Molecular Phylogenetics and Evolution 81 (2014) 195–206

Contents lists available at ScienceDirect

Molecular Phylogenetics and Evolution

journal homepage: www.elsevier .com/ locate /ympev

Global phylogeny and biogeography of grammitid ferns (Polypodiaceae)

http://dx.doi.org/10.1016/j.ympev.2014.08.0171055-7903/� 2014 Elsevier Inc. All rights reserved.

⇑ Corresponding author at: The Pringle Herbarium, Department of Plant Biology,The University of Vermont, 27 Colchester Ave., Burlington, VT 05405, USA.

E-mail address: sundue@gmail.com (M.A. Sundue).

Michael A. Sundue a,c,⇑, Barbara S. Parris b, Tom A. Ranker c, Alan R. Smith d, Erin L. Fujimoto c,Delia Zamora-Crosby a, Clifford W. Morden c, Wen-Liang Chiou e, Cheng-Wei Chen f, Germinal Rouhan g,Regina Y. Hirai h, Jefferson Prado h

a The Pringle Herbarium, Department of Plant Biology, The University of Vermont, 27 Colchester Ave., Burlington, VT 05405, USAb Fern Research Foundation, 21 James Kemp Place, Kerikeri, Bay of Islands, 0230, New Zealandc Department of Botany, University of Hawaii, 3190 Maile Way, Honolulu, HI 96822, USAd University Herbarium, 1001 Valley Life Sciences Bldg. # 2465, University of California, Berkeley, CA 94720-2465, USAe Division of Botanical Garden, Taiwan Forestry Research Institute, 53 Nan-Hai Rd., Taipei 100, Taiwanf Institute of Molecular and Cellular Biology, National Tsing Hua University, Hsinchu 30013, Taiwang Muséum national d’Histoire naturelle, UMR CNRS 7205 ‘Origine, Structure et Evolution de la Biodiversité, Botanique, 16 rue Buffon CP 39, 75005 Paris, Franceh Instituto de Botânica, Caixa Postal 68041, CEP 04045-972 São Paulo, SP, Brazil

a r t i c l e i n f o a b s t r a c t

Article history:Received 23 April 2014Revised 13 August 2014Accepted 15 August 2014Available online 27 August 2014

Keywords:PolypodiaceaeEpiphytesPhylogenyGrammitid fernsBiogeography

We examined the global historical biogeography of grammitid ferns (Polypodiaceae) within a phyloge-netic context. We inferred phylogenetic relationships of 190 species representing 31 of the 33 currentlyrecognized genera of grammitid ferns by analyzing DNA sequence variation of five plastid DNA regions.We estimated the ages of cladogenetic events on an inferred phylogeny using secondary fossil calibrationpoints. Historical biogeographical patterns were inferred via ancestral area reconstruction. Our resultssupported four large-scale phylogenetic and biogeographic patterns: (1) a monophyletic grammitid cladethat arose among Neotropical polypod ancestors about 31.4 Ma; (2) a paraphyletic assemblage of cladesdistributed in the Neotropics and the Afro-Malagasy region; (3) a large clade distributed throughout theAsia–Malesia–Pacific region that originated about 23.4 Ma; and, (4) an Australian or New Zealand originof the circumaustral genus Notogrammitis. Most genera were supported as monophyletic except forGrammitis, Oreogrammitis, Radiogrammitis, and Zygophlebia. Grammitid ferns are a well-supported mono-phyletic group with two biogeographically distinct lineages: a primarily Neotropical grade exhibitingseveral independent successful colonizations to the Afro-Malagasy region and a primarily Paleotropicalclade exhibiting multiple independent dispersals to remote Pacific islands and temperate, austral regions.

� 2014 Elsevier Inc. All rights reserved.

1. Introduction

With close to 1500 species, the cosmopolitan Polypodiaceae areamong the largest families of ferns (Smith et al., 2006; unpubl.).Most species are epiphytes and represent the fourth largest familyof epiphytic vascular plants (Gentry and Dodson, 1987). The fam-ily’s prominence was established during a Cenozoic radiation inwhich leptosporangiate ferns diversified into new niches in other-wise angiosperm-dominated forests (Schuettpelz and Pryer, 2009).

The family includes a large monophyletic clade referred to as the‘‘grammitids’’, which were often treated as a separate family(Grammitidaceae) prior to molecular phylogenetic evidence(Schneider et al., 2004). With ca. 900 species (Perrie and Parris,

2012) treated in 33 genera, the grammitids comprise close totwo-thirds of the diversity in the family; the non-grammitid Poly-podiaceae (referred to here as the polypods) include some 450 spe-cies treated in 40 genera (Smith et al., 2006; unpubl.). Whereas thepolypods are characterized by often round exindusiate sori, reni-form, monolete spores usually lacking chlorophyll at maturity, usu-ally scaly leaves, and dorsiventral rhizomes with well-developedphyllopodia, synapomorphies for the grammitid clade include areduction in the number of cells in the middle of the sporangialstalk from three to one (Wilson, 1959), a reduction in the numberof vascular bundles in the petiole from several (like most eupolypodI ferns) to a single bundle or two that fuse above the base (Parris,1990; Sundue, 2010a), and globose, trilete spores that contain fullydeveloped chlorophyll at maturity (Parris, 1990; Sundue, 2010a).Although both groups often have minute branched hairs (c.0.1 mm long), polypods most often have broad basifixed or peltatescales whereas grammitids do not have these scales but have

196 M.A. Sundue et al. / Molecular Phylogenetics and Evolution 81 (2014) 195–206

pluricellular uniseriate setae (Sundue et al., 2010). Occasionalintermediates between scales and setae suggest that these twotypes of indument may in fact be homologous (Sundue, unpub-lished). Perhaps owing to their relatively larger size, ease of cultiva-tion, and conspicuous adaptations (e.g., myrmecophily, detrituscatching leaves, desiccation tolerance), it is the polypods, as definedabove, that have received most attention in the literature. Thegrammitids, which are generally small, inconspicuous, and notamenable to cultivation, are less well known. This has begun tochange recently, with grammitids becoming the focus of many sys-tematic studies.

Polypodiaceae have undergone massive generic re-circumscrip-tion following molecular phylogenetic studies (e.g., Ranker et al.,2004; Schneider et al., 2004; Kreier et al., 2008; Wang et al.,2010), which demonstrated that many genera were not monophy-letic. Within the grammitids, the classical genera Grammitis, Xip-hopteris, and Ctenopteris were largely based on blade dissection, acharacter that has proven nearly useless in defining monophyleticclades. More recently described genera such as Terpsichore A.R. Sm.and Lellingeria A.R. Sm. & R.C. Moran, have relied instead on suitesof morphological characters including microscopic and anatomicalfeatures (Smith et al., 1991; Smith, 1993). These characters havemore reliably defined monophyletic groups, but are still subjectto cases of morphological homoplasy resulting in polyphyletic gen-era (Ranker et al., 2004; Labiak et al., 2010b). Consequently, thepolyphyly of Lellingeria R.C. Moran & A.R. Sm. (Labiak et al.,2010b) and Terpsichore A.R. Sm. (Ranker et al., 2004; Sundueet al., 2010) has led to the segregation of Alansmia Kessler et al.(Kessler et al., 2011), Ascogrammitis Sundue (Sundue, 2010b),Galactodenia Sundue & Labiak (Sundue et al., 2012), LeucotrichumLabiak (Labiak et al., 2010a), Moranopteris R.Y. Hirai & J. Prado(Hirai et al., 2011), Mycopteris Sundue (Sundue, 2013), and Steno-grammitis Labiak (Labiak, 2011).

Similarly, our understanding of Paleotropical grammitids haschanged radically. Two names that had been widely applied arenow recognized as synonyms of other genera: Ctenopteris Blumeex Kunze = Prosaptia C. Presl (Price, 1982, 1987), and XiphopterisKaulf. = Cochlidium Kaulf. (Bishop, 1978). Moreover, an under-standing that traditional generic boundaries were not well defined(Parris, 1977, 1984), and the determinations of polyphyly of suchlarge historically recognized genera as Grammitis Sw. (in the strictsense including only those species with a distinct black, sclerifiedleaf margin, sensu Parris, 2007), Ctenopteris, and Xiphopteris, whichwere defined mainly on the basis of blade dissection, led to therecent coining of Archigrammitis Parris (Parris, 2013), Chrysogramm-itis Parris (Parris, 1998a), Dasygrammitis Parris (Parris, 2007),Notogrammitis Parris (Perrie and Parris, 2012), Radiogrammitis Par-ris (Parris, 2007), Themelium (T. Moore) Parris (Parris, 1997), andTomophyllum (E. Fourn.) Parris (Parris, 2007), and the reinstatementof Oreogrammitis Copel. (Parris, 2007). With the exception ofNotogrammitis, which Perrie and Parris (2012) distinguished fromGrammitis s.l. using plastid DNA markers, the circumscription ofPaleotropical genera has not had the benefit of densely sampledmolecular phylogenetic studies. Consequently, phylogeneticrelationships among Paleotropical grammitids remain largelyuncertain.

Grammitids are found most abundantly in montane forests(Parris, 2005; Parris et al., 1992). At higher elevations they can bean important component of the fern flora, for example, on Mt.Kinabalu (Sabah, Malaysia) grammitid ferns constitute 25–35% oftotal fern diversity between 2000–4000 m (Kessler et al., 2001).Likewise, Notogrammitis crassior occurs up to 2600 m in New Zea-land (Parris and Given, 1976), higher than any other fern speciesthere. Among vascular epiphytes, the two species thought to growat highest recorded elevations both belong to Melpomene(Sylvester et al., 2014). The majority of grammitid diversity is

confined to the tropics, although some groups extend to the tem-perate zones, reaching 40�N and 56�S (Parris, 2003).

Parris (2003) assigned grammitid distributions to two majorphytogeographic zones: (1) Neotropics–Africa–Madagascar (includ-ing also the Mascarenes, Seychelles, and Comoros) where there areabout 300 species, and (2) Asia–Malesia–Pacific with about 450species. Current species counts are c. 400 species for Neotropics–Africa–Madagascar, and c. 500 for Asia–Malesia–Pacific (Parris,unpubl.), reflecting the progress in grammitid systematics in the lastdecade. Dispersal within each zone appears to be common; all butone of the genera present in Africa–Madagascar also occur in theNeotropics, and at least three species are found in both places(Moran and Smith, 2001). Likewise, the Asian, Malesian, and Pacificareas contain overlapping diversity. Phylogenetic analyses furthercorroborate these two phytogeographic zones, with both Rankeret al. (2004) and Sundue et al. (2010) finding evidence that mostspecies from the Asia, Malesia, and the Pacific belong to a singleclade. Migration between these two zones is rare. As currentlyunderstood, only four genera cross these phytogeographic zones:(1) Stenogrammitis, a primarily Neotropical genus that reachesAfrica–Madagascar, the Hawaiian Islands, and a few south-Pacificislands (Labiak, 2011; Ranker et al., 2010); (2) Ctenopterella, a mainlyMalesian Pacific genus that extends to Africa (Parris, 2007); (3) Noto-grammitis, an austral genus most diverse in Australia and New Zea-land that reaches South Africa and southern South America, and (4)Grammitis s.s., ranging from the western Pacific through the Neo-tropics to Africa, Madagascar, and the Mascarenes. Notably theHawaiian Islands represent an area of overlap, appearing to havebeen colonized via migrations from both phytogeographic zones(Geiger et al., 2007).

The historical biogeography of ferns includes examples of bothvicariance and long-distance dispersal, and discerning betweenthese has been a goal of biogeographic research (Barrington,1993; Wolf et al., 2001; Korall and Pryer, 2014; Labiak et al.,2014). Copeland (1939, 1952) hypothesized that Grammitis s.l.(including all species with simple and entire leaves) were ances-trally Antarctic in distribution, attaining their present distributionvia migration through New Zealand, South Africa and Madagascar,and the southern Andes. While this austral source for the origin ofgrammitids cannot be ruled out, migration via all three routes con-flicts with current inferences. Based on the results of Schuettpelzand Pryer (2009), it appears that the Polypodiaceae diverged fromPaleotropical ancestors �55.8 Ma, and migrated into the Neotrop-ics �43 Ma. Grammitids appear to have evolved from these Neo-tropical polypod ancestors �30.6 Ma. Subsequent migration toAsia/Africa occurred later, and must have operated via long-distance dispersal, at least at the intercontinental level, becausethe continents have not moved appreciably since the cladeevolved. Long-distance dispersal of spores by wind was alsoinvoked by Moran and Smith (2001) to explain floristic similaritiesbetween the Neotropics and Africa/Madagascar. The timing andpatterns of these dispersals remain largely unknown, but puta-tively, it would seem that they are quite recent; some of the samespecies occur in both areas, and there is very little diversification inseveral of the genera that migrated to Africa, namely MelpomeneA.R. Sm. & R.C. Moran, Alansmia, Cochlidium, Leucotrichum, andCeradenia L.E. Bishop. The pattern is similar in some other genera,such as Zygophlebia L.E. Bishop, Enterosora Baker, Stenogrammitis,and Grammitis s.s., but in these genera more diversification hasoccurred in Africa and Madagascar. The origin of the Asian–Malesian–Pacific grammitids; however, is particularly unclear.

With these issues in mind, we undertook a phylogenetic studyof grammitid ferns aiming establishing a global picture of phylog-eny and historical biogeography. Our sampling included 50% ofPaleotropical species and 31 of the 33 currently recognized genera(Archigrammitis Parris and Luisma M.T. Murillo & A.R. Sm. remain

Table 1Type species of grammitid genera sampled.

Genus Species Sampled

Acrosorus exaltatus (Copel.) Copel. = friderici-et-pauli(Christ) Copel

p

M.A. Sundue et al. / Molecular Phylogenetics and Evolution 81 (2014) 195–206 197

unsampled). We addressed two primary questions: Are the Neo-tropical and Afro-Malagasy floras closely related as hypothesizedby Parris (2003)? Do the Asian–Malesian–Pacific species constitutea single clade, or is this Old World diversity the result of multiplemigrations?

Adenophorus tripinnatifidus Gaudich –Alansmia lanigera (Desv.) Moguel & M. Kessler

p

Archigrammitis friderici-et-pauli (Christ) Parris –Ascogrammitis athyrioides (Hook.) Sundue

p

Calymmodon cucullatus (Nees & Blume) C. Presl –Ceradenia curvata (Sw.) L.E. Bishop

p

Chrysogrammitis glandulosa (J. Sm.) Parrisp

Cochlidium graminoides (Sw.) Kaulf –Ctenopterella blechnoides (Grev.) Parris –Dasygrammitis mollicoma (Nees & Blume) Parris

p

Enterosora campbellii Baker –Galactodenia delicatula (M. Martens & Galeotti) Sundue &

Labiak

p

Grammitis marginella (Sw.) Sw –Lellingeria apiculata (Kunze ex Klotzsch) A.R. Sm. & R.C.

Moran

p

Leucotrichum organense (Gardner) Labiakp

Lomaphlebia linearis (Sw.) J. Sm. = graminea (Sw.) Parris –Luisma bivascularis M.T. Murillo & A.R. Sm –Melpomene moniliformis (Lagasca ex Sw.) A.R. Sm. & R.C.

Moran

p

Micropolypodium pseudotrichomanoides (Hayata)Hayata = okuboi (Yatabe) Hayata

p

Moranopteris basiattenuata (Jenman) R.Y. Hirai & J. Prado –Mycopteris taxifolia (L.) Sundue

p

Notogrammitis billardierei (Willd.) Parrisp

Oreogrammitis clemensiae Copel –Prosaptia contigua (G. Forst.) C. Presl

p

2. Materials and methods

2.1. DNA extraction and amplification

Ingroup sampling included 190 species (202 accessions) from31 of the 33 currently recognized genera; samples of Archigramm-itis and Luisma were unavailable. Sampling included the type spe-cies for 20 genera (Table 1). Outgroups included five species ofNeotropical polypods found to be closely related in previous stud-ies (e.g., Schuettpelz and Pryer, 2009). This choice was furtherinformed by our unpublished analysis of available sequences inGenBank that also find these Neotropical polypods as the closestrelative of the grammitid clade. Sequences not generated by uswere downloaded from GenBank. DNA extraction and PCR amplifi-cation protocols followed those of Labiak et al. (2010b). We PCR-amplified five plastid DNA markers: the atpß and rbcL codingregions, and the trnL-trnF, rps4-trnS, and trnG-trnR intergenic spac-ers, and generated 67, 71, 106, 92, and 71 sequences of each,respectively. DNA sequencing was performed at the GreenwoodMolecular Biology Facility at the University of Hawai’i at Manoa,and all sequences were submitted to GenBank (Appendix 1).

Radiogrammitis setigera (Blume) Parrisp

Scleroglossum pusillum (Blume) Alderwp

Stenogrammitis myosuroides (Sw.) Labiakp

Terpsichore asplenifolia (L.) A.R. Smp

Themelium tenuisectum (Blume) Parris –Tomophyllum subsecundodissectum (Zoll.) Parris –Xiphopterella hieronymusii (C. Chr.) Parris

p

Zygophlebia sectifrons (Kunze ex Mett.) L.E. Bishopp

2.2. Alignment and analyses

Sequences were edited and contigs were produced usingGeneious 6.17 (Biomatters Ltd., San Francisco, CA), and the MAAFTplug-in was used to produce alignments (Katoh, 2013). Alignmentswere visually inspected and no areas appeared to be alignedambiguously. For each aligned marker, optimal data partitioningand models of substitution evolution were estimated using AICcin PartitionFinder (Lanfear et al., 2012). We provided PartitionFind-er with subsets for each marker, and for the three coding regionswe provided subsets for each nucleotide position. The resultingbest scheme was a single GTR + I + G model for the dataset. Thiswas implemented in the Bayesian and likelihood tree searches.We conducted tree searches using maximum parsimony (MP),maximum likelihood (ML) and Bayesian (BI) analyses. Maximumparsimony tree searches were performed using TNT (Goloboffet al., 2008) employing two approaches. First, we conducted tradi-tional heuristic searches with 1000 parsimony ratchet replicates(Nixon, 1999) (200 iteration ratchet, the up and down weightsset to 5% each), holding 20 trees per ratchet, followed by tree-bisection-reconnection (TBR)-max branch swapping. We thenimplemented a New Tech search strategy set at level 15 imple-menting tree fusing, sectorial search, and the parsimony ratchetfinding the minimum length tree 10 times. In each case, supportfor nodes was calculated by bootstrap analyses (BS), with 1000replicates using the same methods. Maximum likelihood treesearches were conducted using RAxML (Stamatakis, 2006) throughthe CIPRES portal (Miller et al., 2010). Five independent searchesfor the ‘best tree’ and 10,000 BS replicates were generatedimplementing the best partition scheme determined by Partition-Finder. Bayesian tree searches were conducted using MrBayes(Huelsenbeck and Ronquist, 2001; Ronquist and Huelsenbeck,2003) through the Oslo Lifeportal (https://lifeportal.uio.no/root).We conducted five runs implementing the best partition schemedetermined by PartitionFinder for 10 million generations. Eachrun included four chains (one cold, three heated) with unlinkedparameters, and chain temperature set to 0.2. Priors were uniform

except that rates were allowed to vary among loci (ratepr = vari-able). The posterior was sampled every 1000 generations, andthe first 25% discarded as ‘‘burn-in’’. Convergence was estimatedby examining the standard deviation of split frequencies, plottingthe output parameters in TRACER v 1.5 (Rambaut et al., 2013),and examining tree files in AWTY (Wilgenbush et al., 2004;Nylander et al., 2008).

2.3. Fossil calibration

Known fossil calibration points for the Polypodiaceae are lim-ited. Consequently, we chose to calibrate our chronogram usingnode ages generated by a fossil-calibrated analysis of all leptospo-rangiate ferns (Schuettpelz and Pryer, 2009). Grammitis succineaL.D. Gómez, a fossil in Dominican amber, is the only known fossilthat appears to belong to the grammitids (Gómez, 1982). Whilepurported to have several synapomorphies for the grammitids(namely, stiff, erect uniseriate, pluricellular setae and uniseriatesporangial stalks), this fossil is relatively young (approximately25 Ma) and cannot be placed with confidence in any clade withingrammitids; consequently, it was not useful to our study.

We estimated divergence times using a relaxed molecular clockas implemented in BEAST, using a Markov chain Monte Carlo strat-egy (Drummond et al., 2006, 2013; Drummond and Rambaut,2007). We partitioned the dataset by plastid DNA region, and spec-ified the optimal model for each region as determined by Partition-Finder. We implemented a Yule speciation tree prior and anuncorrelated lognormal model of rate change, with clock models

Table 2Dispersal constraints (Q matrix).

A B C D E F

A 1 0.25 0.01 0.25 0.01 0.25B 0.25 1 0.25 0.25 0.01 0.01C 0.01 0.25 1 0.25 0.01 0.01D 0.25 0.25 0.25 1 0.01 0.01E 0.01 0.01 0.01 0.01 1 0.01F 0.25 0.01 0.01 0.01 0.01 1

Table 3Summary of molecular sequences.

Region Alignedbases

Variablebases

Parsimonyinformativecharacters

No. ofsamples

atpB 1174 394(34%) 295(25%) 173rbcL 1329 503(38%) 320(24%) 200rps4-trnS 778 537(69%) 393(51%) 144trnG-trnR 1572 877(56%) 652(41%) 117trnL-trnF 661 419(61%) 305(45%) 178Total 5514 2730(49%) 1974(35%) 207

198 M.A. Sundue et al. / Molecular Phylogenetics and Evolution 81 (2014) 195–206

unlinked between partitions and a GTR + G substitution model inall cases. We used the following calibration points fromSchuettpelz and Pryer (2009) with normal distribution priors:the most recent common ancestor (MRCA) of the grammitid clade(31.2 Ma, 3.12 s.d.), the MRCA of the large Neotropical cladeincluding Mycopteris–Stenogrammitis (23.3 Ma, 2.33 s.d.), the largeprimarily Paleotropical clade including Moranopteris–Oreogramm-itis (23.4 Ma, 2.34 s.d.), and the MRCA of Serpocaulon A.R. Sm.(15.5 Ma, 1.55 s.d.). The clade corresponding to each calibrationpoint was constrained to be monophyletic. Three analyses wererun, each for 30,000,000 generations, with parameters sampledevery 1000 generations. The program LogCombiner was used topool the resulting files. Tracer v1.5 (Drummond and Rambaut,2007) was used to examine the posterior distribution of all param-eters and their associated statistics including estimated samplesizes (ESS) and 95% highest posterior density (HPD) intervals. Theprogram TreeAnnotator v2.0.2 (Drummond and Rambaut, 2007)was used to summarize the post burn-in trees and produce a max-imum clade credibility chronogram showing mean divergencetime estimates with 95% HPD intervals.

Historical Biogeography – Ancestral area reconstruction (AAR)was conducted using the dispersal-cladogenesis (DEC) model asimplemented in the program Lagrange (Ree and Smith, 2008).We defined the following geographic areas: (A) Neotropical, (B)Africa–Madagascar and islands of the Atlantic (Ascension and St.Helena) and Indian Oceans (Comoros, Réunion, Mauritius, Sey-chelles), (C) tropical Asia and the Pacific including Bougainville,Solomon Islands and Vanuatu, as well as NE Australia and New Cal-edonia and Micronesia (Palau, Pohnpei, Kosrae), (D) a temperatecircumaustral region south of 28�S, (E) Hawaiian Islands and thecentral and eastern South Pacific, and (F) Nearctic. Distributionswere assigned to each species based upon known records of her-barium specimens, and species with broad ranges were scored aspolymorphic. These occurred in Alansmia, Cochlidium, and Steno-grammitis. Given the recent age of grammitid ferns, and vagilityof their spores, we made no restrictions in the adjacency matrix.Dispersal constraints (Q matrix) between these regions werescored as uncommon (0.25) or very uncommon (0.01) and are pre-sented in Table 2; maximum range size was set at 2.

3. Results

3.1. Phylogenetic analyses

The final aligned dataset included 5569 sites of which 1974(35%) were parsimony informative (Table 3). During Bayesian anal-yses, runs converged after the first 1 million generations, ESS val-ues of each parameter were all well above the recommendedthreshold of 200, and the traces of corresponding parameters inindependent runs converged to the same optimum. Maximum par-simony tree searches with TNT found shortest trees of 10,761steps. In a strict consensus, these MP trees retained all backbonerelationships, except for one large polytomy forming in the crowngroup of Oreogrammitis, Radiogrammitis, and Themelium. The over-all topology of the MP trees was very similar to that resulting from

the ML and Bayesian analyses. One minor difference was thatGrammitis s.s. was recovered as monophyletic, but without strongbranch support. Best ML trees resulting from RAxML shared anoverall topology identical to that of the Bayesian trees and there-fore are not discussed in further detail. Results from the Bayesiananalyses were generally well resolved and well supported (Fig. 1).

Results from our analyses supported grammitids as monophy-letic (Fig. 1). The overall topology can be explained in general asa monophyletic tropical Asian clade nested within a primarily Neo-tropical and African grade (Fig. 2). Our results support the Neotrop-ical genus Moranopteris as sister to this tropical Asian clade, andthe tropical Asian genus Chrysogrammitis, which has been difficultto resolve in previous analyses, as sister to the remainder of thetropical Asian clade.

Relationships within the Neotropical and African grade are gen-erally congruent with previous studies, but with improved resolu-tion. Well-supported relationships not previously reportedinclude: (1) Zygophlebia is paraphyletic with Enterosora nestedwithin it; (2) Ceradenia is resolved as two clades correspondingto Bishop’s (1988) subgenera Ceradenia and Filicipecten; and (3)Lomaphlebia J. Sm. is sister to the clade comprising Grammitis s.s.and Cochlidium. In previous studies (e.g., Sundue et al., 2010), Terp-sichore was resolved as sister to all other grammitids, but here it issister to the clade of Adenophorus Gaudich., Cochlidium, Grammitiss.s., and Lomaphlebia. These five genera instead compose a cladesister to all other grammitids. Paraphyly of Grammitis s.s. withrespect to Cochlidium is also a novel result supported by our Bayes-ian and likelihood results.

The remaining tropical Asian genera resolve in two mainclades, one comprising Calymmodon C. Presl, Dasygrammitis,Micropolypodium Hayata, Scleroglossum Alderw., Tomophyllum,and Xiphopterella, and the other comprising Acrosorus, Ctenopterel-la, Notogrammitis, Oreogrammitis, Prosaptia, Radiogrammitis, Theme-lium, and four species currently combined in Grammitis (listed inFigs. 1 and 2 as ‘‘Grammitis’’) that are not placed within currentgeneric concepts.

Most Paleotropical genera were supported as monophyletic,including Calymmodon, Chrysogrammitis, Dasygrammitis, Micropo-lypodium, Notogrammitis, Prosaptia, Tomophyllum, Scleroglossum,and Xiphopterella. In contrast, Oreogrammitis and Radiogrammitisare polyphyletic. The three species of Themelium included wereresolved as monophyletic, but nested within the large clade com-prising Oreogrammitis and Radiogrammitis. Ctenopterella and Acros-orus included only one species each, so the monophyly of thosegenera remains untested.

3.2. Analyses of divergence times and diversification rates

BEAST analyses estimated the MRCA of the grammitids at31.4 Ma and the MRCA of the large Paleotropical clade at 23.4 Ma(Fig. 2a, b). While stem ages varied among Paleotropical genera,many of them began to diversify within the last 8.4 Ma, includingAcrosorus, Calymmodon, Chrysogrammitis, Dasygrammitis,

M.A. Sundue et al. / Molecular Phylogenetics and Evolution 81 (2014) 195–206 199

Micropolypodium, Prosaptia, Scleroglossum, and Xiphopterella. Thelarge predominantly Neotropical clades were estimated to origi-nate from 18 to 23 Ma, with subsequent diversification occurringas early as 14.5 Ma for Ascogrammitis and as recently as 3.8 Ma inMelpomene. The stem age of the Hawaiian Island endemic Adeno-phorus was estimated as 22.5 Ma, and subsequent diversificationof the genus was estimated at 10.6 Ma.

3.3. Ancestral area reconstruction

Lagrange analyses reconstructed a Neotropical ancestral areafor the entire grammitid clade (Fig. 2a). The majority of earlydiverging lineages are distributed in the Neotropics, and a Neo-tropical ancestral area was retained throughout the first six back-bone nodes of the tree. The present distribution of grammitids intropical Asia was explained by a single transition to tropical Asiaat 23.4 Ma (25.0–20.7 Ma). Subsequent backbone nodes after thistransition were reconstructed as having a tropical Asian ancestralarea as well. There was a single transition from within the tropicalAsian clade to the austral region at 14.7 Ma (21.0–8.6) by the genusNotogrammitis.

Based on our sampling, the present distribution of grammitidsin Africa and Madagascar is explained by at least six separatemigrations from the Neotropics: (1) the Grammitis cryptophlebiaclade 6.8 Ma [11.9–3.0]; (2) Alansmia elastica 4.9 Ma [9.9–2.0];(3) Melpomene flabelliformis 0.3 Ma [2.3–0.002]; (4) Stenogrammitisoosora 5.59 Ma [8.9–2.6]; and (5) Zygophlebia 12.6 Ma [22.1–7.5].Migrations of the Grammitis cryptophlebia clade and Zygophlebiadiffer from the others by involving more than a single species.The former is a single event followed by a small radiation of threespecies. The Zygophlebia migration involves two species (threeaccessions) that comprise the first two bifurcations of the Zygo-phlebia-Enterosora clade. This could be interpreted as a singlemigration followed by a subsequent migration back to the Neo-tropics, or as two separate migrations to Africa–Madagascar.

The present distribution of grammitids in the Hawaiian Islandswas explained by three migrations, two from the Neotropics (Sten-ogrammitis 1.5 Ma [4.7–0.38]; Adenophorus 22.5 [29.0–14.1]); andone from tropical Asia (Oreogrammitis hookeri + O. forbesiana2.3 Ma [7.08–0.36]).

4. Discussion

4.1. Phylogenetic results

4.1.1. Overall resultsOur sampling included 31 of the 33 currently recognized gram-

mitid genera. We find that 24 of them are monophyletic (Fig. 1);however, these monophyletic genera are nested within other gen-era in three cases. Ctenopterella (14 spp.) and Acrosorus (11 spp.)are each represented by a single species, and so their monophylyremains untested. Monophyly among the predominantly Neotrop-ical lineages is not surprising as many of these genera wererecently circumscribed following results of molecular phylogeneticstudies. Our finding that most Paleotropical genera are monophy-letic is surprising. Most of these generic concepts were based onmorphological characters alone, which had been shown to beprone to homoplasy among Neotropical lineages (Ranker et al.,2004; Sundue, 2010a,b; Sundue et al., 2010).

4.1.2. Cochlidium, Grammitis s.s., and LomaphlebiaCircumscription of Grammitis, typified by G. marginella, a Neo-

tropical species not included in our analyses (Bishop, 1977), hasbeen a central problem in the systematics of grammitid ferns.One extreme has been to use it to include all or nearly all of the

diversity of the clade (e.g., Tryon and Tryon, 1982; Christenhuszand Chase, 2014). Most recent authors; however, have restrictedit to include the ca. 25 species with simple, entire leaves that haveblack sclerotic margins. The decision to adopt a broad Grammitiscomprising the entire grammitid lineage by Tryon and Tryon(1982) was conservative, but justified by their understanding thatthe other two widely applied names at that time, Ctenopteris andXiphopteris, were patently artificial. However, the problems in gen-eric circumscription that Tryon and Tryon (1982) faced have nowlargely been resolved by subsequent studies. Nonetheless,Christenhusz and Chase (2014) recently advocated treating allgrammitid genera at the subgeneric rank within a single genus,Grammitis, and the clade as a tribe Polypodieae within an unwieldymega-family Polypodiaceae, subfamily Polypodioideae. We viewthis decision as retrogressive, uninformative to the goals ofbiological systematics, and contrary to all other recent advancesin depiction of the relationships of ferns. It is also inconsistent withtheir proposed treatment of the remainder of the family. Their clas-sification has the effect of obscuring understanding rather thanclarifying it. A more informative classification can be found inSmith et al. (2008).

Surprisingly, our results do not support the monophyly ofGrammitis s.s., the black-margined group. Instead, it is paraphylet-ic, with a Neotropical clade and an Afro-Malagasy clade, and thegenus Cochlidium sister to the latter (Fig. 1). Cochlidium differs mor-phologically from Grammitis s.s. by having veins usually ending inhydathodes, lacking the dark sclerotic margin, and in some casesby having a coenosorus (Bishop, 1978). The coenosorus is oftensunken into a thickened lamina, giving Cochlidium a very differentappearance from Grammitis s.s., in spite of having a large numberof similarities that include radially symmetrical rhizomes, conco-lorous scales, and the loss of laminar setae. These results, togetherwith the presence of Grammitis cryptophlebia (Baker) Copel., whichlacks a black sclerotic lamina margin, as sister to G. melanoloma(Cordemoy) Tardieu, which has a black margin, indicate that theblack sclerotic lamina margin, previously considered to be a uniquecharacter, may in fact be homoplastic. However, our sampling ofGrammitis s.s. is limited, and denser sampling is needed to accu-rately estimate the evolution of that character. Sister to all of theseis Lomaphlebia turquina (Maxon) Sundue and Ranker (Appendix 2),one of two species in the genus endemic to the Caribbean andwhich has not been included in previous molecular phylogeneticanalyses. Its position here is supported by its general morphologyin that it is very similar to Grammitis s.s., but lacks the black scle-rotic lamina margin and has a submarginal commissural vein.

4.1.3. Ceradenia, Enterosora, and ZygophlebiaOur results support a monophyletic Ceradenia as sister to a

clade of Zygophlebia with Enterosora nested within the latter(Fig. 1). These largely agree with previous results; however,Sundue et al. (2010) found Zygophlebia nested within Enterosora,or sister to it. Bishop (1988) anticipated the close relationshipbetween Ceradenia and Zygophlebia and subsequent studies havequestioned their circumscription (Rakotondrainibe and Deroin,2006). In contrast, the distinction between Enterosora andZygophlebia has received less attention. The two genera share anumber of characters, and the most prominent character used toseparate them, spongiose mesophyll, is homoplastic throughoutgrammitids (Bishop and Smith, 1992; Sundue, 2010a). Theseresults suggest that maintaining the two as separate genera maynot be tenable.

Ceradenia is resolved into two clades that correspond withBishop’s subgenera: subg. Ceradenia and subg. Filicipecten (Fig. 1).Although several of the species included here were described afterBishop’s (1988) classification, the morphological characters heused still apply. Subgenus Ceradenia is characterized by radially

Fig. 1. Fifty-percent majority rule consensus phylogram from the Bayesian analysis using the combined dataset. Numbers represent the posterior probability (PP) values ofeach branch. The scale bar indicates the number of substitutions per site. The dotted lines indicate where the two portions of the tree are connected.

200 M.A. Sundue et al. / Molecular Phylogenetics and Evolution 81 (2014) 195–206

Fig. 2. Biogeographical hypothesis for the grammitid ferns inferred by the DEC model using the maximum clade credibility chronogram from the BEAST analysis. Blue barsdepict the median divergence time estimates with 95% HPD intervals of each node age. The most likely ancestral areas are indicated at each node by colored squarescorresponding to areas indicated on the map. Two colored squares are present when ancestral ranges include two regions. Numbers at each node represent the probabilitiesof ancestral ranges. Areas depicted are: Neotropical (dark blue), Africa–Madagascar and islands of the Atlantic and Indian Oceans (purple), tropical Asia and the Pacific (red), atemperate circumaustral region south of 28�S (orange), Hawaiian Islands and the central and eastern South Pacific (green), and the Nearctic (light blue). Scale bar indicatesdivergence times in millions of years. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

M.A. Sundue et al. / Molecular Phylogenetics and Evolution 81 (2014) 195–206 201

symmetrical rhizomes, short or absent petioles, and white wax-likeglandular hairs upon the laminar surfaces and paraphyses. Dorsi-ventral rhizomes, elongate petioles, and white wax-like glandularhairs restricted to paraphyses characterize subg. Filicipecten.

4.1.4. Phylogenetic relationships within the tropical Asian cladeA novel relationship revealed here is support of a large clade of

species with radially symmetrical rhizomes comprising c. 140 spp.(Parris, unpublished) belonging to six genera: Dasygrammitis,

Scleroglossum, Tomophyllum, Micropolypodium, Xiphopterella, andCalymmodon (Fig. 1). Parris (2007) argued that Dasygrammitiswas closely allied with Tomophyllum, Scleroglossum, and Calymm-odon based in part upon this character, and she indicated thatXiphopterella was also likely to be related. Parris (2007) also notedthe morphological similarity between Micropolypodium and Xiph-opterella in some characters, including radial rhizome. Radiallysymmetrical rhizomes occur frequently among grammitids(Sundue et al., 2010), particularly in Alansmia, Leucotrichum, and

Fig. 2 (continued)

202 M.A. Sundue et al. / Molecular Phylogenetics and Evolution 81 (2014) 195–206

Radiogrammitis, but the clade of six tropical Asian genera resolvedhere is by far the largest group of taxa sharing this morphology.Polypods, in contrast, exhibit only dorsiventral rhizomes. Thus,

across the family, the change from dorsiventral to radially sym-metrical rhizomes is likely to be a derived trait that arose multipletimes within the grammitid clade. We speculate that this change is

M.A. Sundue et al. / Molecular Phylogenetics and Evolution 81 (2014) 195–206 203

coincident with the reduction in size and with higher niche spe-cialization of grammitids compared to polypods.

4.1.5. Oreogrammitis, Radiogrammitis, and ThemeliumCopeland (1917) described Oreogrammitis to accommodate a

single species from Mt. Kinabalu, Borneo, and cited its confluentsori as unique. Christensen and Holttum (1934) maintained thegenus, but noted that its validity was problematical. Parris (1983,1990) initially sunk the genus into Grammitis because confluentsori occur in other unrelated grammitids, but later resurrected it,making combinations for 107 species when it became apparentthat a broadly circumscribed Grammitis was untenable (Parris,2007). The closely related Radiogrammitis was simultaneouslydescribed and distinguished by its radially symmetrical rhizomesthat sometimes lack scales. By contrast, the rhizomes of Oreo-grammitis are uniformly dorsiventral and scaly. Our results findthat Oreogrammitis is polyphyletic and nested in three placeswithin Radiogrammitis (Fig. 1). In some ways, this result is not sur-prising since the two genera were distinguished using essentially asingle character – rhizome symmetry. It is a bit unexpected; how-ever, because rhizome symmetry effectively circumscribes manygenera of grammitid ferns (Ranker et al., 2004; Sundue et al.,2010). The number of changes in rhizome symmetry within thisclade suggests that there may be an increased rate of change forthis character relative to other grammitid lineages. Further sam-pling is needed to address that issue, however. There are 153 spe-cies of Oreogrammitis and 36 species of Radiogrammitis (Parris,unpubl.), thus the apparently large number of transitions seen inour results may be an artifact of relatively low sampling of bothOreogrammitis and Radiogrammitis.

Relationships within Oreogrammitis are further complicated byThemelium (27 species; Parris, unpubl.), which nests within it(Fig. 1). Themelium is similar to Oreogrammitis in its dorsiventralrhizome with glabrous rhizome scales, but differs in having gla-brous sporangia and usually subclathrate to clathrate rhizomescales. Radiogrammitis has small leaves that are uniformly simpleand entire (or repand) and those of Oreogrammitis are usually alsosimple and entire (or repand, or even pinnate in a few species).Some species of Themelium also have simple laminae; however,most are pinnate or twice-pinnate, sometimes with rigid sclerifiedaxes and reduced laminar tissue (Parris 1997, 2004, 2010). Thistransition from simple to compound leaves is surprising, but notunique. Laminar blade dissection among the 10 species of Adeno-phorus alone ranges from simple to tripinnate (Ranker et al.,2003), in Ceradenia it ranges from simple to bipinnate (Bishop,1988), and in Notogrammitis it ranges from simple to bipinnatifid(Perrie and Parris, 2012). These transitions occur outside of thegrammitids as well, one example being that of Elaphoglossum sect.Squamipedia Mickel & Atehortúa (Vasco et al., 2013) in comparisonto most other Elaphoglossum spp. that have simple, entire leaves.

This is among the largest and least well sampled clades of gram-mitids. With an estimated 216 species, our sampling representsonly 10% of the diversity. Our results suggest that generic recir-cumscription is necessary within this clade, but denser sampling,including the types of all three genera, is warranted before makingtaxonomic changes. If a single genus is to be retained, Oreogramm-itis has nomenclatural priority, but would require alteration of itsmorphological definition.

4.1.6. Orphan speciesNotwithstanding the problems involving the circumscription of

Oreogrammitis, Radiogrammitis, and Themelium, current genericconcepts accommodate all but four species included in our analy-ses. Three of these species reside in a clade sister to Notogrammitisand could potentially be accommodated within an expanded con-cept of that genus. That would require, however, alteration of the

morphological definition of the genus. Two of these species,Grammitis deplanchei and G. pseudaustralis, are endemic to NewCaledonia. The third, G. diminuta, is endemic to Lord Howe Island.Similar phylogenetic relationships were found by Perrie and Parris(2012), who argued to exclude these species from Notogrammitisbecause they depart morphologically and because they lacked suf-ficient support values for those nodes to justify their inclusion inthat genus. Grammitis stenophylla Parris is the fourth unplacedtaxon. In our analyses, it forms a clade with Ctenopterella denticu-lata and the three accessions of Acrosorus friderici-et-pauli. Gramm-itis stenophylla is an Australian endemic that was previouslythought to be related to species now included in Notogrammitis(Parris, 1998b). Perrie and Parris (2012) concluded that it wasunrelated to Notogrammitis, but did not have sufficient samplingto resolve its phylogenetic placement. Its position is well resolvedhere as sister to Acrosorus and Ctenopterella (Fig. 1), from which itdeparts strongly morphologically, and which themselves have littlein common with each other. Long branches lead to each of thesegenera, and this may indicate poor sampling in this clade. Acrosoruscomprises nine species (Parris, unpubl.) and Ctenopterella has 20species (Parris, unpubl.).

4.2. Historical biogeography

4.2.1. General patternsBased on the result of our divergence estimates and ancestral

area reconstructions, we hypothesize that the grammitid lineageevolved in the Neotropics toward the end of the Eocene (37–44.9 Ma) close to the Oligocene boundary (Fig. 2a), when the Ant-arctic ice sheet began to expand rapidly. Subsequent lineagesremained Neotropical through the end of the Oligocene. Frequentmigration to Africa–Madagascar, and infrequent migration to theHawaiian Islands and eastern Pacific occur among these lineages,beginning in the Miocene. Toward the end of the Oligocene, a sin-gle lineage migrated to tropical Asia. Currently, this lineage com-prises well over half of grammitid diversity. Like the Neotropicallineages, migration of the tropical Asian lineage continued duringthe Miocene, but these migrations were to the Hawaiian Islandsand Australasian regions. Further migrations from this lineagedid not occur to Africa–Madagascar. Our evidence does not supportCopeland’s (1939) hypothesis of an Antarctic migration of Gramm-itis s.l. to each of the austral continents. Instead, we reconstruct anaustral lineage as being derived from tropical Asian ancestors.

4.2.2. Origin of the Afro-Malagasy grammitid floraWe infer that migrations to Africa and Madagascar occurred at

least six times in our analyses, all within the last 14 Ma (�21 Ma)including the genera Alansmia, Cochlidium, Grammitis s.s., Melpom-ene, Stenogrammitis, and Zygophlebia (Fig. 2a). Considering theresults of other studies and of species not sampled here, we expectthat the total number of long-distance migrations from the Neotrop-ics to Africa–Madagascar will be considerably higher and willinclude species of Ceradenia, Enterosora, and Leucotrichum, as wellas other species of Grammitis s.s., Stenogrammitis, and Zygophlebia(Ranker et al., 2010; Labiak et al., 2010a; Rouhan et al., 2012).Because of their recent age, these are best explained by long-distance dispersal. Thus, as hypothesized by Moran and Smith(2001), the entire African-Malagasy grammitid flora appears to bethe product of recent migration from the Neotropics, based on cur-rent sampling. Two of the migrations to Africa and/or Madagascarhave led to radiations of species, including Stenogrammitis with 14species in the Neotropics and 13 species in the African-Malagasyregion (Parris, unpublished), and Zygophlebia with seven species inthe Neotropics (Bishop, 1989) and nine species in Africa–Madagas-car (Bishop, 1989; Parris, 2003; Rakotondrainibe and Deroin, 2006).The other large genera in the region, Ceradenia and Grammitis s.s.,

204 M.A. Sundue et al. / Molecular Phylogenetics and Evolution 81 (2014) 195–206

are each represented by eight species, respectively (Bishop, 1988;Parris, 2003). There is currently no phylogenetic evidence that spe-cies have migrated from tropical Asia to Africa–Madagascar or viceversa. However, our sampling does not include any of the six speciesof Ctenopterella from Africa, Madagascar, the Comoros or the Masca-renes (Parris, 2007, 2012; Parris, unpubl.). If Ctenopterella is mono-phyletic, these species would have been derived from tropicalAsian ancestors; however, monophyly of Ctenopterella remains tobe tested.

Over half of the diversity in Zygophlebia resides in Madagascar,and our study is the first to sample taxa from that region (two spp.,three accessions). Their relationships, resolved here as a grade sis-ter to the remainder of the clade, could be interpreted as evidencefor migration to Africa–Madagascar followed by a subsequentmigration back to the Neotropics (Fig. 2a), a result not otherwisesupported by our results. This result could also be derived fromtwo subsequent dispersal events. The latter explanation is consis-tent with what appears to be the predominant direction of migra-tion via long-distance dispersal of spores from West to East in ourresults.

Africa has also received migrants from the austral region. SouthAfrica is home to Notogrammitis angustifolia, which, along with N.crassior on islands to the SW and SE of Africa, is part of a southernmigration out of tropical Asia. Our results, and those of Perrie andParris (2012), suggest that N. angustifolia and N. crassior evolvedfrom ancestors residing in Australia or New Zealand, and not theAmericas.

4.2.3. Origin of the tropical Asian and temperate austral grammitidfloras

The tropical Asian grammitid flora is best explained by a singlemigration from the Neotropics to tropical Asia, an event estimatedto have occurred 24.5 Ma (22.4–27.3 Ma) (Fig. 2b). Because of itsrecent age, this distribution pattern is likely the result of long-distance dispersal. The historical biogeography of grammitid fernsin tropical Asia is quite different than that in Hawai’i and Africa–Madagascar, which are both explained by multiple long-distancedispersal events. Tropical Asia is home to the greatest diversityof grammitid species, with c. 490 species (Parris, unpubl.), and asingle origin of this diversity is a remarkable finding. Grammitidsof tropical Asian origin have migrated to the Hawaiian Islandsand other islands of the central and eastern Pacific, but there isno indication that they have migrated to Africa–Madagascar, orback to the Neotropics. One important migration out of tropicalAsia is that of Notogrammitis (12 spp.), which dominates the gram-mitid flora in the southern hemisphere south of 35�300S (Perrie andParris, 2012). Notogrammitis is estimated to have entered australregions 14.7 Ma (21.0–8.6 Ma) (Fig. 2b). Range expansion waslikely through Zealandia rather than migration across theAustralian plate via the Sahul Shelf, because the sister group toNotogrammitis comprises species currently distributed on New Cal-edonia and Lord Howe Island. Notogrammitis has also migratedaround the southern hemisphere to South Africa and southernSouth America and oceanic islands, including Tristan da Cunha,Gough, Amsterdam, Marion, Crozet, Kerguelen, Falklands, andSouth Georgia Islands. It is the only Paleotropical lineage that hasre-established itself in the Americas. Long-distance dispersal bestexplains this range expansion, which according to our estimateshas occurred during the last 1.5 Ma.

4.2.4. Origin of the Hawaiian grammitid floraThe Hawaiian grammitid flora is also explained by multiple

long-distance migrations (Fig. 2a, b). Unlike Africa–Madagascarand the Austral region whose floras are the product of migrationsfrom single regions, the Hawaiian grammitid flora is the productof migrations from two regions, the Neotropics (Adenophorus and

Stenogrammitis) and tropical Asia (Oreogrammitis). These resultsagree with the results of Geiger et al. (2007). Two of these migra-tions have resulted in local radiations resulting in three species ofOreogrammitis (only two sampled here), and 10 species of Adeno-phorus. The migration of Oreogrammitis to the Hawaiian Islands ispart of a larger pattern of tropical Asian species moving into thecentral and eastern Pacific to Rarotonga (Radiogrammitis cheesema-nii and the Marquesas Islands (O. uapensis). A species from theMoluccas and New Guinea (R. parva) also resides within this cladein our BEAST analysis (Fig. 2), complicating the larger pattern. Weinterpret our results as evidence for multiple migrations from trop-ical Asia to the eastern Pacific, rather than a reversal in direction,and this result is supported by our Bayesian analysis that supportsO. uapensis further outside of this clade (Fig. 1). More sampling isneeded to clarify the pattern. Nonetheless, there is no indicationthat grammitids have migrated from the Hawaiian Islands to otherregions such as the Neotropics.

Our estimated stem age of the Adenophorus clade of 22.5 Ma(29.0–14.1) is effectively identical to what Schneider et al. (2005)estimated for the Hawaiian endemic diellia clade of Asplenium L.of 24.3 Ma (27.0–21.5). Both of these ages effectively coincide withthe estimate of the renewal of Hawaiian terrestrial life at c. 23 Ma(Clague, 1996; Price and Clague, 2002) after a 10 Ma lull in newisland production by the Hawaiian volcanic hotspot. This suggeststhat these two fern lineages were among the first to colonize suc-cessfully the newly emerging islands in the early Miocene. As withthe diellia lineage of Asplenium, the Adenophorus lineage is consid-erably older than any of the current, high Hawaiian Islands (i.e.,Kaua‘i is the oldest at c. 5.2 Ma) and extant or ancestral specieshave arrived at the current islands by dispersing along the islandchain as new islands have been produced. Interestingly, the Adeno-phorus clade at 22.5 Ma and only 10 species is nearly the same ageas the large, species-rich Paleotropical clade at 24.5 Ma and c. 500species (Parris, unpubl.).

Acknowledgments

We thank the Curators of the following herbaria for makingfacilities available for examination of material and/or loans: A,AK, B, BISH, BM, BO, BR, BRI, CANB, CGE, CHR, E, FI, GH, K, KEP,KLU, L, LAE, M, MEL, MO, NHT, NSW, NY, P, PAP, PDA, PE, PTBG,S, SAN, SAR, SING, TAIF, TI, UC, US, VT, WAG, WELT, and EwenCameron (AK) for organizing loans to Barbara Parris. We thankthe University of Oslo Lifeportal for their computational resourcesand the Willi Hennig Society for freely providing TNT. This materialis based upon work supported by the National Science Foundationof the United States under Grant No. DEB-1119695 to T.A. Rankerand C.W. Morden. Michael Sundue was also supported by theH.M. Burkill Fellowship provided by the Singapore Botanic Garden.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.ympev.2014.08.017.

References

Barrington, D.S., 1993. Ecological and historical factors in fern biogeography. J.Biogeogr. 20, 275–279.

Bishop, L.E., 1977. The American species of Grammitis sect. Grammitis. Am. Fern J. 67,101–106.

Bishop, L.E., 1978. Revision of the genus Cochlidium (Grammitidaceae). Am. Fern J.68, 76–94.

Bishop, L.E., 1988. Ceradenia, a new genus of Grammitidaceae. Am. Fern J. 78, 1–5.Bishop, L.E., 1989. Zygophlebia, a new genus of Grammitidaceae. Am. Fern J. 79, 103–

118.

M.A. Sundue et al. / Molecular Phylogenetics and Evolution 81 (2014) 195–206 205

Bishop, L.E., Smith, A.R., 1992. Revision of the fern genus Enterosora(Grammitidaceae) in the New World. Syst. Bot. 17, 345–362.

Christenhusz, M.J.M., Chase, M.W., 2014. Trends and concepts in fern classification.Ann. Bot. 1–24. http://dx.doi.org/10.1093/aob/mct299.

Christensen, C., Holttum, R.E., 1934. The ferns of Mount Kinabalu. Gardens’ Bull.Straits Settlements 7, 191–324.

Clague, D.A., 1996. The growth and subsidence of the Hawaiian–Emperor volcanicchain. In: Keast, A., Miller, S.E. (Eds.), The Origin and Evolution of Pacific IslandBiotas, New Guinea to eastern Polynesia: Patterns and Processes. SPB AcademicPublishing, Amsterdam, pp. 35–50.

Copeland, E.B., 1917. New species and a new genus of Borneo ferns, chiefly from theKinabalu collections of Mrs. Clemens and Mr. Topping. Philipp. J. Sci. 12, 47–65.

Copeland, E.B., 1939. Fern evolution in Antarctica. Philipp. J. Sci. 70, 157–188.Copeland, E.B., 1952. [1951] Grammitis. Philipp. J. Sci. 80, 93–276.Drummond, A.J., Rambaut, A., 2007. BEAST: Bayesian evolutionary analysis by

sampling trees. BMC Evol. Biol. 7, 214–222.Drummond, A.J., Ho, S.Y.W., Phillips, M.J., Rambaut, A., 2006. Relaxed phylogenies

and dating with confidence. PLoS Biol. 4, E88.Drummond, A.J., Rambaut, A., Suchard, M., 2013. BEAST V1.8.0 2002–2013. Bayesian

evolutionary analysis sampling trees. Available at <http://Beast.Bio.Ed.Ac.Uk/>.Geiger, J.M., Ranker, T.A., Ramp Neale, J.M., Klimas, S.T., 2007. Molecular

biogeography and origins of the Hawaiian fern flora. Brittonia 59, 142–158.Gentry, A.H., Dodson, C.H., 1987. Diversity and biogeography of neotropical vascular

epiphytes. Ann. Missouri Bot. Gard. 74, 205–233.Goloboff, P.A., Farris, J.S., Nixon, K.C., 2008. TNT, a free program for phylogenetic

analysis. Cladistics 24, 774–786.Gómez, L.D., 1982. Grammitis succinea, the first New World fern found in amber.

Am. Fern J. 72, 49–52.Hirai, R.Y., Rouhan, G., Labiak, P.H., Ranker, T.A., Prado, J., 2011. Moranopteris: a new

Neotropical genus of grammitid ferns (Polypodiaceae) segregated from AsianMicropolypodium. Taxon 60, 1123–1137.

Huelsenbeck, J.P., Ronquist, F., 2001. MRBAYES: Bayesian inference of phylogeny.Bioinformatics 17, 754–755.

Katoh, S., 2013. MAFFT multiple sequence alignment software version 7:improvements in performance and usability. Mol. Biol. Evol. 30, 772–780.

Kessler, M., Parris, B.S., Kessler, E., 2001. A comparison of the tropical montanepteridophyte floras of Mount Kinabalu, Borneo and Parque Nacional Carrasco,Bolivia. J. Biogeogr. 28, 611–622.

Kessler, M., Moguel-Velazquez, A.L., Sundue, M.A., Labiak, P.H., 2011. Alansmia, anew genus of grammitid ferns (Polypodiaceae) segregated from Terpsichore.Brittonia 63, 233–244.

Korall, P., Pryer, K.M., 2014. Global biogeography of scaly tree ferns (Cyatheaceae):evidence for Gondwanan vicariance and limited transoceanic dispersal. J.Biogeogr. 41, 402–413.

Kreier, H.P., Zhang, X.C., Muth, H., Schneider, H., 2008. The microsoroid ferns:inferring the relationships of a highly diverse lineage of Paleotropical epiphyticferns (Polypodiaceae, Polypodiopsida). Mol. Phylogen. Evol. 48, 1155–1167.

Labiak, P.H., 2011. Stenogrammitis, a new genus of grammitid ferns segregated fromLellingeria (Polypodiaceae). Brittonia 63, 139–149.

Labiak, P.H., Rouhan, G., Sundue, M.A., 2010a. Phylogeny and taxonomy ofLeucotrichum (Polypodiaceae): a new genus of grammitid ferns from theNeotropics. Taxon 59, 911–921.

Labiak, P.H., Sundue, M.A., Rouhan, G., 2010b. Molecular phylogeny, characterevolution, and biogeography of the grammitid fern genus Lellingeria(Polypodiaceae). Am. J. Bot. 97, 1354–1364.

Labiak, P.H., Sundue, M., Rouhan, G., Hanks, J.G., Mickel, J.T., Moran, R.C., 2014.Phylogeny and historical biogeography of the lastreopsid ferns(Dryopteridaceae). Am. J. Bot. 101, 1207–1228.

Lanfear, R., Calcott, B., Ho, S.Y.W., Guindon, S., 2012. PartitionFinder: combinedselection of partitioning schemes and substitution models for phylogeneticanalyses. Mol. Biol. Evol. 29, 1695–1701.

Miller, M.A., Pfeiffer, W., Schwartz, T., 2010. Creating the CIPRES Science Gatewayfor inference of large phylogenetic trees. In: Proceedings of the GatewayComputing Environments Workshop (GCE), 14 November, New Orleans, LA,2010, pp 1–8.

Moran, R.C., Smith, A.R., 2001. Phytogeographic relationships between Neotropicaland African–Madagascan pteridophytes. Brittonia 53, 304–351.

Nixon, K.C., 1999. The parsimony ratchet, a new method for rapid parsimonyanalysis. Cladistics 15, 407–414.

Nylander, J.A., Wilgenbusch, J.C., Warren, D.L., Swofford, D.L., 2008. AWTY (are wethere yet?): a system for graphical exploration of MCMC convergence inBayesian phylogenetics. Bioinformatics 24, 581–583.

Parris, B.S., 1977. A naturally occurring intergeneric hybrid in Grammitidaceae(Filicales): Ctenopteris heterophylla � Grammitis billardierei. N.Z. J. Bot. 15, 597–599.

Parris, B.S., 1983. A taxonomic revision of the genus Grammitis Sw.(Grammitidaceae: Filicales) in New Guinea. Blumea 29, 13–222.

Parris, B.S., 1984. Another intergeneric hybrid in Grammitidaceae: Ctenopterislongiceps � Grammitis sumatrana. Fern Gaz. 12, 337–340.

Parris, B.S., 1990. Grammitidaceae. In: Kramer, K.U., Green, P.S. (Eds.), The Familiesand Genera of Vascular Plants, Pteridophytes and Gymnosperms, vol. 1.Springer, Berlin, pp. 153–157.

Parris, B.S., 1997. Themelium, a new genus of Grammitidaceae (Filicales). Kew. Bull.52, 737–741.

Parris, B.S., 1998a. Chrysogrammitis, a new genus of Grammitidaceae (Filicales).Kew. Bull. 53, 909–918.

Parris, B.S., 1998b. Grammitidaceae. Flora Aust. 48, 450–468.Parris, B.S., 2003. The distribution of Grammitidaceae (Filicales) inside and outside

Malesia. Telopea 10, 451–466.Parris, B.S., 2004. New combinations in Acrosorus, Lellingeria, Prosaptia and

Themelium. Kew Bull. 59, 223–225.Parris, B.S., 2005. Grammitidaceae (Pteridophyta) of Mount Jaya, New Guinea, and

other montane Malesian localities. Fern Gaz. 17, 183–2003.Parris, B.S., 2007. Five new genera and three new species of Grammitidaceae

(Filicales) and the re-establishment of Oreogrammitis. Gard. Bull. Singapore 58,233–274.

Parris, B.S., 2010. New combinations in Oreogrammitis, Prosaptia, Radiogrammitis,Themelium and Tomophyllum (Grammitidaceae: Polypodiopsida: Monilophyta)from Malesia and the Pacific Islands. Kew. Bull. 65, 123–125.

Parris, B.S., 2012. Ctenopterella gabonensis, a new species of grammitid fern(Polypodiaceae) from Gabon, Africa. Fern Gaz. 19, 135–138.

Parris, B.S., 2013. Archigrammitis, a new genus of grammitid fern (Polypodiaceae)from Malesia and Polynesia. Fern Gaz. 19, 135–138.

Parris, B.S., Given, D.R., 1976. A taxonomic revision of the genus Grammitis Sw.(Grammitidaceae: Filicales) in New Zealand. N.Z. J. Bot. 14, 85–111.

Parris, B.S., Beaman, R.S., Beaman, J.H., 1992. The Plants of Mount Kinabalu 1. Fernsand Fern Allies. Royal Botanic Gardens, Kew.

Perrie, L.R., Parris, B.S., 2012. Chloroplast DNA sequences indicate the grammitidferns (Polypodiaceae) in New Zealand belong to a single clade, Notogrammitisgen. nov. N.Z. J. Bot. 50, 457–472.

Price, M.G., 1982. The ferns of Steere and Harrington. Contr. Univ. Michigan Herb.15, 197–204.

Price, M.G., 1987. Sabah ferns collected by John H. Beaman. Contr. Univ. MichiganHerb. 16, 189–200.

Price, J.P., Clague, D.A., 2002. How old is the Hawaiian biota? Geology andphylogeny suggest recent divergences. Proc. R. Soc. London, Ser. B: Biol. Sci. 269,2429–2435.

Rakotondrainibe, F., Deroin, T., 2006. Comparative morphology and rhizomeanatomy of two new species of Zygophlebia (Grammitidaceae) fromMadagascar and notes on the generic circumscription of Zygophlebia andCeradenia. Taxon 55, 145–152.

Rambaut, A., Suchard, M.A., Xie, D., Drummond, A.J., 2013. Tracer v 1.5. Availablefrom <http://beast.bio.ed.ac.uk/Tracer>.

Ranker, T.A., Geiger, J.M.O., Kennedy, S.C., Smith, A.R., Haufler, C.H., Parris, B.S., 2003.Molecular phylogenetics and evolution of the endemic Hawaiian genusAdenophorus (Grammitidaceae). Mol. Phylogen. Evol. 26, 337–347.

Ranker, T.A., Smith, A.R., Parris, B.S., Geiger, J.M.O., Haufler, C.H., Straub, S.C.K.,Schneider, H., 2004. Phylogeny and evolution of grammitid ferns(Grammitidaceae): a case of rampant morphological homoplasy. Taxon 53,415–428.

Ranker, T.A., Sundue, M.A., Labiak, P., Parris, B.S. Rouhan, G., 2010. New insights intothe phylogeny and historical biogeography of the Lellingeria myosuroides clade(Polypodiaceae). PLOS Currents Tree of Life, November 18. Edition 1.Doi:10.1371/currents.RRN1197.

Ree, R.H., Smith, S.A., 2008. Maximum likelihood inference of geographic rangeevolution by dispersal, local extinction, and cladogenesis. Syst. Biol. 57, 4–14.

Ronquist, F., Huelsenbeck, J.P., 2003. MrBayes 3: Bayesian phylogenetic inferenceunder mixed models. Bioinformatics 19, 1572–1574.

Rouhan, G., Labiak, P.H., Randrianjohany, E., Rakotondrainibe, F., 2012. Not soNeotropical after all: the grammitid fern genus Leucotrichum (Polypodiaceae) isalso Paleotropical, as revealed by a new species from Madagascar. Syst. Bot. 37,331–338.

Schneider, H., Smith, A.R., Cranfill, R., Hildebrand, T.J., Haufler, C.H., Ranker, T.A.,2004. Unraveling the phylogeny of polygrammoid ferns (Polypodiaceae andGrammitidaceae): exploring aspects of the diversification of epiphytic plants.Mol. Phylogen. Evol. 31, 1041–1063.

Schneider, H., Ranker, T.R., Russell, S.J., Cranfill, R., Geiger, J.M.O., Aguraiuja, R.,Wood, K.R., Grundmann, M., Kloberdanz, K., Vogel, J.C., 2005. Origin anddiversification of the Hawaiian fern genus Diellia Brack. (Aspleniaceae,Polypodiidae). Proc. R. Soc. London, Ser. B: Biol. Sci. 272, 455–460.

Schuettpelz, E., Pryer, K.M., 2009. Evidence for a Cenozoic radiation of ferns in anangiosperm-dominated canopy. Proc. Natl. Acad. Sci. USA 106, 11200–11205.

Smith, A.R., 1993. Tersichore, a new genus of Grammitidaceae (Pteridophyta). Novon3, 478–489.

Smith, A.R., Moran, R.C., Bishop, L.E., 1991. Lellingeria, a new genus ofGrammitidaceae. Am. Fern J., 76–88.

Smith, A.R., Pryer, K.M., Schuettpelz, E., Korall, P., Schneider, H., Wolf, P.G., 2006. Aclassification for extant ferns. Taxon 55, 705–731.

Smith, A.R., Pryer, K.M., Schuettpelz, E., Korall, P., Schneider, H., Wolf, P.G., 2008.Fern classification. In: Ranker, T.A, Haufler, C.H. (Eds.), Biology and Evolution ofFerns and Lycophytes. Cambridge University Press, Cambridge, pp. 417–462.

Stamatakis, A., 2006. RAxML-VI-HPC: maximum likelihood-based phylogeneticanalyses with thousands of taxa and mixed models. Bioinformatics 22, 2688–2690.

Sundue, M.A., 2010a. A morphological cladistic analysis of Terpsichore(Polypodiaceae). Syst. Bot. 35, 716–729.

Sundue, M.A., 2010b. A monograph of Ascogrammitis, a new genus of grammitidferns (Polypodiaceae). Brittonia 62, 357–399.

Sundue, M.A., 2013. Mycopteris, a new neotropical genus of grammitid ferns(Polypodiaceae). Brittonia 66, 174–185.

Sundue, M.A., Islam, B., Ranker, T.A., 2010. Systematics of grammitid ferns(Polypodiaceae): using morphology and plastid sequence data to resolve the

206 M.A. Sundue et al. / Molecular Phylogenetics and Evolution 81 (2014) 195–206

circumscription of Melpomene and the polyphyletic genera Lellingeria andTerpsichore. Syst. Bot. 35, 701–715.

Sundue, M.A., Labiak, P.H., Mostacero, J., Smith, A.R., 2012. Galactodenia, a newgenus of grammitid ferns segregated from Terpsichore (Polypodiaceae). Syst.Bot. 37, 339–346.

Sylvester, S.P., Sylvester, M.D.P.V., Kessler, M., 2014. The world’s highest vascularepiphytes found in the Peruvian Andes. Alpine Bot., 1–7.

Tryon, R.M., Tryon, A.F., 1982. Ferns and Allied Plants, with Special Reference toTropical America. Springer-Verlag, New York.

Vasco, A., Mickel, J.T., Moran, R.C., 2013. Taxonomic revision of the Neotropicalspecies of Elaphoglossum sect. Squamipedia (Dryopteridaceae). Ann. MissouriBot. Gard. 99, 244–286.

Wang, L., Wu, Z.Q., Xiang, Q.P., Heinrichs, J., Schneider, H., Zhang, X.-C., 2010. Amolecular phylogeny and a revised classification of tribe Lepisoreae(Polypodiaceae) based on an analysis of four plastid DNA regions. Bot. J. Linn.Soc. 162, 28–38.

Wilgenbush, J.C., Warren, D.L., Swofford, D.L., 2004. AWTY: a system for graphicalexploration of MCMC convergence in Bayesian phylogenetic inference.Available at <http://ceb.csit.fsu.edu/awty>.

Wilson, K.A., 1959. Sporangia of the fern genera allied with Polypodium and Vittaria.Contrib. Gray Herb. 185, 97–127.

Wolf, P.G., Schneider, H., Ranker, T.A., 2001. Geographic distributions ofhomosporous ferns: does dispersal obscure evidence of vicariance? J.Biogeogr. 28, 263–270.