Adetogramma (Polypodiaceae), a new monotypic fern genus segregated from Polypodium

Abstract Polypodiaceae is one of the most diverse and abundant families of ferns in tropical and subtropical forests. Despite multiple studies investigating its phylogeny and taxonomy, several generic boundaries within the family still need clarification. One of the most problematic circumscriptions is that of Polypodium L., and one species that still contributes to this uncertainty is Polypodium chrysolepis Hook. The main goal of this study was to use molecular and morphological data to clarify the relationships of P. chrysolepis inside the polygrammoid clade. Sequences from three plastid regions (cpDNA – rbcL, rps4 and rps4–trnS IGS) from fifty species belonging to thirty-two genera of Polypodiaceae were analyzed using maximum likelihood and Bayesian inference. Polypodium chrysolepis constitutes an isolated lineage among the neotropical polygrammoid ferns, close to Serpocaulon and the grammitids, and is recognized here in a new genus. It can be distinguished by its entire leaves with free veins and peltate, pedicellate, lanceolate paraphyses. A new combination, Adetogramma chrysolepis, is proposed and a new taxonomic treatment is presented; its conservation status was assessed using IUCN Red List Categories and Criteria.


Introduction
Polypodiaceae is one of the richest fern families, and one of the most diverse and abundant groups of vascular plants in tropical and subtropical forests (Schneider et al. 2004). The current family circumscription (PPG I 2016) is based on many phylogenetic studies (e.g., Schneider et al. 2004, Schuettpelz andPryer 2007). With this definition, Polypodiaceae includes the large segregate family, Grammitidaceae (sensu Parris 1990), and comprises 65 genera (PPG I 2016). Despite multiple studies investigating the phylogeny and taxonomy of Polypodiaceae, Smith et al. (2008) recognized that certain generic boundaries within the family still need clarification.
One of the most problematic circumscriptions is that of Polypodium L. . As recognized by Hennipman et al. (1990), Polypodium is polyphyletic, and several groups historically included in it (Hennipman et al. 1990) have been recognized as other genera [e.g., Goniophlebium (Blume) C.Presl, Microgramma C.Presl (Salino et al. 2008), Pleopeltis Willd., Phlebodium (R.Br.) J.Sm. (Otto et al. 2009), Serpocaulon A.R.Sm. (Smith et al. 2006a), Synammia C.Presl ]. Nevertheless, recognition of all the above-cited genera still does not render Polypodium monophyletic (Schneider et al. 2004, Otto et al. 2009, Assis et al. 2016. In its latest circumscription (PPG I 2016), the genus is considered presumably monophyletic, but the groups that remain in it [Polypodium dulce group, Polypodium plesiosorum group, and Polypodium colpodes group (Tejero-Díez 2005, Sigel et al. 2014], still need to be comprehensively included in phylogenetic studies. One species that still contributes to this uncertainty is Polypodium chrysolepis Hook., a species occurring in the Andes from northern Argentina to Ecuador. The generic placement of this species has been controversial: it was described in Polypodium by Hooker (1844) in Icones Plantarum, tentatively placed in Lepicystis [now treated within Pleopeltis (Smith and Tejero-Díez 2014)] by Diels (1899), recognized as a distinct entity within Polypodium by de la Sota (1960) and placed in Microgramma by Crabbe (1967), following notes by A.H.G. Alston. The first molecular phylogenetic placement of P. chrysolepis (Schneider et al. 2004) showed that none of the above-mentioned generic positions are acceptable. In Schneider et al. (2004), P. chrysolepis was recovered as sister to the Serpocaulon+grammitids clade, and it is distantly related to the campyloneuroid clade (which includes Campyloneurum, Microgramma, and Niphidium) and to Polypodium s.s. clade. No formal taxonomic changes were proposed for P. chrysolepis by Schneider et al. (2004), as its position was ambiguous, and since then no new studies have been conducted on this species.
The main goal of this study was to employ molecular and morphological data to investigate the relationships of Polypodium chrysolepis within the polygrammoid clade, and to use available morphological and phylogenetic information to formally propose an adequate generic placement for this species in Polypodiaceae.

Alignment and phylogenetic analyses
Sequence electropherograms were edited using the STADEN package software (Bonfield et al. 1995). Edited sequences were submitted to automated alignment with MUSCLE (Edgar 2004) and the resulting alignment was checked manually using MEGA 7 (Kumar et al. 2016).
Datasets were analyzed using maximum likelihood (ML) and Bayesian inference (BI). Maximum likelihood was performed using IQ-TREE (Nguyen et al. 2015) with partitioned matrix (Chernomor et al. 2016), automatic selection of the bestfit substitution model (using Bayesian Information Criterion, Schwarz 1978), and branch support assessed with 10,000 ultrafast bootstrap replicates (Minh et al. 2013). Best-fit models according to BIC were TIM2e+G4 for rbcL, K3Pu+G4 for rps4 gene and TVM+G4 for rps4-trnS IGS. For BI, a model-based phylogenetic analysis using Markov chain Monte Carlo-based was performed using MrBayes v3.2.2 (Ronquist et al. 2012), treating each DNA region (rbcL, rps4 gene and rps4-trnS IGS) as separate partitions. An evolutionary model for each DNA region was selected in jModelTest 2 (Darriba et al. 2012;Guindon and Gascuel 2003), using the Bayesian Information Criterion (Schwarz 1978, Table 1). For the rbcL dataset, the SYM+I+G was selected, and for the rps4 gene and rps4-trnS datasets the GTR+G model was selected. Each analysis consisted of two independent runs with four simultaneous Markov Chains run for 5,000,000 metropolis-coupled generations, starting with random initial trees and sampling one tree every 1000 generations. To improve swapping of chains the temperature parameter for heating the chains was lowered to 0.05. To check the convergence of the runs, ESS (effective sample size) and PSRF (potential scale reduction factor) were examined (Ronquist et al. 2012) using Tracer v.1.6 (Rambaut et al. 2014). Based on the sampled parameter values examined, 2000 trees, including the ones generated during the burn-in phase, were discarded. Remaining trees were used to assess topology and posterior probabilities (PP) in a majority-rule consensus tree. Because PP in Bayesian analysis are not equivalent to bootstrap (BP) (Erixon et al. 2003), we used criteria similar to a standard statistical test, considering groups with PP > 95% as strongly supported, PP 90-95% as moderately supported and PP < 90% as weakly supported. Results were summarized on a majority rule consensus tree.

Taxonomic studies and conservation status
Taxonomic conclusions were based on the study of specimens from the following herbaria: BHCB, BM, BR, G, GH, K, LPB, M, MO, NY, P, Q, QCA, QCNE, QPLS, US, USM, and USZ. Specimens with barcode are cited with herbarium acronym followed by barcode number. Abbreviation of genera and species followed IPNI (ipni. org) and morphological terms follow Lellinger (2002). Scanning electron microscope (SEM) images were made using a FEI Quanta 200 SEM, with an accelerating voltage of 30 kV. Samples were sputter-coated with gold and imaged digitally. Spore terminology follows Tryon and Lugardon (1991). Conservation status was assessed using IUCN Red List Categories and Criteria (IUCN 2016) to calculate the Extent of Occurrence (EOO) and the Area of Occupancy (AOO), using the GeoCAT tool (Bachman et al. 2011). The specimens that did not present coordinates were attributed one, whenever locality information was available. A grid cell of 10 km 2 was used for AOO estimation.  Inside the neotropical clade, Synammia appears as sister to all the other neotropical clades, a result also obtained by Schneider et al. (2006) (Fig. 1). Sister to Synammia, our analyses recovered three main clades: the polypodioid clade including Pecluma, Phlebodium, Pleopeltis, and Polypodium s.s. (1.00 PP, 70% BS,), the campyloneuroid clade, containing Campyloneurum, Microgramma, and Niphidium (1.00 PP, 95% BS), and a clade (1.00 PP, 99% BS) that included the grammitid ferns (1.00 PP, 100% BS), Serpocaulon (1.00 PP, 100% BS), and Polypodium chrysolepis (1.00 PP, 100% BS). Both maximum likelihood and Bayesian inference hypotheses recovered P. chrysolepis as sister to Serpocaulon (0.91 PP, 88% BS).

Discussion
The polygrammoid topology recovered in our analyses agrees with previous results from several studies (Schneider et al. 2004, Smith et al. 2006b, Salino et al. 2008, Otto et al. 2009). The placement of Polypodium chrysolepis as an isolated lineage inside neotropical Polypodiaceae and closely related to Serpocaulon and the grammitid ferns (1.00 PP, 97% BS) confirms results previously found by Schneider et al. (2004); this species is not part of the campyloneuroid clade, containing Microgramma, nor does it belong to the polypodioid clade, containing Polypodium s.s. and Pleopeltis. The afore-mentioned genera have all been used as a "home" for Polypodium chrysolepis by various authors in the past (e.g., Hooker 1844, de la Sota 1960, Crabbe 1967).
In our analyses, Serpocaulon, a group segregated from Polypodium (Smith et al. 2006a), appears as sister to P. chrysolepis (0.97 PP, 88% BS), supporting the results of Schneider et al. (2004). Morphologically Polypodium chrysolepis can be readily distinguished from Serpocaulon by having free veins, non-clathrate scales, peltate scales covering the laminar surfaces, and peltate scales as paraphyses. These characters contrast with features of Serpocaulon species: regular anastomosing veins (goniophlebioid venation), clathrate scales, and the absence of scales in veins or laminar surface between veins or paraphyses, or if paraphyses are present, they are 2-3 celled glands (Smith et al. 2006a) (Table 2). Hernández et al. (2014) reported similarities in anatomical features of root cortical cells between P. chrysolepis and Serpocaulon gilliesii (C.Chr.) A.R.Sm, the only Serpocaulon species sampled in their study. Of all Serpocaulon species, S. levigatum (Cav.) A.R.Sm. is the only one that has entire leaves, and therefore the single species resembling P. chrysolepis morphologically. However, in addition to the characteristics mentioned above, S. levigatum differs from P. chrysolepis by having multiple rows of sori on each side of midvein and by rhizomes be- ing covered by whitish wax-like deposits, with few roundish scales, not covering the entire surface of the rhizome (Smith et al. 2006a, Labiak andPrado 2008), while in P. chrysolepis only one row of sori is found on each side of midvein, and rhizomes lack whitish wax-like deposits and are covered by lanceolate rhizome scales (Figs 2,  3). Smith et al. (2006a) highlighted a possible trend of increasing pinnation in Serpocaulon, but the fact their results showed S. levigatum to be closely related to species with completely free, non-adnate pinnae would make the entire lamina shared with P. chrysolepis a homoplastic feature. Sanín (2014Sanín ( , 2015 described a species of Serpocaulon with free veins (S. tayronae D.Sanín), but no phylogenetic evidence is presented to support its placement in Serpocaulon. Sanín also stated that S. eleuterophlebium (Fée) A.R.Sm. and S. patentissimum (Mett.) A.R.Sm. have free veins, although Hensen (1990) describes their venation as regularly anastomosing in the taxonomic treatment of the group (treated as the Polypodium loriceum-complex). More evidence is needed to elucidate the generic position of S. obscurinervium and S. tayronae, as is the case with several other groups morphologically related to Polypodium (Tejero-Díez 2005, Assis et al. 2016).
Polypodium chrysolepis was combined by Crabbe (1967) in Microgramma, but it exhibits morphological characters distinct from that genus. It has free venation while Microgramma, as circumscribed by Almeida et al. (unpublished data) and Salino et al. (2008), has anastomosing veins on the sterile fronds. Moreover, P. chrysolepis has lanceolate, peltate, long-stalked paraphyses (Fig. 2E) while in Microgramma, paraphyses (if present) are hairlike or lanceolate and sessile, never stalked (Table 2). Polypodium chrysolepis resembles some Pleopeltis species in having entire laminae, long-creeping rhizomes, and peltate paraphyses. The main difference between Pleopeltis and Polypodium chrysolepis lies in the shape of the paraphyses -ovate-lanceolate in P. chrysolepis vs. roundish in Pleopeltis. Furthermore, species of Pleopeltis with entire blades always have anastomosing veins (Table 2).
Grammitid ferns, the lineage sister to the Polypodium chrysolepis+Serpocaulon clade, are a very distinct group of species inside the polygrammoid ferns. Once recognized as a separated family (Parris 1990) or a subfamily inside Polypodiaceae (Tryon and Tryon 1982), this lineage can be distinguished by usually tetrahedral chlorophyllous spores, sporangia stalk reduced to a single cell wide in the middle, absence of scales in the fronds, and free or occasionally anastomosing without free included veinlets (Parris 1990, Sundue et al. 2010. Although sharing the free veins with Polypodium chrysolepis, we can distinguish the latter from grammitids by the bilateral spores, and the presence of scales in the blades (Table 2).
Polypodium s.s., following Tejero-Díez (2005), includes the species groups of Polypodium plesiosorum, Polypodium colpodes, Polypodium dulce (sensu Moran 1995) and Polypodium vulgare species group (sensu . Using this circumscription, Polypodium s.s. remains a polyphyletic assemblage of species, with species from the Polypodium dulce complex apparently closer to Pecluma than to Polypodium s.s. (Otto et al. 2009, Assis et al. 2016; in fact, some species in this group were moved to Pecluma (Assis et al. 2016). Further studies are needed to define a monophyletic Polypodium. Nevertheless, following this circumscription, Polypodium s.s. can be distinguished from P. chrysolepis by its deeply-pinnatifid to pinnate leaves with free (in the Polypodium dulce complex species) to anastomosing veins, with a single included veinlet in each areole. Indument is also a useful character for separating P. chrysolepis: Polypodium s.s. shows glabrous to pilose laminar surfaces and the paraphyses, when present, are filamentous or branched, while P. chrysolepis has scaly laminae and the paraphyses are pedicellate scales. Table 2 summarizes the features and main differences amongst the related genera.
Our results do not support the inclusion of Polypodium chrysolepis in any genus previously recognized, including Microgramma, Pleopeltis, or Polypodium s.s. Therefore, we consider this species as constituting a separated, isolated lineage inside the polygrammoid clade. Because the species also has a morphology distinct from that of all other known genera in Polypodiaceae, we believe it merits recognition as a genus, and is described below.

Taxonomic treatment
Adetogramma T.E.Almeida, gen. nov. urn:lsid:ipni.org:names:77161864-1 Note. Adetogramma is similar to Microgramma and Pleopeltis in its epiphytic habit, long-creeping rhizomes and in having entire leaves with one row of sori on each side of the midrib, but differs from these genera by having free veins (vs. veins anastomosing in Microgramma and Pleopeltis) and peltate, pedicellate, lanceolate paraphyses (vs. hairlike or lanceolate and sessile paraphyses in Microgramma, and round, peltate, pedicellate paraphyses in Pleopeltis).

Adetogramma chrysolepis (Hook.) T.E.Almeida
Distribution and ecology. Restricted to central and southern Andes, with known collections from Ecuador, Peru, Bolivia, and Argentina (Fig. 5). It occurs preferably in high elevation formations, ranging from 1,800 to 4,100 meters above sea level, with most collections between 3,000-4,000 m in the central part of the Andes, but found at lower elevations further south in Southern Bolivia and Argentina. Adetogramma chrysolepis is mostly epiphytic or epipetric, rarely terrestrial. Epiphytic specimens usually grow on Polylepis, Berberis, or Buddleja, inside highland humid montane forests (Yungas). Epipetric ones were recorded as growing on rocks inside Yungas forest or in sub-alpine grasslands with scattered shrubs, normally associated with mosses (de la Sota 1960). According to herbarium sheet labels, the species also occurs in secondary forests and modified areas.
Conservation status. Least Concern (LC -IUCN 2016). Adetogramma chrysolepis presents EOO of 1,108,559.773 km 2 and AOO of 5,300 km 2 , respectively, and its conservation status is considered Least Concern. However, the species occurs in a fragile environment that is undergoing an increasing pressure due to human settlement, and extensive grazing by cattle and camelids. Although it is known from at least 40 localities and occurs in several protected areas (e.g., Parque Nacional Carrasco, Cocha-  bamba, Bolivia; Reserva Nacional de Flora e Fauna de Tariquía, Tarija, Bolivia; Parque Nacional Huascarán, Ancash, Peru; Santuario Nacional de Ampay, Apurimac, Peru; Parque Nacional del Rio Abiseo, San Martín, Peru) no information on population fluctuations is available. Decline in the quality and area of occupancy in the near future seems feasible and it is advised that the species be monitored.
Etymology. The generic epithet refers to the most distinctive character of the species, the free venation (Fig. 2C); from the Greek adetos-, free and -gramma, line.
Notes. Adetogramma is a monotypic genus, and although its sole species, A. chrysolepis, varies in laminar size and shape and stipe length, all other characters, such as the rhizome and stipe scales, venation, and paraphyses, are constant. Specimens from the Argentinean Provinces of Tucumán and Salta and Bolivian Provinces of Tarija and Chuquisaca have very long, linear laminae, and longer stipes, but in other characters are congruent with the circumscription here adopted for A. chrysolepis. This variation may reflect colonization of drier, seasonal habitats in a subtropical region.
Morphologically, Adetogramma chrysolepis shares features with several neotropical genera of Polypodiaceae (Table 2), while presenting unique characteristics within the group. Free veins (Figs 2C, 4C) are shared with the Polypodium dulce species group, some Pecluma species, a few Pleopeltis species, few pinnatisected species of Serpocaulon, S. tayrona and most grammitids. Presence of peltate scales on the receptacle is shared between Adetogramma and Pleopeltis (Figs 2D-E, 3A, 4A-B). In both genera, paraphyses may almost completely cover the sporangia in immature sori, and presumably have a protective function. Kreier et al. (2008) hypothesized that Serpocaulon, sister group to Adetogramma, has the Bolivian Andes and adjacent southeastern Brazil as its ancestral area. According to this hypothesis, the Bolivian Andes formed a path for migration and successful colo-nization of the Northern Andes, with subsequent migration into Central America and Caribbean regions. With Adetogramma as sister to Serpocaulon, we believe the hypothesis of the Bolivian Andes as putative ancestral area of the Adetogramma+Serpocaulon clade common ancestor to be more likely. From Bolivia, Adetogramma could have dispersed southwards where colder climate allowed it to colonize lower elevation habitats, and also migrated northwards to Peru and Ecuador. Adetogramma chrysolepis may represent a single relictual species from a previously more diverse and geographically widespread group, or a single, undiversified lineage that colonized high elevation environments. More detailed phylogenetic or phylogeographic studies are required to support or refute these hypotheses.