-Sixteen new sequences of the chloroplast gene rbcL were used to assess evolutionary relationships in Dennstaedtiaceae. Sequence data from nuclear 18S rRNA genes from several dennstaedtioid ferns were also analyzed and found to support some of the inferences based on rbcL data. Of the 1860 nucleotide sites of the 18S rRNA gene examined, 215 sites (11.6%) were variable and 71 sites (3.8%) were phylogenetically informative. The 18S rRNA gene appeared to be evolving at about one ninth of the rate of rbcL, and therefore 18S data should provide increased resolution for phylogenetic studies of the early branches in fern evolution. In the context of those species sampled, the following phylogenetic patterns were evident from maximum parsimony and maximum likelihood analyses of rbcL: 1) Dennstaedtia was the only genus that was not monophyletic: Leptolepia and Microlepia diverged from within Dennstaedtia; 2) Tapeinidium did not emerge with the lindsaeoid genera, in which it is usually treated, but instead diverged at the base of the dryopteroid clade; 3) Lonchitis diverged at the base of the lindsaeoid clade, as in previous analyses of rbcL, a pattern supported by phylogenetic analysis of 18S rDNA; 4) Saccoloma emerged within the Hypolepis clade; 5) Orthiopteris (a segregate genus of Saccoloma) diverged after Dicksoniaceae but before all other clades of higher indusiate ferns; 6) Coptodipteris (usually treated as Dennstaedtia) emerged as a sister to Saccoloma; 7) Dennstaedtiaceae sensu lato appear to be polyphyletic, with the lindsaeoid genera (plus Lonchitis) emerging as a separate clade to Dennstaedtiaceae sensu stricto; 8) Monachosorum diverged from within Dennstaedtiaceae sensu stricto. Dennstaedtiaceae, like many other fern groups, have suffered a restless taxonomic history. The family name was first used by Herter (1940), but not published validly until 1970 (Pichi Sermolli, 1970a). The number of constituent genera in Dennstaedtiaceae has ranged from 8 (Ching, 1940) to over 90 (Holttum, 1947), but more recently has converged to an approximate consensus of 16 to 18 genera (Lovis, 1977; Tryon and Tryon, 1982; Kramer, 1990a). Unless otherwise indicated, I refer to the Dennstaedtiaceae of Kramer (1990a), who illustrates the uncertainty of current systematics with a description of the family that does not include a single defining characteristic. Mickel (1973), who described a Dennstaedtiaceae similar to that of Kramer (1990a), included those genera with, Marginal sori with two indusia, tetrahedral spores, creeping rhizome, solenostele, rhizome indument of hairs, terrestrial inhabitants of wet forests. This definition is rather restrictive in that it excludes some of the lindsaeoid genera (e.g. Lindsaea, Odontosoria, and Ormoloma), which generally have a characteristic protostele. Mickel (1973) noted that some authors have treated these genera as a separate family: Lindsaeaceae (Ching, 1940, Kramer, 1957, Pichi Sermolli, 1977), whereas others combine Dennstaedtiaceae sensu stricto and the lindsaeoid genera (Lovis, 1977; Tryon and Tryon, 1982; Kramer, 1990a) into Dennstaedtiaceae sensu lato. Regardless of taxonomic This content downloaded from 157.55.39.157 on Tue, 17 May 2016 05:53:26 UTC All use subject to http://about.jstor.org/terms WOLF: PHYLOGENETIC ANALYSES OF DENNSTAEDTIACEAE