Proc. Nail. Acad. Sci. USA Vol. 91, pp. 5730-5734, June 1994 Evolution rbcL gene sequences provide evidence for the evolutionary lineages of leptosporangiate ferns (pteridophyte/molecular systematics/phylogenetls/ribulose-blsphosphate carboxylase) MITSUYASU HASEBE*, TOMOYUKI OMORI, MIYUKI NAKAZAWA, TOSHIO SANO, MASAHIRO KATO, AND KUNIO IWATSUKI Botanical Gardens, Faculty of Science, University of Tokyo, 3-7-1 Hakusan, Bunkyo-ku, Tokyo 112, Japan Communicated by Warren H. Wagner, Jr., March 15, 1994 ABSTRACT Pteridophytes have a longer evolutionary his- gies, both live in aquatic habitats and are characterized by tory than any other vascular land plant and, therefore, have heterospory, the latter being extremely rare in the leptospo- endured greater loss of phylogenetically informative informa- rangiate ferns. The phylogenetic relationship ofthese aquatic tion. This factor has resulted in substantial disagreements in ferns to the homosporus ferns remains unsolved (8). evaluating characters and, thus, controversy in establishing a Classification and phylogenetic relationships of the stable classification. To compare competing casicatis, we leptosporangiate ferns above the family level are controver- obtained DNA sequences ofa chloroplast gene. The sequence of sial (1, 5, 6, 8-10). The reasons for discrepancy among 1206 nt of the large subunit of the ribulose-bisphosphate classification schemes include disagreements in evaluation of carboxylase gene (rbcL) was determined from 58 species, morphological characters used. It is often difficult to identify representing almost all families of leptosporangiate ferns. homologous characters because similar characters are found Phylogenetic trees were inferred by the neighbor-joining and in apparently different phylogenetic lineages; convergent or the parsimony methods. The two methods produced almost parallel evolution probably often occurred during the long identical phylogenetic trees that provided insights concerning evolutionary history of ferns (11). Furthermore, frequent major general evolutionary trends in the leptosporangiate extinctions produced missing links, which have resulted in ferns. Interesting findings were as follows: (i) two morpholog- difficulties elucidating phylogenetic interrelationships of ma- ically distinct heterosporous water ferns, Marula and Sal- jor groups (1, 11). Micromolecular information (12) also is not vinia, are sister genera; (iu) the tree ferns (Cyatheaceae, useful to infer familial relationships for the same reasons. Dicksoniaceae, and Metaxyaceae) are monophyletic; and (ui) Recently molecular systematics in plants has progressed polypodiolds are distantly related to the gleicheniolds in spite rapidly with in vitro DNA amplification (polymerase chain of the imilarit of their exindusiate soral morphology and are reaction, PCR) mediated by thermostable DNA polymerase close to the higher indusiate ferns In addition, the affiities of and the direct sequencing methods. In angiosperm system- several "problematic genera" were assessed. atics, this molecular approach has been effective in address- ing many phylogenetic questions that had not been solved The extant ferns include -10,000 species and 250 genera in using phenotypic characters (13). The gene for the large the world (1). They are the most conspicuous spore-bearing subunit of the ribulose-bisphosphate carboxylase (rbcL), land plants and the principal members of land flora after the located on the chloroplast genome, is an appropriate choice flowering plants. Ferns range' widely from tropical to cold for inference of phylogenetic relationships at higher taxo- temperate regions and from lowland to alpine zones, and their nomic levels (13-15). Because of its slow synonymous nu- habitats vary from xeric to aquatic conditions, although the cleotide substitution rate in comparison with nuclear genes center of their distribution is wet tropical and subtropical and its functional constraint that reduces the evolutionary mountains. rate ofnonsynonymous substitutions (16), rbcL is considered Ferns, or megaphyllus pteridophytes, are usually classified to be more useful than the isozymes (e.g., ref. 17) and the into three major groups: the Ophioglossaceae, the Maratti- restriction fragment length polymorphisms (e.g., ref. 18) at aceae, and the leptosporangiate ferns. Although the former these taxonomic levels. Whereas rbcL sequence data have two eusporangiate families have formerly been classified in a been accumulated for angiosperms, only a few rbcL se- single group, recent analyses ofmorphological and molecular quences have been reported for ferns (3), because of the characters revealed that they are not monophyletic (2, 3). In difficulty of finding appropriate primers to amplify or se- contrast, leptosporangiate ferns were inferred to be a mono- quence the gene. Ferns include much more ancient groups phyletic group, because leptosporangia are present only in than angiosperms (19), and their nucleotide sequences are leptosporangiate ferns and this is considered an apomorphic diversified among fern groups. character (2). The eusporangiate condition, however, is a 'In this study, we produced effective primers for fern rbcL plesiomorphic character, observed in the other vascular gene sequencing, with which we could sequence represen- plants. Monophyly of the leptosporangiate ferns is also tatives from >90% ofextant fern families sensu Kramer (8).t supported by unusual gene arrangements on the chloroplast We attempted to (i) identify major evolutionary lineages of genome (4). ferns and infer relationships of the families, (ii) evaluate Based on morphology, the leptosporangiate ferns are usu- previous taxonomic schemes and propose a working hypoth- ally classified into three major groups, Marsileaceae, Salvin- esis for future studies, and (iii) determine the phylogenetic iaceae including Azollaceae, and the rest, which are often positions of problematic taxa. treated as different orders (5-7). Although both the Mar- sileaceae and the Salviniaceae have distinctive morpholo- Abbreviation: NJ, neighbor joining. *To whom reprint requests should be sent at present address: Department of Botany and Plant Pathology, Purdue University, The publication costs of this article were defrayed in part by page charge Lilly Hall of Life Sciences, West Lafayette, IN 47907. payment. This article must therefore be hereby marked "advertisement" tThe sequences reported in this paper have been deposited in the in accordance with 18 U.S.C. §1734 solely to indicate this fact. GenBank data base (accession nos. U05601-U05658). 5730 Downloaded by guest on September 26, 2021 Evolution: Hasebe et al. Proc. Natl. Acad. Sci. USA 91 (1994) 5731 MATERIALS AND METHODS PARS on, STEEPEST DESCENT on, and NNI branch swapping. The trees obtained in the analysis were then used as starting Sixty-four species were selected (see Fig. 2) representing trees to search more parsimonious trees under the equal >90%o offern families sensu Kramer (8). Voucher specimens weighting criterion using MULPARS on, STEEPEST DESCENT sequenced in this study have been deposited at the University on, and TBR branch swapping. We also used the heuristic of Tokyo (TI). Metaxia total DNA was kindly provided by search with the weighting criterion of Albert et al. (32) with Diana Stein (Mount Holyoke College) and David Conant 10 random sequence additions, MULPARS on, STEEPEST DE- (Lyndon State College, Lyndonville, VT), and the voucher SCENT on, and TBR branch swapping. The search was re- specimen is deposited at Lyndon State College. peated five times with different random seeds. We used the Total DNA was isolated (20) from single plants and usually bootstrap analyses (27, 28) and the decay analyses (33) to purified by CsCl density gradient centrifugation (21). Three measure the degree of support for given branches. The overlapping fragments, which cover most of the rbcL gene, bootstrap analysis was performed under unweighting crite- were amplified (22) by Taq polymerase-mediated PCRs. Two rion with 228 bootstrap replicates with simple sequence synthetic primers for each region (aF and aR, bF and bR, and additions, MULPARS on, STEEPEST DESCENT on, and NNI cF and cR in Fig. 1) were designed based on the reported rbcL branch swapping. sequences (3). We usually obtained a single amplified product using these primers. If amplification failed, we used other primers (effective primer arrangements for each taxon can be RESULTS AND DISCUSSION obtained from authors upon request). The amplified frag- Phylogenetic Analyses. We sequenced PCR-amplified frag- ments were electrophoresed on 1% agarose gel, sliced out, ments of rbcL gene from 58 leptosporangiate ferns and used and purified by Geneclean II (Bio 101). The purified double- previously published sequences from Angiopteris, Adiantum stranded DNA fragments were directly sequenced using the capillus-veneris, Botrypus, and Osmunda (3). Sequences of AutoCycle sequencing kit (Pharmacia). The same primers two species in the Aspleniaceae were kindly provided by N. were employed as those used in amplifications but their 5' Murakami (University of Tokyo). In our previous studies of ends were chemically labeled by the fluorescein amidite land plants (3, 34), we used the translated amino acid se- (FluoroPrime; Pharmacia) followed by column purification quences to construct phylogenetic trees, because this elim- (NAP-10 column; Pharmacia). We sequenced the fragments inates bias caused by variation in the GC content of land in both directions