1272OLIVIER DE CLERCK ET AL.TABLE 1. Genera of the Dictyoteae with indication of the respective generitypes, defining characters, and number of species.Genus Type species Defining charactersDictyota D. dichotoma (Hudson) Lamouroux Medullary layers 1, cortical layers 1 ~ 46Dilophus Dil. gunnianus J. Agardh Medullary layers 41, cortical layers 1 16Glossophora G. kunthii (C. Agardh) J. Agardh Surface proliferations, medullary and 3cortical layers 1 to variableGlossophorella G. dhofarensis Nizamuddin and Campbell Surface proliferations, medullary1( 2)and cortical layers 41 near the basePachydictyon P. furcellatum J. Agardh [ 5 P. polycladum(Kützing) Womersley]Medullary layers 1, cortical layers 41 4Currently recognizedspeciesDictyotopsis and Scoresbyella, that are each provisionallyassigned to a separate family, the Dictyotopsidaceae(Allender 1980) and Scoresbyellaceae (Womersley1987), respectively. The family Dictyotaceae is subdividedinto the two tribes Dictyoteae and Zonarieae onthe basis of the number of meristematic cells at thefrond apices. In contrast the Zonarieae have a row or asmall group of such cells, members of the Dictyoteaeare characterized by a single, transversely oriented,lenticular apical cell. Recent molecular phylogenieshave largely confirmed this traditional tribal classification(Lee and Bae 2002, Hoshina et al. 2004, Kraftet al. 2004). There is much less consensus, however,about genus delineations within the tribe Dictyoteae. J.Agardh (1882, 1894) originally recognized four genera(Dictyota, Dilophus, Glossophora, and Pachydictyon) thatwere distinguished by the relative number of corticaland medullary layers and by the presence or absenceof surface proliferations (Table 1, Fig. 1). In Agardh’ssystem, Dictyota comprised species with a unilayeredcortex and medulla, whereas those with a multilayeredmedulla, in at least some part of the thallus, were assignedto Dilophus. Species with a unilayered medullabut a cortex that is at least locally composed of severallayers were placed in Pachydictyon. Glossophora, a genuscomprising only three species restricted to Australia,New Zealand, the Pacific coast of South America, andthe Galapagos Archipelago, was chiefly characterizedby the presence of multiple surface proliferations thateven occasionally may bear reproductive structures.Glossophora is only to a lesser extent defined on thenumber of cortical and medullary layers, which arereported to be variable even within the respective species(Womersley 1987). Surface proliferations alsocharacterize several Dictyota and Dilophus species, butthey are always less abundant than in Glossophora andnever serve as sporophylls. Recently a fifth genus,Glossophorella, was added to the Dictyoteae. It is characterizedby multiple cortical layers, duplication of medullarycells near the margins, and the presence of surfaceproliferations (Nizamuddin and Campbell 1995).The distinction among these four or five genera hasbeen the subject of considerable debate, as some speciesare particularly hard to assign to one or anothergenus (Setchell and Gardner 1925, Taylor 1945, Dawson1950). Hörnig et al. (1992a, b) demonstratedexperimentally that the number of medullary layerscan be altered in many species depending on theTribeDictyoteaeDictyotaGlossophora+ surfaceproliferationsGeneraGlossophorellaDilophusPachydictyon+ surfaceproliferationsFIG. 1. Schematic representation ofthe traditional defining characters ofgenera in the tribe Dictyoteae.
MOLECULAR SYSTEMATICS OF THE DICTYOTEAE 1273culture conditions. They concluded that Dilophus didnot warrant recognition at the generic level and henceproposed a merger of Dilophus with Dictyota. Althoughseveral authors have accepted the merger, others continueto recognize Dilophus as a separate genus (Phillips1992, Huisman 2000). Moreover, recently publishedmolecular phylogenies (Lee and Bae 2002, Hoshinaet al. 2004, Kraft et al. 2004) indicate that the separationof Dilophus and Dictyota is justified, but taxonsampling in these studies was on the low side for drawingsound taxonomic conclusions. The status of Pachydictyon,Glossophora, and Glossophorella has receivedmuch less attention. The morphological data presentedon Dictyota naevosa (Suhr) Montagne andD. radicans Harvey by De Clerck and Coppejans(2003) show that generic boundaries between Dictyotaand the above-mentioned genera are equally fuzzy, indicatingthat these genera also need to be consideredin a global evaluation of generic concepts in the Dictyoteae.Additionally, Hwang et al. (2004) showed thatat least one species of Pachydictyon, P. coriaceum (Holmes)Okamura, is resolved within the genus Dictyotausing a combination of three different chloroplast-encodedgenes (rbcL, psaA, and psbA).In recent times, authors have attempted to findmore stable and pylogenetically informative reproductivecharacters to separate genera. Phillips (1992)showed that marked differences exist in the arrangementand structure of sporangia among the AustralianDilophus species, several of which (Dil. fastigiatus,Dil. robustus, and Dil. marginatus) are characterized bysporangia subtended by two stalk cells rather than oneas is the rule in the other Dilophus species and throughoutthe Dictyoteae generally. Coppejans et al. (2001)and De Clerck (2003) observed that male reproductivestructures (antheridia) in D. crispata Lamouroux andD. magneana De Clerck et Coppejans are not surroundedby hyaline, unicellular paraphyses but by denselypigmented multicellular filaments. De Paula et al.(2001) highlighted the aberrant nature of the D. crispata–cervicornis–magneanagroup in respect to bioactivesecondary metabolites, as these show unique diterpenesignatures that are markedly different from those ofthe remaining Dictyota and Dilophus species examined.In the absence of reliable support from gene-sequencedata, the transfer of the D. crispata–cervicornis–magneanagroup to a separate genus has been suggested but noteffected (see De Clerck 2003).This study aims to: (1) assess the status of the generacurrently placed in the Dictyoteae based on sequencedata of the plastid encoded large subunit RUBISCOgene (rbcL) and nuclear encoded large subunit rDNA(26S, partial); (2) evaluate the taxonomic utility of traditionalmorphological characters at the generic level;(3) assess the concordance of recently introducedanatomical characters with the molecular phylogeneticresults; and (4) interpret the molecular phylogenyin a morphological context and propose arevised classification of the tribe Dictyoteae. Forclarity reasons, names of taxa follow the traditionalclassification recognizing all five genera of theDictyoteae.MATERIALS AND METHODSTaxon sampling and laboratory protocols. Broad taxon samplingwas carried out to ensure as complete a representationas possible of the Dictyoteae (Table 2). This sampling includedgeneritypes of all but one genus, as well as severalspecies that are questionably placed in their present generabased on morphological inconsistencies. Sequences of theDictyoteae were supplemented by a wide selection of taxaspanning the entire phylogenetic spectrum of the Zonarieae.The latter were defined as out-group. The sole representativeof the family Scoresbyellaceae, S. profunda, was also included.Following Chase and Hillis (1991), DNA was extracted fromsamples desiccated in the field in silica gel, and vouchers weredeposited in GENT. Total genomic DNA was extracted using astandard CTAB-extraction method and subsequent purificationwith a Wizard s DNA Clean-Up System (Promega Inc.,Madison, WI, USA) following the manufacturer’s protocol.The rbcL gene was amplified as a single or as two overlappingproducts. In order to amplify the rbcL gene in a single stretch,the rbcL68F primer (Draisma et al. 2001) was combined withthe reverse primer S3R (Siemer et al. 1998). If two separateproducts were needed, rbcL68F was combined with rbcL1380Rand rbcL496F with S3R (all primer sequences listed in Draismaet al. 2001). The 5 0 -end of the 26S rDNA [approximately 1200base pairs (bp)] was amplified as a single product using primersAB28 (Draisma et al. 2001) and T13N (Harper and Saunders2001). The PCR conditions of all primer combinations consistedof an initial denaturation at 941 C for 3 min, followed by941 C for 1 min, annealing at 461 C for 1 min, extension at721 C for 2 min for 28 cycles, followed by a final extension of10 min at 721 C. Excess primer and dNTP were removed withExoSAP-IT s (USB Corp., Cleveland, OH, USA) for 15 min at371 C, followed by 15 min at 801 C to inactivate the enzymes.The resulting products were used for cycle sequencing with theprimers of the initial PCR reactions using an ABI Prism DyeTerminator Cycle Sequencing Ready Reaction kit followingthe manufacturer’s instructions. Sequencing products wereanalyzed with an ABI 3100 Prism Genetic Analyzer (PEApplied Biosystems, Foster City, CA, USA). Sequences wereedited and assembled with Autoassembler version 1.4.0.The rbcL gene, lacking indels, was aligned by eye in BioEdit7.0.4.1 (Hall 1999). Owing to missing data at the 5 0 and 3 0termini of the rbcL sequences, a 1207 bp fragment was selectedfor analysis excluding the first 105 and last 155 bp of the1467 bp gene.The 26S sequences were aligned on the basis of secondarystructure information with DCSE v. 2.60 (De Rijk and DeWachter 1993). The rationale for using secondary-structuremodels for aligning rRNA sequences is based on the fact thatthe conservation of secondary structures exceeds that of nucleotides(Kjer 1995). The 26S sequence of Scytosiphon lomentaria(Lyngbye) Link, the only representative of thePhaeophyceae incorporated in the European Ribosomal RNAdata base (http://www.psb.ugent.be/rRNA/, Wuyts et al. 2004),was used as a model for building our alignment. The alignmentof the variable B13-1, B14, B15, D5, and D5-1 helicesand the highly variable region enclosed by the C1 helix (see DeRijk et al. 1999 for nomenclature) was aided by folding the sequencesof each sample using the Mfold software (http://www.bioinfo.rpi.edu/, Zuker 2003). Foldings were conductedat 371 C using a search within 5% of thermodynamic suboptimality.The different optimal and suboptimal secondary structureswere screened for common motifs. The aligned partialrDNA sequences were 1363 sites in total. The region enclosed