A new species of terrestrial-breeding frog (Amphibia, Craugastoridae, Pristimantis) from high elevations of the Pui Pui Protected Forest in central Peru

Abstract We describe a new species of Pristimantis from upper montane forests and high Andean grasslands of the Pui Pui Protected Forest and its close surroundings, Región Junín, central Peru. The description of the new species is based on 34 specimens found at elevations between 3400 and 3936 m a.s.l. Pristimantis attenboroughi sp. n. is characterized by a snout–vent length of 14.6–19.2 mm in adult males (n = 21), 19.2–23.0 mm in adult females (n = 10), and is compared morphologically and genetically with other taxonomically and biogeographically relevant species of Pristimantis. The new species is characterized by having narrow digits that lack circumferential grooves, irregularly shaped, discontinuous dorsolateral folds, and absence of both tympanic membrane and tympanic annulus. The high similarity in morphology between P. attenboroughi sp. n. and members of the Andean genera Phrynopus and Bryophryne provides an example for convergent evolution, and highlights the importance of using molecular data to justify generic assignment. Pristimantis attenboroughi sp. n. is most similar to Phrynopus chaparroi from the Región Junín, suggesting that the generic placement of this species needs to be revised. Phylogenetically the new species belongs to the Pristimantis danae species Group, a clade that includes several Pristimantis species distributed in the montane forests of central Peru, including P. albertus, P. aniptopalmatus, P. ornatus, and P. stictogaster.


Introduction
The Pui Pui Protected Forest (Bosque de Protección Pui Pui, hereafter PPPF, Fig. 1) is located in the Selva Central of Peru and is one of twelve natural protected areas with different levels of legal protection such as national parks, national sanctuaries, and national reserves in the regions of Pasco and Junín (SERNANP 2010). The PPPF, located in the Región Junín, was established in 1985 and covers 60,000 hectares encompassing montane forest (30%) and high Andean grassland (Puna; 70%) habitats (SERNANP 2010). The area protects the upper watershed of several rivers and includes elevations between 1700 and 4500 m a.s.l. (SERNANP 2010).
In 2012-2014, we conducted herpetological surveys in montane forests and Puna of the PPPF to catalog the amphibian and reptile species and to evaluate their conservation status. As a result, we found several new species of frogs (Craugastoridae) as well as new species of lizards (Gymnophthalmidae). All new species were compared morphologically and genetically with other taxonomically and biogeographically relevant taxa mostly from Ecuador, Peru, and Bolivia. Herein we describe a new species of Pristimantis from upper montane and Puna habitats collected between 2012 and 2013.

Materials and methods
Fieldwork. Because of its remote location, the PPPF is difficult to reach and is only accessible through a few entrances located ca. 1-2 days of walking distance from the nearest villages. The upper montane forests and Puna of the PPPF were reached from Toldopampa (11°30'15.4"S, 74°55'32.7"W, 3670 m a.s.l., ca. 45 km SW from Satipo) with the help of local guides by walking in 1.5 days (ca. 11 km airline). In 2012 fieldwork was conducted between May 8 and 21 by EL and RvM, and in 2013 between June 21 and July 8 by EL, J. Moravec, and J.C. Cusi. Amphibians were preserved in 96% ethanol and stored in 70% ethanol. Deposited eggs were stored in 70% ethanol.
Morphological characters. The format for the description follows Lynch and Duellman (1997), except that the term dentigerous processes of vomers is used instead of vomerine odontophores (Duellman et al. 2006), and diagnostic characters are those of Duellman and Lehr (2009). Taxonomic classification follows Hedges et al. (2008), except that we followed Pyron and Wiens (2011) for family placement and Padial et al. (2014) for names of Pristimantis species groups. Sex and maturity of specimens were identified by observing gonads through dissections. Specimens were considered juveniles when gonads were too small to distinguish between sexes. The tympanic region of two specimens (MUSM 31199, NMP6V 75534) was opened to see if a tympanic annulus is present under the skin. We measured the following variables to the nearest 0.1 mm with digital calipers under a stereomicroscope: snout-vent length (SVL, straight length distance from tip of snout to vent), tibia length (TL, distance from the knee to the distal end of the tibia), foot length (FL, distance from proximal margin of inner metatarsal tubercle to tip of Toe IV), head length (HL, from angle of jaw to tip of snout), head width (HW, at level of angle of jaw), horizontal eye diameter (ED), interorbital distance (IOD), upper eyelid width (EW), internarial distance (IND), eye-nostril distance (E-N, straight line distance between anterior corner of orbit and posterior margin of narial opening), and egg diameter. Fingers and toes are numbered preaxially to postaxially from I-IV and I-V, respectively. We compared the lengths of toes III and V by adpressing both toes against Toe IV; lengths of fingers I and II were compared by adpressing the fingers against each other. All drawings were made by EL using a stereomicroscope and a camera lucida. Photographs taken by EL and RvM were used for descriptions of coloration in life. Comparisons of congeners focus on species in similar habitats from Ecuador and Peru and those with close phylogenetic relationships as recovered in our phylogenetic trees. Information on species for comparative diagnoses was obtained from Duellman and Lehr (2009)  Molecular phylogenetic analysis. The phylogenetic position of the new species with respect to other morphologically similar species was determined through analysis of DNA sequence data. This analysis included two mitochondrial genes, 16S rRNA (16S) and 12S rRNA (12S). We used tissue samples from specimens collected in central Peru (Región Junín) to obtain DNA sequences for the new species and several other Pristimantis species (Table 1). Additionally, we downloaded selected sequences of morphologically similar taxa (Bryophyrne, Lynchius, Phrynopus, Oreobates) distributed at high elevations (> 2000 m a.s.l.) from Genbank (Table 1). We included Hamptophryne boliviana, Ischnocnema guentheri, and Bufo melanostictus as outgroup taxa (Padial et al. 2014).
Extraction, amplification, and sequencing of DNA followed protocols previously used for Neotropical terrestrial breeding frogs (Lehr et al. 2005. We used the 16SA (forward) primer (5'-3' sequence: CGCCTGTTTATCAAAAACAT) and the 16SB (reverse) primer (5'-3' sequence: CCGGTCTGAACTCAGATCACGT) to amplify a fragment of the 16S gene (Palumbi et al. 1991), and we employed the following thermocycling conditions to amplify DNA using the polymerase chain reaction (PCR): 1 cycle of 96°C/3 min; 35 cycles of 95°C/30 s, 55°C/45 s, 72°C/1.5 min; 1 cycle 72°C/7 min. Additionally, we used the L25195 (forward) primer (5'-3' sequence: AAACTGGGATTAGATACCCCACTA) and the H2916 (reverse) primer (5'-3' sequence: GAGGGTGACGGGCGGTGTGT) to amplify a fragment of the 12S gene (Palumbi et al. 1991, Vences et al. 2000, and we employed the following thermocycling conditions to amplify DNA using PCR: 1 cycle of 94°C/1.5 min; 35 cycles of 94°C/45 s, 50°C/1 min., 74°C/2 min; 1 cycle 72°C/10 min. We completed the cycle sequencing reactions by using the corresponding PCR primers and the Big-Dye Terminator 3.1 (Applied Biosystems), and obtained sequence data by running the purified reaction products in an ABI 3730 Sequence Analyzer (Applied Biosystems). The newly obtained sequences are deposited in GenBank (Table 1). Geneious R6, version 6.1.8 (Biomatters 2013; http://www.geneious.com/) was used to align the sequences. Within Geneious, we used the MAFFT, version 7.017 (Katoh and Standley 2013) alignment program. Prior to conducting phylogenetic analysis, we used PartitionFinder, version 1.1.1 (Lanfear et al. 2012) to select the appropriate models of nucleotide evolution and used the Bayesian information criterion (BIC) to determine the best partitioning scheme and substitution model for each gene. According to PartitionFinder, the best scheme included one partition combining both 12S and 16S and the best model of nucleotide substitution was GTR + I + Γ. Phylogenetic analysis was done using Maximum Likelihood (ML) approach using RaxML version 8.2.4 (Stamatakis 2006), where the "f-a" function was employed to conduct a bootstrap analysis and search for the optimal likelihood tree. Our analysis included 82 terminals and a 922 bp concatenated alignment that included the 16S and 12S dataset. The GTR + I + Γ model of nucleotide substitution was used to perform 200 trees searches; node support was assessed using 1000 bootstrap replicates. Additionally, we used the R package 'APE' (Paradis et al. 2004) to estimate uncorrected p-distances (i.e., the proportion of nucleotide sites at which any two sequences are different).

Molecular phylogenetic analysis.
The Maximum Likelihood (ML) tree ( Fig. 2) was generally congruent with a previous molecular phylogeny (Padial et al. 2014) and supported the distinctiveness of the new species from other closely related taxa. Placement of Pristimantis attenboroughi sp. n. in the genus Pristimantis Jiménez de la Espada, 1871 was strongly supported and, based on the available data, the new species is most closely related to P. albertus Duellman and Hedges, 2007, P. aniptopalmatus (Duellman and Hedges, 2005), P. ornatus (Lehr, Lundberg, Aguilar, and von May, 2006), and P. stictogaster (Duellman and Hedges, 2005) (Fig. 2). Table 2 compares uncorrected pdistances of a 542 bp (including gaps) fragment of the 16S mitochondrial rRNA gene of Pristimantis species included in our analyses. The lowest distance occurs between the new species and P. aniptopalmatus (uncorrected p-distance 4.3 %) while the uncorrected p-distances between the new species and the other three species in the same clade of the Pristimantis danae species Group (P. albertus, P. ornatus, P. reichlei Padial and De la Riva, 2009, P. rhabdolaemus [Duellman, 1978a], P. stictogaster [Duellman and Hedges, 2005], P. sagittulus [Lehr, Aguilar, and Duellman, 2004], P. toftae [Duellman, 1978b]) vary between 5.2 to 11.8 %.        Generic placement. We assign this species to Pristimantis based on our molecular data (Fig. 2).

Pristimantis attenboroughi
Diagnosis. A new species of Pristimantis assigned to the danae species Group having the following combination of characters: (1) Skin on dorsum shagreen with low scattered tubercles, skin on flanks tuberculate, skin on venter areolate; discoidal fold absent, thoracic fold present; irregularly shaped, discontinuous dorsolateral folds present; (2) tympanic membrane and tympanic annulus absent; (3) snout short, rounded in dorsal and in lateral views; (4) upper eyelid without enlarged conical tubercles; EW shorter than IOD; cranial crests absent; (5) dentigerous processes of vomers present; (6) males without vocal slits, nuptial pads absent; (7) Finger I shorter than Finger II; tips of digits narrow, rounded, lacking circumferential grooves; (8) fingers without lateral fringes; (9) small conical ulnar and tarsal tubercles present; (10) heel with a small conical tubercle; inner tarsal fold usually absent; (11) inner metatarsal tubercle ovoid, 1.5 times as large as outer; outer metatarsal tubercle small, rounded; vie low supernumerary plantar tubercles; (12) toes without lateral fringes; basal toe webbing absent; Toe V longer than Toe III; tips of digits narrow, rounded, lacking circumferential grooves, toe tips slightly smaller than those on fingers; (13) in life, dorsal ground coloration pale or dark gray, reddish brown or brownish olive with dark gray scattered flecks, some with X-shaped mark on scapular and ill-defined diagonal bars on flanks; dark grayish-brown canthal and supratympanic stripes usually present; groin dark gray or pale reddish brown with a pale red to pink tint in some; venter dark gray, pale gray, grayish brown or pale grayish green and in some dark gray mottled; iris pale grayish green with fine black vermiculation and brownish-orange horizontal streak across pupil and lower half of iris; (14) SVL in adult males 14.6-19.2 mm (n = 21), in adult females 19.2-23.0 mm (n = 10).
Members of the Pristimantis orestes species Group are terrestrial and inhabit high elevations in southern Ecuador and in Peru (Duellman and Lehr, 2009) and have narrow digits, and only one of the 17 species (Guayasamin and Artega 2013) lacks circumferential grooves (P. simonsii), and only two (P. seorsus, P. simonsii) lack a tympanum. Furthermore P. attenboroughi is phylogenetically distant from members of this group which is considered to be not monophyletic (Duellman and Lehr 2009, Fig . 2).
Among the three other new species of Pristimantis from the upper montane forests and Puna of the PPPF, only Pristimantis sp. n. E lacks circumferential grooves and a tympanum. However, P. attenboroughi and P. sp. n. E both differ regarding other morphological traits, coloration, and genetically.
Description of the holotype. Head about as long as wide; head length 39.7% of SVL; head width 38.6% of SVL; cranial crests absent; snout short, rounded in dorsal view, rounded in lateral view (Fig. 3A, B); eye-nostril distance 70% of eye diameter; nostrils slightly protuberant, directed dorsolaterally; canthus rostralis short, rounded in lateral view, weakly concave in dorsal view; loreal region concave; lips rounded; outer margin of upper eyelid each with few slightly enlarged conical tubercles; upper eyelid width 51.9% of IOD (see photo in life Fig. 3); supratympanic fold short and broad, extending from posterior margin of upper eyelid slightly curved to insertion of arm; tympanic membrane and annulus absent; distinct conical postrictal tubercles present bilaterally. Choanae small, ovoid, not concealed by palatal shelf of maxilla; dentigerous processes of vomers positioned posterior to level of choanae, oblique, narrowly separated; tongue long, oval, about three times as long as wide, not notched posteriorly, posterior half free.
Skin on dorsum shagreen with low scattered tubercles, skin on flanks tuberculate, irregularly shaped, discontinuous dorsolateral folds present extending from posterior level of tympanic area to level of hind limb insertion; skin on throat, chest, and belly areolate; discoidal fold absent, thoracic fold present; cloacal sheath short.
Outer ulnar surface each with a row of four minute low tubercles; palmar tubercle bifid; thenar tubercle ovoid; subarticular tubercles well defined, most prominent on base of fingers, round in ventral view, subconical in lateral view; supernumerary tubercles indistinct; fingers short and stout lacking lateral fringes, Finger I shorter than Finger II; tips of digits of fingers narrow, round, lacking circumferential grooves (Fig. 4A).
Hind limbs short, slender, tibia length 40.2% of SVL; foot length 41.3% of SVL; dorsal surfaces of hind limbs tuberculate; inner surface of thighs smooth, posterior surfaces of thighs tuberculate, ventral surfaces of thighs areolate; heels each with a small conical tubercle; outer surface of tarsus with few scattered minute low tubercles; inner tarsal fold absent, but small tubercle proximal to metatarsal tubercle; inner metatarsal tubercle ovoid, one and a half times the size of round outer metatarsal tubercle; subarticular tubercles well defined, round in ventral view, subconical in lateral view; few plantar supernumerary tubercles, about one third the size of subarticular tubercles; toes without lateral fringes; basal webbing absent; tips of digits narrow, round, less expanded than those on fingers, lacking circumferential grooves; relative length of toes: 1<2<5<3<4; Toe V slightly longer than Toe III (tip of digit of Toe III and Toe V not reaching distal subarticular tubercle on Toe IV; Fig. 4B).
Coloration of the holotype in life (Fig. 3). The dorsal ground coloration is pale reddish brown with few dark brown flecks; narrow dark brown canthal and supratympanic stripes; flanks pale reddish brown with dark brown flecks forming irregularly shaped diagonal bars; groin and anterior surfaces of thighs reddish brown with dark brown flecks and pale reddish tint; chest, belly, and ventral surfaces of thighs dark grayish brown, throat pale reddish brown and pale gray mottled; palmar and plantar surfaces, and fingers and toes dark grayish brown; iris pale grayish green with fine black vermiculation and brownish-orange horizontal streak across pupil and lower half of iris.
Coloration of the holotype in preservative. The dorsal ground coloration is pale brown with few dark brown flecks; narrow dark brown canthal and supratympanic stripes; flanks pale brown with many dark brown flecks forming irregularly shaped diagonal bars; groin and anterior surfaces of thighs brown with dark brown flecks; chest, belly, and ventral surfaces of thighs dark brown, throat pale brown and pale gray mottled; palmar and plantar surfaces, and fingers and toes dark brown; iris pale gray.
Juveniles (MUSM 31187, 31990, 31200, Fig. 7) have a paler coloration (yellowish to reddish brown) with contrasting dark brown flecks and distinct canthal and supratympanic stripes. All have the iris pale grayish green with fine black vermiculation and brownish-orange horizontal streak across pupil and lower half of iris, and usually a narrow vertical dark gray streak from pupil through middle of lower iris.  Etymology. We dedicate this species to Sir David Frederick Attenborough in honor for his educational documentaries on wildlife, especially on amphibians (e.g., Life in Cold Blood, Fabulous Frogs), and for raising awareness about the importance of wildlife conservation. The specific epithet is used as noun in apposition.
Distribution, natural history, and conservation status. Pristimantis attenboroughi is known from six localities inside the PPPF (Puna of Quebrada Tarhuish at Laguna Udrecocha, Fig. 8A; upper montane forest of Quebrada Tarhuish on the left bank "Shiusha" of Antuyo River; Antuyo; Antuyo Bajo; Hatunpata, and Trancapampa, Figs 8B, 9) and from two outside the PPPF (upper part of Quebrada Tasta close to Laguna Luichococha; in Polylepis forest of first mountain peak next to trail from Tasta to Tarhuish), and is distributed at elevations between 3400 and 3936 m a.s.l., Fig. 9. The type locality (Figs 8A,9), upper part of Quebrada Tarhuish, on the east side of Laguna Udrecocha at 3936 m a.s.l., belongs to the Puna ecoregion (Brack 1986). The vegetation consists of Peruvian feather grass (Stipa ichu), mosses, and small bushes. The holotype was found inside moss in the afternoon on 17 May 2012. No sympatric anurans were found at the type locality. At the upper montane forest of Quebrada Tarhuish on the left bank "Shiusha" of Antuyo River, P. attenboroughi was found deep inside large moss layers. Sympatric anurans are Gastrotheca griswoldi (MUSM 31193), A female Pristimantis attenboroughi (MUSM 31980, Fig. 8C) guarding 20 eggs was found at Hatunpata inside moss, 3710 m a.s.l., on 28 June 2013. The eggs were pale cream colored and had an average diameter of 3.5 ± 0.1 mm (3.3-3.6 mm, n = 20).
The IUCN Red List criteria (IUCN 2001) consider that if a species occurs in fewer than 10 threat-defined locations and the extent of occurrence (EOO) is < 20,000 km 2 , it should be classified as Vulnerable or Endangered. Pristimantis attenboroughi is known from seven localities distributed in the PPPF and its buffer zone (Fig. 9), with an estimated EOO of 66.54 km 2 . As such, this new species might be classified as Vulnerable if we take into account these criteria. However, given that the PPPF may host a greater number of locations and most of them are inside the protected area, we propose that Pristimantis attenboroughi should likely be categorized as Near Threatened (NT).
Given that the known distribution of Pristimantis attenboroughi overlaps with the PPPF, a substantial portion of the habitat of this species is formally protected. However, other factors such as fungal infections, climate change, pollution, and man-made fires (used to expand grazing areas for livestock) continue to be threats for many Andean amphibians even inside protected areas (Catenazzi and von May 2014).

Discussion
When we encountered the first specimen of Pristimantis attenboroughi in the field both of us were sure that we had found a new species of Phrynopus because of its overall morphological appearance: most species in the genus Phrynopus usually lack tympanum, have narrow digits without circumferential grooves and are distributed at high elevations. However, following an integrative taxonomy approach that included molecular and morphological data, we realized that Pristimantis attenboroughi is not a Phrynopus species. Our analysis also revealed that Pristimantis attenboroughi is not closely related to other Pristimantis species that have narrow digits (e.g., members of the P. orestes species group), an assumption that could have been made if only morphological data were available. In other words, Pristimantis attenboroughi displays convergence that easily could have led to an incorrect generic assignment. Pristimantis attenboroughi is morphologically most similar to Phrynopus chaparroi (Mamani and Malqui 2014) and we assume that the latter species might belong to Pristimantis and to the danae species group. Thus, molecular data are needed to determine whether the current generic placement of Phrynopus chaparroi is correct.
With Pristimantis attenboroughi, seven species of Pristimantis are known from the Puna (> 3000 m a.s.l.) of Peru. Of these, six occur in northern Peru (P. atrabracus [Duellman and Pramuk, 1999], 2963-3330 m a.s.l.; P. bellator Lehr, Aguilar, Siu-Ting, Jordán, 2007, 1900-3100 m a.s.l.; P. cordovae [Lehr and Duellman, 2007b], 3400-4100 m a.s.l.; P. mariaelenae Venegas and Duellman, 2012, 3596 m a.s.l.; P. pinguis [Duellman and Pramuk, 1999], 3000-3916 m a.s.l.; P. stipa Venegas and Duellman, 2012, 3596 m a.s.l.), and only one species in central Peru (P. attenboroughi, 3400-3936 m a.s.l.), Lehr 2009. Navarrete et al. (2016) pointed out the disparity in species richness of Pristimantis at high elevation between Ecuador (18 species of Pristimantis) and Peru (5 species of Pristimantis). Whilst the Páramo in Ecuador is more humid than the drier Puna in Peru, it is likely that, besides climatic differences between the two regions, the lower species richness of Pristimantis in the Puna of Peru is an artifact of lower survey effort and the presence of other high-elevation clades not present in Ecuador. Thus, we hypothesize that the occurrence of the genus Phrynopus at high elevations (28 species from elevations between 2200-4400 m a.s.l., Amphibi-aWeb 2016, Duellman and Lehr 2009) in central Peru might restrict the number of niches available for Pristimantis at high elevations.
Additional new species of terrestrial-breeding frogs from montane forests and Puna of the PPPF will be described in the near future.