galley

SYSTEMATIC BOTANY Monday Mar 27 2006 05:06 PM Allen Press • DTPro System

sbot 31_205 Mp_271 File # 05TQ

GALLEY

Systematic Botany (2006), 31(2): pp. 271–284 q Copyright 2006 by the American Society of Plant Taxonomists

New Insights into the Phylogeny of the Genus Hymenophyllum s.l. (Hymenophyllaceae): Revealing the Polyphyly of Mecodium SABINE HENNEQUIN,1,4 ATSUSHI EBIHARA,2 MOTOMI ITO,2 KUNIO IWATSUKI,3 and JEAN-YVES DUBUISSON1 Laboratoire de Paleóbotanique et Paleóećologie, UMR 5143 Paleóbiodiversite´ et Paleóenvironnements, Universite´ Pierre et Marie Curie, 12 rue Cuvier, F-75005 Paris, France; 2 Department of System Sciences, Graduate School of Arts and Sciences, the University of Tokyo, 3-8-1 Komaba, Meguro-ku, Tokyo 153-8902, Japan; 3 The Museum of Nature and Human Activities, Hyogo, Yayoigaoka 6, Sanda 669-1546, Japan 4 Author for correspondence ([email protected])

1

Communicating Editor: Thomas A. Ranker ABSTRACT. With more than 100 species, Mecodium is the largest infrageneric taxon of Hymenophyllum s.l. It was long considered a natural and homogeneous group, but recent phylogenetic studies have questioned this assertion. Using rbcL, rbcL-accD, and rps4-trnS sequences, we demonstrate that Mecodium is highly polyphyletic. Several species of Mecodium form the derived clade ‘‘H. polyanthos’’; one species is nested within a second derived clade; and the remaining species are assigned to five basal clades including taxa regarded as distantly related. These clades are strongly supported both by parsimony and Bayesian analyses, but the relative placement of the basalmost clades lacks support. We show that the members of ‘‘basal Mecodium’’ are characterized by features that are plesiomorphic for Hymenophyllum s.l.—a reduced or dorsi-ventral stele, a lamina generally at least partially thickened, and a chromosome number based on x 5 36, whereas taxa in the ‘‘H. polyanthos’’ clade have a subcollateral stele, the one-cell thick lamina typical of the family, and x 5 28. There is a high level of variation among the basal species, and, notably, the rhizome indumentum is shown to be an interesting character for distinguishing among the basal clades. These new findings stress the need for further studies on Hymenophyllum s.l., and reassessment of its classification.

Recent phylogenetic studies on the Hymenophyllaceae, or ‘‘filmy ferns’’ (Pryer et al. 2001; Ebihara et al. 2002, 2003, 2004; Dubuisson et al. 2003a; Hennequin 2003; Hennequin et al. 2003), have gradually clarified the controversial systematics of the family (Copeland 1938; Morton 1968; Pichi Sermolli 1977; Iwatsuki 1984, 1990). These analyses have indicated a basal dichotomy in the family, with two clades roughly corresponding to the traditionally recognized genera Trichomanes L. and Hymenophyllum Sm. The Hymenophyllum clade appears less precisely circumscribed than the Trichomanes clade, as it includes the four segregate monotypic genera proposed by Morton (1968) (Cardiomanes C. Presl, Serpyllopsis Bosch, Rosenstockia Copel., and Hymenoglossum C. Presl), along with several species formerly included in Trichomanes by most authors. In addition, the phylogenetic relationships within the Hymenophyllum clade remain poorly understood compared to those within Trichomanes (Dubuisson 1997a,b; Dubuisson et al. 1998, 2003a). Using a preliminary taxon sample including representatives of the five subgenera recognized by Morton (1968), Hennequin et al. (2003) obtained three major clades corresponding broadly to the subgenera Hymenophyllum Sm., Sphaerocionium (C. Presl) C. Chr., and Mecodium Copel. Nevertheless, the placement of one species of Mecodium as sister to Cardiomanes reniforme (G. Forst.) C. Presl at the base of the tree indicated that this subgenus may be polyphyletic. This preliminary result was also obtained from a different taxonomic sampling in phylo-

genetic studies based on other molecular data (Ebihara et al. 2003), as well as on morphology and cytology (Hennequin 2003). Hennequin (2003) and Hennequin et al. (2003) found the heterogeneity of the anatomical and the cytological data in Mecodium consistent with this result. They suggested that species having a chromosome number n 5 36 and a dorsi-ventral stele may be basal in the genus, whereas species having n 5 28 and a subcollateral stele may form a more derived clade, and concluded that it was necessary to expand taxonomic sampling within this taxon. With more than 100 species described, Mecodium is by far the largest subgenus in the Hymenophyllaceae. It is distributed in temperate wet forests and tropical highland mossy forests and includes species that were placed in various taxa by the 19th century workers (e.g., Diploophyllum, van den Bosch 1861; Euhymenophyllum, Presl 1843, and other authors; Mecodium, Presl 1849). These species were regrouped by Copeland (1937, 1938) on the basis of entire margins and generally glabrous fronds, bivalved sori cleft to the base or, if partly immersed, to the lamina, and receptacles generally included. Mecodium was then recognized as a natural monophyletic group by the authors of the 20th century, and its circumscription has not varied much since Copeland (1937, 1938). There are, however, divergent views about the taxonomic rank assigned to Mecodium, in relation to the system of classification used for the family. Like Copeland (1938), Pichi Sermolli (1977) treated Mecodium as a genus, whereas

271



GALLEY

272

[Volume 31

SYSTEMATIC BOTANY

TABLE 1. Classifications of the subgenus Mecodium sensu Morton (1968) and sensu Iwatsuki (1984, 1985). * 5 taxa included in the genus Mecodium sensu Copeland (1938, 1947) and sensu Pichi Sermolli (1977). Copeland did not assign H. levingei to any genus. Subgenus Mecodium sensu Iwatsuki (1984) (renamed subgenus Craspedophyllum in 1985)

Subgenus Mecodium sensu Morton (1968) Sections

Mecodium

Subsections

Mecodium*

Type

H. polyanthos Sw.

Diplophyllum*

H. dilatatum (G.Forst.) Sw. Amphipterum H. fuscum (Blume) Bosch (subg. Craspedophyllum)

Number of species

.100

Sections

Type

Mecodium* (Integra, 1985)

H. polyanthos Sw. (H. asplenioides (Sw.) Sw., 1985) H. dilatatum (G.Forst.) Sw.

4

Diplophyllum*

4

(included in subg. Chilodium)

(included in subg. Mecodium) (included in subg. Mecodium)

both Morton (1968) and Iwatsuki (1984) proposed a subgenus Mecodium including different additional taxa (Table 1). Morton defined a single section within the subgenus, Mecodium, and further divided it into three subsections Amphipterum, Diplophyllum, and Mecodium (more than 100 species). Iwatsuki (1984) divided Mecodium into five sections: Corrugatae, Diplophyllum, Mecodium, Pachyloma, and Plumosa. In 1985, he changed the subgeneric name from Mecodium to Craspedophyllum, which was described by Presl (1843), and changed the sectional name from Mecodium to Integra (Table 1). In addition, Iwatsuki (1984, 1985) excluded from Mecodium a few species displaying an obconic involucre, and placed them in subgenus Chilodium, section Pseudomecodium K. Iwats. In summary, the genus Mecodium sensu Copeland corresponds to Morton’s subsections Mecodium and Diplophyllum, and to Iwatsuki’s sections Mecodium (or Integra), Diplophyllum, and Corrugatae, as well as to some species of the section Pseudomecodium. Hereafter, we will generally address the taxon Mecodium as defined by Copeland (1938) and refer to it as such without specifying a taxonomic rank. We will make a distinction, whenever needed, between the genus Mecodium sensu Copeland and the subgenus, section, or subsection Mecodium sensu Morton or Iwatsuki. A further issue in the systematics of Mecodium lies in the designation of the type species. Copeland (1937) first based Mecodium on the very common H. polyanthos (Sw.) Sw., whose type specimen is from the Neotropics. However, he changed the type species in 1938 and in 1947 to H. sanguinolentum (G. Forst.) Sw., from New Zealand. Meanwhile, Pichi Sermolli (1977), Morton (1968), and Iwatsuki (1984, 1990) maintained H. polyanthos as type. These two species are included in this study. In previous molecular studies of Hymenophyllum, the chloroplast gene rbcL was combined either with the rps4-trnS region (Hennequin et al. 2003) or with the rbcL-accD region (Ebihara et al. 2002, 2003). Robust

Pachyloma Plumosa Corrugatae*

H. marginatum Hook. & Grev. H. levingei Clarke H. ooides F. Muell. & Baker

Number of species

.100

4

1 1 2

support was shown for the clades obtained by Ebihara et al. (2002, 2003), but based on a restricted sampling. With broader sampling (Hennequin et al. 2003), several clades lacked robust support, especially at the base of the Hymenophyllum clade. With a view to improving clade support, we acquired rbcL, rps4-trnS, and rbcLaccD sequence data for each taxon included in the current study, and used the three markers in a combined analysis. In addition, we studied the anatomy and morphology of these taxa and compiled a morphological matrix based on one used in an earlier study (Hennequin 2003). The aims of this study were to: (1) further test the monophyly of Mecodium as defined by Copeland; (2) determine the close relatives of Mecodium and compare our results with the circumscription of this group as a subgenus as described by Morton (1968) and Iwatsuki (1984, 1985); and (3) discover which characters appear congruent with the molecular data. MATERIALS

AND

METHODS

Taxonomic Sampling. Taxonomic sampling was based on Hennequin et al. (2003) plus species of Mecodium, especially some with a haploid chromosome number of 36 or multiples and/or a dorsiventral stele. In addition, we included the type species of Iwatsuki’s (1984, 1985) section Pachyloma or Morton’s (1968) subgenus Craspedophyllum, Hymenophyllum marginatum Hook & Grev., the type species of Morton’s (1968) subsection Amphipterum, H. fuscum (Blume) Bosch, one species of Iwatsuki’s (1984, 1985) section Pseudomecodium, H. oligosorum Makino, and the two species of his section Corrugatae, H. ooides F. Muell. & Baker and H. corrugatum H. Christ. We could not obtain material for the single species of his section Plumosa. The subtlety of species distinctions within Mecodium makes it difficult to delimit some species. As Morton (1968) noted: ‘‘. . . [the section Mecodium] is also the most difficult group, for lacking characters of toothing and pubescence many of the species look quite alike.’’ This has resulted in a ‘‘H. polyanthos’’ species complex, which has many tropical representatives from distant geographic regions. Therefore, in this study, different geographic populations of H. polyanthos were also selected. In order to confirm that the use of a broader sampling did not alter the monophyly of the two main clades within the family, we first performed an analysis based on rbcL with a broad sampling


2006]


GALLEY

HENNEQUIN ET AL.: PHYLOGENY OF HYMENOPHYLLUM

273

FIG. 1. Strict consensus of the 10 most parsimonious trees obtained in the rbcL analysis. Numbers above branches are bootstrap values . 50%; branches with thick lines are the most robustly supported (BS $ 80%).

including five Trichomanes taxa and using three non-Hymenophyllaceae taxa as outgroups, as in Hennequin et al. (2003). The sampling was then reduced to 50 in-group taxa and three Trichomanes as outgroups for the analyses based on rbcL-accD and rps4trnS (Appendix 1). The final data matrixes have been deposited at TreeBASE (study accession number S1346, matrixes M2376 and M2377). For the purpose of presenting the results and discussion, we adopted the classification of Morton (1968) as in previous studies on the family. On the trees (Figs. 1–3) and whenever needed in the text, we inserted the name of the section sensu Morton in parentheses between the generic abbreviation and the specific ep-

ithet. For H. fuscum we preferred to maintain the name Amphipterum instead of Mecodium because this subsection was treated as a genus by Copeland (1938, 1947) and Pichi Sermolli (1977). DNA Sequencing. All procedures for DNA extraction, amplification and sequencing followed Hennequin et al. (2003). We used primers rbcL1195F (59-TTCTACAGTTCGGTGGTGG-39; newly designed) and accD816R (Ebihara et al. 2003) to amplify rbcL-accD, and newly designed internal primers accDHIF3 (59-TGTCAGGTT CTAACATGTGATTG-39) and accDHIR3 (59-CCTATACCTGTTT GAACAGCATC-39) to sequence this marker. Phylogenetic Analyses. The sequences were aligned using ClustalW (Thompson et al. 1994) and were adjusted when needed


274

GALLEY

SYSTEMATIC BOTANY


[Volume 31

FIG. 2. Single most parsimonious tree obtained in the analysis of the combined three molecular data sets. Numbers above branches are bootstrap values . 50%; branches with thick lines are the most robustly supported (BS $ 80%).

using MacClade (Maddison and Maddison 2001). As in Hennequin et al. (2003), indels of equal length were treated as binary presence/absence characters using the program ID coding (Barriel 1994; run with barcod, graphical interface provided by C. Gallut, Universite´ Paris 6, France). We performed Maximum Parsimony (MP) analyses using PAUP*4.0b10 (Swofford 2001), and Bayesian Metropolis coupled Markov chain Monte Carlo (MC/B) analyses using MrBayes 3.0 (Ronquist and Huelsenbeck 2003). This version of MrBayes allows the integration of data other than nucleotide or protein sequences, and indel characters were integrated in both the Maximum Parsimony and the Bayesian inference analyses. For MP, we conducted both equally weighted and unequally weighted analyses as described in Pryer et al. (2001) and Hennequin et al.

(2003). In the unequally weighted analyses, for each data set, character state changes were weighted using a symmetrical step matrix obtained using STMatrix (Lutzoni and Zoller, Duke University), in which the frequencies of each possible and reciprocal character state change are converted to costs with the negative natural logarithm. All searches used a heuristic approach (TBR branch-swapping, 100 replicates of random sequence addition, MulTrees option on). The robustness of each branch was assessed by bootstrap analysis (1000 replicates; Felsenstein 1985). For MC/B, the combined data set was assumed to have six partitions (rbcL, rps4, and accD genes; rps4-trnS and rbcL-accD intergenic spacers; and characters resulting from the treatment of indels). Each molecular partition was assigned its own model of nucleotide substitution (GTR


2006]


GALLEY


275

FIG. 3. Phylogram of the Bayesian topology showing the number of substitutions/site for each branch. Values at nodes are a posteriori probabilities; branches with thick lines are the most robustly supported (posterior probability $ 0.99). 1 I 1 G) as determined using ModelTest 3.06 (Posada and Crandall 2000). For indel characters, we used the Mk (Markov K) model of Lewis (2001). Clade credibility values were estimated calculating the posterior probability for each node using a Bayesian procedure as implemented in MrBayes 3.0. Using a random tree, we performed three separate runs of 10,000,000 generations each. 100,000 trees were sampled for each run and the consensus tree was computed (PAUP*4.0b10; Swofford 2001) on the last 98,000 trees from one of the runs chosen at random, excluding the 2,000 trees found in the ‘‘burn-in period.’’ We evaluated congruence between the three data sets by comparing tree topology obtained in the separate MP analyses, using a reciprocal 70% bootstrap threshold (Mason-Gamer and Kellogg 1996). Character Evolution. Patterns of morphological and cytological evolution were assessed in MacClade by mapping characters of interest (Table 2) onto the most parsimonious tree obtained from the unequally weighted combined parsimony analysis. Both ACCTRAN and DELTRAN optimizations options were performed and compared. The cytological data were extracted from the literature. The two morphological characters selected here were ac-

quired from personal observations of herbarium specimens from P, BM, K, REU, TI and a personal collection, and compared to data in the literature when available.

RESULTS Table 3 summarizes, for each analysis, the number of total, informative, and indel characters, the number of trees, the tree length, the consistency index (CI) and the retention index (RI) (Farris 1989). DNA Sequences and Alignments. The final length of the combined alignment for the 53 taxa was 3,697 sites (1,206 rbcL sites, 1,423 rbcL-accD sites, and 1,068 rps4trnS sites), with only three missing cells (0.07%). Two hundred and three characters resulting from the treatment of indels were added to the data matrix (0 for rbcL, 80 for rbcL-accD, and 123 for rps4-trnS). This re-



GALLEY

276

[Volume 31

SYSTEMATIC BOTANY

TABLE 2. Cytological and morphological characters examined. 1. Chromosome number: n 5 36, 72 (0); n 5 28, 56 (1); n 5 13, 26 or 12 or 11, 22 (2); n 5 21, 42 (3); n 5 32 or 33 (4). The haploid chromosome numbers reported and their multiples were treated as a single character state (Brownlie 1954, 1958; Walker 1966; Braithwaite 1969, 1975; Tindale and Roy 2002; Hennequin 2003, 2004). The chromosome numbers n 5 32 and n 5 33 are observed in Trichomanes. The chromosome numbers n 5 13, 12, and 11 are supposed to be connected by dysploidy and are reported only for a derived clade (Hymenophyllum s.s.; Hennequin 2003, 2004). Due to the restricted sampling within this clade, and as evolution within it is not the subject of this paper, we treated these numbers as a single character state. 2. Type of hairs on the rhizome (more frequently around the base of stipes—all hairs are simple): Numerous, dark brown or reddish, more or less adpressed, 1.5–3.5 mm long, composed of six to 20 cells of medium size with thick cell walls, cross section circular (0); numerous, dark brown, 1.5–3.5 mm long, more or less adpressed, composed of six to 20 cells of small to medium size with thick cell walls, cross section compressed (1); sparse, pale, more or less adpressed, composed of up to 20 very short cells, cross section compressed (2); stiff, pale brown to rusty brown, 1.5 mm long, more or less adpressed, composed of ca. ten cells of medium size with thick cell walls, cross section more or less compressed (3); numerous, stiff, rusty brown, 2.5 mm long, adpressed, composed of up to 50 small cells with thick cell walls, cross section circular (4); numerous, thin, more or less curled, woolly, transparent to light brown, up to 3 mm long, not adpressed, composed of six to 15 cells of medium to long size, with thin cell walls, cross section compressed (5); numerous, thin, stiff, transparent to rusty brown, up to 1.5 mm long, not adpressed, composed of one to ca. six very long cells with thin cell walls, cross section circular (6); sparse, brown to reddish, up to 1 mm, more or less adpressed, composed of up to six cells with thick cell walls, cross section circular (7). 3. Stele anatomy: Massive protostele (0); reduced protostele with more or less xylem parenchyma (protoxylem endarch lying within parenchyma) to dorsi-ventral protostele (with lateral xylem elements reduced) (1); subcollateral (ventral band of xylem completely reduced) (2). Boodle 1900; Ogura 1938; Le Thomas 1961; Hennequin 2004.

sulted in a total of 3,900 characters, of which 1,444 were variable and 872 parsimony-informative. Maximum Parsimony Analyses. Only the consensus trees of the most parsimonious trees resulting from the unequally weighted rbcL analysis and the un-

equally weighted one combining all data sets are shown, but bootstrap values (BS) for the major clades obtained in the combined analysis are also reported for all separate analyses in Table 4. The analysis based on rbcL alone yielded ten MP trees. The strict consensus tree (Fig. 1) supports the monophyly of the Hymenophyllaceae (BS 5 100%), of Trichomanes (BS 5 53%) and of Hymenophyllum (BS 5 100%) including Cardiomanes, Serpyllopsis, Hymenoglossum, Rosenstockia, two species of the section Microtrichomanes, and one Trichomanes species of the section Pleuromanes. The topology is well resolved but very few clades are strongly supported. Nevertheless, there is strong evidence for the polyphyly of Mecodium: several species of the taxon are dispersed at the base of the tree, while others form a more derived, but unsupported clade. The separate unequally weighted analyses of rbcLaccD and rps4-trnS yielded 12 and one trees, respectively, the consensus trees of which are well resolved. Increased bootstrap support is observed for many clades (Table 4). The few conflicting clades are not supported. The unequally weighted analysis combining all three markers yielded a single MP tree (Fig. 2). The clades retrieved are well-supported except for the nodes at the base of the tree and within the derived clade Hymenophyllum s.s. Eight major clades were obtained, three corresponding to previously described clades (Hennequin et al. 2003): 1) Hymenophyllum s.s. (BS 5 93%) including one species of subgenus Mecodium and H. (Amphipterum) fuscum; 2) the ‘‘H. polyanthos’’ clade (BS 5 100%) sister to Hymenophyllum s.s. (BS 5 98%); 3) the ‘‘H. australe’’ clade (BS 5 100%), which includes six species of Mecodium; 4) Sphaerocionium s.l. (BS , 50%); 5) the ‘‘H. flabellatum’’ clade (BS 5 96%) containing two species of Mecodium and T. pallidum; 6) the ‘‘H. cruentum’’ clade (BS 5 100%) with Hymenoglossum cruentum and H. (Mecodium) heimii; 7) the ‘‘C. reniforme’’ clade (BS 5 92%) comprising Cardiomanes reniforme and three species of Mecodium; 8) the ‘‘H. sanguinolentum’’ clade (BS 5 86%) composed of three species of Mecodium and the newly-described species H. paniense Ebihara et K. Iwats. The species traditionally attributed to the subgenus Mecodium by Morton are thus placed in seven different clades. Nev-

TABLE 3. Statistics for the separated and combined Maximum Parsimony analyses. * 5 broad taxonomic sampling (62 taxa). Data sets

rbcL*

rbcL-accD

rps4-trnS

Combined

Number of sites Number of indel characters Number of informative sites Number of MP trees Number of steps CI RI

1206 0 327 (27.1%) 10 1921.00 0.5003 0.5459

1503 80 359 (23.9%) 12 1827.74 0.6215 0.6514

1191 123 330 (27.7%) 1 1921.84 0.6241 0.6237

3900 203 872 (22.4%) 1 4967.26 0.5999 0.6235


2006]


GALLEY

277


TABLE 4. Bootstrap support and a posteriori probabilities for the major clades obtained in the combined analyses. These clades include the taxa shown in Figs. 2 and 3. MP 5 Maximum Parsimony analyses, MC/B 5 Bayesian analysis. ‘‘pres’’ 5 clade obtained but BS , 50, a posteriori probability ,0.90; ‘‘—’’ 5 clade not obtained. MP Clades/data sets

rbcL

rbcL-accD

rps4-trnS

Combined

MC/B Combined

Hymenophyllum s.s. (1) ‘‘H. polyanthos’’ clade (2) clade (1) 1 (2) ‘‘H. australe’’ clade (3) Sphaerocionium s.l. (4) clade (3) 1 (4) clade (1) 1 (2) 1 (3) 1 (4) ‘‘H. flabellatum’’ clade (5) ‘‘H. cruentum’’ clade (6) clade (5) 1 (6) clade (1) 1(2) 1 (3) 1 (4) 1 (5) 1 (6) ‘‘C. reniforme’’ clade (7) clade (1) 1(2) 1 (3) 1 (4) 1 (5) 1 (6) 1 (7) ‘‘H. sanguinolentum’’ clade (8) Hymenophyllum s.l.

pres pres pres 61 — pres pres 62 — — — — — — 100

60 99 — 99 50 pres 51 — 93 — — pres — 98 100

66 81 83 73 — — — 77 94 — — 90 — 63 100

93 100 98 100 pres pres pres 96 100 80 pres 92 pres 86 100

1 1 1 1 1 0.94 1 1 1 0.99 pres 1 pres 1 1

ertheless, the support for the nodes determining the relative placement of the five last clades is very weak. When character-state changes were equally weighted (‘‘equally weighted analysis,’’ results not shown), the MP analysis yielded three most parsimonious trees. The consensus tree differs in that the species of Sphaeocionium have a basal position and do not form a clade. The seven other clades obtained in the unequally weighted analysis are retrieved. Bayesian Analyses. The topology (Fig. 3) obtained in the Bayesian analysis combining all three markers is nearly identical to the one using unequally weighted MP. The eight clades cited above are recovered with high support (posterior probability P 5 1). Increased support is shown for internal nodes, notably the clade including the ‘‘H. australe’’ clade and Sphaerocionium s.l. (P 5 0.94), and the one joining these two clades with ‘‘H. polyanthos’’ and Hymenophyllum s.s. clades (P 5 1). The relative position of the basalmost clades (‘‘C. reniforme’’ and ‘‘H. sanguinolentum’’) is still not supported. Evolution of Characters. CHROMOSOME NUMBER (FIG. 4A). The ancestral state of this character for the family appears equivocal, due to the different character states of the outgroup species used. A chromosome number based on 36 is inferred at the base of Hymenophyllum s.l. and could be proposed as ancestral for the family, as many Trichomanes basal species also display n 5 36 chromosomes. The chromosome number becomes equivocal on the branch leading to Hymenophyllum s.s. 1 ‘‘H. polyanthos’’ clade. One transition to n 5 28 (or 56) is proposed for the ‘‘H. polyanthos’’ clade and one to n 5 11, 12, or 13 (or multiples) for Hymenophyllum s.s. Two further transitions to n 5 28 (56) and n 5 21 (42) are then proposed within Hymenophyllum s.s. RHIZOME HAIRS (FIG. 4B). The ancestral states of

this character for the family and for the basal branches of Hymenophyllum are equivocal. One transition to state (3) is proposed on the branch leading to the ‘‘H. sanguinolentum’’ clade, with a further transition to state (4) in H. scabrum. One transition to state (1) is inferred on the branch leading to the ‘‘C. reniforme’’ clade, one to state (5) on the branch leading to the ‘‘H. flabellatum’’ 1 ‘‘H. cruentum’’ clades, one to state (2) for the ‘‘H. australe’’ clade, one to state (6) for Sphaerocionium s.l, and one to (7) for the ‘‘H. polyanthos’’ clade 1 Hymenophyllum s.s. STELE ANATOMY (FIG. 4C). The ancestral state of this character for the family is equivocal. A reduced or dorsi-ventral stele is inferred at the base of Hymenophyllum s.l. Three transitions to a subcollateral stele are then inferred in the basal taxa H. heimii, H. leratii, and H. paniense. The state is equivocal on the branch leading to the more derived clades, with transitions toward a subcollateral stele occurring on the branches leading to Sphaerocionium s.l. and to Hymenophyllum s.s. 1 ‘‘H. polyanthos’’ clade. DISCUSSION We will focus here on the results of the unequally weighted MP analysis and of the Bayesian analysis, as they yielded a quite similar topology. The slight differences in the tree obtained in the equally weighted parsimony analysis do not alter our conclusions. Polyphyly of Subgenus Mecodium sensu Morton (1968) and Iwatsuki (1984, 1990). Regardless of the analytical approach, this study provides strong evidence for the polyphyly of Mecodium as a subgenus sensu Morton (1968) and Iwatsuki (1984), as highlighted in the previous studies on Hymenophyllum (Ebihara et al. 2003; Hennequin 2003; Hennequin et al. 2003). In


278

GALLEY

SYSTEMATIC BOTANY


[Volume 31

FIG. 4. Inferred evolution of selected characters on the single MP tree obtained in the combined analysis. A. Chromosome number. B. Type of rhizome hairs. C. Stem stele anatomy.

agreement with Ebihara et al. (2002), H. fuscum, type species of Amphipterum, is nested within Hymenophyllum s.s. This result corroborates Copeland’s (1938), Pichi Sermolli’s (1977), and Iwatsuki’s (1984) assumptions that Amphipterum is closely related to subgenus Hymenophyllum, and more precisely to Ptychophyllum

sensu Morton. Iwatsuki (1984) included Amphipterum in his subgenus Chilodium, a treatment supported here. In addition, our results invalidate all suggestions of affinity between Craspedophyllum and Mecodium (e.g., Copeland 1938), put forth by Iwatsuki (1984) who included H. marginatum Hook. & Grev. in Mecodium (un-


2006]


GALLEY


der the sectional name Pachyloma, see Table 1). Morton (1968) treated Craspedophyllum as a subgenus with two species, H. marginatum and H. armstrongii (Baker) Kirk. Based on our results, these two species are indeed closely related, but they are here included in Hymenophyllum s.s., so Craspedophyllum may not deserve a subgeneric rank. We also refute the hypothesis of affinity between Mecodium and Hemicyatheon (Copeland 1938; Pichi Sermolli 1977), as Hemicyatheon is shown nested within Hymenophyllum s.s. Polyphyly of Genus Mecodium sensu Copeland (1938) and Pichi Sermolli (1977). Copeland’s (1938) and Pichi Sermolli’s (1977) genus Mecodium, as well as Morton’s (1968) subsection and Iwatsuki’s (1984) section Mecodium, more narrowly defined, are equally polyphyletic. Some species of Mecodium come together as a well-supported clade named here the ‘‘H. polyanthos’’ clade, sister to Hymenophyllum s.s.; some form a clade named ‘‘H. australe,’’ sister to Sphaerocionium s.l.; one species is included in Hymenophyllum s.s.; and the other species are positioned in several of the basalmost clades. The placement of H. oligosorum Makino within Hymenophyllum s.s., as sister to H. barbatum Baker, was also obtained by Ebihara et al. (2002) and is in agreement with Iwatsuki (1984, 1985). Ebihara et al. (2002) proposed several morphological characters supporting this association, notably the type of hairs on the axes of fronds, the structure of internal cell walls, and the chromosome number based on x 5 21 (Fig. 4A). The relationships within Hymenophyllum s.s. will be addressed in a future study. The ‘‘H. polyanthos’’ clade corresponds to the clade Mecodium p.p. retrieved by Hennequin et al. (2003), here confirmed with a broader sampling. It includes several species of Mecodium, among which is the type species H. polyanthos according to Pichi Sermolli (1977), Morton (1968), and Iwatsuki (1984, 1990). An illustration of a representative of this clade, H. inaequale (Poir.) Desv., is provided in Fig. 5A. The two species assigned to the section Corrugatae by Iwatsuki (1984, 1985) are also included in the ‘‘H. polyanthos’’ clade, but not as sister species (Fig. 1). All species in the ‘‘H. polyanthos’’ clade for which chromosome counts were made have a chromosome number of n 5 28 or multiples (also reported for H. fucoides with n 5 56), so that a basic chromosome number x 5 28 can be proposed for this clade, while other Mecodium species have n 5 36 or multiples (Fig. 4A; Brownlie 1958; Walker 1966; Braithwaite 1969, 1975). Both n 5 36 and n 5 56 were reported for H. rarum R. Br. (Brownlie 1954; Tindale and Roy 2002, respectively); the specimen used for the first count may have been misidentified. The ‘‘H. polyanthos’’ clade notably shares with Hymenophyllum s.s. the same type of rhizome hairs (sparse, brown to reddish, up to 1 mm, more or less adpressed, composed of up

279

to six cells with thick cell walls, cross section circular; Fig. 4B), and shares with Hymenophyllum s.s. and Sphaerocionium s.l. a subcollateral stele (Figs. 4C, 5B; Table 5). The ‘‘H. polyanthos’’ clade can be divided further into two clades based on biogeography, showing interesting biogeographical patterns: one clade groups species from the Mascarene Islands (H. inaequale and H. polyanthos from La Reúnion), Chile (H. cuneatum Kunze), and Australasia (H. mnioides Baker and H. polyanthos, plus H. rarum and H. ooides in the rbcL tree (Fig. 1)), while the other clade groups species from the Neotropics (H. apiculatum Mett. Ex Kuhn and H. polyanthos from Bolivia) and North Asia (H. wrightii Bosch and H. polyanthos from Japan, plus H. corrugatum in the rbcL tree), possibly also from North America. The four H. polyanthos individuals sampled here from Bolivia, Japan, Tahiti, and La Reúnion all fall into the ‘‘H. polyanthos’’ clade, but they appear more closely related to other species with similar geographic origins than with other H. polyanthos individuals. The ‘‘H. polyanthos’’ species complex is thus clearly polyphyletic and requires taxonomic revision. A further significant finding of our study is that not only are there numerous Mecodium species in basal positions compared to the ‘‘H. polyanthos’’ clade, but also they are positioned in five different clades that include taxa considered remotely related to them by all authors (i.e., C. reniforme, H. cruentum, and T. pallidum). The ‘‘H. australe’’ clade, composed strictly of species of Mecodium, can be distinguished from the other clades by the type of rhizome hairs (sparse, pale, more or less adpressed, composed of up to 20 very short cells, cross section compressed, Fig. 4B) and the absence of indumentum on the axes of fronds. This clade is restricted to Australasia, Asia, and southern South America. It is resolved as sister to Sphaerocionium s.l., but this association is weakly supported (BS , 50 %, P 5 0.94) and no morphological synapomorphy seems to support it either. The inclusion in the ‘‘H. flabellatum’’ clade of one species of Trichomanes, T. pallidum Blume, is supported by two major characters: a dorsi-ventral stele (personal observation; Fig. 4C), never reported for Trichomanes, and the same type of rhizome hairs (numerous, thin, more or less curled, woolly, transparent to light brown, up to 3 mm long, not adpressed, composed of six to 15 cells of medium to long size, with thin cell walls, cross section compressed; Fig. 4B). The rhizome hairs are also similar in the sister clade, ‘‘H. cruentum’’. Interestingly, this latter clade groups two species that were considered distantly related despite their sharing simple fronds, a rare feature in Hymenophyllum. These two species are Hymenoglossum cruentum (Cav.) C. Presl, endemic to Chile, and H. (Mecodium) heimii Tardieu, endemic to Madagascar. The shared simple fronds thus appear to be an homology. This result raises the issue of the biogeographical origin of these


280

GALLEY

SYSTEMATIC BOTANY


[Volume 31

FIG. 5. Examples of Mecodium species. A. H. inaequale (Poir.) Desv., a species close to H. polyanthos, in situ in La Reúnion. B. H. apiculatum Mett. ex Kuhn., subcollateral stele with the ventral band of xylem completely reduced. C. H. dilatatum, dorsiventral stele with two bands of xylem (ventral and dorsal). D. H. dilatatum in situ in New Zealand. E. H. inaequale sorus, showing lateral veinlets running up to the margin and cylindrical receptacle (observed by transparency). F. H. dilatatum sorus, showing reduced lateral veinlets and extruding sporangia born on a voluminous globose receptacle (not visible).

two distant sister species. A similar pattern was already observed in the ‘‘H. polyanthos’’ clade between H. inaequale and H. polyanthos from La Reúnion and H. cuneatum from Chile. The taxa of the ‘‘C. reniforme’’ clade share the same type of rhizome hairs (numerous, dark brown, 1.5–3.5 mm long, more or less adpressed, composed of six to 20 cells of small to medium size with thick cell walls, cross section compressed; Fig. 4B), which is close to the type observed in the outgroup species, but with smaller cells and compressed cross section. An illustration of H. dilatatum Sw. is provided in Fig. 5D. The sister species H. (Mecodium) pulcherrimum (Colenso) Copel. and H. (Mecodium) fucifor-

me Sw. share a feature that is exceptional in the genus: a very short and erect rhizome with short internodes. The distribution of the ‘‘C. reniforme’’ clade is restricted to austral regions: New Zealand and southern Chile. The last basal clade, ‘‘H. sanguinolentum,’’ is named from the species considered the type of Mecodium by Copeland (1938, 1947). It includes three species from New Zealand together with the New Caledonian H. paniense. This clade is characterized by the rhizome indumentum of stiff, pale brown to rusty brown hairs, composed of ca. 10 cells with thick cell walls, with further evolution to particular hairs with many cells in H. scabrum A. Rich (Fig. 4B), and displays the same


2006]


GALLEY

281


TABLE 5. Summary of morphological, cytological, and biogeographical characteristics of the ‘‘basal Mecodium’’ species (including the ‘‘H. australe’’ clade) in comparison to the species of the ‘‘H. polyanthos’’ clade (based on Hennequin 2003, 2004). 1 5 also displayed by Sphaerocionium s.l. and Hymenophyllum s.s.; 2 5 also displayed by Hymenophyllum s.s. only; 3 5 also displayed by Sphaerocionium s.l. only. Characteristics

Basic chromosome number Rhizome stele Rhizome hairs

Frond hairs Lamina thickening Frond (petiole 1 lamina) size Rhizome diameter Sorus Receptacle Sporangiophores (receptacle extensions) Distribution

‘‘Basal Mecodium’’

‘‘H. polyanthos’’ clade

x 5 36 mainly reduced to dorsi-ventral, sometimes subcollateral very variable 3

variable, sometimes absent absent to uniformly 2 cells thick generally $15 cm

x 5 28 subcollateral1 sparse, brown to reddish, up to 1 mm, more or less adpressed, composed of up to six cells with thick cell walls, cross section circular2 absent absent1 generally ,15 cm1

generally $1 mm bivalved, generally not immersed in the lamina and with no lateral veinlets variable but mainly globose or capitate3 generally present3

,1 mm1 bivalved, often with lateral veinlets

Palaeotropics and austral temperate areas (Chile, New Zealand, Australia); absent from the Neotropics, no data for continental Africa

Pantropics and austral temperate areas (Chile, New Zealand, Australia)

type of hairs on the frond axes. Hymenophyllum sanguinolentum, H. villosum Colenso, and H. scabrum are also characterized by the presence of oil in their fronds, giving the dried specimens a strong odor, and causing the impression of the fronds on the sheets on old herbarium specimens. The phylogram obtained (Fig. 3) reveals very short branch lengths at the base of the tree, suggesting a rapid radiation at this stage in the history of Hymenophyllum. The systematics of Mecodium thus appears much more complex than previously supposed. This conclusion is not surprising as the description of Mecodium was actually based on the absence, rather than on the presence, of peculiar characters (lack of denticulation, lack of peculiar hairs), and thus on characteristics that appear plesiomorphic for Hymenophyllum according to Hennequin (2003, 2004). It thus seems logical that such a number of species of the taxon are at the base of the Hymenophyllum tree. Copeland (1938) opined that ‘‘. . . both Presl and van den Bosch [had] failed to grasp’ Mecodium. Contrastingly, our results corroborate in part former classifications such as Presl’s (1843, 1849), where some species of Mecodium sensu Copeland (such as H. polyanthos Sw.) were included in his genus Hymenophyllum, but in a separate infrageneric taxon, while others (including, for example, H. dilatatum and H. australe Willd.) were placed in an infrageneric taxon included in his genus Sphaerocionium. The high variability observed among the basalmost species is even more conspicuous when studying the type of hairs present on the rhizome. Indumentum has rarely been used in systematic studies of Hymenophyl-

cylindrical2 absent1

lum, and the recent studies of the genus (Ebihara et al. 2002, 2003) have shown that, indeed, it appears quite promising for its systematics. Morphological and Cytological Evolution. Although they do not form a clade, the basalmost species of Mecodium, including species retrieved in the ‘‘H. australe’’ clade, are clearly distinguished from the ‘‘H. polyanthos’’ clade by a chromosome number based on 36 and a reduced or dorsi-ventral stele (Figs. 4A, C, 5C; Table 5), confirming an hypothesis postulated by Hennequin (2003) and Hennequin et al. (2003). The stele of the Hymenophyllaceae is small and always protostelic. However, there is considerable variation between two extremes, which are a massive protostele, and an extremely reduced one. Of the seven types of protostele described for the family, only four are reported in Hymenophyllum: reduced, further reduced, dorsi-ventral, and subcollateral (Boodle 1900; Ogura 1938; Le Thomas 1961; Hennequin 2004). The inference of the evolution of this character shows reduction of the vascular system. Furthermore, the basalmost species of Mecodium generally display a more robust habit, with thicker rhizomes and larger fronds. All these characters appear plesiomorphic for the genus (Hennequin 2003). Some of these species have a lamina uniformly thicker than the usually one-cell thick lamina of the Hymenophyllaceae (i.e., four cells thick in Cardiomanes reniforme, three cells thick in H. scabrum and H. dilatatum), and many species have a partially thickened lamina, especially at the margins and along the veins. However, the ancestral state for this feature remains ambiguous.


282

GALLEY


[Volume 31

SYSTEMATIC BOTANY

Many authors have proposed a regressive evolution in the family in relation to the hygrophilous habit; this has been well-illustrated recently in Trichomanes by Dubuisson et al. (2003b). This hypothesis appears corroborated in Hymenophyllum as well. A few basal species of Mecodium included in this study have evolved a subcollateral stele; this is likewise an indication of regressive evolution within the basal taxa. Finally, both Morton (1968) and Iwatsuki (1984) separated the species with thicker lamina (i.e., H. dilatatum and H. scabrum, along with two other species, H. australe and H. demissum (G. Forst.) Sw.) from Mecodium and placed them in the segregate subsection or section Diplophyllum. Copeland (1938) did not reassign them, arguing that these species seemed independently related to other species of Mecodium. We show here that, in agreement with Copeland’s (1938) idea, these four species are not closely related but are instead included in three different clades. Even though no author had suggested the polyphyly of Mecodium, a few had pointed out the variability in sorus morphology and in chromosome number. Braithwaite (1975) postulated that the cytological heterogeneity within Mecodium was associated with variation in sorus morphology. He noted that in species with n 5 36 chromosomes ‘‘. . . the involucre is deeply bivalved, cleft often almost to the base, and the vascular supply passes directly into the receptacle’, and that in species with n 5 28 the involucre is bivalved but ‘‘. . . cleft less deeply, with a shallow obconic base, [. . . ] and the vascular supply gives off two traces which pass along either side of the obconic base of the involucre’. The issue of sorus morphology in Hymenophyllum seems indeed a difficult one. Rather than using the two or three (Iwatsuki 1977) sorus types proposed for the family, Hennequin (2003, 2004) proposed a combination of characters to describe sorus morphology. The Mecodium species studied here (H. oligosorum excluded) all display bivalved sori, but significant variation in morphology can be observed (Hennequin 2003, 2004). The two traces corresponding to the bifurcation of the vascular supply described by Braithwaite (1975), named ‘‘lateral veinlets’’ by Hennequin (2003), are observed both in basal Mecodium species and in the ‘‘H. polyanthos’’ clade. In agreement with Braithwaite (1975), however, they seem to be the minority in the basal taxa (displayed here only by H. heimii, H. villosum, H. dilatatum [Fig. 5F], H. cruentum, and C. reniforme), while they seem to be the majority in the ‘‘H. polyanthos’’ clade (see Fig. 5E; pers. obs.). In addition, in many basal Mecodium the involucre is deeply cleft, while it is often one quarter to half immersed in the lamina in the ‘‘H. polyanthos’’ clade. The cytological heterogeneity in Mecodium may thus be associated with some differences in sorus morphology, but these are only tendencies, which are sometimes difficult to

assess. Developmental studies of the sorus are needed to further address this issue. We also observed considerable variability in the shape and size of the receptacles, which bear the sporangia (see Table 5). In conclusion, we have shown here that the taxon Mecodium, even in its narrowest circumscription, is polyphyletic and therefore refute all previous treatments of the taxon performed by various specialists of the Hymenophyllaceae. We have revealed the basal placement, in the Hymenophyllum clade, of many species of Mecodium, and have shown that this position is supported by characters that are plesiomorphic for Hymenophyllum. This significant finding stresses even more the need for a new classification of Hymenophyllum and provides a basis for this revision (Ebihara et al., in prep.). It may be difficult to get more signal for the basal nodes of our tree, but the acquisition of additional molecular data and their combination with morphological data may improve the resolution. Finally, the austral distribution of the basalmost taxa points to Gondwanan connections, nevertheless the most derived clades (e.g., Hymenophyllum s.s.) also have many representatives in austral areas. A broader taxonomic sampling is needed to address a biogeographic study of Hymenophyllum. ACKNOWLEDGEMENTS. The authors thank the staffs of the herbaria of the Museúm d’Histoire Naturelle, Paris, the Natural History Museum, London, and the Royal Botanic Gardens, Kew, for access to the collections; the Royal Botanic Gardens of Edinburgh, Takeshi A. Ohsawa, Sabine Miehe, France Rakodondrainibe, Philippe Danton, Alan Smith, and Je´roˆme Muzinger for providing material; David R. Given, the Direction of Natural Resources and Direction of Economic Development (New Caledonia), the Department of Conservation Te Papa Atawhai (New Zealand), LIPI (Indonesia), and CONAF (Chile) for collecting permits; Molly McMullen, Nathalie Natalingum, Kathleen Pryer and Eric Schuettpelz for their comments on an earlier version of the manuscript; and the reviewers of the manuscript. This research was supported in part by grants from the IFR 101 CNRS ‘‘Institut d’Ecologie Fondamentale et Appliqueé’, the FR 1541 CNRS ‘‘Institut de Syste´matique’’, the SYS-Resource program of the European Commission, and the JSPS Grant for Oversea Research No. 12575012.

LITERATURE CITED BARRIEL, V. 1994. Phylogeńies molećulaires et insertions-de´le´tions de nucleótides. Comptes Rendus de l’Acade´mie des Sciences de Paris, Sciences de la Vie 317: 693–701. BOODLE, L. A. 1900. Comparative anatomy of the Hymenophyllaceae, Schizaeaceae and Gleicheniaceae. I. On the anatomy of Hymenophyllum, Annals of Botany 14: 455–496. BRAITHWAITE, A. F. 1969. The cytology of some Hymenophyllaceae from the Solomon Islands. British Fern Gazette 10: 81–91. ———. 1975. Cytotaxonomic observations of some Hymenophyllaceae from the New Hebrides, Fiji and New Caledonia. Botanical Journal of the Linnean Society 71: 167–189. BROWNLIE, G. 1954. Introductory note to cyto-taxonomic studies of New Zealand ferns. Transactions of the Royal Society of New Zealand, Botany 82: 665–666. ———. 1958. Chromosome numbers in New Zealand ferns. Transactions of the Royal Society of New Zealand, Botany 85: 213–216.


2006]


GALLEY


COPELAND, E. B. 1937. Hymenophyllum. The Philippine Journal of Science 64: 1–196. ———. 1938. Genera Hymenophyllacearum. The Philippine Journal of Science 67: 1–110. ———. 1947. Genera Filicum—the genera of ferns. New York: The Ronald Press Company. DUBUISSON, J.-Y. 1997a. Systematic relationships within the genus Trichomanes sensu lato (Hymenophyllaceae, Filicopsida): cladistic analysis based on anatomical and morphological data. Botanical Journal of the Linnean Society 123: 265–296. ———. 1997b. rbcL sequences: a promising tool for the study of the molecular systematics of the fern genus Trichomanes (Hymenophyllaceae)? Molecular Phylogenetics and Evolution 8: 128– 138. ———, R. HE´BANT-MAURI, and J. GALTIER. 1998. Molecules and morphology: conflicts and congruence within the fern genus Trichomanes (Hymenophyllaceae). Molecular Phylogenetics and Evolution 9: 390–397. ———, S. HENNEQUIN, E. J. P. DOUZERY, R. B. CRANFILL, A. R. SMITH, and K. M. PRYER. 2003a. rbcL phylogeny of the fern genus Trichomanes (Hymenophyllaceae), with special reference to neotropical taxa. International Journal of Plant Sciences 164: 753–761. ———, ———, F. RAKOTONDRAINIBE, and H. SCHNEIDER. 2003b. Ecological diversity and adaptative tendencies in the tropical fern Trichomanes L. (Hymenophyllaceae) with special reference to epiphytic and climbing habits. Botanical Journal of the Linnean Society 42: 41–63. EBIHARA, A., S. HENNEQUIN, K. IWATSUKI, P. D. BOSTOCK, S. MATSUMOTO, R. JAMAN, J.-Y. DUBUISSON, and M. ITO. 2004. Polyphyletic origin of Microtrichomanes (Prantl) Copel. (Hymenophyllaceae), with a revision of the species. Taxon 53: 935–948. ———, K. IWATSUKI, S. KURITA, and M. ITO. 2002. Systematic position of Hymenophyllum rolandi-principis Rosenst. or a monotypic genus Rosenstockia Copel. (Hymenophyllaceae) endemic to New Caledonia. Acta Phytotaxonica et Geobotanica 53: 35–49. ———, ———, T. A. OHSAWA, and M. ITO. 2003. Hymenophyllum paniense, a new species of filmy fern (Hymenophyllaceae) from New Caledonia. Systematic Botany 28: 228–235. FARRIS, J. S. 1989. The retention index and the rescaled consistency index. Cladistics 5: 417–419. FELSENSTEIN, J. 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39: 783–791. HASEBE, M., P. G. WOLF, K. M. PRYER, K. UEDA, M. ITO, R. SANO, G. J. GASTONY, J. YOKOYAMA, J. R. MANHART, N. MURAKAMI, E. H. CRANE, C. H. HAUFLER, and W. D. HAUK. 1995. Fern phylogeny based on rbcL nucleotide sequences. American Fern Journal 85: 134–181. HENNEQUIN, S. 2003. Phylogenetic relationships within the fern genus Hymenophyllum s.l. (Hymenophyllaceae, Filicopsida): contribution of morphology and cytology. Comptes Rendus Biologies 326: 599–611. ———. 2004. Le genre Hymenophyllum Sm. (Hymenophyllaceae, Filicopsida): syste´matique phylogeńe´tique, e´volution morphologique et histoire biogeógraphique. The`se de Doctorat de l’Universite´ Pierre et Marie Curie, Paris, France. 266 pp. ———, A. EBIHARA, M. ITO, K. IWATSUKI, and J.-Y. DUBUISSON. 2003. Molecular systematics of the fern genus Hymenophyllum s.l. (Hymenophyllaceae) based on chloroplastic coding and noncoding regions. Molecular Phylogenetics and Evolution 27: 283–301. IWATSUKI, K. 1977. Studies in the systematics of filmy ferns: III. An observation on the involucres. The Botanical Magazine Tokyo 90: 259–267. ———. 1984. Studies in the systematics of filmy ferns: VII. A scheme of classification based chiefly on the Asiatic species. Acta Phytotaxonica et Geobotanica 35: 165–179. ———. 1985. The Hymenophyllaceae of Asia, excluding Malesia.

283

Journal of the Faculty of Science, University of Tokyo, section III, Botany 13: 501–551. ———. 1990. Hymenophyllaceae. Pp. 157–163 in The families and genera of vascular plants, Vol. I: Pteridophytes and Gymnosperms, eds. K. U. Kramer and P. S. Green. Berlin: Springer-Verlag. LE THOMAS, A. 1961. Etude anatomique du rhizome et du pe´tiole des Hymenophyllaceae d’Afrique Occidentale et de la re´gion malgache. Bulletin de la Socie´te´ Scientifique de Bretagne 36: 217– 264. LEWIS, P. O. 2001. A likelihood approach to estimating phylogeny from discrete morphological character data. Systematic Biology 50: 913–925. MADDISON, W. P. and D. R. MADDISON. 2001. MacClade: analysis of phylogeny and character evolution, version 4.06. Sunderland: Sinauer Associates. MASON-GAMER, R. J. and E. A. KELLOGG. 1996. Testing for phylogenetic conflict among molecular data sets in the tribe Triticeae (Gramineae). Systematic Biology 45: 524–545. MORTON, C. V. 1968. The genera, subgenera and sections of the Hymenophyllaceae. Contributions from the United States National Herbarium 38: 153–214. OGURA, Y. 1938. Anatomie der Vegetationsorgane der Pteridophyten. Berlin: Gebru¨der Borntra¨ger. PICHI SERMOLLI, R. E. G. 1977. Tentamen Pteridophytorum genera in taxonomicum ordinem redigendi. Webbia 31: 313–512. POSADA, D. and K. A. CRANDALL. 2000. Selecting the best-fit model of nucleotide substitution. Systematic Biology 50: 580–601. PRESL, K. B. 1843. Hymenophyllaceae. Prag: Eine botanische Abhandlung. ———. 1849. Epimeliae Botanicae. Prag: A. Haase. PRYER, K. M., A. R. SMITH, J. S. HUNT, and J.-Y. DUBUISSON. 2001. rbcL data reveal two monophyletic groups of filmy ferns (Filicopsida: Hymenophyllaceae). American Journal of Botany 88: 1118–1130. RONQUIST, F. and J. P. HUELSENBECK. 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19: 1572–1574. SWOFFORD, D. L. 2001. PAUP*, phylogenetic analysis using parsimony (* and other methods), ver.4. Sunderland: Sinauer Associates. THOMPSON, J. D., D. G. HIGGINS, and T. J. GIBSON. 1994. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position specific gap penalties and weight matrix choice. Nucleic Acids Research 22: 4673–4680. TINDALE, M. D. and S. K. ROY. 2002. A cytotaxonomic survey of the Pteridophyta of Australia. Australian Systematic Botany 15: 839–937. VAN DEN BOSCH, R. 1861. Eerste Bijdrage tot de Kennis der Hymenophyllaceae. Verslagen en mededeelingen van de Afdeeling Natuurkunde; Koninklijke Akademie van Wetenschappen 11: 300– 330. WALKER, T. G. 1966. A cytotaxonomic survey of the Pteridophytes of Jamaica. Transactions of the Royal Society of Edinburgh 66: 27–237.

APPENDIX 1 Filmy ferns taxa used in this study, with distribution, GenBank accession numbers (in sequence: rbcL, rbcL-accD, rps4 1 rps4-trnS), and source and/or reference. 1) Distribution: A 5 Asia, AUS 5 Australia, C 5 Cosmopolitan, Ch 5 Chile; Ch-Arg 5 Chile and Argentina, E 5 Western Europe, H 5 Hawaii, M 5 Madagascar 1 Mascarene Islands, MP 5 Melanesia-Polynesia, N 5 Neotropics, NC 5 New Caledonia, NZ 5 New Zealand. 2) LPB 5 Laboratoire de Paleóbotanique et Paleóećologie, Paris. 3) Deposited in GenBank as Hymenophyllum tunbridgense. 4) Deposited in GenBank as Trichomanes sp. ‘ISEM-H2901’; voucher (see Dubuisson 1997a) reidentified here as T. tamarisciforme Jacq.


284

GALLEY

SYSTEMATIC BOTANY

Cardiomanes reniforme (G. Forst.) C. Presl, NZ, U30833 (Hasebe et al. 1995), AB08329; A. Ebihara 011222-07, New Zealand (TI), AY095132 (Hennequin et al. 2003); H. apiculatum Mett. ex Kuhn, N, AF275642 (Pryer et al. 2001), AY775438; J.-Y. Dubuisson HV 199723, Venezuela (LPB, Duke), AY095131 (Hennequin et al. 2003); H. armstrongii (Baker) Kirk, NZ, AY095109 (Hennequin et al. 2003), AB162691; A. Ebihara 011219-09, New Zealand (TI), AY095128 (Hennequin et al. 2003); H. australe Willd., AUS, AB191439; T. A. Ohsawa 001125-03, Australia (TI), AB191439; T. A. Ohsawa 00112503, Australia (TI), AY775412; T. A. Ohsawa 001125- 03, Australia (TI); H. badium Hook & Grev., A, AB191440; A. Ebihara 991121-06, Japan (TI), AB191440; A. Ebihara 991121-06, Japan (TI), AY775413; A. Ebihara 991121-06, Japan (TI); H. baileyanum Domin., AUS, AF275643 (Pryer et al. 2001), AB191441; A. Ebihara 010909-02, Australia (TI), AY095129 (Hennequin et al. 2003); H. barbatum (Bosch) Baker, A, AB064287 (Ebihara et al. 2002), AB064299 (Ebihara et al. 2002), AY095124 (Hennequin et al. 2003); H. caudiculatum Mart. var. productum (C. Presl) C. Chr., Ch, AB191442; T. A. Ohsawa 2019, Chile (TI), AY775439; T. A. Ohsawa 2019, Chile (TI), AY775414; T. A. Ohsawa 2019, Chile (TI); H. corrugatum H. Christ, A, AB191443; G. Miehe & U. Wuendisch 94-220-11, Tibet, China (TI), -, -; H. cuneatum Kunze, Ch, AY775401; P. Danton s. n., Juan Fernandez Is. (LPB), AY775440; P. Danton s. n., Juan Fernandez Is. (LPB), AY775415; P. Danton s. n., Juan Fernandez Is. (LPB); H. demissum (G. Forst) Sw., NZ, AY775402; Glasgow B. G. 830, cult. RBG Edinburgh, AY775441; Glasgow B. G. 830, cult. RBG Edinburgh, AY775416; Glasgow B. G. 830, cult. RBG Edinburgh; H. deplanchei Mett. ex Kuhn, NC, AB064288 (Ebihara et al. 2002), AB064300 (Ebihara et al. 2002), AY095136 (Hennequin et al. 2003); H. dilatatum (G. Forst.) Sw., NZ, AY095111 (Hennequin et al. 2003), AB191444; A. Ebihara 011219-06, New Zealand (TI), AY095138 (Hennequin et al. 2003); H. ferrugineum Colla, N, AF275644 (Pryer et al. 2001), AB191445; A. Ebihara 021224-02, Chile (TI), AF537124 (Hennequin et al. 2003); H. flabellatum Labill., MP, NZ, AY775403; unknown collector 42, Tahiti (LPB), AY775442; unknown collector 42, Tahiti (LPB), AY775417; unknown collector 42, Tahiti (LPB); H. flexuosum A. Cunn., NZ, AB217850; A. Ebihara 011219-03, New Zealand (TI), AB217851; D. Callen s.n., New Zealand (LPB), DQ077943; D. Callen s.n., New Zealand (LPB); H. fuciforme Sw., Ch, AB191446; A. Ebihara 021226-02, Chile (TI), AB191446; A. Ebihara 021226-02, Chile (TI), AY775418; A. Ebihara 021226-02, Chile (TI); H. fucoides (Sw.) Sw., N, U20933 (Hasebe et al. 1995), AY775449; J.-Y. Dubuisson HV-19979, Venezuela (Duke), AY095142 (Hennequin et al. 2003); H. fuscum (Blume) Bosch, A, AB064292 (Ebihara et al. 2002), AB064304 (Ebihara et al. 2002), AY775408; M. Ito 2000 0210-16, Java (TI); H. heimii Tardieu, M, AY775404; F. Rakotondrainibe 6008, Madagascar (P), AY775443; F. Rakotondrainibe 6008, Madagascar (P), AY775419; F. Rakotondrainibe 6008, Madagascar (P); H. hirsutum (L.) Sw., M, AY775407; J.-Y. Dubuisson HR-1999-6, La Reúnion (LPB, Duke), AY775450; J.-Y. Dubuisson HR-1999-6, La Reúnion (LPB, Duke), AY775432; J.-Y. Dubuisson HR-1999-6, La Reúnion (LPB, Duke); H. hygrometricum (Poir.) Desv., M, AY095113 (Hennequin et al. 2003), AY775451; J.-Y. Dubuisson HR-1999-13, La Reúnion (LPB, Duke), AY095118 (Hennequin et al. 2003); H. inaequale (Poir.) Desv., M, AY095112 (Hennequin et al. 2003), AB217848; J.-Y. Dubuisson HR1999-9, La Reúnion (LPB, Duke), AY095122 (Hennequin et al. 2003); H. javanicum Spreng., A, MP, AB191447; A. Ebihara 01090901, Australia (TI), AB191447; A. Ebihara 010909-01, Australia (TI), DQ077945; A. Ebihara 010909-01, Australia (TI); H. lanceolatum Hook. & Arn., H, AF275646 (Pryer et al. 2001), AY775452; T. O’Brien s.n., Hawaii (UC), AY095119; T. O’Brien s.n., Hawaii (UC); H. leratii Rosenst., NC, AB191448; J. Muzinger 767, New Caledonia (P, MO), AY775444; J. Muzinger 767, New Caledonia (P, MO), AY775421; J. Muzinger 767, New Caledonia (P, MO); H. marginatum Hook. & Grev., AUS, AB162692; A. Ebihara 010915-03, Australia (TI), AY775435; A. Ebihara 010915-03, Australia (TI), AY775409; A. Ebihara 010915-03, Australia (TI); H. mnioides Baker, NC, AB217849; A. Ebihara 001228-03, New Caledonia (TI), AB217849; A. Ebihara


[Volume 31

001228-03, New Caledonia (TI), DQ077944; A. Ebihara 001228-03, New Caledonia (TI); H. oligosorum Makino, A, AB064293 (Ebihara et al. 2002), AB064305 (Ebihara et al. 2002), AY775422; A. Ebihara 001105-01, Japan (TI); H. ooides F. Muell. & Baker, A, AB191449; M.H.A. 21589, New Guinea (K), -, -; H. paniense Ebihara et K. Iwats., NC, AB083275 (Ebihara et al. 2003), AB083275; A. Ebihara 001225-02, New Caledonia (P, TI, KYO, NOU), AY775410 (Ebihara et al. 2003); H. pectinatum Cav., Ch-Arg, AY095115 (Hennequin et al. 2003), AB191450; Asakawa 2017, Chile (TI), AY095134 (Hennequin et al. 2003); H. polyanthos (Sw.) Sw., N, AF275647 (Pryer et al. 2001), AB217847; Asakawa 174-4, Bolivia (UC), AY095139 (Hennequin et al. 2003) ; H. polyanthos (Sw.) Sw., A, AB064295 (Ebihara et al. 2002), AB064307 (Ebihara et al. 2002), AY775423; A. Ebihara 991122-01, Japan (TI); H. polyanthos (Sw.) Sw., M, AY775405; J.-Y. Dubuisson s.n., La Reúnion (LPB), AY775445; J.-Y. Dubuisson s.n., La Reúnion (LPB), AY775424; J.-Y. Dubuisson s.n., La Reúnion (LPB); H. polyanthos (Sw.) Sw., MP, AY775406; unknown collector 40, Tahiti (LPB), AB217846; unknown collector 40, Tahiti (LPB), AY775425; unknown collector 40, Tahiti (LPB); H. pulcherrimum Colenso, NZ, AB191451; A. Ebihara 011221-03, New Zealand (TI), AB191451; A. Ebihara 011221-03, New Zealand (TI), AY775426; A. Ebihara 01122103, New Zealand (TI); H. rarum R. Br., AUS, NZ, AB217845; A. Ebihara 011217-09, New Zealand (TI), -, -; H. sanguinolentum (G. Forst.) Sw., NZ, AB191452; A. Ebihara 011222-02, New Zealand (TI), AY775446; A. Ebihara 011222-02, New Zealand (TI), AY775427; A. Ebihara 011222-02, New Zealand (TI); H. scabrum A. Rich, NZ, AB083278 (Ebihara et al. 2003), AB083278 (Ebihara et al. 2003), AY775428; A. Ebihara 011223-05, New Zealand (TI); H. secundum Hook. & Grev., Ch-Arg, AF275648 (Pryer et al. 2001), AY775436; W. C. Taylor 6075, Chile (UC), AY095125 (Hennequin et al. 2003); H. sibthorpioides (Bory ex Willd.) Mett., M, AY095117 (Hennequin et al. 2003), AB162688; J.-Y. Dubuisson HR-1999-1, La Reúnion (P, Duke), AY095127 (Hennequin et al. 2003); H. subdimidiatum Rosenst., AUS, MP, NC, AB064290 (Ebihara et al. 2002), AB064302 (Ebihara et al. 2002), AY095140 (Hennequin et al. 2003); H. tenellum (Jacq.) Kuhn., M, AY095116 (Hennequin et al. 2003), AB191453; J.Y. Dubuisson HR-1999-27, La Reúnion (LPB, Duke), AY095126 (Hennequin et al. 2003); H. tunbrigense (L.) Sm., E, Y09203 (Dubuisson 1997b), AY775437; J.-Y. Dubuisson NV. 2.1, France (LPB, Duke), AY095123 (Hennequin et al. 2003); H. villosum Colenso, NZ, AB191454; A. Ebihara 011223-01, New Zealand (TI), AB191454; A. Ebihara 011223-01, New Zealand (TI), AY775429; D. Callen s.n., New Zealand; H. wrightii Bosch, A, AB064294 (Ebihara et al. 2002), AB064306 (Ebihara et al. 2002), AY775430; A. Ebihara 000901-01, Japan (TI); Hymenoglossum cruentum (Cav.) C. Presl., Pa, AY095107 (Hennequin et al. 2003), AB191455; T. A. Ohsawa 2015, Chile (TI), AY095133 (Hennequin et al. 2003); Rosenstockia rolandi-principis (Rosenst.) Copel., NC, AY095110 (Hennequin et al. 2003), AB083286 (Ebihara et al. 2002), AY095143 (Hennequin et al. 2003); Serpyllopsis caespitosa (Gaudich.) C. Chr, Ch-Arg, AF275649 (Pryer et al. 2001), AB191456; T. A. Ohsawa 2014, Chile (TI), AY095130 (Hennequin et al. 2003); T. cuspidatum Willd., M, AF537122 (Hennequin et al. 2003), -, AY095144 (Hennequin et al. 2003); T. digitatum Sw., M, AB162676 (Ebihara et al. 2004), AB162676 (Ebihara et al. 2004), AY095120 (Hennequin et al. 2003); T. javanicum Blume, A, MP, Y09195 (Dubuisson 1997b), AY775453; S. Hennequin 20017, cult. Indonesia (LPB), AY095141 (Hennequin et al. 2003) ; T. pallidum Blume, A, MP, NC, AUS, AB191457; A. Ebihara 001228-01, New Caledonia (TI), AB191457; A. Ebihara 001228-01, New Caledonia (TI), AY775431; A. Ebihara 001228-01, New Caledonia (TI); T. radicans Sw., N, AF275650 (Pryer et al. 2001), -, AF537123 (Hennequin et al. 2003); T. rigidum Sw., N, AY095108 (Hennequin et al. 2003), AY775447; J.-Y. Dubuisson HV-1997-3, Venezuela (Duke), AY095137 (Hennequin et al. 2003); T. taeniatum Copel., MP, AF275651 (Pryer et al. 2001), AB162681 (Ebihara et al. 2004), AY095121 (Hennequin et al. 2003); T. tamarisciforme Jacq., M, Y092024 (Dubuisson 1997b), AY775448; J.-Y. Dubuisson HR-1999-32, La Reúnion (LPB, Duke), AY095135 (Hennequin et al. 2003)