Tryonia, a new taenitidoid fern genus segregated from Jamesonia and Eriosorus (Pteridaceae)

Abstract The Neotropical fern genera Eriosorus and Jamesonia have long been thought of as close relatives. Molecular phylogenetic studies have confirmed this notion but have also revealed that neither genus is monophyletic with respect to the other. As a result, all known species of Eriosorus were recently subsumed under the older generic name Jamesonia. Here, through an analysis of a four-gene plastid dataset, we show that several species traditionally treated in Eriosorus are in fact more closely related to other taenitidoid fern genera (namely Austrogramme, Pterozonium, Syngramma, and Taenitis) than they are to the large Jamesonia sensu lato clade. Tryonia Schuettp., J.Prado & A.T.Cochran gen. nov. is described to accommodate these species and four new combinations are provided. Tryonia is confined to southeastern Brazil and adjacent Uruguay; it is distinct (from most species of Jamesonia) in having stramineous rachises.


Tryonia, a new taenitidoid fern genus segregated from Jamesonia and Eriosorus (Pteridaceae) Introduction
Th e Neotropical genus Jamesonia Hook. & Grev. sensu stricto is among the most distinctive of all fern genera. It has linear, indeterminate leaves bearing highly reduced, coriaceous pinnae covered with dense pubescence (Tryon 1962; Fig. 1). Th ese morphological characteristics are generally considered to be an adaptation to the high- elevation Andean páramo habitats where most Jamesonia species reside (Tryon et al. 1990). Based on reproductive and other cryptic morphological characteristics, Jamesonia has long been thought to be closely related to the genus Eriosorus Fée (Tryon 1962, 1970, Tryon and Tryon 1982. Eriosorus mostly occupies middle-elevation habitats in the Andes and its leaves are much more typical of ferns, usually being very dissected and rather delicate in texture (Tryon 1970;Figs 2, 3). Recent analyses have demonstrated that Jamesonia is both nested within Eriosorus and polyphyletic (Prado et al. 2007, Sánchez-Baracaldo 2004a, 2004b, Schneider et al. 2013, supporting the hypothesis of Tryon (1962Tryon ( , 1970) that the unique morphology of Jamesonia evolved independently multiple times. Th is fi nding prompted the recent recombination of all known species of Eriosorus into Jamesonia (sensu lato, Christenhusz et al. 2011).
Although it is clear that species of Jamesonia sensu stricto are intermixed with those previously assigned to Eriosorus, relationships remain rather poorly supported and additional studies are needed to better resolve the evolutionary history of this group. With that said, the isolated phylogenetic position revealed for one Brazilian species requires special attention. In the most comprehensive study of Jamesonia sensu lato to date (Sánchez-Baracaldo 2004b), two accessions of E. myriophyllus (Sw.) Copel. (Fig. 4) were resolved together and well supported as sister to the remainder of Jamesonia sensu lato. However, it is clear from the phylogram included in the Sánchez- Baracaldo (2004b) study that these accessions are genetically more similar to the outgroup used than they are to the remainder of the ingroup, suggesting that the phylogenetic position of E. myriophyllus may be an artifact of including a single outgroup genus (Pterozonium Fée). Subsequent analyses with a broader phylogenetic context but including fewer exemplars from within Jamesonia sensu lato, actually found E. myriophyllus to be most closely related to the genus Taenitis Willd. ex Schkuhr (Prado et al. 2007, Schneider et al. 2013. Here, through analyses of a four-gene (atpA, chlL, rbcL, and rps4) plastid dataset that incorporates many Eriosorus and Jamesonia sensu stricto species, as well as a broad sampling of related genera, we aim to better resolve the phylogenetic position of E. myriophyllus and allied species. Based on our results, we describe a new genus, Tryonia Schuettp., J.Prado & A.T.Cochran, to accommodate this species and its closest allies.
Sequencing reactions were carried out, in both directions, with the amplifi cation primers, following a standard protocol . For rbcL, two additional (internal) sequencing primers were utilized (Table 2). Sequencing reactions were cleaned using the ZR-96 DNA Sequencing Clean-up Kit (Zymo Research), according to the manufacturer's protocol. Sealed plates were submitted to Operon (Huntsville, Alabama) for sequencing.
Sequencing reads were independently (for each PCR product) assembled and edited using Sequencher (Gene Codes Corporation). Th e 110 new consensus sequences were added to the Fern Lab Database (http://fernlab.biology.duke.edu) and deposited into GenBank (Table 1). For four (of thirty-eight) collections, we could only obtain three of the four gene regions targeted (Table 1). For six collections, an atpA and/or  (Maddison and Maddison 2011). Th e fi nal atpA, chlL, rbcL, and rps4 datasets included 30, 31, 31, and 35 taxa, respectively (see Table 3 for additional details concerning our alignments).

Phylogenetic analyses
Bayesian phylogenetic analyses were conducted independently for each of the four single-gene datasets using MRBAYES version 3.2.1 (Huelsenbeck andRonquist 2001, Ronquist andHuelsenbeck 2003). Th ese Bayesian analyses utilized the GTR+Γ+I model of sequence evolution (the most complex model available) and consisted of four independent runs per dataset, each utilizing four chains and proceeding for fi ve million generations, with trees sampled every 4000 generations. After completion of each analysis, we examined the standard deviation of split frequencies among the runs, plot-ted the output parameter estimates using Tracer 1.5 (Rambaut and Drummond 2009), and very conservatively excluded the fi rst 250 trees (one million generations) from each run. A majority-rule consensus phylogeny with clade posterior probabilities was then calculated from the remaining 4000 trees, for each gene. Based on earlier studies with broader sampling (Prado et al. 2007, Sánchez-Baracaldo 2004a, we rooted our resulting gene trees with Actiniopteris and Onychium.
We compared the results of our single-gene analyses, looking for confl icts that were supported by a Bayesian posterior probability ≥ 0.95. Finding none, we concatenated the four datasets. Th e resulting 38-taxon combined dataset was analyzed as above, but with model parameters estimated and optimized separately for each gene and each run proceeding for 20 million generations. We sampled trees every 16,000 generations and excluded the fi rst four million generations from each run prior to calculating a majority-rule consensus phylogeny with clade posterior probabilities.

Results
Th e four single-gene (atpA, chlL, rbcL, and rps4) datasets contained varying amounts of phylogenetic signal, providing signifi cant support (Bayesian posterior probability, BPP ≥ 0.95) for as few as 11 and as many as 17 bipartitions (Table 3). Th e single-gene trees were largely consistent in their resolved relationships (trees not shown) and there were no well-supported (BPP ≥ 0.95) confl icts among them.
Our combined four-gene dataset comprised a total of 4514 characters, of which 660 were variable (Table 3). Analysis of this dataset resulted in a phylogeny with considerably improved support relative to the single-gene phylogenies; 25 bipartitions had a BPP ≥ 0.95 (Fig. 5). Th e separation of Actiniopteris and Onychium from the remaining taenitidoid genera was well supported (BPP = 1.00). Anogramma, Cosentinia, and Pityrogramma formed a well-supported clade that was, in turn, well-supported as sister to a robust clade including Austrogramme, Pterozonium, Syngramma, Taenitis, and all sampled species previously assigned to either Jamesonia or Eriosorus (Fig. 5).
Th e vast majority of our Jamesonia sensu lato collections come together in a clade on a rather long branch; within this clade branches are short and support is frequently lacking. Six samples previously included within Jamesonia sensu lato are not allied to that larger clade, but rather are embedded within a well-supported clade that also contains Austrogramme, Pterozonium, Syngramma, and Taenitis (Fig. 5).

Discussion
Most species previously assigned to Eriosorus and Jamesonia sensu stricto have been consistently resolved together in a well-supported clade (Prado et al. 2007, Sánchez-Baracaldo 2004a, 2004b, Schneider et al. 2013). And, although support for relationships within this large clade has been generally lacking, the hypothesis that Jamesonia sensu stricto was derived from within Eriosorus (Tryon 1962(Tryon , 1970 has received considerable backing. In our combined analysis, we too fi nd strong support for a clade containing most sampled Eriosorus and Jamesonia sensu stricto species (Fig. 5). Additionally, we fi nd strong support for some of its constituent internal nodes, which indicate that neither Eriosorus nor Jamesonia sensu stricto is monophyletic. Phylogenetic analyses incorporating a more comprehensive sample of taxa and a greater number of markers will ultimately be necessary to fully understand evolutionary relationships within this clade. However, based solely on the evidence to date, it is abundantly clear that Jamesonia and Eriosorus (as typically circumscribed) cannot both be recognized, assuming monophyly as a criterion for generic delimitation. With Jamesonia being the older name (published in 1830, versus 1852 for Eriosorus), the recombination of all known species of Eriosorus into Jamesonia in Christenhusz et al. (2011) was mostly warranted.
Eriosorus myriophyllus was shown by Prado et al. (2007), Sánchez-Baracaldo (2004b), and Schneider et al. (2013) to be isolated relative to most other species previously assigned to Eriosorus or Jamesonia sensu stricto. Here, we fi nd E. myriophyllus and two previously unsampled species of Eriosorus to be more closely related to Austrogramme, Pterozonium, Syngramma, and Taenitis than to Jamesonia (as newly circumscribed herein, Fig. 5). Support for this relationship is strong (BPP = 1.00) and the implications are signifi cant if monophyly is used as a criterion for generic delimitation. Because the type of Jamesonia (Jamesonia pulchra Hook. & Grev.) is resolved well within the large Jamesonia clade and the type of Eriosorus (E. aureonitens (Hook.) Copel.) shows clear morphological and geographical affi nities to this clade, and because there are no other generic names available for the E. myriophyllus group, we here describe a new genus-Tryonia (see below)-to accommodate the isolated species.
In her monograph of Eriosorus, Tryon (1970) identifi ed several small groups of closely allied species. Among these was the species pair of E. myriophyllus and E. sellowianus (with E. schwackeanus considered by her to be a synonym of E. sellowianus). Th is group corresponds perfectly to our proposed circumscription of Tryonia. We fi nd E. myriophyllus, E. schwackeanus (which we consider to be distinct from E. sellowianus), and the recently described E. areniticola (Schwartsburd and Labiak 2008) to form a genetically isolated clade of closely related species (Fig. 5). New combinations for these species, along with the unsampled E. sellowianus, are provided below.
Based on our current dataset, we do not consider the precise phylogenetic position of Tryonia (within the Austrogramme, Pterozonium, Syngramma, Taenitis, and Tryonia clade) to be fully resolved. Although our combined analysis clearly places Tryonia sister to Austrogramme, Syngramma, and Taenitis (collectively), this relationship is not well supported in any single-gene analysis. Th e atpA and rbcL datasets do place Tryonia sister to Taenitis (atpA and rbcL sequences were not available for Austrogramme and Syngramma), but support is lacking (BPP = 0.61 and 0.83, respectively). Likewise, the rps4 dataset resolves Tryonia as sister to Austrogramme, Syngramma, and Taenitis without signifi cant support (BPP = 0.88). Strong single-gene support for the precise position of Tryonia only comes from the chlL dataset, where Tryonia is most closely related to Pterozonium (BPP = 1.00).
Two of the species of Tryonia included in our phylogenetic analysis (T. areniticola and T. schwackeana) are endemic to Brazil; the third sampled species (T. myriophylla) also occurs in Uruguay, near its border with the Brazilian state of Rio Grande do Sul. Although the Andes are the center of diversity for Jamesonia (as newly circumscribed herein), this genus is not entirely geographically distinct from Tryonia. In the recently published Catálogo de Plantas e Fungos do Brasil, a total of nine species are ascribed to Eriosorus or Jamesonia (Prado 2010). Only three of these species noted for Brazil (E. areniticola, E. myriophyllus, and E. schwackeanus) are resolved as sister to Austrogramme, Syngramma, and Taenitis. We found Eriosorus cheilanthoides, E. insignis, and J. brasiliensis to be embedded within the Jamesonia clade (Fig. 5) and E. rufescens was resolved within Jamesonia in an earlier study (Sánchez-Baracaldo 2004b). As for the remaining Brazilian species that have yet to be included in a phylogenetic study, one (E. sellowianus) shows clear morphological affi nities to, and is here considered to be a member of, Tryonia; the other (E. biardii) appears, based on morphology, to be best accommodated in Jamesonia. Regardless of the ultimate phylogenetic placement of these two unsampled species, the genus Tryonia can be described as wholly endemic to Brazil and Uruguay. Similar to some species of Jamesonia, but with stramineous rather than castaneous rachises.
Etymology. Th e generic name honors Dr. Alice Faber Tryon, who made extraordinary contributions to fern systematics and published taxonomic revisions of both Jamesonia sensu stricto and Eriosorus (from which Tryonia is segregated herein).
Distribution. Tryonia occurs primarily in southeastern Brazil. However, one species (T. myriophylla) can also be found in Uruguay (Cerro Largo: Sierra Souza), near the Brazilian border. Th e genus is mostly restricted to the Atlantic Forest, along shaded streams, on damp shaded sandstone, or in more open places (but here shaded by shrubs); 600-2300 m.
Discussion. Tryonia can be distinguished most readily from Jamesonia by its stramineous rachises, but its gross morphology is also reasonably distinct. Tryon (1970) referred to the leaves of T. myriophylla as "generalized" (i.e., elongate-triangular and well developed). She drew a distinction between them and the "specialized" (i.e., either complex and scandent or compact and linear) leaves of Jamesonia sensu stricto and many other species at the time placed in Eriosorus, as well as between them and the "intermediate" (i.e., falling between the two extremes) leaves of other species she treated in Eriosorus. Although the Andean Jamesonia congesta also has "generalized" leaves, it is readily distinguished from Tryonia by its rachis color. Th e only species of Jamesonia with occasionally stramineous rachises (J. fl exuosa) has "specialized" (complex and scandent) leaves. Spores of Tryonia (Fig. 9) and Jamesonia are basically indistinguishable.
Tryonia comprises the following species. Discussion. Based on the gene regions included in our analysis, we found Tryonia areniticola to be genetically indistinguishable from T. myriophylla, despite the presence of several morphological diff erences (Schwartsburd and Labiak 2008). Further studies that include nuclear markers will be necessary.