Comparison of molecular and morphological data on St Helena: Elaphoglossum

Comparison of molecular and morphological data

on St Helena: Elaphoglossum

A. Eastwood1,2, Q. C. B. Cronk

3, J. C. Vogel

4, A. Hemp

5, and M. Gibby

1

1Royal Botanic Garden Edinburgh, Edinburgh, UK2Institute of Cell and Molecular Biology, University of Edinburgh, Edinburgh, UK3Botanical Garden and Centre for Plant Research, University of British Columbia,Vancouver, British Columbia, Canada4Botany Department, Natural History Museum, Cromwell Road, London, UK5Department of Plant Systematics, University of Bayreuth, Bayreuth, Germany

Received March 27, 2003; accepted November 20, 2003Published online: February 23, 2004� Springer-Verlag 2004

Abstract. The endemic elaphoglossoid ferns,Elaphoglossum dimorphum, E. nervosum andMicrostaphyla furcata of St Helena, form aclosely related group within section Lepidoglossawhen analysed phylogenetically using sequencesfrom the chloroplast trnL intron (partial) andtrnL-F intergenic spacer. Microstaphyla furcata,traditionally placed in its own genus, is clearlyshown to belong to Elaphoglossum confirming theprevious transfer of this species to Elaphoglossumas E. bifurcatum. There is hardly any trnL-Fsequence divergence between the species, in factsequences of E. nervosum and E. dimorphum areidentical. These results are consistent with thepossible origin of E. dimorphum as a hybridbetween E. bifurcatum and E. nervosum or withthe view that the three species are the result of arecent radiation. The potential conflict betweenphylogenetic and morphological distinctness indetermining species conservation priorities is dis-cussed.

Key words: Elaphoglossum,Microstaphyla, St Helena,phylogeny, morphological divergence, conservationpriorities, trnL intron, trnL-F spacer, molecularsystematics, Lomariopsidaceae, ferns.

St Helena is a small, sub-tropical oceanic island(15�56¢S, 5�42¢W) in the South Atlantic Ocean.This isolated volcanic island has a remarkableflora with 49 endemic plant species, 13 of whichare ferns (Cronk 2000). Of the 13 endemic fernspecies, 11 merit inclusion in the IUCN RedList of Threatened Plants (Walter and Gillett1998). The main threat to these endemic ferns isthe encroachment of their habitat from invasiveexotic species such as Phormium tenax Forst.(New Zealand Flax). Three of the endemicferns are in the family Lomariopsidaceae andinclude two species of Elaphoglossum Schott exJ.Sm., E. dimorphum, E. nervosum and therelated Microstaphyla furcata.

The relationship of Microstaphyla C. Preslto Elaphoglossum, in particular to E. dimor-phum andE. nervosum from StHelena, has longbeen debated (Mickel 1980). The genus Elaph-oglossum, which contains over 500 species, isremarkably uniform with regards to morpho-logical characters. The majority of species havea characteristic simple blade, free veins andacrostichoid sori (Mickel and Atehortua 1980).

Plant Syst. Evol. 245: 93–106 (2004)DOI 10.1007/s00606-003-0116-9

Microstaphyla furcata, however, has a dis-tinctly pinnate frond, a distinction, which, inthe morphologically uniform genus Elapho-glossum, has warranted separate generic statusby some taxonomists (Maxon 1923; Copeland1947; Pichi Sermolli 1968, 1977). The originaldescription of the genus Microstaphyla wasin fact based on this pinnate species fromSt Helena (Presl 1849). Microstaphyla latercame to include another two species both fromthe Andes, M. moorei (Britton) Underw.(Underwood 1905) and M. columbiana Maxon(Maxon 1923), but all these were later trans-ferred into Peltapteris (Gomez 1975) and sub-sequently into Elaphoglossum (Mickel 1980).

The blade of E. dimorphum has a toothedmargin. In fact, Fee (1852, 1857) proposed thatE. dimorphum and Microstaphyla furcata werethe same species, having found intermediatesbetween the two. However, despite the factthat the species occur sympatrically at somelocalities we have not found evidence of anyintermediate forms on St Helena, and, likeHooker (1861), reject this hypothesis. Elapho-glossum nervosum has an entire blade withprominent veins. These veins fork near themargin, and then unite laterally to form acommissural vein. In his discussion on therelationships of dissected elaphoglossoid fernsMickel (1980) concluded, based on morpho-logical characters such as rhizome scales andhabit, that E. dimorphum, E. nervosum andM. furcata are closely related, arguing thatfrond architecture is a weak taxonomic char-acter at the generic level. Mickel (1980)suggested that E. dimorphum arose as a lacer-ated form of E. nervosum, the more extremedissected form being M. furcata, but did notdismiss the possibility of E. dimorphum being ahybrid of E. nervosum and M. furcata. He alsosuggested they could be forms of the samespecies. Like Christ (1899) and Christensen(1906), Mickel (1980) treated Microstaphylaunder Elaphoglossum, as E. bifurcatum (Jacq.)Mickel. In contrast, Cronk (2000) in hisendemic Flora of St Helena, maintained thegeneric distinction. For a full synonymy ofM. furcata (E. bifurcatum) see Mickel (1980).

Unfortunately, these interesting ferns,which have stimulated much interest fromtaxonomists and evolutionary biologists, areseverely threatened on St Helena. Elaphoglos-sum dimorphum is restricted to two localitieswithin Diana’s Peak National Park, MountActaeon and Cuckold’s Point. The nationalpark represents the largest area of tree fernthicket remaining on St Helena, coveringapproximately 1 km2. Occasional plants havealso been noted further west on the centralridge, at High Peak and the Depot (Cronk2000), but the species has not been seenat these localities recently. Elaphoglossumdimorphum is a terrestrial fern, found growingon stone steps and surrounding rocks andbanks. Because the total population ofE. dimorphum is less than 50 individuals, it isclassified as ‘‘Critically Endangered’’ (CR, D)according to the IUCN Categories of Threat(IUCN 1994).

Elaphoglossum nervosum, like E. dimor-phum, is also restricted to Diana’s PeakNational Park, growing locally at Cuckold’sPoint, Diana’s Peak and Mount Actaeon. Thespecies is predominantly epiphytic, growing onthe endemic tree fern, Dicksonia arborescensL’Her. (Cyatheaceae), and the endemic tree,Melanodendron integrifolium (Roxb.) DC. (As-teraceae). It is also occasionally found on rockfaces and mossy banks. The total populationof E. nervosum does not exceed 100 plants,and so the species is considered ‘‘Endangered’’(E, D).

Microstaphyla furcata has a wider distribu-tion on St Helena than either E. dimorphum orE. nervosum. There are least ten populations ofM. furcata on St Helena, all in the uplands ataltitudes greater than 650 m. The species isterrestrial, growing on shaded rocks, rockcrevices and earth banks. Although not underany immediate threat, the species occupies lessthan 100 km2 and would therefore be classifiedas ‘‘Vulnerable’’ (V, D) (IUCN 1994).

The use of morphology in reconstructingphylogenies of ferns is often complicated bythe lack of phylogenetically informative char-acters (Haufler andRanker 1995). For example,

94 A. Eastwood et al.: Evolution of endemic elaphoglossoid ferns on St Helena

taxonomic complexity in the cheilanthoid fernsis attributed to morphological homoplasyby convergent evolution to xeric habitats(Gastony and Rollo 1995). The lack of infor-mative characters coupled with disagreementby authors in character interpretation has ledto taxonomic controversy in a number of ferngroups (Crane et al. 1995, Hauk 1995). Inresponse to the lack of informative charactersfern systematists have started to search for newsources of characters in molecular data,including restriction site and nucleotidesequence data, to infer phylogenetic relation-ships. The use of molecular data to inferphylogeny has yielded valuable insights intothe relationships and evolution of ferns, somewith taxonomic implications (Hasebe et al.1994, Gastony and Ungerer 1997, Murakamiet al. 1999).

In angiosperms, the comparatively slowlyevolving chloroplast rbcL gene, which encodesthe large subunit of ribulose 1,5-biphosphatecarboxylase/oxygenase, has been used primar-ily to infer phylogenies at higher taxonomiclevels, especially at the interfamily or ordinallevel (Soltis and Soltis 1998). Its range ofapplicability in ferns appears to be wider andhas been shown to resolve closely relatedgenera (Gastony and Ungerer 1997) and spe-cies within genera (Haufler and Ranker 1995,Hauk 1995, Murakami et al. 1999). However,in Elaphoglossum, rbcL failed to resolve closelyrelated species (Mickel pers. comm. 2001) as itdid in closely related species of Botrychium(Ophioglossaceae) (Hauk 1995). Nucleotidesequences of the non-coding regions of trnL-F gene have been widely applied to resolvephylogenetic relationships in many groups ofangiosperms at the generic level. Unfortu-nately, the application of these sequences inphylogenetic reconstruction in ferns has beenlimited to only a few studies (Pinter et al. 2002,Ranker et al. 2003), possibly due to limitationsof the ‘‘universal’’ primers of Taberlet et al.(1991) commonly used. Hauk et al. (1996)found that sequence divergence rates in thetrnL-trnF intergenic spacer in Ophioglossaceaewere five times greater than in rbcL.

This study uses phylogenetic analysis ofsequences from two non-coding regions ofchloroplast DNA, the trnL intron (partial) andtrnL-F intergenic spacer to infer relationshipsin the endemic elaphoglossoid ferns fromSt Helena.

Materials and methods

Plant material. The plant material used in thisstudy was all dried in silica gel, apart fromRumohra adiantiformis which was taken from theliving collection at the Royal Botanic GardenEdinburgh (RBGE). The ingroup included thethree endemic elaphoglossoid species fromSt Helena and the native, but not endemic,Elaphoglossum conforme. As a comparison, sixother species of Elaphoglossum from East Africawere included in the ingroup, as well as E. semicy-lindricum from Madeira. The 11 elaphoglossoidspecies in the ingroup represent four out of the ninesections of Elaphoglossum (Lepidoglossa H.Christ,Eximia Mickel & Atehortua, Setosa H.Christ, andElaphoglossum) according to Mickel and Atehortua(1980). Table 1 lists the species used in the phylo-genetic analysis with details of authorities, localityand habitat. Voucher herbarium specimens of thetaxa from St Helena and Rumohra adiantiformis areheld at the Royal Botanic Garden Edinburgh (E).Voucher specimens for the East African species arewith Andreas Hemp at the University of Bayreuth(UBT) in Germany. Rumohra adiantiformis waschosen to represent the outgroup as recent molec-ular and morphological studies have shown it to bea sister taxon to Elaphoglossum (Hasebe et al. 1995,Pryer et al. 1995).

DNA extraction and PCR. Total genomicDNA was extracted from one individual of eachtaxon using the modified CTAB procedure fromDoyle and Doyle (1987). Two non-coding regionsof chloroplast DNA were amplified using thepolymerase chain reaction (PCR): 1) the intron ofthe trnL (UAA) gene, and 2) the intergenic spacerbetween the trnL (UAA) 3¢ exon and trnF(GAA). To amplify these regions the universalprimers d, e and f of Taberlet et al. (1991) wereused. The usual c primer of Taberlet et al. (1991)was replaced with FERN1, 5¢ GGC AGC CCCCAR ATT CAG GGR AACC 3¢, a primerdesigned for Dryopteris at the Natural HistoryMuseum, London. The FERN1 primer lies within

A. Eastwood et al.: Evolution of endemic elaphoglossoid ferns on St Helena 95

the intron of the trnL (UAA) gene, approximately50–80 base pairs downstream from the trnL(UAA) 5¢ exon. Each PCR reaction contained:2.5 ll of x10 NH4 buffer (Bioline, UK), 2.5 ll of2 mM dNTPs (Bioline, UK), 1 ll of 10 lM

forward primer (FERN1 or e), 1 ll of 10 lM ofreverse primer (d or f) , 1 ll of 50 mM MgCl2,0.1 ll of Biotaq polymerase (Bioline, UK) and1–2 ll DNA. The reaction volume was made upto 25 ll with sterile ultra-pure water.

Table 1. List of species with collection, locality, habitat and GenBank accession details used in thephylogenetic analysis

Species Collectingno.

Locality and habitat GenBankaccession

Elaphoglossum conforme(Sw.) ex J. Sm.

AEAST211 St Helena, Diana’s Peak NationalPark, Mt Actaeon: Epiphyte onDicksonia arborescens. 780 m

AY194078

Elaphoglossum dimorphum(Hook. & Grev.) Moore

AEAST302 St Helena, Diana’s Peak NationalPark, Mt Actaeon: On rock facewith M. furcata andE. nervosum. 780 m

AY194068

Elaphoglossum nervosum(Bory) H. Christ

AEAST358 St Helena, Diana’s Peak NationalPark, Mt Actaeon: Epiphyte onDicksonia arborescens. 780 m

AY194069

Microstaphyla furcata(L. f.) Fee

AEAST721 St Helena, Diana’s Peak NationalPark, Cuckold’s Point: On largerock boulder, amongst flax withE. nervosum amd M. furcata. 750 m

AY194070

Elaphoglossum angulatum(Blume) Moore

AHEMP7 Tanzania, Mt Kilimanjaro, betweenWeru-Weru and Lanza river:Epiphyte in montaneOcotea-Podocarpus forest, 2320 m

AY194076

Elaphoglossum aubertii(Desv.) Moore


AY194074

Elaphoglossum deckenii(Kuhn) C. Chr.


AY194072

Elaphoglossum lastii(Baker) C. Chr.

AHEMP23 Tanzania, Mt Kilimanjaro, Marangugate, epiphyte on Agauria salicifoliain Agauria-Macaranga forest,1800 m

AY194077

Elaphoglossum semicylindricum(Bowdich) Benl

JVOGE-ELA-7 Madeira AY194071

Elaphoglossum spathulatum(Bory) Moore

AHEMP1 Tanzania, Mt Kilimanjaro, onboulders in stream bed, 1700 m,Weru-Weru river system

AY194075

Elaphoglossum subcinnamomeum(H. Christ) Hieron.

AHEMP19 Tanzania, Mt Kilimanjaro Maua:semishaded rock in small rivergorge, 2800 m

AY194073

Rumohra adiantiformis(G. Forst.) Ching

19932232 South Africa, cultivated at RBGE. AY194067


PCR was performed with the following condi-tions for all taxa: i) one initial pre-step of 94 �C for4 minutes, followed by ii) 30 cycles of denaturationat 94 �C (45 seconds), annealing at 55 �C (45 sec-onds), and extension at 72 �C (3 minutes) and, iii) aterminal extension of 72 �C for 10 minutes. Aftervisualisation using a 2% agarose gel, the PCRproducts were purified using QIAquick� purifica-tion columns supplied by Qiagen Ltd., UK. Anannealing temperature of 55 �C was required forhigh specificity to avoid multiple bands. However,this temperature resulted in no product for the trnLintron in M. furcata and E. subcinnamomeum. Theannealing temperature was therefore reduced to51 �C in these two species. This lower temperatureresulted in two PCR products, seen as one brightband (�600 bp) and a faint band (�420 bp) whenvisualised on a 2% agarose gel. The remainingPCR product (20 ll) was loaded and visualised ona gel of 1% low melting point agarose (Sigma). Thebright band, corresponding to the trnL intron inthe other Elaphoglossum species, was extractedusing the QIAquick gel extraction kit (QiagenLtd. UK) and sequenced directly.

The PCR products were sequenced directlyusing premixed Thermo Sequenase II reagent(Amersham Pharmacia, UK) according to themanufacturer’s protocol. The sequencing reactionwas conducted using the following PCR conditions:25 cycles of denaturation at 96 �C (10 seconds),annealing at 50 �C (5 seconds) and extension at60 �C (4 minutes). DyeEx� spin columns fromQiagen Ltd. were used according to the recom-mended protocol to remove unincorporated dyefrom the sequence reactions. For confirmation bothforward and reverse sequences were performed foreach taxon. Sequence analysis was conducted on anautomated ABI Prism� 377 sequencer.

Sequence alignment and analysis. Sequenceswere initially edited using Sequence Navigator�v.1.01 (Applied Biosystems) and subsequentlyexported into AutoAssembler� v.2.1 (AppliedBiosystems) for final editing and assembling.Sequences were aligned using ClustalX v.1.8 (Hig-gins et al. 1992) with some minor manual adjust-ment. The first 30 bp downstream from theFERN1 primer were unreadable and were thereforeexcluded from further analysis. Sequence diver-gence between the taxa was calculated withinPAUP* version 4.0b7 (Swofford 2001) whilst thetransition/transversion ratios were calculated using

MacClade version 3.05 (Maddison and Maddison1992). The sequences in this study have beensubmitted to GenBank (see Table 1 for accessionnumbers) and the aligned matrix can be obtainedfrom the correspondence author.

Phylogenetic analysis. A phylogenetic analysiswas performed on the aligned data matrix, usingthe branch-and-bound option in PAUP* version4.0b7 (Swofford 2001). Gaps in the sequence datawere treated as missing data. Insertion/deletionevents (indels) were scored according to the simplegap coding method of Simmons and Ochoterena(2000) and added to a separate gap matrix at theend of the sequence data. Ambiguous regions,which yielded a number of alternative interpreta-tions, were excluded from the analysis. Bootstrapvalues for each clade were calculated from 10,000replicate parsimony analyses using the ‘‘branch-and-bound’’ option and ‘‘furthest’’ addition se-quence of taxa. The decay index for each clade wascalculated using Autodecay 4.0.2 (Eriksson 1999).

Results

Sequence analysis. A complete sequence forthe intergenic spacer and a partial intronsequence of the trnL-F gene was obtained forall 12 taxa in this study. This gave an aligneddata matrix of 960 base pairs. The sequenceand indel characteristics are shown in Table 2.

The sequences of E. dimorphum andE. nervosum are identical. Microstaphyla fur-cata differed in only two autapomorphicindels, one 8bp in length the other 1 bp (the1 bp indel was in the ambiguous region andtherefore excluded from the analysis). Whilethese three species showed no divergence, thedivergence between the other ingroup taxaranges from 1.3 to 7.2%. The pairwisesequence divergences between Elaphoglossumsemicylindricum and E. deckenii, and theSt. Helenian endemics are 2.37% and 1.85%,respectively. Elaphoglossum semicylindricumand E. deckenii are considered to be in thesame section (Lepidoglossa) as the St. Helenianendemics, but different subsections accordingto Mickel and Atehortua (1980). The averagepairwise sequence divergence in the ingroup is4.7%. Both the intron (partial) and spacer


sequences provide a substantial number ofphylogenetically informative sites, 31 and 30respectively. The two chloroplast regions alsoprovide a number of phylogenetically infor-mative indels (see Table 2).

Phylogenetic analysis. The parsimonyanalysis of unambiguously aligned trnL-Fsequences yielded only one most parsimonioustree, represented as a cladogram in Fig. 1. Thetree has a length of 261 steps with a consis-tency index of 0.931 and a retention index of0.885 (including all characters). The threeendemic elaphoglossoid ferns from St Helena,E. nervosum, E. dimorphum and M. furcataform a well-supported monophyletic clade,

with 100% bootstrap support. The two spe-cies, E. semicylindricum from Madeira andE. deckenii from East Africa, are the sistergroup to the St. Helenian endemics. The otherspecies of Elaphoglossum of St Helena, thenative E. conforme, is distantly related toE. nervosum, E. dimorphum and M. furcata.

The parsimony analysis of unambiguouslyaligned trnL-F sequences combined with thegap matrix also yielded one tree of 300 steps(CI¼ 0.923, RI¼ 0.787). The tree is repre-sented as a phylogram in Fig. 2. The topol-ogy of the tree is identical to that of the mostparsimonious tree based on sequences alone(Fig. 1) except for the position of E. lastii. In

Table 2. Sequence and indel characteristics for 10 species of Elaphoglossum (ingroup), Microstaphylafurcata (ingroup) and Rumohra adiantiformis (outgroup) based on trnLUAA partial intron and trnLUAA-trnFGAA intergenic spacer. In addition, sequence divergence (%) is given for i) the endemic St Helena taxa(E. dimorphum, E. nervosum and M. furcata) and ii) between the endemic St Helena taxa and the otherElaphoglossum species in the ingroup

Parameter Partialintron

Intergenicspacer

Partial intron+ spacer

Length range (total) (bp) 474–589 308–321 795–906Length mean (total) (bp) 552.7 315.3 868Length range ingroup (bp) 531–589 311–318 839–906Length mean ingroup (bp) 559.8 314.8 874.6Length outgroup (bp) 474 321 795Aligned length 610 350 960G + C content range (%) 45.1–48.2 48.3–52.0 47.8–48.9G + C content mean (%) 47.2 50.5 48.4Number (%) of excluded sites 53 (9%) 26 (7%) 79 (8%)Number of indels (ingroup) 12 12 24Number of indels (total) 18 16 34Number (%) of informative indels (ingroup) 9 (50%) 6 (37.5%) 15 (44.1%)Size of indels (ingroup) 1–36 1–6 1–36Size of indels (total) 1–55 1–11 1–55Number of sites after exclusion of ambiguous 557 324 881Number (%) of variable sites 129 (23.2%) 95 (29.3%) 224 (25.4%)Number (%) of constant sites 428 (76.8%) 229 (70.7%) 657 (74.6%)Number (%) of uninformative sites 98 (17.6%) 65 (20.1%) 163 (18.5%)Number (%) of informative sites 31 (5.6%) 30 (9.3 %) 61 (6.9%)Transitions (unambiguous) 35 27 66Transversions (unambiguous) 29 17 48Transitions/tranversions 1.21 1.6 1.36Sequence divergence (St Helena) (%) 0 0 0Sequence divergence(St Helena/African Elaphoglossum) (%)

1.7–5.9 1.7–7.3 1.9–6.4


Fig. 2 (sequences and gap matrix) E. lastiiforms a clade with E. angulatum and E. con-forme (80% bootstrap support) with all theother species forming a monophyletic sistergroup. When sequences are analysed alone asin Fig. 1, E. angulatum and E. conforme form

a small clade whilst E. lastii and the otherElaphoglossum species form another cladewith relatively weak (57%) bootstrapsupport.

The sections of Elaphoglossum accordingto Mickel and Atehortua (1980) are indicated

Fig. 1. A cladogram of the single most parsimonious tree based on an analysis of sequences of the trnL intron(partial) and trnL-F intergenic spacer in 10 species of Elaphoglossum and Microstaphyla furcata. The tree has alength of 261 steps; CI¼ 0.931; RI¼ 0.885. Figures above branches are the branch lengths. The first figurebelow branches is the bootstrap value (10,000 replicates), the second figure is the decay index for each clade.S¼ simple leaves, D¼ divided leaves


on Fig. 2. The clade that includes the St.Helenian endemics (E. dimorphum, E. nervo-sum and M. furcata) and their sister group(E. semicylindricum and E. deckenii) corre-sponds to the subsections Pilosa Christ andPolylepidea Christ of section Lepidoglossa,respectively. Elaphoglossum subcinnamomeum

is also considered to be in subsection Pilosaaccording to Mickel and Atehortua (1980),but, in the trnL-F phylogeny, it is moredistantly related to the St. Helenian endemicsof subsection Pilosa than are E. semicylindri-cum and E. deckenii of subsection Poly-lepidea.

Fig. 2. The single most parsimonious tree of 10 Elaphoglossum species andMicrostaphyla furcata based on trnLpartial intron and trnL-F intergenic sequences including a gap matrix. The tree has a length of 300 steps, aCI¼ 0.923 and a RI¼ 0.875. Bootstrap values (10,000 replicates) are above each branch. Decay indices for eachclade are shown below. The different sections of Elaphoglossum according to Mickel and Atehortua (1980) areindicated with L¼Lepidoglossa H.Christ, EX¼Eximia Mickel & Atehortua, S¼Setosa H.Christ andEL¼Elaphoglossum


Discussion

Special relationships. The molecular phyl-ogeny clearly indicates that E. dimorphum,E. nervosum and M. furcata are a closelyrelated group, strongly supporting Mickel’s(1980) hypothesis of their close relationship.The evidence from molecular data is thusconsistent with Mickel’s (1980) treatment ofMicrostaphyla furcata under Elaphoglossum, asE. bifurcatum. As a result of these findingsM. furcata will now be referred to asE. bifurcatum.

The low levels of sequence divergencebetween the endemic Elaphoglossum speciesindicate recent speciation events on St Helenafrom a common ancestor, which was probablyAfrican in origin, or recent speciation com-bined with hybridisation, as discussed below.These events must have occurred less than14 million years ago, the estimated age ofSt Helena from potassium-argon dating (Baker1967). The common ancestor speciated indifferent ecological niches into taxa withdistinct blade morphology: epiphytic E. nervo-sum with an entire blade and prominentvenation; terrestrial E. dimorphum with lacer-ated frond; and terrestrial E. bifurcatum with adistinctly pinnate frond, the cause of muchtaxonomic debate. Extensive ecological andmorphological variation (which can obscurephylogenetic relationships, particularly withcontinental relatives) with little genetic diver-gence has been reported for a number ofinsular plant groups (Sang et al. 1994, Baldwin1998). Marked morphological differences canarise through changes in one or a few genes.For example, two mutant alleles of the Unifo-liata gene dramatically affect the pinnate leafof pea (Pisum sativum) (DeMason and Schmidt2001). The majority of species of Aspleniaceaeare pinnatifid but a small number have anentire leaf. In a recent molecular phylogeny ofAspleniaceae, using chloroplast rbcL nucleo-tide sequences, neither simple nor pinnatifidleaf construction was found to be phylogenet-ically informative (Murakami et al. 1999).Interestingly, the genus Neottopteris, which

was defined by its special simple leaf wasparaphyletic and formed a clade with Aspleni-um prolongatum Hook. whose leaves are pin-natifid 3–4 times. This was confirmed in astudy by Gastony and Johnson (2001) that alsofound the 2-pinnate to 2-pinnate-pinnatifidleaves ofLoxoscape thecifera (Kunth) T.Moorerobustly nested among the simple-leavedspecies of Neottopteris. Evidence from themolecular phylogeny of Elaphoglossum onSt Helena and that by Murakami et al.(1999) and Gastony and Johnson (2001) onAspleniaceae indicates that leaf dissection canbe a weak taxonomic character in ferns at thegeneric level, as suggested by Mickel (1980).

There are two hypotheses concerning therelationships of E. dimorphum, E. nervosumand E. bifurcatum. The first, as suggested byMickel (1980), is that E. dimorphum is ofhybrid origin involving E. nervosum andE. bifurcatum. The intermediate morphologyof E. dimorphum supports this hypothesis.Although there have been no studies on theinheritance of chloroplast DNA in Elaphoglos-sum, maternal inheritance has been reported inAsplenium (Vogel et al. 1998) and demon-strated in the cheilanthoid ferns (Gastonyand Yatskievych 1992). The trnL-F sequencesof E. dimorphum and E. nervosum are identi-cal, which does not conflict with the hypoth-esis of a hybrid origin of E. dimorphum. IfE. dimorphum were of hybrid origin, E. nervo-sum would be the maternal parent. In a recentrbcL phylogeny of Aspleniaceae by Murakamiet al. (1999), sequences of A. x kenzoi Kurata,which is thought to be hybrid of A. prolong-atum, and A. wrightii Eaton ex Hook. fromYaku Island, were found to be identical tothose of A. prolongatum, indicating thatA. prolongatum is the maternal parent. Furtheranalysis using allozymes revealed that thepaternal parent of A. x kenzoi was A. antiquumMakino (Neottopteris antique (Makino)Masam.), not A. wrightii (Murakami et al.1999). The molecular phylogeny first indi-cated the close relationship of A. prolongatum(pinnate) to A. antiquum (simple blade), despite


their different gross morphologies. There is noreproductive evidence to support the hypoth-esis that E. dimorphum is of hybrid origin. Thespores are fully fertile (unlike those of thehybrid A. x kenzoi) and progeny show nosegregation into the putative parental types.Nor have we observed any morphologicalevidence for segregation in the field. The hybridhypothesis for the origin of E. dimorphum hasbeen investigated further using allozymes andis discussed in the light of some preliminarycytological data in Eastwood et al. (2004b).

The second hypothesis explaining the rela-tionships of these three taxa is that they haveundergone a recent divergence. Although thechloroplast trnL intron and trnL-F spacerregions are non-coding they may not evolvefast enough to differentiate between E. dimor-phum and E. nervosum if their divergence isrecent. A molecular phylogeny of Commiden-drum (Asteraceae), a genus endemic toSt Helena, using the internal transcribedspacers of nuclear ribosomal DNA also failedto resolve relationships completely, indicatingrecent speciation and radiation (Eastwoodet al. 2004a). A recent molecular phylogeneticanalysis of the endemic Hawaiian genusAdenophorus (Grammitidaceae) using threechloroplast DNA fragments, rbcL, atpB andthe trnL-F intergenic spacer, also showed lowlevels of sequence divergence between somespecies groups, resulting in unresolved rela-tionships in the separate analyses (Rankeret al. 2003).

Utility of trnL intron and trnL-F spacer

sequences. The sequence divergence valuesamong the 11 Elaphoglossum species (repre-senting four sections) were relatively high(average value 4.7%). There are no publishedsequence divergence values of rbcL in Elapho-glossum at present and so no direct compari-sons can be made. Sequence divergence rates inrbcL for other intrageneric fern studies varyconsiderably, from an average of 0.6% rbcLin Botrychium subgenus Botrychium (Hauk1995), 1.87% in sister species of Polypodium(Haufler and Ranker 1995), and to 9.8%between different sections of Trichomanes

(Hymenophyllaceae) (Dubuisson 1997). Differ-ences between sequence divergence values at agiven rank will vary greatly in different groupsof plants depending on different rates ofmutation and contrasting concepts of taxon-omy (Hauk 1995). Hauk et al. (1996) foundthat sequence divergence in the trnL-F inter-genic spacer was three to five times higher thanrbcL in 40 species of Ophioglossaceae. Inaddition to higher sequence divergence ratesthe trnL-F spacer also revealed 20 phylogenet-ically informative indels that reduced the num-ber of most parsimonious trees and providedgreater resolution (Hauk et al. 1996). The trnLintron (partial) and trnL-F spacer successfullyresolved relationships between most of theElaphoglossum species in this study, indicatingits applicability to the whole genus. It was,however, unable to resolve the close relation-ship between E. dimorphum, E. nervosum andE. bifurcatum, which may reflect their recentspeciation and/or hybridisation.

The use of non-coding regions of trnL-Fgene has been limited in ferns. Although theobjective of our study was primarily to inves-tigate the phylogenetic relationships of theendemic elaphoglossoid taxa from St Helena,it has also revealed the general utility of thesenon-coding regions of chloroplast genesin resolving relationships among species ofElaphoglossum. It would be interesting toinvestigate the phylogenetic relationshipsof South American Elaphoglossum species,in particular those in section Lepidoglossa,subsection Pilosa, with the St Helenianendemics. The location of the FERN1 primerused to sequence the trnL intron is positionedin the intron rather than the exon, and so onlya partial intron sequence was obtained for eachtaxon. Although this provided useful phyloge-netic signal, it would be advantageous forfuture studies to design fern specific primerswithin trnL exon.

Implications for conservation. As Daugh-erty et al. (1990) stated, taxonomic classifica-tion is a primary determinant of managementpriorities for endangered species. Outdatedclassifications may not reflect phylogenetic


diversity and could lead to the misguidedmanagement of rare species (Avise 1989,Daugherty et al. 1990, Rojas 1992, Hibbettand Donoghue 1996). This is particularlyrelevant for cryptic species, but could con-versely be as important when devising conser-vation strategies for endemic island plantgroups whose phylogenetic relationships maybe obscured by extensive morphological orecological divergence. As Soltis and Gitzen-danner (1999) emphasised, phylogeneticallybased species, for which one can ascertain i)distinguishing synapomorphy (-ies), ii) compo-nent populations and individuals, and iii)closest relatives, should be the target ofconservation efforts. Byrne et al. (2001)recently conducted a phylogenetic analysis oftwo threatened Australian Acacia species,A. sciophanes Maslin and A. lobulata Cowan& Maslin, and their widespread relatives usingRFLP analysis of cpDNA. The study revealedthat both threatened species were phylogenet-ically distinct despite being morphologicallysimilar to their widespread relatives. In factA. lobulata represents an ancient lineage andmost likely a relict species.

It has been suggested that cryptic speciesshould be conserved on the basis of theirphylogenetic divergence (even though morpho-logical divergence is limited) (Avise 1989,Daugherty et al. 1990). However, argumentsfor phylogenetic divergence would tend todiminish the value of recently diverged islandspecies that may be morphologically distinc-tive. Clearly there may be a conflict betweenconservation on the basis of phylogenetichistory versus morphological diversity. Erwin(1991) went so far as to recommend theprioritisation of recent evolution (‘‘evolution-ary front’’) in order to secure future diversity.In the case of Microstaphyla (i.e. E. bifurca-tum) the morphological feature of dissectedfronds is of great interest among elaphoglos-soids, arguably justifying conservation priori-tisation. However, in terms of evolutionaryhistory there is no justification for prioritisingits conservation as an endemic genus Micro-staphyla.

Species conservation priorities, based ontaxonomic, often typological framework, arepredominantly governed by the degree ofthreat to a species as emphasised by the IUCNRed Lists of threatened species (IUCN 1996,Oldfield et al. 1998, Walter and Gillett 1998,Hilton-Taylor 2000). Species conservation pri-orities are usually assigned with the assump-tion that all species have an equal biodiversityvalue. However, this is obviously not the caseand to address this issue a number of authorshave proposed the use of phylogenetic ortaxonomic diversity to select taxa for conser-vation priority (Vane-Wright et al. 1991, Faith1992). In terms of rarity-based prioritisationE. dimorphum is clearly the most threatened.However, if E. dimorphum is of recent hybridorigin involving E. nervosum and E. bifurca-tum, its conservation priority based on geneticdivergence would be lower.

The authors acknowledge support and assistancefrom a number of colleagues and institutes. At theRoyal Botanic Garden Edinburgh we thankMichelle Hollingsworth and Alex Ponge for assis-tance in the laboratory, Michael Moller for his helpwith data analysis and Andrew Ensoll for propa-gating and cultivating the Elaphoglossum species.From St Helena we thank past and present staffof the Conservation and Environmental Section,Agriculture and Natural Resources Department, inparticular Vanessa Thomas, Hazel Bowers,Rebecca Cairns-Wicks and George Benjamin.Many thanks also to John Mickel for his commentson the manuscript. This study was conducted aspart of a Ph.D. dissertation by A.E. funded by aBiotechnological and Biological Sciences ResearchCouncil (BBSRC) Case Scholarship with the Nat-ural History Museum, London, UK.

References

Avise J. C. (1989) A role for molecular genetics inthe recognition and conservation of endangeredspecies. Trends Ecol. Evol. 4: 279–281.

Baker I., Gale N. H., Simons J. (1967) Geochro-nology of the St Helena volcanoes. Nature 215:1451–1456.

Baldwin B. G. (1998) Adaptive radiation of theHawaiian silversword alliance: congruence and


conflict of phylogenetic evidence from molecularand non-molecular investigations. In: GivnishT. J., Sytsma K. J. (eds.) Molecular evolutionand adaptive radiation. Cambridge UniversityPress, Cambridge, pp. 103–128.

Byrne M., Tischler G., Macdonald B., Coates D. J.,McComb J. (2001) Phylogenetic relationshipsbetween two rare acacias and their common,widespread relatives in south-western Australia.Conserv. Genet. 2: 157–166.

Christ H. (1899) Monographie des Genus Elapho-glossum. Denkschr. Schweiz. Naturf. Ges. 36:1–159.

Christensen C. (1906) Index filicum. Hagerup,Copenhagen.

CopelandE. B. (1947)Genera Filicum. In: VerdoornF. (ed.) Ann. Cryptog. Phytopathol. V. ChronicaBotanica Company, Waltham, Massachusetts.

Crane E. H., Farrar D. R., Wendel J. F. (1995)Phylogeny of the Vittariaceae: convergent sim-plification leads to a polyphyletic Vittaria. Amer.Fern J. 85: 283–305.

Cronk Q. C. B. (2000) The endemic flora ofSt Helena. Anthony Nelson, Oswestry.

Daugherty C. H., Cree A., Hay J. M., ThompsonM. B. (1990) Neglected taxonomy and contin-uing extinctions of tuatara (Sphenodon). Nature347: 177–179.

DeMason D. A., Schmidt R. J. (2001) Roles of theuni gene in shoot and leaf development of pea(Pisum sativum): phenotypic characterisationand leaf development in the uni and uni-tacmutants. Int. J. Plant Sci. 162: 1033–1051.

Doyle J. J., Doyle J. L. (1987) A rapid DNAisolation procedure for small quantities of freshleaf tissue. Phytochem. Bull. 19: 11–15.

Dubuisson J.-Y. (1997) rbcL sequences: a promis-ing tool for the molecular systematics of the ferngenus Trichomanes (Hymenophyllaceae)? Mol.Phylogenet. Evol. 8: 128–138.

Eastwood A., Gibby M., Cronk Q. C. B. (2004a)Evolution of St Helena arborescent Astereae(Asteraceae): relationships of genera Commiden-drum and Melanodendron. Bot. J. Linn. Soc. 144:69–83.

Eastwood A., Vogel J. C., Gibby M., Cronk Q. C. B.(2004b) Relationships and genetic diversity ofendemic Elaphoglossum from St Helena. PlantSyst. Evol. 245: 107–118.

Eriksson T. (1999) Autodecay ver. 4.0 (programdistributed by author). Bergius Foundation,

Royal Swedish Academy of Science, Stock-holm.

Erwin T. L. (1991) An evolutionary basis forconservation strategies. Science 253: 750–752.

Faith D. P. (1992) Conservation evaluation andphylogenetic diversity. Bio. Conservation 61:1–10.

Fee A. L. A. (1852) Memoires sur la famille desFougeres. 5: 1–388, Veuve Berger-Levraultet Fils Libraires, Strasbourg.

Fee A. L. A. (1857) Memoires sur la famille desFougeres. 6–8: 1–138, Veuve Berger-Levraultet Fils Libraires, Strasbourg.

Gastony G. J., Johnson W. P. (2001) Phylogeneticplacements of Loxoscaphe thecifera (Asplenia-ceae) and Actiniopteris radiata (Pteridaceae)based on analysis of rbcL nucleotide sequences.Amer. Fer. J. 91: 197–213.

Gastony G. J., Rollo D. R. (1995) Phylogeny andgeneric circumscriptions of cheilanthoid ferns(Pteridaceae: Cheilanthoideae) inferred fromrbcL nucleotide sequences. Amer. Fern J. 85:341–360.

Gastony G. J., Ungerer M. C. (1997) Molecularsystematics and a revised taxonomy of theonocleoid ferns (Dryopteridaceae: Onocleeae).Amer. J. Bot. 84: 840–849.

Gastony G. J., Yatskievych G. (1992) Maternalinheritance of the chloroplast and mitochondrialgenomes in cheilanthoid ferns. Amer. J. Bot. 79:716–722.

Gomez P. L. (1975) Contribuciones a la pteridolo-gia Costarricense. VI, El genero Peltapteris Linken Costa Rica. Brenesia 6: 25–31.

Hasebe M., Oumori T., Nakazawa M., IwatsukiK., Kato M. (1994) rbcL gene sequences gavenew clue to evolutionary lineage for leptospo-rangiate ferns. Proc. Natl. Acad. Sci. U.S.A. 91:5730–5734.

Hasebe M., Wolf P. G., Pryer K. M., Ueda K., ItoM., Sano R., Gastony G. J., Yokoyama J.,Manhart J. R., Murakami N., Crane, E. H.,Haufler C. H., Hauk W. D. (1995) Fernphylogeny based on rbcL nucleotide sequences.Amer. Fern J. 85: 134–181.

Haufler C. H., Ranker T. A. (1995) rbcL sequencesprovide phylogenetic insights among sister spe-cies of the fern genus Polypodium. Amer. FernJ. 85: 361–374.

Hauk W. D. (1995) A molecular assessment ofrelationships among cryptic species of Botrych-


ium subgenus Botrychium (Ophioglossaceae).Amer. Fern J. 85: 375–394.

Hauk W. D., Parks C. R., Chase M. W. (1996) Acomparison between trnL-F intergenic spacer andrbcL DNA sequence data: an example fromOphioglossaceae. Amer. J. Bot. (Suppl.) 83: 126.

Hibbet D. S., Donoghue M. J. (1996) Implicationsof phylogenetic studies for conservation ofgenetic diversity in Shiitake mushrooms. Con-servation Biol. 10: 1321–1327.

Higgins D. G., Bleasby A. J., Fuchs R. (1992)CLUSTAL: a new multiple sequence alignmentprogram. Computer Applic. Biosci. 8: 189–191.

Hilton-Taylor C. (compiler) (2000) 2000 IUCN redlist of threatened species. IUCN, Gland, Swit-zerland and Cambridge, UK. Xviii + 61pp.

Hooker W. L. (1861) A second century of ferns.William Pamplin, London.

IUCN (1994) IUCN red list categories. Prepared bythe IUCN Species Survival Commission, IUCN,Gland, Switzerland and Cambridge, UK.

IUCN (1996) The 1996 IUCN red list of threatenedanimals. IUCN, Gland, Switzerland.

Maddison W. P., Maddison D. R. (1992) Mac-Clade version 3.05. Sinauer Associates, Sunder-land, Massachusetts.

Maxon W. R. (1923) The genus MicrostaphylaJ. Wash. Acad. Sci. 13: 28–31.

Mickel J. T. (1980) Relationships of the dissectedElaphoglossoid ferns. Brittonia 32: 109–117.

Mickel J. T., Atehortua L. G. (1980) The subdivi-sion of the genus Elaphoglossum. Amer. FernJ. 70: 47–68.

Murakami N., Nogami S., Watanabe M., IwatsukiK. (1999) Phylogeny of Aspleniaceae inferredfrom rbcLnucleotide sequences. Amer. Fern J. 89:232–243.

Oldfield S., Lusty C., Mackinven A. (eds.) (1998)The world list of threatened trees. World Con-servation Press, Cambridge.

Pichi Sermoli R. E. G. (1968) Adumbratio FloraeAethiopicae 15: Elaphoglossaceae. Webbia 23:209–246.

Pichi Sermoli R. E. G. (1977) Tentamen pterido-phytorum genera in taxonomicum ordinemredigendi. Webbia 31: 313–512.

Pinter I., Bakker F., Barret J., Cox C., Gibby M.,Henderson S., Morgan-Richards M., Rumsey F.,Russell S., Trewick S., Schneider H., Vogel J.(2002) Phylogenetic and biosystematic relation-ships in four highly distinct polyploidy complexes

in the subgenera Ceterach and Phyllitis in Asple-nium (Aspleniaceae). Org. Divers. Evol. 2:299–311.

Presl C. B. (1849) Epimeliae botanica. AmadeiHaase, Prague.

Pryer K. M., Smith A. R., Skog J. E. (1995)Phylogenetic relationships of extant ferns basedon evidence from morphology and rbcLsequences. Amer. Fern J. 85: 205–282.

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 theendemic Hawaiian genus Adenophorus (Gram-mitidaceae). Mol. Phyl. Evol. 28: 337–347

Rojas M. (1992) The species problem and conser-vation: what are we protecting? ConservationBiol. 6: 170–178.

Sang T., Crawford D. J., Kim S., Stuessy T. (1994)Radiation of the endemic genus Dendroseris(Asteraceae) on the Juan Fernandez Islands:evidence from sequences of the ITS regions ofnuclear ribosomal DNA. Amer. J. Bot. 81: 1494–1501.

Simmons M. P., Ochoterena H. (2000) Gaps ascharacters in sequence-based phylogenetic anal-yses. Syst. Biol. 49: 369–381.

Soltis D. E., Soltis P. S. (1998) Choosing anapproach and an appropriate gene for phylo-genetic analysis. In: Soltis D. E., Soltis P. S.,Doyle J. J. (eds.) Molecular systematics ofplants II: DNA sequencing. Kluwer AcademicPublishers, Boston, pp. 1–42.

Soltis P. S., Gitzendanner M. A. (1999) Molecularsystematics and the conservation of rare species.Conservation Biol. 13: 471–483.

Swofford D. L. (2001) PAUP*: phylogenetic anal-ysis using parsimony (* and other methods),version 4. Sinauer Associates, Sunderland, Mas-sachusetts.

Taberlet P., Gielly L., Pautou G., Bouvet J. (1991)Universal primers for the amplification of threenon-coding regions of chloroplast DNA. Pl.Molec. Biol. 17: 1105–1109.

Underwood L. M. (1905) A much named fern.Torreya 5: 87–89.

Vane-Wright R. I., Humphries C. J., Williams P. H.(1991) What to protect? – systematics and theagony of choice. Biol. Conservation 55: 235–254.

Vogel J. C., Russell S. J., Rumsey F. J., BarrettJ. A., Gibby M. (1998) Evidence for maternaltransmission of chloroplast DNA in the genus


Asplenium (Aspleniaceae, Pteridophyta). Bot.Acta 111: 247–249.

Walter K. S., Gillett H. J. (eds.) (1998) 1997 IUCNred list of threatened plants. Compiled by theWorld Conservation Monitoring Centre. IUCN,The World Conservation Union, Gland,Switzerland and Cambridge, UK.

Addresses of the authors: Antonia Eastwood,Mary Gibby (e-mail: [email protected]),

Royal Botanic Garden Edinburgh, 20A InverleithRow, Edinburgh EH3 5LR, UK. Johannes Vogel,Botany Department, Natural History Museum,Cromwell Road, London SW7 5BD, UK. Quen-tin Cronk, Botanical Garden and Centre for PlantResearch, University of British Columbia, 6804SW Marine Drive, Vancouver, British Columbia,Canada. Andreas Hemp, Department of PlantSystematics, University of Bayreuth, 95440,Bayreuth, Germany.


Documents

Comparison of molecular and morphological data on St Helena: Elaphoglossum