Laminal cells in stem leaves: 0—smooth; 1—papillose. 38. Laminal cells: 0—weakly prorate; 1—distinctly prorate. This character refers to distinctly and sharply exserted upper ends of cells (Bryhnia, Kindbergia, etc.). Minutely prorate ends, as in some Rhynchostegium species, are coded as 0. Sometimes this character is expressed in branch leaves only. 39. Basal laminal cells to the base of stem leaves: 0—slightly wider than laminal cells, more or less transparent; 1—1.5 times wider than mid-leaf cells in no less than five rows across the base (at least in two leaves from one shoot, from apical 5 mm), lax and transparent; 2—equal to slightly wider than laminal cells, in opaque group across the base. 40. Alar cells: 0—not different from juxtacostal cells; 1—in group of larger, thin-walled cells; 2—in group of small, opaque cells. Character state 2 also includes: (A) rather large, but thick-walled, porose, resulting in opaque appearance, cf. Meteoridium and (B) a group that is a widening of the thick-wall celled area across the base, as in Brachythecium plumosum. This character was coded according to the most differentiated cases. For example, in Rhynchostegium species angular cells are often not differentiated, but when they are, the alar cells are larger and thin-walled, and hence were coded as 1. Similarly, Scorpiurium was coded as 2, although there are sometimes opaque cells throughout the base, and the angular group is poorly differentiated. 41. Alar group shape: 0—triangular to subquadrate; 1—elongate along the margin. 42. Branch leaf shape: 0—acuminate with acumen no less than 1 ⁄ 5th of leaf length; 1—leaves obtuse, acute, apiculate, or very shortly acuminate with acumen less than 1 ⁄ 5th of leaf length; 2—abruptly piliferous from rather rounded part; 3—attenuate. 43. Costa in branch leaves ending: 0—without spine; 1—with spine. 44. Costa in branch leaves by the apex: 0—not toothed; 1—toothed. This character does not include the single spine at the very apex. 45. Sexual condition: 0—autoicous; 1—dioicous, including dioicous + rarely autoicous cases (for example Brachythecium rivulare); 2—phyllodioicous, incuding dioicous + phyllodioicous case; 3—partly synoicous. In Aerobryum speciosum phyllodioicousity has not been reported before and this was observed in 3 specimens: Bhutan, Griffin 462 (BM!); Sikkim, Himalaya, Levier (BM!). 46. Perichaetial leaves: 0—reflexed; 1—straight. 47. Seta: 0—longer than branches; 1—shorter than branches, but longer than two urn lengths; 2—shorter than two urn lengths. Branch length is measured in place 183 where branching is regularly pinnate and branches show rather stable length. 48. Seta: 0—smooth; 1—rough or at least slightly rough, including cases such as Clasmatodon. 49. Capsule orientation: 0—inclined to horizontal; 1—erect. In latter case the urn is sometimes plainly curved in the middle, as occasionally in Leskea polycarpa, but never below mouth. In Homalothecium lutescens and some other species with capsule orientation varying from almost erect to distinctly inclined (and curved also below mouth) are coded as 0. 50. Stomatal pore: 1—round; 0—elongate. 51. Operculum: 0—beak > 2· the height of the rest of the operculum; 1—beak present but < 2· the height of the rest of the operculum; 2—conic. 52. Annulus: 0—consisting of 2–3 rows of large cells, partly separated; 1—consisting of 1()2) rows of cells, almost nonseparated; 2—absent. 53. Exostome: 0—xerocastique; 1—hygrocastique. Aerolindigia was coded as xerocastique. The exostome teeth in this species turn straight when wet, but form a high conus enclosing the capsule mouth, and hence it is regarded as functionally xerocastique. 54. Basal part of exostome teeth: 0—with red color; 1—without red color. 55. Exostome teeth transitional zone: 0—short (1–4 plates); 1—expanded. 56. Transversal striae between dorsal trabeculae in proximal part of tooth: 0—present; 1—absent. 57. Striae in lower part of exostome teeth: 0—nonpapillose including sometimes reticulating striae of Rhytidiadelphus; 1—papillose above striae, giving an appearance of papillose surface in the basal part under a light microscope. 58. Basal membrane of endostome: 0—higher than 1 ⁄ 4 of exostome; 1—lower than 1 ⁄ 4 of exostome. 59. Endostome segments: 0—at least some segments broadly and more or less regularly perforated; 1—narrowly to not perforated; 2—absent or adherent to teeth. 60. Cilia: 0— > 0.8· length of free part of segment; 1— 0.3–0.8; 2— < 0.3. 61. Spores: 0—diameter never > 25 lm; 1—at least sometimes > 25 lm. 62. Calyptra: 0—naked; 1—at least sparsely hairy. Appendix 2 Command line used in all POY analyses: poy -parallel -solospawn 7 -maxtrees 5 -holdmaxtrees 20 -random 15 -multibuild 14 -treefuse -fuselimit 50 -fusingrounds 1 -driftspr -numdriftspr 5 -drifttbr -numdrifttbr 5 -slop 1 -checkslop 3 -seed -1 -noleading -fitchtrees -norandomizeoutgroup -indices -impliedalignment

RELATED PAPERS

RELATED TOPICS

Log In

Phylogeny of the Brachytheciaceae (Bryophyta) based on morphology and sequence level data

Phylogeny of the Brachytheciaceae (Bryophyta) based on morphology and sequence level data

Related Papers

RELATED PAPERS

RELATED TOPICS