An updated checklist and key to the open-panicled species of Poa L. (Poaceae) in Peru including three new species, Poa ramoniana, Poa tayacajaensis, and Poa urubambensis

Abstract We provide an updated checklist and key to the 30 Poa species with open panicles from Peru which includes previously circumscribed Dissanthelium and Aphanelytrum species, new taxon records, and three undescribed species. Poa compressa, Poa grisebachii, and Poa leioclada are recorded from Peru for the first time. A number of species are placed in synonymy: Poa carazensis, Poa ferreyrae and Poa tovarii are synonymized under the name Poa fibrifera; Poa adusta (tentatively) and Poa pilgeri are synonymized under Poa candamoana; Poa superata is synonymized under Poa grisebachii; and Poa paramoensis is synonymized under Poa huancavelicae. Included within this treatment are three new species, Poa ramoniana, Poa tayacajaensis and Poa urubambensis, which are described and illustrated. Poa ramoniana, found growing near lakes in high elevation Puna grasslands of Junín, is similar to a small form of Poa glaberrima, but differs in having rhizomes and growing to only 5 cm tall. Poa tayacajaensis, found from shrublands on Andean slopes of Huancavelica and Huánuco, bears similarities to Poa aequatoriensis but differs in having shorter lemmas which are pubescent between the veins, densely scabrous sheaths with smooth, glabrous throats, and shorter ligules. Poa urubambensis, a common element of the undisturbed Polylepis forest understory of the Cordillera Urubamba, Cusco, is distinct from all other members of open-panicled Poa’s by having glabrous lemmas with a smooth and glabrous callus, and notably small anthers. The type material for the name Poa adusta is discussed and a lectotype is selected.


Introduction
Th e genus Poa L. is the largest genus of the Poaceae, containing over 500 species with a large distribution across temperate areas of the globe (Soreng et al. 2003(Soreng et al. , 2015. Th e fi rst taxonomic treatment of Poa from Peru comes from Hitchcock's (1927) 'Th e grasses of Ecuador, Peru and Bolivia' in which he mentions 17 species. Standley (1936), in the Flora of Peru series, produced the fi rst treatment of purely Peruvian grasses in which he added P. aequatoriensis to the country records, raising the total number of Peruvian Poa to 18. Following this, the renowned Peruvian agrostologist, Óscar Tovar Serpa, began his life-long work on Peruvian grasses producing a number of publications related to Poa (Tovar , 1974(Tovar , 1984(Tovar , 1986 until his largest cumulatory work in 1993 where he provided a concise treatment of all grasses from Peru ).  most up-to-date taxonomic treatment, and the Checklist for Peru (Brako and Zarucchi 1993) that was published in the same year, considered the country to have either 40 or 41 species of Poa, respectively. Th is number has since varied due to taxonomic revision placing certain names in synonymy (Soreng et al. 2003, and discoveries of undescribed species (Negritto and Antón 2006;Davidse et al. 2010;Soreng and Peterson 2010;Peterson and Soreng 2016). Additionally, DNA studies (Gillespie et al. 2007(Gillespie et al. , 2008Refulio-Rodríguez et al. 2012;Peterson and Soreng 2016) found four small, mostly closed-panicled, genera to be nested within Poa subg. Poa supersect. Homalopoa (Dumort.) Soreng & L.Gillespie. Species of Anthochloa Nees & Meyen, Tovarochloa T.D. Macfarl. & P. But, and Dissanthelium Trin., were thus transferred to Poa by Gillespie et al. (2007) and Refulio-Rodríguez et al. (2012), while species of Aphanelytrum (Hack.) Hack. were recently transferred to Poa by Peterson and Soreng (2016).
To this point, 51 species had been accepted in Poa for Peru ), four of which are considered to be exotic (P. annua L., P. infi rma Kunth, P. pratensis L. [subsp. alpigena (Lindm.) Hiitonen is indigenous in North America and Patagonia], P. trivialis L.). Of these 51 species, 32 have open-panicled infl orescences. Panicle characteristics are good for separating Peruvian Poa into two distinct groups. All taxonomic treatments of Poa in Peru (Hitchcock 1927;Standley 1936;, Bolivia (Renvoize 1998) and Ecuador (Hjorth 1991) have provided diagnostic keys that, in the fi rst couplet, separate species into those with a congested spike-like panicle, with panicle branches appressed, and those with an open-panicled infl orescence, the branches spreading. We believe this artifi cial character to be convenient and reliable in separating Peruvian Poa.
Our objective is to provide an up-to-date summary of the open-panicled species of Poa in Peru including locality information using verifi ed specimens and discussion of nomenclatural and taxonomic attributes, with the new species, Poa ramoniana, P. tayacajaensis and P. urubambensis, being described and illustrated. Two keys are provided to aid with identifi cation of the open-panicled Poa. Th e main key fi rst uses anther length to separate taxa while the Suppl. material 1 fi rst uses lemma indumentum. We hope users of our keys will have a better chance of accurately identifying Peruvian specimens of Poa.

Materials and methods
In this treatment, glabrous means without pubescence (in the sense of slender, relatively soft hairs). Smooth indicates no prickle-hairs with broad bases and/or hooked or pointed apices (i.e., pubescence can occur on a smooth surface, and a rough or scabrous surface can be glabrous). Specimen localities in the checklist are cited by political region (also historically called 'departamento') (capital letters) and then province. Only herbaria where specimens have been checked and verifi ed by the authors have been cited (acronyms following Th iers, continuously updated): Mainly MO (material on loan to US, c. 240 collections of Poa from Peru [Peterson duplicates excepted], and many more from across South America), and US, but USM specimens and types were examined in-situ by RJS (in 2006 and, and CUZ and Z specimens were examined by SPS. Almost all P.M. Peterson Poa collections (fi rst set at US, c. 460 collections from Peru) are duplicated at USM, although the USM duplicates have not been re-checked for this paper. Excluded species are discussed at the end of the checklist.

Results
Of the 32 species of Poa with open-panicles previously recognized in Peru , we consider three of these species records to be erroneous and have placed fi ve species names in synonymy. Following the discovery of three undescribed species and three new country records, we now recognize 30 species of Poa with open-panicles in Peru.
Th e new species, Poa urubambensis, was found in remote areas of the Cordillera Urubamba, southern Peru, during recent fi eldwork by the fi rst author. While reviewing Peruvian specimens of open-panicled Poa from collections in the United States National Herbarium and Missouri Botanical Garden, a further two undescribed species, Poa ramoniana and Poa tayacajaensis, were discovered. Poa ramoniana was discovered from collections by eminent Peruvian botanist, Ramón Alejandro Ferreyra, from Junín. Poa tayacajaensis was discovered from collections by the renowned Peruvian agrostologist, Óscar Tovar Serpa, from the province of Tayacaja, Huancavelica.
We report new species records for: Poa cf. leioclada Hack., previously considered endemic to Ecuador (León Yánez et al. 2011), and P. grisebachii R.E. Fr., previously considered endemic to Argentina . Poa compressa L., originating from Europe, is also reported for the fi rst time. Th ree species: Poa carazensis Pilg., P. ferreyrae Tovar, and P. tovarii Soreng, appear to be morphologically indistinct from P. fi brifera Pilg. and have been placed under that name. Poa adusta J. Presl, known only from the type collection, and the recently described P. pilgeri Negritto and Antón (2006: 87) are synonymized under P. candamoana Pilg. Specimens of Poa superata Hack., previously known only from Argentina and Chile, have been collected from Peru and were determined to be morphologically indistinct from P. grisebachii R.E. Fr. Poa paramoensis Laegaard, previously considered endemic to Ecuador (Laegaard 1998), has been found to be morphologically indistinct from P. huancavelicae Tovar. Reports of Poa lilloi Hack. and P. supina Schrad. for Peru are considered erroneous, and P. bromoides Vahl (accepted by Brako and Zarucchi 1993) is currently accepted as Eragrostis bromoides (Vahl) Steud. by Soreng et al. (2016). Th ese species were removed from the checklist.
Key to the open-panicled species of Poa in Peru Poa aequatoriensis Hack. Ref: Standley (1936: 125); Tovar ( : 45, 1993. Ill: Hjorth (1991: fi g. 6), Tovar (1965: lam Peterson 15175 (US). Discussion: Poa aequatoriensis occurs from northern Peru, Ecuador, and Colombia, although one collection is known from Cusco. Brako and Zarucchi (1993) and  also state P. aequatoriensis to occur in ANCASH, HUANCAVELICA and LIMA. Commonly misidentifi ed as P. trivialis and vice-versa. Reports of P. aequatoriensis from Bolivia by Hjorth (1991) are, most likely, another taxon. Th is report was probably based on material called Poa umbrosa Trin. by Renvoize (1998;Renvoize & Cope 4071, K, US!), which RJS redetermined as Poa bradei Pilger, a species otherwise known only from Brazil, which has spikelets with only perfect fl owers and short anthers (0.5-1 mm). Poa annua L. Syn: Ochlopoa annua (L.) H. Scholz. Ref: Standley (1936: 125); Tovar ( : 61, 1993.   Swallen & Tovar (1965: 370-371); Tovar (1993: 154-156). Ill: Renvoize (1998 Th ere has long been uncertainty regarding the identity of Poa adusta (Standley 1936, p. 129;, p. 61, Soreng et al. 2003, which is known only from the type col-lection, T. Haenke s.n. (PR, HAL, W). A report from Colombia (Giraldo-Cañas 2011) represents a diff erent species. Th e type description of P. adusta can be considered erroneous regarding the lemma indumentum, which was stated to be scabrous while the type specimens examined had pubescence present on the lemma keel marginal and intermediate veins (and at least sparingly between them). After studying the HAL isotype, a solitary fl owering shoot with one full leaf, without base, we felt it most likely represented P. candamoana. A photo of the W isotype (which has only an infl orescence and a bit of upper culm) is also a match for the HAL specimen. Upon studying the PR isotype sheet of P. adusta, it was found to comprise two separate leafy shoots that diff er from each other. Th e PR left-hand plant, has a base (with basal-most sheaths characteristic of P. candamoana; ie, the base slightly infl ated and lustrous), and it otherwise matches the HAL sample in details. Th e spikelets of these plants diff er from P. candamoana in the broad, blunt, somewhat distorted lemmas, and dark coloration (adustus means blackened or scorched), and do not match any species we know of. Although the dark coloration might derive from poor preservation (coming as they did from moldy bundles), we have seen similarly distorted and discolored spikelets in a few specimens of Poa from the region that we expect resulted from disease (given that in one case other spikelets in the same plant infl orescence were normal in shape, color, and pubescence). Th e PR right-hand plant, although bearing similar characteristics of lemma indumentum, is obviously quite pubescent between the veins and the lemmas are acute and not so discolored (a few spikelets are discolored and distorted to a lesser degree than in the left hand plant), lacks a base, has blade apices that are not navicular as is normal for P. candamoana, and the upper culm blade is far too long (longer than the panicle) and inserted too high on the culm for that species (> 10 cm and exceeding the panicle). Th e right hand plant is almost certainly P. horridula. Th e left hand plant could be either the result of hybridization between P. candamoana and P. horridula (see P. horridula discussion below), diseased material of P. candamoana or another species, or a rare species not yet rediscovered. Our choice is to lectotypify Poa adusta on the left hand plant of specimen PR-495759. We will propose conservation of Poa candamoana (Pilger 1906) over P. adusta (J. Presl 1830), rather than outright rejection, since the former name has been widely used for this commonly collected species, and the identity of the lectotype of the latter is still in doubt.
Th e origin of the P. adusta specimens has been uncertain as they might not have been collected from Peru, even though J. Presl (1830: 271) mentioned 'Peruvia' as place of origin in the protologue. Th e P. adusta specimen at PR was part of a bundle of specimens which were tagged with the note 'Aus verfault P.' (translated as 'from rotten bundle'). Th e tag, written by Bohemian botanist K.M. Sternberg (sometime between1821 and 1822), indicates that the specimen came from a damaged bundle of plants, for which the country of origin was not indicated. Haenke's herbarium originally consisted of specimens without labels and, when the Czech National Museum purchased the bundle from the compatriot handling company of Hiecke, Ziencke & Co. in 1821, the origin was indicated only on the top of every bundle of specimens. Th is is the reason why Haenke's handwriting is generally missing from all of his specimens. 'Peruvia', mentioned in the protologue, is J. Presl's opinion about the origin of these specimens rather than the real place of origin for some portion of them. Hae-nke's collections (made between 1790 and 1792) may have come from his crossing from Buenos Aires and Rio La Plata to Santiago (where he fi nally caught up with the Malaspina Expedition), or any (suitable) place along the Pacifi c Coast of America from Santiago (Chile), Lima (Peru), Acapulco (Mexico), to Monterey (California, then part of Mexico), Nootka Sound (now British Columbia), and Yakutat Bay (now Alaska), where the expedition landed (see Češka 2002). Our determination of the right hand plant at PR as P. horridula provides a location for the Poa adusta lectotype as the central Andes, where P. horridula and P. candamoana are common.
Although the set of Haenke's collections that made their way to Prague were purchased by the Czech National Museum, a substantial part of it ended up in the Prague University herbarium (now the herbarium of Charles University in Prague, PRC). J. Presl's brother (who wrote up the Gramineae, Cyperaceae, and Taccaceae), C.B. Presl, was custodian of the PR herbarium (where all or most of Haenke's Poaceae from the expedition now reside), and together, while writing the Reliquiae Haenkeanae (C.B. Presl 1825-1835), they off ered surplus duplicate specimens for sale to other botanical institutions, which is likely how they arrived at HAL and W (Češka 2002;Otakar Sida [PR], pers. communication).
Poa compressa L. Ref: Soreng and Peterson (2012: 31). Ill: Giussani et al. (2012: sp. 11, p. 300 Poa tovarii Soreng. Ref: Standley (1936: 126); Tovar ( : 37-44, 1984Tovar ( : 8, 1993; Soreng (1998: 200). Ill: Tovar (1965: lam  It is impossible to say if either of these specimens were accepted by Brako and Zarucchi (1993) as only one specimen was cited for the country. Poa gilgiana Pilg. Syn: Melica expansa Steud. ex Lechl.. Ref: Standley (1936: 126); Tovar ( : 36, 1993. Ill: Tovar (1965: lam   ) and this is the fi rst recording of the species from Peru. It is likely that, with further revision of Bolivian Poa, this species will also be found to occur in Bolivia. Specimens of P. superata were found to be morphologically indistinct from the type of P. grisebachii and so have been grouped as a single taxon. All material examined of this taxon exhibited soft pubescence on at least the upper lemmas of the spikelets, with the lower fl orets being glabrous or softly pubescent. Poa grisebachii can be highly variable in terms of lemma pubescence. Th e lectotype of P. grisebachii p.p. has glabrous and smooth lemmas while the isolectotype and other syntypes of P. grisebachii p.p. at US fragms. ex UPS, and the type specimens of P. superata p.p., were all observed with at least sparsely pubescent lemmas. All the Peruvian material of P. grisebachii has pubescent lemmas, at least on the distal fl orets.
Distribution. Known only from a single locality in Junín.
Habitat. Puna grassland, 4200-4300 m, in wet margins/shore of lakes, in moss. Etymology. Th e species is named in recognition of the eminent Peruvian botanist, Ramón Alejandro Ferreyra   Discussion. Th is new species appears like a small form of P. glaberrima, but diff ers by being extensively rhizomatous and reaching only 5 cm tall, versus densely tufted and 12-45 cm tall for P. glaberrima. Th e US holotype of P. ramoniana is a mixed collection, with a second taxon p.p. "b", which appears to be Poa gymnantha Pilg. that is sterile, tightly tufted, with intravaginally branching shoots and involute leaf blades that are adaxially scabrous. Th e USM isotype also contains two species: the small rhizomatous plant is P. ramoniana; the taller plants appear to be P. glaberrima. Tovar originally determined the USM type and paratype as Poa lilloi , which, among other diff erences, has a dense habit, without rhizomes, and ascending panicles branches, densely scabrous lemmas with narrow white, scarious margins, and sometimes a web on the callus. Tovar also identifi ed the US type as Poa ovata Tovar (1965: 17), which RJS considers to be a rare to uncommon sexually reproducing phase of the small form of P. gymnantha, a species that is otherwise predominantly pistillate and apomictic (Negritto et al. 2008). Other material determined as P. lilloi in Peru has been referred to P. glaberrima and P. candamoana, or small P. kurtzii (see excluded species, below).  Plants gynomonoecious. Perennials; tufted, without lateral or downward tending shoots. Tillers intravaginal. Culms 55-65 cm tall, erect or decumbent (when decumbent sometimes extravaginally branching at the lower culm nodes, i.e. Peterson et al. 20369); Culm nodes 3-4, terete, smooth, 2-3 nodes exposed at fl owering; Culm internodes terete, smooth. Leaves; Sheaths slightly laterally compressed, keeled, lower culm and lateral ones densely scaberulous distally; Butt sheaths thin papery, somewhat loosely investing the shoots; Uppermost culm sheaths 10-13 cm long, margins fused 60-75 % their length, slightly shorter than their blades; Collars and throats smooth, glabrous; Ligules 2.0-3.5 mm long, not decurrent, abaxially sparsely to moderately densely scabrous, apices obtuse to subacute, margins densely scabrous, ligules of sterile shoots and lower culm leaves 0.5-1 mm long; Cauline blades 6-15 cm long, 3-5 mm wide, well developed, longer than their sheaths, generally fl at, keeled, thin, lax, abaxially, marginally, and adaxially scabrous mainly along the veins, folded near the apex, apex gradually tapered to a slender point; Blades gradually increasing in length up the culm, fl ag leaf blade 10-15 cm long; Sterile shoot blades like those of the culm, but somewhat shorter and smoother. Panicles 18-20 cm long, loose, open, exerted, slightly lax, to 5 cm wide, with more than 100 spikelets, proximal internodes 3.5-4 cm long, smooth; Rachis with (3) 5-6 branches at lower nodes; Primary branches slender, mostly laxly ascending, sometimes spreading, one sometimes refl exed, angled, proximally smooth to moderately scabrous mainly on the angles; Lateral pedicels mostly < 1 mm long, scabrous; Longest branches 6-8 cm long, with 14-22 spikelets in the distal half, slightly overlapping. Spikelets 4.5-6 mm long, c. 2 × longer than wide, lanceolate, laterally compressed, not bulbiferous, greyish-green to somewhat anthocyanic at maturity; Florets (3-)4(-5), proximal fl orets hermaphroditic and distal one pistillate; Rachilla internodes terete, distal internodes 0.7-1 mm long, terete, smooth, glabrous; Glumes unequal, narrow lanceolate to lanceolate, herbaceous and pale green below, sometimes anthocyanic in margins and apex, veins distinct, distinctly keeled, keels sparsely short scabrous distally, surfaces smooth, margins scarious-hyaline, edges entire smooth, apices sharply acute, entire; Lower glumes 1.7-2.5 mm, 2/3-4/5 as long as adjacent lemmas, 1-veined, very narrow, slightly sickle shaped; Upper glume 2.4-3 mm, c. 2 × wider than the lower, 3-veined; Calluses webbed, with a dense, long dorsal tuft of wooly hairs; Lemmas (the lowest) 2.8-3.7 mm long, 5-veined, lanceolate in side view, the proximal one c. 5 × longer than wide at maturity, greyish-green, to strongly anthocyanic at maturity, strongly laterally compressed, distinctly keeled, thin, keel to 3/4 the length and marginal veins and sometimes the intermediate veins to 1/2 the length, loosely sericious to villous, between veins sparsely to moderately densely appressed pubescent or occasionally glabrous on the proximal lemma, keel distally weekly scabrous, intermediate veins distinct, not extending to near the margin, margins inrolling below at maturity, very narrowly hyaline above, edges smooth or with a few hooks, apices acute, briefl y hyaline; Paleas shorter than the lemmas by c. 0.5 mm, keels scabrous distally, sometimes weakly so, sparsely puberulent medially or nearly so, glabrous. Flowers chasmogamous; Lodicules c. 0.25 mm long, obscurely lobed; Anthers 1.2-1.4 mm long, vestigial in the upper fl oret. Caryopsis 1.8-2 mm long, strongly laterally compressed, sulcate, honey brown, fi rm, adherent to the lemma and palea, hilum 0.2 mm long, elliptical. 2n = unknown.

Poa tayacajaensis
Distribution. Endemic to the central Andes of Peru. Known from Huancavelica and Huánuco, although the Huánuco specimen is only tentatively placed.
Habitat. Shrublands on Andean slopes at mid elevations. Conservation status. Data insuffi cient. Additional specimens examined. One other specimen appears to represent this species but is too immature to be certain. Th e specimen in question has extravaginal shoots branching from lower culm nodes; PERU: Región HUÁNUCO. Prov. Pachitea: Distr. Chaggla, canyon of the Rio Grande, c. 20 km above confl uence with Rio Huallaga, E of Huánuco c. 44 air km, 1.7 air km SSW of Estación Huacachay Discussion. Th ese plants bear similarities to Poa aequatoriensis but diff er by having lemmas which are generally shorter (2.8-3.7 mm long), pubescent between the veins, and by more densely scabrous sheaths, with more-or-less smooth glabrous throats, ligules generally shorter.  placed his voucher of this form (2038) in P. aequatoriensis. However, among the 20 sheets and the US isotype reviewed of P. aequatoriensis, all have lemmas that are smooth and glabrous between the veins (consistent with the description of Ecuadorian material by Hjorth, 1991), and the keel and marginal veins can be glabrous or sparsely puberulent. Th e species also bears some slight similarity to Poa myriantha Hack. and P. hieronymi Hack. from the Yungas cloudforests, Argentina, that diff er by size of the anthers and ligule being much smaller (anthers <1 mm long, ligules <1 mm long) and overall habit being larger (culms 60-350 cm long with 10-15 internodes, and panicles 20-36 cm long) and glumes having both antrorse and retrorse scabrocities.
Distribution. Restricted to undisturbed areas of Polylepis woodland in hard to access areas throughout the Cordillera Urubamba, Cusco, Peru, at 4390-4802 m. Known from three localities; 1) Cliff ledges of the prominent SW facing cliff face 1.5 km S (170°) of Cancha Cancha village, Huarán. 2) Ledges of the prominent tower known by locals as "Kontorqayku", 5 km NE of Huarán. 3) Ridgeline to the W of Laguna Manalloqsa, Área de Conservación Privada (ACP) Mantanay, 10 km up the valley from Yanahuara in the small valley 3 km E of Laguna Ipsaycocha.
Th is species was found during a large scale ecological study attempting to reconstruct the potential natural vegetation (PNV) and soils of the high-elevation Puna grasslands (see Heitkamp et al. 2014 andSylvester et al. 2014 for pilot studies). In this research, pristine zonal vegetation, only accessible with mountaineering equipment, was compared with surrounding slopes which have been grazed and burnt consistently over millennia (Th ompson et al. 1988;Chepstow-Lusty et al. 1996, 2009Kuentz et al. 2011). Poa urubambensis was a common element in undisturbed Puna vegetation in the Cordillera Urubamba, being found associated with Polylepis Ruiz & Pav. forests from three diff erent sites and also found growing alongside other species new to science, e.g. Bartsia lydiae S.P. Sylvester (2014: 41). Following indicator species analyses, Poa urubambensis has been found as an indicator species for the PNV, due to its frequency and abundance within relict patches of near natural vegetation (Sylvester et al., unpubl. data). Th is species has not been found in accessible, disturbed or secondary, vegetation at similar or lower elevations in the Andes of the Cuzco region, despite a more thorough botanical exploration. Th is may relate to its susceptibility to disturbance from grazing and burning (Sylvester, pers. observation).
Poa supina Schrader Ref: Tovar (1993: 126); Soreng 2007: 529. Ill:  p. 529). Discussion: Th is taxon is highly unlikely to occur in South America with all specimens identifi ed as P. supina, so far encountered, pertaining to Poa annua. Th ese include all specimens encountered from Ecuador and Bolivia (Simon Laegaard pers. communication).  includes P. supina in his treatment of Peru separating it from P. annua and P. infi rma by having glabrous lemmas, or only lightly pubescent on the nerves, and swollen culms. However, the main distinction between P. supina and P. annua, aside from sparser lemma pubescence, is in the length of the anthers with P. supina having anthers (1.2-)1.6-2(-2.5) mm while P. annua has anthers 0.7-1(-1.2) mm. Specimens examined of P. supina from Peru (J. Espinoza 2 [US]) were redetermined as P. annua, but the voucher collections Tovar & Rivas Martinez 7720 and Tovar 7855 accepted by  have not been seen by us. Also see note under Poa infi rma about the origin of P. annua.