Phylogeny of Nassella (Stipeae, Pooideae, Poaceae) Based on Analyses of Chloroplast and Nuclear Ribosomal DNA and Morphology

The genus Nassella, as currently circumscribed, includes 116–117 American species. It is characterized by florets with a strongly convolute lemma, a conspicuous or inconspicuous crown, and a short palea. Using 53 species of Nassella and 22 outgroup species we conducted phylogenetic analyses to test the monophyly of Nassella and relationships among species. Two plastid (trnT-trnL and rpl32-trnL) and two nuclear ribosomal (ITS and ETS) regions and morphology were used. Our DNA data alone and combined with morphology showed Nassella to be paraphyletic with respect to a monophyletic Amelichloa. Two main clades were recovered: one with species of Nassella distributed in regions of high elevation from Mexico to northwestern Argentina and one composed of the remaining species of Nassella and those of Amelichloa. The latter is mainly concentrated in southern South America in a variety of habitats with generally lower elevation than the other clade. The monophyly of the close relative of Nassella, the South American genus Jarava s. s., was rejected. None of the groups previously circumscribed as subgenera of Stipa, that are now considered to be composed of species in Nassella, were recovered as monophyletic. The close phylogenetic relationship of Nassella and Amelichloa is supported by only one morphological synapomorphy: the lemma margins flat and strongly overlapping. Keywords—ETS, ITS, morphology, phylogeny, plastid DNA, Stipeae. The tribe Stipeae s. s. (Romaschenko et al. 2012) includes between 572 and 670 species, depending on how the Asian taxa are treated. The species are distributed in temperate and warm temperate grasslands of Africa, Australia, Eurasia, and America (Barkworth 1993). Stipeae includes mostly perennial plants, with loose or dense panicles of one-flowered spikelets, without a rachilla extension, and lemmas usually with a single, terminal awn or with a thickened point, that is entered by the keel and lateral veins. The circumscription of the Stipeae has undergone several changes through the removal of unrelated genera (Barkworth and Everett 1987). Moreover, delimitation of the genera within the tribe has also experienced important changes. The genus Stipa L. was, as traditionally circumscribed, the largest genus within the tribe. In South America it was studied by Spegazzini (1901, 1925), who revised the Argentinean species of Stipa, including some species from Uruguay and Chile, and recognized several subgenera based on morphological characters of the floret. Some of those subgenera were subsequently removed from Stipa and placed into synonymy, such as S. subg. Parastipa Speg. under Ortachne Nees ex Steud. (Soreng et al. 2003), or raised to generic rank, such as Anatherostipa (Hack. ex Kuntze) Peñail. (Peñailillo 1996), Nassella (Trin.) E. Desv. emend. Barkworth (Barkworth 1990, comprising seven subgenera of Stipa), Jarava Ruiz et Pav. (including S. subg. Ptilostipa Speg. and S. subg. Pappostipa Speg., according to Peñailillo 2002), and Pappostipa (Speg.) Romasch., P. M. Peterson & Soreng (Romaschenko et al. 2008, 2011, 2012). As a result of these taxonomic changes, a new delimitation of genera has been suggested, considering morphological and anatomical characters (Romaschenko et al. 2012). Currently, the Stipeae s. s. comprises about 29 genera, 15 or 16 of which are represented by species native to the Americas (Soreng et al. 2003; Arriaga and Barkworth 2006; Romaschenko et al. 2008, 2011, 2012; Cialdella et al. 2010): Amelichloa Arriaga et Torres, Nassella, and Piptochaetium J. Presl, have a wide geographical distribution from Canada to South America; Achnatherum sensu Barkworth, Eriocoma Nutt., Hesperostipa (M. K. Elias) Barkworth, Oryzopsis Michx., Patis Ohwi, Piptatheropsis Romasch., P. M. Peterson & R. J. Soreng, and Ptilagrostis Griseb., are restricted to North America or also occur in Eurasia; and Anatherostipa, Aciachne Benth., Lorenzochloa Reeder & C. Reeder, Ortachne, Pappostipa, and Jarava are South American genera. Aristella (Trin.) Bertol., Austrostipa S. W. L. Jacobs & J. Everett, Celtica F. M. Vázquez & Barkworth, Macrochloa Kunth, Oloptum Röser & H. R. Hamasha, Stipa L., and Stipella (Tzvelev) Röser & H. R. Hamasha, are cultivated or introduced in the Americas (Soreng et al. 2003; Romaschenko et al. 2011). Recent studies, based on combined morphological and molecular data, showed that Achnatherum, Anatherostipa, Aciachne, Jarava, Nassella, Ptilagrostis, and Stipa are polyphyletic groups, while Aciachne is polyphyletic or paraphyletic, and Piptatherum, Piptochaetium, Austrostipa, Piptatheropsis, and Hesperostipa are monophyletic (Cialdella et al. 2010; Romaschenko et al. 2008, 2011, 2012). Nassella was originally described as a subgenus of Stipa (Trinius 1830), then as a subgenus of Urachne (Trinius 1834), and finally raised to generic rank by Desvaux (1854). Later on, Spegazzini (1901, 1925) treated Nassella as a subgenus of Stipa. Parodi (1947) and Clayton and Renvoize (1986) viewed the genus as containing nine or 15 species with short florets, respectively, but Barkworth (1990) expanded Nassella, based on morphological and anatomical characters, to also include species of Stipa with long florets. Although some authors (Zanı́n and Longhi-Wagner 1990; Renvoize 1998) did not agree with this expansion and preferred the traditional treatment of the species under Stipa, the new delimitation of Nassella has been accepted in several recent treatments (Rojas 1998; Torres 1997; Peñailillo 1998; Jørgensen and León-Yañez 1999; Jacobs et al. 2000; Barkworth and Torres 2001; Soreng et al. 2003; Barkworth et al. 2008; Romaschenko et al. 2012). As a result, Nassella is one of the largest genera in the tribe, including, as now circumscribed, 116–117 species (Barkworth 1990; Jacobs et al. 2000; Barkworth and Torres 2001; Soreng et al. 2003; Romaschenko et al. 2012). According to Barkworth and Torres (2001), Nassella is characterized by a strongly convolute lemma, with a conspicuous or inconspicuous crown at the apex, short palea almost always glabrous and without veins. According to Romaschenko et al. (2012), Nassella also has a “ladder-like” lemma epidermal pattern that is believed to be unique within Stipeae. This genus is widely distributed from Canada to Argentina and Chile, also present in Bolivia, Uruguay, Brazil, Colombia, Ecuador, Paraguay, Peru, Venezuela, Guatemala, Costa Rica, Mexico, and U. S. A. Fewer than eight species are native to North America. It is well represented in two South American regions: the central Andean region (Peru, Bolivia, northern and central Chile, and northwestern Argentina), and from northern Patagonia, the Pampas, central and northeastern Argentina to Uruguay, and southern Brazil. Argentina includes 70 species, and more than half of them are concentrated in the northwestern region of the country (Jujuy, Salta, Tucumán, Catamarca, and La Rioja). Earlier phylogenetic studies rendered Nassella monophyletic (Jacobs et al. 2000; Cialdella et al. 2007; Barber et al. 2009), while our recent studies suggested that Nassella is polyphyletic (Cialdella et al. 2010), with some species most closely related to Jarava and other species most closely related to Amelichloa (Romaschenko et al. 2008, 2012). Nassella, Jarava s. s. (excluding species now in Pappostipa), Amelichloa, and the American species of Achnatherum are grouped together in a clade called the Major American Clade (MAC) or New World Subclade (Romaschenko et al. 2008; Cialdella et al. 2010; Romaschenko et al. 2012). Based on a chloroplast DNA-derived phylogram, Jarava was resolved as sister to a clade of Nassella, Amelichloa, and a subset of Mexican species of Achnatherum s. l. (Romaschenko et al. 2008, 2012). We chose four regions, two from plastid DNA (trnT-trnL and rpl32-trnL) and two from nuclear ribosomal DNA (nrDNA), the internal transcribed regions (ITS) and the external transcribed spacer (ETS). Plastid regions and ITS were selected to complete matrices used in previous contributions (trnT-trnL in Cialdella et al. 2010; rpl32-trnL and ITS in Romaschenko et al. 2012). ETS was chosen to test another highly informative nuclear DNA region. Potential pitfalls of the ITS region for inferring phylogenies have been widely addressed (Mayol and Roselló 2001; Nieto Feliner et al. 2001; Álvarez and Wendel 2003; Small et al. 2004; Nieto Feliner and Roselló 2007). The ribosomal region is composed of the 18S, 5.8S, and 26S genes, the internal spacers (ITS-1 and ITS-2), and the intergenic spacer (IGS), which includes the external spacer (ETS). This transcriptional unit is in potentially thousands of tandem copies. Although identical sequences in all ribosomal copies are expected (due to concerted evolution process, see Arnheim 1983) the mechanism of gene conversion sometimes fails to homogenize copies in the face of introgression and/or recent interspecific hybridization (Nieto Feliner and Roselló 2007). As a consequence, nonfunctional ribosomal loci (pseudogenes) may lead to wrong phylogenetic inferences. Moreover, the ribosomal ITS regions are prone to evolutionary constraints to maintain the secondary structures for the accurate processing of mature RNAs. As a result, compensatory base mutations could occur, hence violating previous assumptions of neutrality and independence of characters (Liu and Schardl 1994). Using sequences from four DNA regions and morphological characters from 53 species of Nassella, we conducted a phylogenetic analysis using 22 outgroup species to test the monophyly of the genus Nassella, to study relationships among Nassella and allied genera, especially Amelichloa and Jarava, and to test the monophyly of the infrageneric taxa previously included in Stipa (now under the synonymy of Nassella). Materials and Methods Taxon Sampling—Fifty-three species of Nassella were included in the study, and the samples were either collected in the field or obtained from herbarium material (Appendix 1). All five species of Amelichloa, together with Aristella bromoides (L.) Bertol., Achnatherum eminens (Cav.) Barkworth, A. inebrians (Hance) Keng, A. multimode (Scribn. ex Beal) Valdés-Reyna & Barkworth, Aciachne flagellifera Lægaard, A. acicularis Lægaard, Anatherostipa rigidiseta (Pilg.) Peñail., A. obtusa (Nees & Meyen) Peñail., Austrostipa campylachne (Nees) S. W. L. Jacobs & J. Everett , A. nodosa (S. T. Blake) S. W. L. Jacobs & J. Everett, Jarava media (Speg.) Peñail., J. ichu Ruiz & Pav., J. castellanosii (F. A. Roig) Peñail., J. leptostachya (Griseb.) F. Rojas, J. plumosula (Nees ex Steud.) F. Rojas, and J. scabrifolia (Torres) Peñail., were also included in the matrix and were selected according to previous phylogenetic analyses to represent the phylogenetic diversity of the tribe as best as possible (Cialdella et al. 2010; Romaschenko et al. 2008, 2012). Piptochaetium montevidense was used to root the tree because this taxon is part of a sister clade to the clade that includes MAC in the topology proposed by Cialdella et al. (2010). Morphological Characters—A total of 14 morphological characters were included in the matrix; they were selected from Cialdella et al. (2007, 2010). These characters were chosen to represent diagnostic features for the genera and to elucidate phylogenetic groups (Table S1, Appendix 2). DNA Isolation, Amplification and Sequencing—Genomic DNA was isolated from silica-dried leaf tissue following a CTAB protocol (Doyle and Doyle 1987) and from herbarium material with the DNeasy plant mini kit (Qiagen, Hilden, Germany). We chose two plastid regions to complete matrices used in previous contributions: the trnT-trnL and rpl32-trnL intergenic spacers. These non-coding regions were amplified using primers TabA and TabB (Taberlet et al. 1991) and primers trnL and rpl32F (Shaw et al. 2007), respectively. Additionally, two nrDNA regions were selected: the ITS from nrDNA, including spacer-1, the 5.8S subunit, and spacer-2, and the ETS. The regions were amplified using primers ITS4 (White et al. 1990) and ITS5A (Stanford et al. 2000) and primers RETS4-F (Gillespie et al. 2010) and 18S-IGS (Baldwin and Markos 1998), respectively. The amplification profile consisted of 94 C for 3 min followed by 30 cycles of 94 C for 1 min, 52 C (58 C for nrDNA) for 1 min, and 72 C for 1 min. The PCR reactions were performed in 25 ml final volume with 50–100 ng of DNA template, 0.2 mM of each primer, 25 mM of dNTPs, 5 mM MgCl2, 1 + Taq buffer, and 1.5 units of Taq polymerase (Invitrogen, Life Technologies Sao Paulo, Brazil). Automated sequencing was performed by Macrogen Inc. (Seoul, South Korea). Electropherograms were edited and assembled using BioEdit 5.0.9 (Hall 1999). All sequences were deposited in GenBank (Appendix 1). We combined data from different individuals to represent taxa when it was not possible to obtain sequences from the same individual. Characterization of ITS Sequences—Evidence for ITS paralogous sequences or pseudogene candidates includes length variation, decreased GC content, low stability of secondary structures and absence of conserved motifs (Mayol and Roselló 2001; Bailey et al. 2003; Nieto Feliner and Roselló 2007). Length variation and GC content was determined using BioEdit version 5.0.9 (Hall 1999). Free energy of RNA transcripts and predicted secondary structures was determined at the DINAMelt web server (http://mfold.rit.albany.edu/?q=DINAMelt/Zipfold) by use of the Zipfold application (Markham and Zuker 2008). The presence of the conserved motif (Liu and Schardl 1994) GGCRY-(4–7n)GYGYCAAGGAA was searched at the spacer 1 of ITS region. Sequence Alignment and Phylogenetic Analysis—Sequences were aligned using the program MAFFT version 6 (Katoh and Toh 2008; http://mafft.cbrc.jp/alignment/server/). Indels were coded as binary SYSTEMATIC BOTANY [Volume 39