Phylogenetic Relationships of Extant Ferns Based on Evidence from Morphology and rbcL Sequences

-We present the first cladistic analysis of extant ferns based on morphological characters. Our data set consisted of 77 vegetative and reproductive morphological/anatomical characters recorded on a broad sampling of 50 extant pteridophyte taxa, with representatives of all major fern groups, and one seed plant (Cycas). An annotated list of both retained and excluded morphological characters is presented. Results from the morphological analysis are compared with an independent analysis of rbcL data carried out here for the same set of pteridophyte taxa. Finally, we analyze a combined (morphological and molecular) data set. All three data sets were analyzed using maximum parsimony. N o separate sets of analyses using different taxon combinations were conducted on each of the three data sets. Analysis 1 focused on phylogenetic relationships of ferns only (Filicopsida, Botrychium, and Angiopteris), using Cycas as an outgroup representative from the seed plants. Analysis 2 focused on phylogenetic relationships of pteridophytes (Filicopsida, Angiopteris, Botrychium, Equisetum, Psilotum), using Lycopodium as the outgroup. In both sets of analyses, the combined data set provided the most robustly supported hypothesis of relationships. Results from the combined data set in Analysis 1 provided strong bootstrap support for the monophyly of the following clades: leptosporangiate ferns (with Osmunda as the most basal leptosporangiate fern], heterosporous ferns, Cheiropleuria-Dipteris, Diplopterygium-Stromafopteris, tree ferns, schizaeoid ferns, pteridoid ferns, and a large clade consisting of a derived group of leptosporangiate ferns that excludes dennstaedtioids and pteridoids. Various smaller clades within some of these larger clades also have strong support. The dennstaedtioid ferns are paraphyletic. We use the results of the combined data set in Analysis 1to examine character evolution within the leptosporangiate ferns. Results from the combined data set in Analysis 2 indicated robust support for essentially the same fern clades as the combined data set in Analysis 1.In both Analyses 1 and 2, bootstrap support for the leptosporangiate fern clade is much greater using the combined data set than when either the morphological or, particularly, the molecular data set is analyzed separately. Relationships among major groups of pteridophytes at the base of the tree (Botrychium, Angiopteris, Psilotum, Equisetum, Lycopodium) were poorly supported by the combined data in Analysis 2, except for a weak association between Botrychium and Psilotum. We are convinced from this study of the value of using both molecular and morphological data sets in combination as well as separately. A synthetic approach that integrates paleobotanical and neobotanical data will be of greatest interest in further elucidating the phylogenetic relationships of pteridophytes. Our understanding of the phylogeny of pteridophytes has lagged behind the considerable progress made recently in clarifying the phylogenetic relationships of other green plants, especially seed plants (Crane, 1985a, 1985b; Doyle Present address: Department of Botany, Field Museum of Natural History, Roosevelt Rd. at Lakeshore Dr., Chicago, IL 60605. 206 AMERICAN FERN JOURNAL: VOLUME 85 NUMBER 4 (1995) and Donoghue, 1986a, 1986b, 1992; Doyle et al., 1994; Loconte and Stevenson, 1990, 1991; Nixon et al., 1994; Rothwell and Serbet, 1994). Often-cited reasons for this discrepancy are that ferns have, relatively speaking, far fewer morphological characters, and that determining homologous character states, particularly when dealing with organisms with such a long geological record, is not feasible. These arguments have become less persuasive since the publication of a number of comprehensive studies on bryophytes and green algae and their relationships to tracheophytes (Garbary et al., 1993; Graham et al., 1991; Mishler and Churchill, 1984, 1985; Mishler et al., 1994). An overview of extant tracheophyte relationships based on recent analyses of green plant phylogeny is shown in Figure 1, indicating major clades that are resolved and areas of greatest uncertainty (unresolved polychotomies). Although numerous phylogenetic studies have been devoted to the seed plants (spermatophytes), higher-level relationships among some of the major extant lines (cycads, Ginkgo, conifers, gnetales, angiosperms) still are not resolved convincingly (Doyle et al., 1994). For example, the position of the cycads as the most basal group in the spermatophytes is supported by Crane (1985b), Loconte and Stevenson (1990), and Nixon et al. (1994), whereas Ginkgo is the most basal spermatophyte in the analysis by Rothwell and Serbet (1994). Other authors have obtained still different arrangements. Within tracheophytes, relationships among pteridophytes are the least understood (Kenrick and Crane, 1991; Nayar, 1970; Rothwell, 1994). The major clades of extant pteridophytes are: lycopodiophytes (Lycopodiaceae, Selaginellaceae, Isoetaceae); psilotophytes (Psilotum, Tmesipteris); equisetophytes (Equisetum); and ferns (Fig. 1).Ferns comprise three classes: the Ophioglossopsida and Marattiopsida (eusporangiate ferns), and the Filicopsida (leptosporangiate ferns). It has been hypothesized that the psilotophytes are the most basal lineage of extant pteridophytes, and indeed of all extant tracheophytes (Bremer, 1985; Bremer et al., 1987; Parenti, 1980; Pichi Sermolli, 1959). However, more recent evidence supports the placement of the lycopodiophytes at the base of the extant tracheophyte clade (DiMichele and Skog, 1992; Donoghue, 1994; Kenrick and Crane, 1991; Raubeson and Jansen, 1992). Contemporary estimates of higher-level relationships in ferns are mostly intuitive and founded largely on the phenetic concept of overall similarity of morphological/anatomical characters. These characters most often include the sorus and its associated structures (sporangia, indusia, spores), leaf architecture and venation, rhizomes, stipes, and chromosome numbers. In 1969, Wagner attempted a more objective approach to resolving fern phylogeny by applying his "ground plantdivergence method" to homosporous ferns. This method consisted of first inferring the primitive and advanced states of various characters based on the assumption that the commonest character state is usually also that which is primitive. How these states were correlated was then determined so as to organize the taxa graphically in an evolutionary pattern (Wagner, 1969, 1980). A slightly modified version of Wagner's scheme of evolutionary relationships is still presented in morphology and evolution textbooks (e.g., Gifford and Foster, 1988). Smith (1995) reviewed modern ideas on PRYER ET AL.: FERN PHYLOGENY BASED ON MORPHOLOGY AND rbcL SEQUENCES 207