Next to the angiosperms, the Filicopsida, or leptosporangiate ferns, are the most diverse group of living land plants (129). Recent estimates place their diversity at about 11,000 species in 300 genera. The earliest-known occurrence of the Filicopsida is in the Lower Carboniferous (40, 41). By the end of the Carboniferous, six families were represented, all of which became extinct by the Lower Permian. In a second major filicalean evolutionary radiation in the Permian, Triassic, and Jurassic, the Carboniferous families were replaced by several families with extant representatives in the Osmundaceae, Schizaeaceae, Matoniaceae, Dipteridaceae, Dicksoniaceae, and Cyatheaceae (118). A subsequent filicalean radiation among the "polypodiaceous" ferns began in the Upper Cretaceous, after flowering plants had become dominant over much of the land surface.
A well-supported and detailed phylogenetic framework for green plants is critical to our understanding of major evolutionary events, such as the origin of multicellularity, the conquest of land, diversification of life-history patterns, the nature of the relationship between ontogeny and phylogeny, and modes of evolution at the molecular level (46, 96, 97). Recent years have seen an explosion of interest in estimating land plant phylogeny (7, 16, 17, 28, 29, 30, 32, 42, 47, 76, 77, 95, 96, 97, 98, 122), but primary emphasis has been on the seed plants. For purposes of outgroup comparison in studies of tracheophyte phylogeny, the availability of a well-supported framework of "deep" relationships is essential (18, 44). Despite having the longest history of all tracheophyte groups, phylogenetic relationships of pteridophytes (including the ferns) are among the least understood (70, 119, 134, 135). This limits our understanding of pteridophyte diversification and the evolution of terrestrial ecosystems.
Progress has been made in recent years, however, to arrive at an approximate hypothesis for fern relationships that is based on a rigorous and explicit approach. A major collaborative effort by thirteen researchers (53) to expand upon recent rbcL fern studies (51, 52, 143) brought together rbcL data for 99 genera of leptosporangiate ferns representing 31 of the 33 extant families. This large analysis provides an initial framework for elaborating future studies using both molecular and nonmolecular characters. Pryer et al. (109) presented the first cladistic analysis of extant ferns (sampled from all major groups) based on morphological characters and compared results with an independent analysis of rbcL data for the same set of fern taxa (Fig. 1). These morphological analyses confirmed that several results of the recent rbcL analyses that contradicted long-standing pteridological dogma were neither flukes nor demonstrations of the worthlessness of morphology, but are rather well supported by morphology as well. These results include the monophyly of the heterosporous ferns (also simultaneously confirmed by paleobotanical discoveries; 123) and the status of the "Polypodicaceae" sensu lato as a strongly supported monophyletic group, rather than as several unrelated lines derived from Gleicheniaceae, Schizeaceae, etc., a view that seems in hindsight to have been based on overinterpretation and overweighting of a few characters involving sori and/or indusia, while dismissing conspicuous synapomorphies in the sporangia and antheridia. Other recent papers focused on the utility of 16S, rbcL, and 18S sequences, as well as genome mapping, in more specific taxonomic groups within the ferns (13, 15, 43, 55, 56, 90, 110, 142). Rothwell (119) discussed ongoing, preliminary cladistic analyses of living and fossil ferns; these results are consistent with those of other recent studies (18, 97) and support the monophyly of ferns. His evidence for fern monophyly, however, was based only on a small data set and additional data have already cast doubt on it (Rothwell, pers. comm.).
From these recent studies bearing on fern relationships, it is clear that new insight has not been tied to the use of any one type of evidence. In fact, progress in fern phylogeny stems from "making sense" of all of the data, including information on fossil and extant organisms, morphology, and molecules. Although a workable framework of higher-level fern phylogeny is now in place, replacing former intuitive estimates of relationships founded largely on concepts of overall similarity (127), much more data are needed to test adequately the strengths of these new hypotheses based on morphology, fossils, and single gene trees. A glance at our current best estimate of higher-level fern phylogeny, based on morphology and rbcL (Fig. 1), demonstrates that only a few internal nodes have robust bootstrap support: leptosporangiate ferns (89%) and the "polypodiaceous" ferns (86%). All internal nodes at the base of the fern topology are very weakly supported, making it impossible to say with certainty how any of these basal groups of ferns are related to one another. For example, what are the closest relatives to the tree ferns: heterosporous ferns, schizaeoid ferns, or the "polypodiaceous" ferns? These basal branches of the fern topology are critically important not only for stabilizing the overall framework for ferns, but also for resolving relationships among the "deep" branches of land plants (lycophytes, sphenophytes, eusporangiate ferns) and elucidating which of these taxa are the sister group to seed plants. A robust phylogeny for extant ferns will permit clarification of character patterns at the base of the leptosporangiate group that will likely also have an important impact on the interpretation of character patterns of Mesozoic and late Paleozoic taxa.
We propose here a collaborative research study to investigate further the phylogenetic relationships of extant ferns, with a particular emphasis on the basal groups, using additional chloroplast genes, nuclear genes, and morphology. Although it is clear there is also much work to be done within the well-supported derived "polypodiaceous" clade (Fig. 1), it is not the focus of this proposal, since recent studies have clearly established it as a strongly supported monophyletic group. Our aim is to provide a robust overall framework for ferns that provides strong support for relationships at the base of the topology, and our preliminary results from fern sequence data for newly sampled genes suggest we can achieve this objective. Such a well-supported phylogeny will enable us to work towards understanding patterns and sequences of character evolution that gave rise to the Cretaceous radiation and diversification of "polypodiaceous" ferns and also will provide a foundation for future family and generic-level taxonomic studies in ferns.
N.B. Numbers in parentheses refer to literature citations listed in References.