Phylogeny and Character Evolution of the Bolbitidoid Ferns (Dryopteridaceae) Author(s): Robbin C. Moran, Paulo H. Labiak, and Michael Sundue Source: International Journal of Plant Sciences, Vol. 171, No. 5 (June 2010), pp. 547-559 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/10.1086/652191 . Accessed: 06/06/2013 16:22 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. . The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to International Journal of Plant Sciences. http://www.jstor.org This content downloaded from 69.74.186.251 on Thu, 6 Jun 2013 16:22:22 PM All use subject to JSTOR Terms and Conditions Int. J. Plant Sci. 171(5):547–559. 2010. Ó 2010 by The University of Chicago. All rights reserved. 1058-5893/2010/17105-0009$15.00 DOI: 10.1086/652191 PHYLOGENY AND CHARACTER EVOLUTION OF THE BOLBITIDOID FERNS (DRYOPTERIDACEAE) Robbin C. Moran,1 ,* Paulo H. Labiak,y and Michael Sundue* *New York Botanical Garden, Bronx, New York 10456-5126, U.S.A.; and yUniversidade Federal do Paraná, Departamento de Botânica, CP 19031, Curitiba PR 81531-980, Brazil We performed a phylogenetic analysis of the traditionally recognized genera of bolbitidoid ferns (i.e., Arthrobotrya, Bolbitis, Elaphoglossum, Lomagramma, and Teratophyllum) using two noncoding chloroplast spacers: trnL-trnF and rps4-trnS. The sampling included 57 species, of which 55 had not been sequenced previously. The results supported the monophyly of bolbitidoid ferns and of Arthrobotrya, Elaphoglossum, Lomagramma, and Teratophyllum; however, Bolbitis was resolved as polyphyletic. A clade of eight Neotropical species currently placed in Bolbitis is sister to Elaphoglossum, not the other species of Bolbitis. We refer to this group of species as the Bolbitis nicotianifolia clade. Lomagramma (or Bolbitis) guianensis, whose generic placement has been uncertain, was found to belong to the B. nicotianifolia clade. Bolbitis s.s. was resolved sister to the rest of the bolbitidoid ferns, which are in turn divided into two clades, one consisting of Elaphoglossum and the B. nicotianifolia clade and the other of Lomagramma, Teratophyllum, and Arthrobotrya. We optimized 34 morphological characters on the resulting phylogenetic tree. The characters found to be synapomorphic for bolbitidoid ferns were ventral root insertion, elongated ventral meristeles, sterile-fertile leaf dimorphism, acrostichoid sori, and the absence of hairs on the leaves. Other characters, such as articulate pinnae, venation patterns, laminar buds, paraphyses, and growth habit, are discussed in relation to the clades they support at different nodes on the tree. The bolbitidoid ferns show a transition series from terrestrial (Bolbitis) to hemiepiphytic (the B. nicotianifolia clade, Arthrobotrya, Lomagramma, and Teratophyllum) to epiphytic (Elaphoglossum). A sister-species relationship between the Neotropical Bolbitis serratifolia and the African Bolbitis acrostichoides was recovered, supporting their relationship as previously postulated on the basis of morphology. Keywords: pteridophytes, Arthrobotrya, Bolbitis, Elaphoglossum, Lomagramma, Teratophyllum. Online enhancements: appendix tables. Introduction within the Dryopteridaceae, a family to which Lomariopsis does not belong. Instead, Lomariopsis forms a clade with Cyclopeltis and Nephrolepis, a clade now recognized as the Lomariopsidaceae (Smith et al. 2006; Schuettpelz and Pryer 2007). Three previous studies have dealt with the phylogeny of the Dryopteridaceae, but they included no species or only a few species of bolbitidoid ferns. The first study (Li and Lu 2006), based primarily on Chinese species, did not include any bolbitidoid ferns. The second study (Liu et al. 2007) included six species (two species of Bolbitis, two of Elaphoglossum, one Lomagramma, and one Arthrobotrya [as Teratophyllum]). The third study included 29 species (Schuettpelz and Pryer 2007), 25 of which were Elaphoglossum, with the remaining four Bolbitis auriculata (Lam.) Alston, Bolbitis nicotianifolia (Sw.) Alston, Teratophyllum wilkesianum Holttum, and Lomagramma guianensis (Aubl.) Ching. Both studies that included bolbitidoid ferns supported the group’s monophyly. Noteworthy was the finding by Schuettpelz and Pryer (2007) that Bolbitis was polyphyletic—a result that in part prompted our study. Given the small sampling of bolbitidoid ferns in the previous studies, the purpose of our article is to infer the phylogeny of this group, using more species, The bolbitidoid ferns are a clade of dryopteroid ferns termed the ‘‘former lomariopsids’’ by Schuettpelz and Pryer (2007). The clade is pantropical and consists of five commonly recognized genera: Arthrobotrya J. Sm. (3 spp.), Bolbitis Schott (;55 spp.), Elaphoglossum Schott (;600 spp.), Lomagramma J. Sm. (22 spp.), and Teratophyllum Mett. (11 spp.). These genera contain ;680 species, or ;40% of the estimated 1700 species of Dryopteridaceae (Smith et al. 2006). Previously, these five genera of bolbitidoid ferns were classified in the Lomariopsidaceae (e.g., Holttum 1978; Moran 1995) because they share with Lomariopsis an elongated ventral meristele (as seen in cross section) that produces all the roots, sterile-fertile leaf dimorphy, and acrostichoid sori (Holttum 1978). Phylogenetic studies based on DNA sequences, however, reveal that the bolbitidoids are nested 1 Author for correspondence; e-mail: rmoran@nybg.org. Manuscript received November 2009; revised manuscript received January 2010. 547 This content downloaded from 69.74.186.251 on Thu, 6 Jun 2013 16:22:22 PM All use subject to JSTOR Terms and Conditions 548 INTERNATIONAL JOURNAL OF PLANT SCIENCES and to examine the evolution of its morphological characters by optimizing these characters onto the resulting cladogram. Methods Taxon Sampling Within the ingroup (i.e., bolbitidoid ferns) we sampled 49 species representing all four of the traditionally recognized genera: Elaphoglossum (10 species sampled, representing ;2% of the species in the genus), Bolbitis (24, ;45%), Lomagramma (10, 44%), and Teratophyllum (5, ;55%). Although species sampling within Elaphoglossum was low, the 10 species included represent all of the major clades in the genus recognized in the molecular phylogenetic study by Rouhan et al. (2004). We did not consider it worthwhile to include more species of Elaphoglossum because a previous study by Rouhan et al. (2004) had already included 123 species. The species we sampled in Bolbitis represent seven of the 10 series recognized by Hennipman (1977) as well as five species that Hennipman considered incertae sedis or that were described after the publication of his monograph. For outgroups, we used four genera: Stigmatopteris (represented in our analysis by one species), Rumohra (one species), Megalastrum (three species), and Lastreopsis (three species). We chose these genera as outgroups because, according to Schuettpelz and Pryer (2007), they are the ones most closely related to the ingroup. Voucher information and GenBank accession numbers are listed in appendix A (available in the online edition of the International Journal of Plant Sciences). DNA Extraction DNA extractions were performed using the Qiagen DNeasy Kit (Valencia, CA) for samples either dried in silica gel or taken from herbarium specimens. Because many of the specimens were from herbarium collections, we changed some of the steps suggested in the manufacturer’s protocol. The samples (;1 cm2) were disrupted using a small bead and sterile garnet in a FastPrep machine at 5 m/s for 15 s. Then, for the tissue lysis, we added 30 mL of proteinase K (20 mg/ mL) and 30 mL of B-mercaptoethanol (98%) to the lysis buffer per tube and incubated the solution at 42°C for 12 h. The following steps were then the same as the Qiagen DNeasy kit protocol except by the final elutions: we used two final elutions, of 75 mL each, into the same Eppendorf tube. PCRs and Sequencing PCRs were performed using 1 mL of genomic DNA. For the trnL-trnF spacer we used the universal primers e and f, designed by Taberlet et al. (1991), and the primers rps4-3r.f (Skog et al. 2004) and trnSr for the rps4-trnS spacer (SouzaChies et al. 1997). PCR amplifications were performed using 1 mL of nondiluted genomic DNA, 2.5 mL of 10X Taq buffer with 15 mM of MgCl2 added, 2.5 mL of dNTPs, 5 mL of Q solution, 2.5 mL of 2.5 mg/mL BSA solution, 1 mL of each primer at 10 mM, 0.2 mL of Taq DNA polymerase, and 9.3 mL of purified water. For both spacers we used the same amplification protocol, beginning with an initial denaturation cycle of 5 min at 94°C and then 35 cycles of 1 min at 94°C, 30 s at 50°C, 1 min at 72°C, and a final extension period of 7 min at 72°C. The resulting PCR products were then checked on a 1% agarose gel with ethidium bromide. The PCR products were sequenced by the High-Throughput Genomics Unit at the University of Washington (http://www.htseq.org/index.html), using the same primers that were used for amplification. Alignment and Phylogenetic Analysis Sequences were visually edited using Sequencher 4.9 (Gene Codes), and the consensus sequences were aligned using Muscle (ver. 3.6; Edgar 2004). The resulting alignment was manually revised when necessary. Once aligned, the resulting gaps were coded following the simple coding model suggested by Simmons and Ochoterena (2000), using the program 2xread (Little 2005). The two data matrices were constructed using Mesquite 2.6 (Maddison and Maddison 2009) and analyzed using equally weighted maximum parsimony and Bayesian inference. All trees were rooted using one of the outgroups, Stigmatopteris prionites. Maximum parsimony (MP). These analyses were performed using the TNT software (Goloboff et al. 2008). An initial MP analysis was performed on each of the two markers separately (table 1), and the resulting topologies were compared to test the null hypothesis that both markers are congruent in their phylogenetic information. In all MP analyses, heuristic searches were performed using TNT (Goloboff et al. 2008), with 1000 parsimony ratchets replicates (Nixon 1999; 200-iteration ratchet, the up and down weights set to 5% each), holding 20 trees per ratchet, with TBR-max branch swapping. Relative support for each node (bootstrap support) was calculated for the combined data set, with 1000 bootstrap replicates doing 10 ratchets per replicate, holding 20 trees per ratchet. Bayesian analysis. Bayesian analysis was performed using MrBayes (ver. 3.1.2; Ronquist and Huelsenbeck 2003). To select the model of DNA substitution, we first used PAUP* (ver. 4.10b; Swofford 2002) with a neighbor-joining tree to calculate the likelihood values, and then the evolutionary model that best fit the data was identified using the hierarchical likelihood ratio test and the Akaike information criterion (Akaike 1973), as implemented in MrModelTest (Nylander et al. 2004). For the selected model, three separate runs were started from random trees. For each run, we used four different chains, one cold and three heated, with the temperature parameter set to 0.05 to ensure a good mixing. The gap characters were also included in the analysis and were set to follow the model implemented in MrBayes for binary data (l set coding ¼ variable). The parameters for each partition were allowed to evolve independently, using the ‘‘unlink’’ command. The analysis was run for 10 million generations, sampling every thousandth generation, with ‘‘burn-in’’ fraction set at 25%. In order to assess if MCMC reached stationarity after the ‘‘burn-in’’ period, we examined the loglikelihood (lnL) plots using Tracer (ver. 1.3; Rambaut and Drummond 2003). This content downloaded from 69.74.186.251 on Thu, 6 Jun 2013 16:22:22 PM All use subject to JSTOR Terms and Conditions MORAN ET AL.—BOLBITIDOID FERN PHYLOGENY 549 Table 1 Maximum Parsimony Statistics after the Heuristic Search for Each Marker and for the Combined Data Set Number of taxa Aligned length Parsimony informative (%) Confidence index Retention index Number of trees Tree length trnL-trnF Gaps coded for trnL-trnF rps4-trnS Gaps coded for rps4-trnS Combined without coded gaps Combined with coded gaps 73 558 55 .49 .81 60 811 73 122 48 ... ... ... ... 76 688 53 .53 .83 30 912 76 89 51 ... ... ... ... 77 1246 54 .51 .82 22 1728 77 1457 54 .51 .82 24 1927 Also, the convergence between the different runs was examined by looking at the posterior probability of each clade, as suggested by Huelsenbeck and Bollback (2001), using the online program AWTY (http://king2.scs.fsu.edu/CEBProjects/ awty/awty_start.php; Wilgenbusch et al. 2004). Morphological Analysis Thirty-eight characters were scored (app. B [available in the online edition of the International Journal of Plant Sciences]) from herbarium specimens and photos, and, for some taxa (Bolbitis bipinnatifida, Bolbitis fluviatilis, Bolbitis major, Bolbitis tibetica), partially from the literature (Nayar 1966; Hennipman 1977; Holttum 1978; Ching and Wu 1983). Winclada (Nixon 2004) was used to build the morphological data matrix, draw the tree, and optimize characters on the tree resulting from the molecular phylogenetic analysis. Characters are unordered, and optimizations are reported as unambiguous except where noted in the text. Morphological Characters Habit 1. Habit—(0) terrestrial, epipetric, or rheophytic, (1) scandent, (2) epiphytic. Plant habits were determined from herbarium labels and personal observations in the field. The scandent state includes both low- and high-climbing plants Rhizomes 2. Root insertion—(0) radial, (1) ventral. 3. Ground tissue (pith and cortex) color of freshly cut rhizome—(0) white, (1) green, (2) red. This character was scored only for taxa where fresh material or photographs of the sectioned rhizome were available. 4. Elongate ventral meristele (fig. 1A–1C)—(0) absent, (1) present. 5. Rhizome with scattered strands of sclerenchyma present in the ground tissue—(0) absent, (1) present. 6. Rhizome aculeate—(0) nonaculeate, (1) aculeate. Rhizome Indument 7. Rhizome scale attachment—(0) basal, (1) peltate. 8. Rhizome scale margin—(0) nonglandular, (1) glandular. 9. Rhizome scale cell lumens—(0) clathrate, (1) nonclathrate. Petioles Phyllopodia were considered present when there was an abrupt change in color or width (or both) at the base of the petioles. 10. Petiole articulate to rhizome—(0) absent, (1) present. 11. Phyllopodium—(0) absent, (1) present. Leaves In some bolbitidoid genera (i.e., Lomagramma and Teratophyllum), sterile leaves on the climbing rhizomes are differentiated into bathyphylls and acrophylls, which differ in size, cutting, and shape (Holttum 1978). Bathyphylls are present on the lower portions of the rhizomes; the upper portions bear acrophylls. 12. Number of ranks of leaves—(0) two ranks, (1) more than two ranks. The number of leaf ranks in bolbitidoids varies from two to six. We found that for plants with more than two ranks of leaves, it was not possible to define discrete character states; thus, we combined all numbers of ranks greater than two into a single character state ‘‘more than two.’’ 13. Leaf division of acrophylls—(0) simple, (1) 1-pinnate, (2) 1-pinnate-pinnatifid or -pinnatisect, (3) 2-pinnate, (4) 2-pinnatepinnatifid or -pinnatisect, (5) 3-pinnate, (6) 3-pinnate-pinnatifid or -pinnatisect. 14. Apical pinna shape—(0) conform (resembling lateral pinnae), (1) pinnatifid. 15. Sterile leaf differentiation—(0) absent: sterile leaves of mature plants all similar, (1) present: sterile leaves of mature plants differentiated into bathyphylls and acrophylls. 16. Dimorphy of fertile leaves—(0) absent: fertile leaves similar to sterile leaves, (1) present: fertile leaves differentiated from sterile leaves (with reduced green laminar tissue). 17. Lateral pinna articulation—(0) not articulate, (1) articulate, (2) indistinctly articulate. Pinnae were scored as articulate when both darkened tissue and a line of articulation were present at the base of the petiole. Pinnae did not have to be deciduous to be considered articulate. Pinnae were scored as indistinctly articulate when only one of the two criteria was present. 18. Apical pinna articulation—(0) not articulate, (1) articulate, (2) indistinctly articulate. This character was scored using the same criteria as character 19. This content downloaded from 69.74.186.251 on Thu, 6 Jun 2013 16:22:22 PM All use subject to JSTOR Terms and Conditions 550 INTERNATIONAL JOURNAL OF PLANT SCIENCES Fig. 1 Rhizome cross sections and paraphyses of some bolbitidoid ferns. A–C, Rhizome cross sections. Ventral side of rhizome at bottom of photo. Note the elongated ventral meristele, a synapomorphy for the bolbitidoid clade. A, Bolbitis nicotianifolia. B, Bolbitis portoricensis. C, Elaphoglossum productum. Scale bars ¼ 5 mm. D–F, Paraphyses of Teratophyllum, Arthrobotrya, and Lomagramma. D, Teratophyllum aculeatum (Corner 2593, NY). E, Arthrobotrya wilkesiana (Bandonin 60, NY). F, Lomagramma cordipinna (Yuncker 9300a, NY). Scale bars ¼ 100 mm. 19. Pinna margins of acrophylls—(0) entire, (1) toothed or lobed. 20. Bathyphyll dissection—(0) 2-pinnate, (1) 1-pinnate. 21. Filiform teeth in the sinuses of pinna lobes—(0) absent, (1) present. This character not applicable for taxa with simple leaves. 22. Proliferous buds—(0) absent, (1) present. 23. Proliferous bud position (fig. 2)—(0) distal on adaxial surface of pinnae or segments, (1) proximal on acroscopic margin. Adaxial buds are usually associated with the terminal leaf segment, whereas proximal buds occur on the acroscopic side of a petiolule where it joins the green laminar tissue. This character is inapplicable for species without buds. character not applicable for taxa without free included veinlets. 27. Arcuate secondary cross-veins between main lateral veins—(0) absent, (1) present. Arcuate cross-veins are thicker than tertiary veins. 28. Hydathodes—(0) absent, (1) present. Leaf Indument All Dryopteridaceae have scales on their leaves. Hairs are only considered present when they appear to be distinct from laminar scales and not serially homologous with them. 29. Hairs—(0) absent, (1) present. Sori Leaf Venation 24. Venation—(0) free, (1) anastomosing, (2) free except for a single row of costal areoles. 25. Free veinlets included in the areoles—(0) absent, (1) present. 26. Direction of free included veinlets in areoles—(0) excurrent, (1) recurrent, (2) both recurrent and excurrent. This 30. Sorus type—(0) acrostichoid, (1) discrete. 31. Soral paraphyses—(0) absent, (1) present. 32. Soral paraphysis type (fig. 1D–1F)—(0) peltate, (1) branched, (2) simple. These are scored as states of a single character because they appear to have been derived from scales. 33. Indusium—(0) absent, (1) present. This content downloaded from 69.74.186.251 on Thu, 6 Jun 2013 16:22:22 PM All use subject to JSTOR Terms and Conditions MORAN ET AL.—BOLBITIDOID FERN PHYLOGENY 551 Fig. 2 Position of proliferous buds in bolbitidoid ferns. A, Bolbitis portoricensis (McVaugh 18983, MICH). B, C, Bolbitis hastata (Hallberg 1597, NY). D, Bolbitis hemiotis (Hart 6924, NY). Scale bar ¼ 5 cm. 34. Indusium shape—(0) reniform, (1) peltate. This character not applicable for taxa that do not have an indusium. Results The results of our molecular analysis support the monophyly of the bolbitidoid ferns (fig. 3; bootstrap support [BS] 100; posterior probability [PP] 1; five morphological synapomorphies). They also support the monophyly of three of the five genera, namely, Arthrobotrya (BS 99, PP 1; one morpho- logical synapomorphy), Elaphoglossum (BS 100; PP 1; five morphological synapomorphies), and Teratophyllum (BS 100; PP 1; one morphological synapomorphy). The two remaining genera, Bolbitis and Lomagramma, were resolved as polyphyletic: Bolbitis consisted of two clades, one American (Bolbitis oligarchica–Acrostichum scandens, as the Bolbitis nicotianifolia clade, BS 95; PP 1) and the other pantropical (Bolbitis heteroclita–Bolbitis humblotii, as Bolbitis, BS 100; PP 1; one morphological synapomorphy), which includes the type of Bolbitis. Lomagramma was resolved as polyphyletic This content downloaded from 69.74.186.251 on Thu, 6 Jun 2013 16:22:22 PM All use subject to JSTOR Terms and Conditions 552 INTERNATIONAL JOURNAL OF PLANT SCIENCES Fig. 3 Fifty-percent majority rule consensus tree from the Bayesian analyses of the combined data set (rps4-trnS þ trnL-trnF þ coded gaps). The clades representing the genera in the bolbitidoid clade are indicated by vertical lines, as well as the outgroups and Elaphoglossum. Numbers following species names indicate the different accessions for the same species (see app. A in the online edition of the International Journal of Plant Sciences). Stars indicate values of posterior probabilities (above the branches) and bootstrap (below the branches) equal to 1 and 100%, respectively. Species in boldface are the types of the genera indicated at right. because its sole American species, Lomagramma guianensis, was not resolved with other species in its genus; instead, it was sister to the entirely American clade of Bolbitis. Deeper nodes resolve Bolbitis s.s. as sister to the remaining bolbitidoids (BS 100; PP 1). Among the remaining genera, there are two main clades. The first includes Lomagramma, which is sister to Teratophyllum þ Arthrobotrya (BS 100; PP 1). The second includes Elaphoglossum and the American clade segregated from Bolbitis (as the B. nicotianifolia clade; fig. 3; BS 100; PP 0.95). This content downloaded from 69.74.186.251 on Thu, 6 Jun 2013 16:22:22 PM All use subject to JSTOR Terms and Conditions MORAN ET AL.—BOLBITIDOID FERN PHYLOGENY Discussion 553 species relationship previously postulated on the basis of morphology (Hennipman 1977; Moran and Smith 2001). Bolbitidoid Clade The monophyly recovered here for the bolbitidoid ferns agrees with previous studies (Liu et al. 2007; Schuettpelz and Pryer 2007). Morphological synapomorphies uniting the bolbitidoids include ventral root insertion (character 2, state 1), the presence of an elongate ventral meristele (4, 1), dimorphic fertile leaves (16, 1), the absence of hairs on the leaves (29, 0), and acrostichoid sori (30, 0; fig. 4). These character states occur without losses within the bolbitidoid ferns. Outside of the bolbitidoids, however, these characters evolved independently, most notably in Lomariopsis, which shares all of the above character states. Articulate pinnae (17, 1), which are an apparent synapomorphy for Lomariopsidaceae (R. Moran, personal observation), are also common among bolbitidoids (fig. 7). These examples of homoplasy are perhaps best explained as convergent evolution in response to the hemiepiphytic habit that is found in Lomariopsis and commonly among bolbitidoid ferns (Bolbitis nicotianifolia clade, Arthrobotrya, Lomagramma, Teratophyllum; fig. 5). As presented here, the bolbitidoid ferns comprise six genera. With the exception of Bolbitis (Schuettpelz and Pryer 2007) and Elaphoglossum (Rouhan et al. 2004; Skog et al. 2004; Schuettpelz and Pryer 2007), the monophyly of these remaining genera had not been tested previously. In previous phylogenetic studies (Liu et al. 2007; Schuettpelz and Pryer 2007), only one species of Arthrobotrya, Lomagramma, and Teratophyllum were included. Our results are the first to support the monophyly of each of these three genera and reveal that collectively they form a clade. In contrast, Bolbitis formed two clades in our analysis (fig. 3). This agrees with the finding of Schuettpelz and Pryer (2007) for the two species of Bolbitis in their analysis (i.e., Bolbitis auriculata and B. nicotianifolia, also included in our study). Of the two clades (fig. 3), the name Bolbitis applies to the clade Bolbitis heteroclita–Bolbitis humblotii (fig. 3) because its type species, Bolbitis serratifolia (Kaulf.) Schott. (included in our analysis), belongs here. Bolbitis Bolbitis s.s. can usually be easily diagnosed. Most of its species have anastomosing veins (character 24, state 1), serrate pinnae (19, 1), and proliferous buds (22, 1) adaxially on the apex of terminal segments (23, 0; the apices, in some species, are flagelliform; fig. 2A–2C). In our analysis using an unambiguous optimization, the presence of two leaf ranks (12, 0) is the only morphological synapomorphy uniting Bolbitis s.s.; however, many losses among derived species reduce the diagnostic power of this character state. Proliferous buds (22, 1) and anastomosing veins (24, 1) act as additional synapomorphies for Bolbitis under ACCTRAN and DELTRAN, respectively. Bolbitis s.s. comprises ;55 species and is pantropical (fig. 6C). As circumscribed here, it is either terrestrial or epipetric but rarely can be found as low-climbing hemiepiphytic plants. It is frequently associated with riparian habitats, and several species are rheophytic. In our analysis, the Neotropical Bolbitis serratifolia and the African Bolbitis acrostichoides were recovered as sister species (fig. 3). This supports their sister- Bolbitis nicotianifolia Clade The second clade of Bolbitis is formed by Acrostichum scandens–Bolbitis oligarchica (fig. 3). (NB: Acrostichum scandens has long been considered a synonym of Lomagramma guianensis; however, we believe it is a distinct species and plan to provide a combination for it.) We found no morphological synapomorphies for this clade using an unambiguous optimization; however, the genus can often be distinguished from Bolbitis s.s. because many of the species have conform terminal pinnae (i.e., the terminal segment that resembles the lateral pinnae; character 14, state 0), anastomosing venation (24, 1) with free included veinlets (25, 1), and arcuate secondary cross-veins between the main lateral veins (27, 1). The species of the B. nicotianifolia clade are climbing (1, 1), except for Bolbitis hemiotis (Maxon) C. Chr. and B. oligarchica (Baker) Hennipman, which are terrestrial (1, 0; fig. 5). These two species bear laminar buds, as do many species of Bolbitis. But unlike that genus, the buds occur on the acroscopic side of the pinna stalks (23, 1) where the stalks join the green laminar tissue (fig. 2D). The buds are never associated with the leaf apex as in Bolbitis. The B. nicotianifolia clade represents a new genus. In a subsequent paper we plan to make the necessary new combinations for the species belonging to this new genus and provide a key to these species and a synopsis of each. In addition to those in the analysis (fig. 3), two other species belong to the B. nicotianifolia clade: Bolbitis lindigii (Mett.) C. Chr. and Bolbitis pergamentacea. Both have characters typical of the clade, namely, high-climbing habit, terminal segment resembling the lateral pinnae, and anastomosing veins between the main lateral ones. Moreover, B. lindigii greatly resembles Bolbitis bernoullii, and B. pergamentacea resembles B. hemiotis, both of which are in the B. nicotianifolia clade in this analysis (fig. 3). Following this circumscription, the B. nicotianifolia clade comprises ;10 species, all Neotropical (fig. 6A). Except for the terrestrial B. hemiotis and B. oligarchica, all species climb 2–4 m up tree trunks. It is unknown whether these species are primary or secondary hemiepiphytes or lianas. Sometimes, as in B. nicotianifolia, they also grow on fallen logs, where they produce abundant fertile leaves (R. Moran, personal observation). Clade of Hemiepiphytic and Epiphytic Genera Sister to Bolbitis s.s. (fig. 3) are the B. nicotianifolia clade, Arthrobotrya, Elaphoglossum, Lomagramma, and Teratophyllum. These latter genera form a clade (BS 73; PP 80) united by their hemiepiphytic habit (character 1, state 1), a terminal segment resembling the lateral pinnae (14, 0), and pinnae articulate to the rachis (17, 1). Elaphoglossum, however, has numerous character state changes and does not retain any of these character states. The clade of Arthrobotrya, Lomagramma s.s., and Teratophyllum is defined morphologically by having peltate rhizome scales (7, 1), mature sterile leaves differentiated into This content downloaded from 69.74.186.251 on Thu, 6 Jun 2013 16:22:22 PM All use subject to JSTOR Terms and Conditions 554 INTERNATIONAL JOURNAL OF PLANT SCIENCES Fig. 4 Optimizations of five morphological characters (root insertion, elongate ventral meristele, hairs on the leaves, leaf dimorphism, and type of sori) for bolbitidoid ferns. Numbers in parentheses refer to the character numbers in the text. bathyphylls and acrophylls (15, 1), articulate apical pinnae (18, 1), and the presence of paraphyses (31, 1; fig. 1D–1F). The affinity of these genera was suggested by Holttum (1978). In our results, Lomagramma is sister to the clade of Teratophyllum þ Arthrobotrya (BS 100; PP 1). Lomagramma is defined morphologically by having clathrate scales (9, 0) and slender-stalked, peltate, and scalelike paraphyses (32, 0; figs. 1F, 7). Also, under DELTRAN optimization, anastomosing veins (24, 1) are synapomorphic for Lomagramma. That L. guianensis is unrelated to other species of the genus comes as no surprise. Holttum (1937), who monographed Lomagramma, omitted this species from the genus, and authorities such as Copeland (1947) and Kramer (1954) have agreed. It resembles other species of Lomagramma by 1-pinnate leaves, conform terminal pinnae, and veins that are areolate throughout (i.e., no main lateral veins); however, it differs from all other species of Lomagramma by having flat (nonbullate) laminar scales, no paraphyses (vs. present and peltate), and cristate (vs. smooth) spores (Holttum 1978). In these characters, it agrees with the B. nicotianifolia clade. The anatomy and unusual rhizome architecture of L. guianensis were studied by Hebant-Mauri and Gay (1993) and Gay (1993). Some characteristics found by them, such as reiterative positively geotropic rhizome branches, might be synapomorphic for the B. nicotianifolia clade, but field studies are needed to determine this. Old World Hemiepiphytic Clade The sister relation of Teratopyllum and Arthrobotrya is supported by petioles articulate to rhizomes (character 10, state 1), and (under ACCTRAN) free veins (24, 0). Teratophyllum This content downloaded from 69.74.186.251 on Thu, 6 Jun 2013 16:22:22 PM All use subject to JSTOR Terms and Conditions MORAN ET AL.—BOLBITIDOID FERN PHYLOGENY 555 Fig. 5 Optimizations of four morphological characters for bolbitidoid ferns and (at right) nine of the series recognized in Bolbitis by Hennipman (1977). Numbers in parentheses refer to the character numbers in the text. is defined by having two ranks of leaves (12, 0), whereas Arthrobotrya has more than two ranks (12, 1). Bipinnate leaves are an often helpful diagnostic character for Arthrobotrya, but this character does not optimize as a synapomorphy for this genus in our analysis. Both genera are high-climbing hemiepiphytes (this character is a synapomorphy lower on the tree, uniting these two genera and the B. nicotianifolia clade). Arthrobotrya (fig. 3) consists of three species: Arthrobotrya articulata J. Sm., Arthrobotrya brightiae F. v. Mueller, and Arthrobotrya wilkesiana (Brackenr.) Copel. (Copeland 1947; Holttum 1978). It occurs from the Philippines to the Solomon Islands, Australia, and Tahiti (fig. 6B). Some pteridologists have subsumed Arthrobotrya in Teratophyllum (e.g., Holttum 1978, as sect. Polyseriatae Holttum). Teratophyllum s.s. (fig. 3) comprises 11 species in southeastern Asia and Malesia (fig. 6F). Lomagramma s.s. is characterized by the synapomorphy of clathrate rhizome scales (9, 1). It is further diagnosed by the nonsynapomorphic characters of differentiated acrophylls and bathyphylls (15, 1), 1-pinnate leaves (20, 1), anastomosing veins This content downloaded from 69.74.186.251 on Thu, 6 Jun 2013 16:22:22 PM All use subject to JSTOR Terms and Conditions 556 INTERNATIONAL JOURNAL OF PLANT SCIENCES Fig. 6 Distribution of the bolbitidoid fern genera and their number of species. (24, 1), and paraphyses consisting of slender-stalked scales (32, 0; fig. 1F). The genus comprises ;22 species and is restricted to southeastern Asia and Malesia (fig. 6E). Like Arthrobotrya, Teratophyllum, and the B. nicotianifolia clade, it consists of high-climbing hemiepiphytic species (Holttum 1978). Elaphoglossum The results show strong support for Elaphoglossum being sister to the B. nicotianifolia clade (BS 100; PP 95). The monophyly of Elaphoglossum has high branch support (BS 100; PP 1) and is further supported by the morphological synapomorphies of phyllopodia present (character 11, state 1) and simple leaves (13, 0) with entire margins (19, 0). The epiphytic habit (1, 2) is a synapomorphy for the genus. Our species sampling of Elaphoglossum is a limited but representative sample of the major lineages in the genus. A detailed phylogenetic study of Elaphoglossum including ;120 species was presented by Rouhan et al. (2004). The genus comprises ;600 species and is pantropical (fig. 6D), with greatest diversity in the Neotropics. It consists primarily of epiphytic plants growing in wet forest, but many species are also found growing terrestrially or epipetrically at high elevations. Infrageneric Classification of Bolbitis In his monograph of Bolbitis, Hennipman (1977) recognized 10 series within the genus. Of these series, eight are represented in our study, and seven are represented by more than one species, which would allow a test for their monophyly. Three series were resolved as monophyletic (series Alienae, Bolbitidae, and Euryostichae). Also, series Heteroclitae was resolved as monophyletic, but Bolbitis lonchophora, a species treated by Hennipman as incertae sedis, was nested within it (fig. 5). All the species assigned to series Euryostichae by Hennipman belonged to the B. nicotianifolia clade (fig. 5) and were sister to B. bernoullii, a species placed by Hennipman in incertae sedis. If, as we believe, B. lindigii belongs to the B. nicotianifolia clade, then the monotypic series Lindigianae would also be part of that clade. Thus, some of the series are monophyletic and others are not. More inclusive sampling of Bolbitis species is needed to better resolve the issue of monophyly of the series recognized by Hennipman. This content downloaded from 69.74.186.251 on Thu, 6 Jun 2013 16:22:22 PM All use subject to JSTOR Terms and Conditions MORAN ET AL.—BOLBITIDOID FERN PHYLOGENY 557 Fig. 7 Optimizations of five morphological characters (rhizome scale attachment, pinnae articulation, paraphyses, secondary cross-veins, and sterile leaf differentiation [bathyphylls and acrophylls]) for bolbitidoid ferns. Numbers in parentheses refer to the character numbers in the text. Ecological Adaptations Bolbitis, which is sister to the rest of the bolbitidoids, is often associated with riparian habitats (Hennipman 1977; Holttum 1978), either growing as rheophytes (e.g., many Bolbitis) or on rocks near stream banks. Holttum (1978) suggested that the dorsiventral rhizomes and ventral root insertion exhibited by these ferns were ecological adaptations to growing on rocks in periodically inundated habitats. He further hypothesized that this suite of characters was a precondition for evolution of the hemiepiphytic and epiphytic plants, which need to be firmly attached to their substrate. Within the bolbitidoids, our phylogeny suggests that this transition in habit type does in fact occur as hypothesized by Holttum (1978). Habit changes from terrestrial among outgroups and Bolbitis to hemiepiphytic in the B. nicotianifolia clade, Lo- magramma, and Teratophyllum, followed by a second transition to epiphytic in Elaphoglossum (fig. 5). This transition is interesting because it suggests that Elaphoglossum, one of the most species-rich genera of epiphytic ferns (;600 spp.), has evolved from hemiepiphytic ancestors. The articulation of leaf parts occurs several different ways among bolbitidoid ferns, and its occurrence might have adaptive significance. Petioles may be articulate to rhizomes (e.g., Arthrobotrya, L. guianensis), lateral and terminal pinnae to rachises (fig. 7), and petioles to phyllopodia (in Elaphoglossum). Whereas Holttum (1978) thought that articulate pinnae were absent in Bolbitis, Hennipman (1977) pointed out that articulate pinnae do occur in that genus, although the articulations are sometimes indistinct. Holttum (1978) suggested that the articulate and deciduous pinnae of high-climbing ferns are This content downloaded from 69.74.186.251 on Thu, 6 Jun 2013 16:22:22 PM All use subject to JSTOR Terms and Conditions 558 INTERNATIONAL JOURNAL OF PLANT SCIENCES an adaptation to the drier conditions above the ground; the pinnae may fall during dry weather, thus reducing water stress to the plant. The pinnae might also disarticulate in response to insect damage. Although our results show that articulate pinnae also occur in terrestrial species of Bolbitis (fig. 7), articulate leaves and pinnae are more widespread within the clade of climbers (i.e., the B. nicotianifolia clade, Arthrobotrya, Lomagramma, and Teratophyllum). The pinnae of Arthrobotrya and Teratophyllum readily disarticulate in herbarium specimens, but in other bolbitidoids with articulate pinnae, the articulation does not seem to function (R. Moran, personal observation). The deciduousness of these pinnae, if any, and the relationship between disarticulation and climbing habit require further investigation. Potential Synapomorphies Several other characters might represent synapomorphies for the bolbitidoid ferns, but more observations are needed. One character is the position and arrangement of branch buds along the rhizomes. Although not surveyed for all bolbitidoid ferns, this character has been recorded in the B. nicotianifolia clade (Hebant-Mauri and Gay 1993), Bolbitis (Nayar and Kaur 1964a, 1964b, 1965; Hennipman 1977), and Elaphoglossum (Bell 1950, 1951a, 1951b, 1955, 1956). These studies have found that the buds are located ventrolaterally in one row on each side of the rhizome and between the adjacent leaf bases. The buds are often inconspicuous and typically do not develop unless the apex of the rhizome is removed or damaged. If the rhizome has more than two ranks of leaves—that is, if it bears dorsal rows of leaves between the two lateral ones—branch buds do not occur between the leaves in the dorsal rows. To our knowledge, this situation is unique among ferns. All bolbitidoids studied have such an arrangement, but we refrain from optimizing it on the tree until all of the species in figure 3 can be scored for the character. This is best done with living material and therefore is beyond the scope of this study. Another possible synapomorphy for bolbitidoid ferns is diplodesmic veins (i.e., in fertile pinnae, an extra set of veins below [abaxially to] the normal ones that supply the sporangia). Nayar (1966) surveyed the presence of these veins in all bolbitidoid ferns except the B. nicotianifolia clade (specifically, in Arthrobotrya, one species; Bolbitis, 16 species; Elaphoglossum, one species; Lomagramma, five species; Teratophyllum, three species) and two genera of Lomariopsidaceae (Lomariopsis, four species; Thysanosoria, one species). He found diplodesmic veins present in all bolbitidoid genera except in the one species of Elaphoglossum, where they were absent. Subsequently, Hennipman (1977) claimed, without explanation, that Nayar had incorrectly attributed diplodesmic veins to Bolbitis. In any case, because this character needs a more thorough survey among the bolbitidoid ferns, we refrain from optimizing it on our tree or calling it a synapomorphy. Even if not present in Elaphoglossum and, as Hennipman (1977) claimed, in Bolbitis, it might eventually prove to be a synapomorphy for the clade formed by Arthrobotrya, Teratophyllum, and Lomagramma. The rhizome ground tissue of most ferns is white (R. Moran, personal observation), but in five species of the B. nicotianifolia clade it is greenish (R. Moran, personal observation of Acrostichum scandens, B. lindigii, B. nicotianifolia, and B. pergamentacea). This suggests that green ground tissue (fig. 1A) might be a synapomorphy for the B. nicotianifolia clade, but other species of the genus and more bolbitidoid ferns need to be examined to assess this character. Acknowledgments This research was funded by a grant to Robbin Moran from the U.S. National Science Foundation (DEB 0717056). Paulo Labiak’s research at the NYBG was partially funded by the Brazilian government (CNPq/PDE 201782/2008-1). We thank Alan Smith (UC) and Gregory McKee (US) for sending us herbarium samples, Jefferson Prado (SP) and Fernando Matos (UPCB) for helpful observations of living plants and for discussions, and John Mickel (NY) for permission to use the illustrations in figure 2. In that figure, A was drawn by Karen Douthit, B and C by Haruto Fukuda, and D by Bobbi Angell. Literature Cited Akaike H 1973 Information theory as an extension of the maximum likelihood principle. Pages 267–281 in BN Petrov, F Csaki, eds. Second International Symposium on Information Theory. Akademiai Kiado, Budapest. Bell PR 1950 Studies in the genus Elaphoglossum Schott. I. Stelar structure in relation to habit. Ann Bot, NS, 14:545–555. ——— 1951a Studies in the genus Elaphoglossum Schott. II. The root and bud traces. Ann Bot, NS, 15:333–346. ——— 1951b Studies in the genus Elaphoglossum Schott. III. Anatomy of the rhizome and frond. Ann Bot, NS, 15:347–357. ——— 1955 Studies in the genus Elaphoglossum. IV. The morphological series in the genus and their phylogenetic interpretation, pt 1. Ann Bot, NS, 19:173–199. ——— 1956 Studies in the genus Elaphoglossum. IV. The morphological series in the genus and their phylogenetic interpretation, pt 2. Ann Bot, NS, 20:69–88. Ching RC, SK Wu 1983 Flora Xizangica. Vol 1. Science, Beijing. Copeland EB 1947 Genera filicum: the genera of ferns. Chronica Botanica, Waltham, MA. Edgar RC 2004 MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res 32:1792–1797. Gay H 1993 The architecture of a dimorphic clonal fern, Lomagramma guianensis (Aublet) Ching (Dryopteridaceae). Bot J Linn Soc 111:343–358. Goloboff PA, JS Farris, KC Nixon 2008 TNT, a free program for phylogenetic analysis. Cladistics 24:774–786. Hebant-Mauri R, H Gay 1993 Morphogenesis and its relation to architecture in the dimorphic clonal fern Lomagramma guianensis (Aublet) Ching (Dryopteridaceae). Bot J Linn Soc 112:257–276. Hennipman E 1977 A monograph of the fern genus Bolbitis (Lomariopsidaceae). Leiden Botanical Series 2. Leiden University Press, Leiden. Holttum RE 1937 The genus Lomagramma. Gard Bull Straits Settl 9: 190–221. This content downloaded from 69.74.186.251 on Thu, 6 Jun 2013 16:22:22 PM All use subject to JSTOR Terms and Conditions MORAN ET AL.—BOLBITIDOID FERN PHYLOGENY ——— 1978 Lomariopsis group. Pages 255–330 in GGJ van Steenis, RE Holttum, eds. Pteridophyta, ferns and fern allies. Flora Malesiana Series II, vol 1, pt 4. Huelsenbeck JP, JP Bollback 2001 Empirical and hierarchical Bayesian estimation of ancestral states. Syst Biol 50:351–366. Kramer KU 1954 A contribution to the fern flora of French Guiana. Acta Bot Neerl 3:481–494. Li C-X, S-G Lu 2006 Phylogenetic analysis of Dryopteridaceae based on chloroplast rbcL sequences. Acta Phytotax Sin 44:503–515. Little DP 2005 2xread: a simple indel coding tool. Program distributed by the author. http://www.nybg.org/files/scientists/ 2xread.html. Liu H-M, XC Zhang, W Wang, YL Qui, ZD Chen 2007 Molecular phylogeny of the fern family Dryopteridaceae inferred from chloroplast rbcL and atpB genes. Int J Plant Sci 168:1311–1323. Maddison WP, DR Maddison 2009 Mesquite: a modular system for evolutionary analysis, version 2.6. http://mesquiteproject.org. Moran RC 1995 Lomariopsidaceae. Page 247 in G Davidse, M Sousa, S Knapp, eds. Flora Mesoamericana. Vol 1. Universidad Nacional Autónoma de México, Ciudad Universitaria. Moran RC, AR Smith 2001 Phytogeographical relationships between Neotropical and African-Madagascan pteridophytes. Brittonia 53: 304–351. Nayar BK 1966 Morphology of the fertile leaves of the Lomariopsidaceae, with special reference to the venation. New Phytol 65: 221–239. Nayar BK, S Kaur 1964a Ferns of India. XI. Bolbitis. Bull Natl Bot Gard India 88:1–74. ——— 1964b Ferns of India. XIII. Egenolfia. Bull Natl Bot Gard India 100:1–38. _____ 1965 Studies on the fern genera Bolbitis and Egenolfia. I. Morphology of the sporophytes. J Linn Soc Lond Bot 59:127–140. Nixon KC 1999 The Parsimony Ratchet, a new method for rapid parsimony analysis. Cladistics 15:407–414. ——— 2004 ASADO, version 1.5 Beta. Program and documentation distributed by the author. Ithaca, NY. 559 Nylander JAA, F Ronquist, JP Huelsenbeck, JL Nieves-Aldrey 2004 Bayesian phylogenetic analysis of combined data. Syst Biol 53: 47–67. Rambaut A, AJ Drummond 2003 Tracer, version 1.3. http://tree .bio.ed.ac.uk/software/tracer/. Ronquist F, JP Huelsenbeck 2003 MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19:1572–1574. Rouhan G, J-Y Dubuisson, F Rakotondrainible, TJ Motley, JT Mickel, J-N Labat, RC Moran 2004 Molecular phylogeny of the fern genus Elaphoglossum (Elaphoglossaceae) based on chloroplast noncoding DNA sequences: contributions of species from the Indian Ocean area. Mol Phylogenet Evol 33:745–763. Schuettpelz E, KM Pryer 2007 Fern phylogeny inferred from 400 leptosporangiate species and three plastid genes. Taxon 56:1037– 1050. Simmons MP, H Ochoterena 2000 Gaps as characters in sequencebased phylogenetic analyses. Syst Biol 49:369–381. Skog JE, JT Mickel, RC Moran, M Volovsek, EA Zimmer 2004 Molecular studies of representative species in the fern genus Elaphoglossum (Dryopteridaceae) based on cpDNA sequences rbcL, trnL-F and rps4-TRNS. Int J Plant Sci 165:1063–1075. Smith AR, KM Pryer, E Schuettpelz, P Korall, H Schneider, PG Wolf 2006 A classification for extant ferns. Taxon 55:705–731. Souza-Chies TT, G Bittar, S Nadot, S Carter, E Besin, B Lejeune 1997 Phylogenetic analysis of Iridaceae with parsimony and distance methods using the plastid gene rps4. Plant Syst Evol 204:109–123. Swofford DL 2002 PAUP*: phylogenetic analyses using parsimony (*and other methods), version 4.10b. Sinauer, Sunderland, MA. Taberlet P, L Gielly, G Pautou, J Bouvet 1991 Universal primers for amplification of three non-coding regions of chloroplast DNA. Plant Mol Biol 17:1105–1109. Wilgenbusch JC, DL Warren, DL Swofford 2004 AWTY: a system for graphical exploration of MCMC convergence in Bayesian phylogenetic inference. http://king2.scs.fsu.edu/CEBProjects/awty/awty_ start.php. This content downloaded from 69.74.186.251 on Thu, 6 Jun 2013 16:22:22 PM All use subject to JSTOR Terms and Conditions