Beauty is in the eye of the beholder: Cruciform eye reveals new species of direct-developing frog (Strabomantidae, Pristimantis) in the Amazonian Andes

We describe a new species of frog from the eastern slopes of the Andes in central Peru. Pristimantis sira sp. nov. has a distinctive crossing mark on the iris and no tympanum. The new species is closely related to P. antisuyu Catenazzi & Lehr, 2018, P. cruciocularis Lehr, Lundberg, Aguilar & von May, 2006, and P. erythroinguinis Catenazzi & Lehr, 2018, but is easily differentiable by lacking colour blotches on groins. Pristimantis sira sp. nov. inhabits the mountain forests from 1550 to 2200 m a.s.l., inside a national reserve threatened by illegal mining.


Introduction
Expeditions to remote places in the eastern slopes of the Peruvian Andes have shown that unexplored neotropical mountains are sources of unknown and endemic herpetofauna (Duellman et al. 2014;Catenazzi et al. 2015;Echevarría et al. 2015;Chávez and Catenazzi 2016;Chávez et al. 2017;Lehr and Moravec 2017;Lehr and von May 2017;Lehr et al. 2017aLehr et al. , 2017b. Frogs are one the most diverse herpetofaunal orders, particularly in Andean ecosystems where endemism is high (Swenson et al. 2012). Furthermore, many of the discoveries of new amphibians are frogs (Duellman et al. 2014;Catenazzi et al. 2015;Chávez and Catenazzi 2016;Lehr and Moravec 2017;Lehr and von May 2017;Lehr et al. 2017aLehr et al. , 2017b. Among recently named frogs, most of the new Peruvian Andean species belong to the genus Pristimantis Jiménez de la Espada, 1870, which is the most diverse clade in the Neotropics (Padial et al. 2014). Over the last ten years, herpetologists have named seven new species of Pristimantis from the eastern Andes of central Peru (Shepack et al. 2016;Lehr and Moravec 2017;Lehr and von May 2017;Lehr et al. 2017a, b), showing that these mountains host a still unknown anuran diversity, and that Pristimantis are one of the most diverse clades of these amphibian communities.
El Sira Communal Reserve is located on the eastern slopes of the Andes in central Peru, and protects about 616413 hectares of primary forest. El Sira is the highest cordillera adjacent to the Ucayali River, going from 200 m to 2200 m a.s.l., and is bordered by the Pachitea river (which eventually flows into the Ucayali), further isolating the El Sira mountains as the eastern branch of the Andes in the Ucayali basin (see Figure 5). Because of the hard access and rough topography, El Sira is likely one of the most unexplored places of Peru. Only a handful of herpetological expeditions (Duellman and Toft 1979;Aichinger 1991;Whitworth et al. 2016aWhitworth et al. , 2016b have reached these forests, resulting in the description of four anuran species (Duellman and Toft 1979;Aichinger 1991;Lötters and Henzl 2000), all of them endemic from El Sira.
Nevertheless, none of the four endemic frogs described so far is a Pristimantis, in contrast to other mountain chains in Peru where most of the recently described species belong to this genus (Shepack et al. 2016;Lehr and Moravec 2017;Lehr and von May 2017;Lehr et al. 2017aLehr et al. , 2017b. The systematics of Pristimantis is challenging because it often requires an integrative taxonomy approach for species delimitation, and because genetic sequences are not available for many species.

During fieldwork we performed in El Sira Communal
Reserve from 2013 to 2014, we collected a series of unidentified Pristimantis frogs. Genetic and morphological analysis revealed that these specimens belong to an unnamed species, which we describe below.

Material and methods
We follow Lynch and Duellman (1997) for the format of diagnosis and description of the new species. For systematics of Strabomantidae we follow Hedges et al. (2008), Blackburn andWake (2011) andPadial et al. (2014). We collected specimens during the night while conducting Visual Encounter Surveys (Crump and Scott Jr 1994). We euthanized specimens with an 8% benzocaine solution, fixed them in 10% formalin, and stored them in 70% ethanol. We deposited all specimens in the herpetological collection of Centro de Ornitología y Biodiversidad (CORBIDI). The Servicio Nacional de Areas Protegidas de Peru issued collecting permit RJ N° 003-2014-SERNANP-RCS-JEF029-2016-SERFOR-DGGSPFFS. We measured the following variables to the nearest 0.1 mm with digital callipers under a stereoscope, as described in Duellman and Lehr (2009): snout-vent length (SVL); eye-nostril distance (E-N); head length (HL); head width (HW); interorbital distance (IOD); internarial distance (IND); tibia length (TL); foot length (FL); eye diameter (ED); upper eyelid width (EW). Fingers and toes are numbered preaxially to postaxially from I-IV and I-V respectively. We determined comparative lengths of toes III and V by adpressing both toes against Toe IV; lengths of fingers I and II were determined by adpressing the fingers against each other. Specimens were sexed based on external sexual characteristics (e.g., presence of vocal sacs in males), and internal dissection of the gonads. Photographs were taken in the field by GC, and in the laboratory by LAGA. We used these photos for descriptions of coloration in life and in preserved condition, respectively. In addition to the type series of the new species, we examined specimens of related congeners (Suppl. material 1) or obtained morphological data from the original description for diagnostic comparisons.
We downloaded sequences of closely related (on the basis of BLAST results for 16S) or morphologically similar, congeneric species and of two species of Oreobates (used as outgroup taxa) from GenBank (Suppl. material 2). We used Geneious R11, version 11.1.5 (Biomatters, http://www.geneious.com/) to assemble pairend reads, to generate a consensus sequence, to align our novel and GenBank sequences with the alignment program MAFFT v7.017 (Katoh and Standley 2013), and also to concatenate sequences of the three genes. We trimmed aligned sequences to a length of 571 bp for 16S, 678 for COI, and 645 bp for RAG1 (total length 1894 bp), after removing regions of ambiguous alignment for 16S. Our analysis included 56 terminals. We used PartitionFinder, v. 1.1.1 to select the best partitioning scheme and substitution model for each gene using the Bayesian information criterion (BIC). The best partitioning scheme included six partitions: subset 1 for 16S with the model GTR + I + G, subset 2 for the first codon position of COI with SYM+I+G, subset 3 for the second codon position of COI with TVM+I, subset 4 for the third codon position of COI with TIM+G, subset 5 for the first and second codon positions of RAG1 with HKY+G, and subset 6 with the third codon positions of RAG1 with K81+G.
We inferred phylogenetic relationships with Maximum Likelihood (ML) inference. We conducted the analysis with IQ-TREE v1.6.12 (Nguyen et al. 2015) using our concatenated dataset, the partitioning and substitution models determined by PartitionFinder, and the ultrafast bootstrap method (10000 bootstrap alignments). We also estimated genetic distances for the 16S rRNA mitochondrial fragment to provide further support of species delimitation. Although there is not a set threshold for delimiting species, Fouquet et al. (2007) suggested that 3% distance for 16S is a reasonable criterion to identify putative new species. The benefit of using 16S is that this fragment is the most frequently sequenced marker for anuran taxonomy (Fouquet et al. 2007;Padial et al. 2009;Vences et al. 2005), including for species of Holoadeninae (Hedges, Duellman & Heinicke, 2008). We estimated uncorrected p-distances (i.e., the proportion of nucleotide sites at which any two sequences are different) with the R package "ape" (Paradis et al. 2004), and uploaded the table to Figshare (https://doi.org/10.6084/ m9.figshare.13640615).
The electronic version of this article in portable document format will represent a published work according to the International Commission on Zoological Nomenclature (ICZN), and hence the new names contained in the electronic version are published under that Code from the electronic edition alone. This published work and the nomenclatural acts it contains have been registered in ZooBank, the online registration system for the ICZN. The ZooBank LSIDs (Life Science Identifiers) and the associated information can be viewed through any standard web browser at http://zoobank.org/117AA0B4-99A2-4F4F-85A9-B9FAE8E47BB1.

Generic placement
Our study (Figure 1) supports the generic placement of the new species and its distinctiveness with respect to similar species. According to phylogenetic analyses, the closest related species are P. antisuyu Catenazzi and Lehr 2018, P. cruciocularis Lehr, Lundberg, Aguilar, von May 2006, and P. erythroinguinis Catenazzi and Lehr 2018. These four species form a well-supported clade of mostly Andean foothill and montane forest small Pristimantis with cruciform eyes and, except for P. sira sp. nov., yellow or red coloration on groin and belly. The close relationships among P. antisuyu, P. cruciocularis, P. erythroinguinis and P. sira sp. nov. are supported by 16S rRNA genetic distances (Figshare https://doi.org/10.6084/ m9.figshare.13640615), which are smallest for pairwise comparisons among these four species (range 7.2-10.2%) than they are for comparisons with other species of Pristimantis (all distances >10.2%).

Description
Pristimantis sira sp. nov. Diagnosis. The new species is diagnosed by the following combination of characters: (1) skin on dorsum finely shagreen with a few scattered subconical tubercles, that on venter areolate, W-shaped scapular fold present, discoidal fold absent, dorsolateral folds absent; (2) tympanic membrane and tympanic annulus absent, supratympanic fold absent; (3) snout acutely rounded from dorsal view, moderate in length and rounded from lateral view, canthus rostralis weakly concave in dorsal view, angular in lateral view, loreal region concave, rostral papilla or keel absent; (4) upper eyelid bearing two or three sub conical small tubercles, narrower than IOD, cranial crests absent; (5) dentigerous process of vomers absent; (6) males with vocal sacs and vocal slits, nuptial excrescences absent; (7) heels lacking tubercles; (8) finger I shorter than finger II, tips of digits expanded, bearing circumferential grooves, discs about 1.5 times wider than digits in fingers I, II and III, finger IV bearing a rounded disc about twice as wide as its digit; (9) fingers with narrow lateral fringes; (10) antebrachial tubercle absent; (11) ulnar and tarsal tubercles absent (12) inner metatarsal tubercle oval twice as long as round outer metatarsal tubercle, low supernumerary plantar tubercles at the base of toes II, III, IV and V; (13) toes without lateral fringes, webbing absent, toe V longer than toe III; (14) in life, dorsum yellowish-brown, dark brown or olive brown with dark transversal bands; interorbital bar dark brown; canthus rostralis paler than loreal region and dark bordered; dark labial bars present; throat, chest, and belly dark brown or dark grey with scattered white flecks; groins, posterior surfaces of thighs, and shanks dark brown; iris copper yellow with a vertical black line and dark reticulations, black pupil surrounded by a copper orange ring; (15) SVL 12.9-14.7 mm in males; 19.0-20.4 mm in females.
Description of the holotype. An adult female (COR-BIDI 14430; Figure 2A-E, 3A, B) with a SVL of 19.6 mm, head as wide as long ( Fig. 2A, B); snout acutely rounded in dorsal view and rounded in lateral view, short (eyenostril distance 9.8% of SVL); canthus rostralis distinct in lateral view; loreal region concave; nostrils protuberant, directed anteriorly; interorbital area flat, slightly broader than upper eyelid (upper eyelid width 89% of interorbital distance); cranial crests absent; upper eyelid bearing two small subconical tubercles; tympanic membrane absent; tympanic annulus absent (Fig. 2E); postrictal tubercles absent. Choanae small, rounded, not concealed by palatal shelf of maxillary; tongue longer than wide and granular, dentigerous processes of vomers absent. Skin texture on dorsum and flanks finely shagreen; dorsolateral folds absent; venter areolate; thoracic fold present, discoidal fold absent, cloacal sheath absent. Forearm slender; ulnar tubercles absent, ulnar fold absent; radio-ulnar length 22% of SVL; fingers with narrow lateral fringes; relative lengths of fingers I ≤ II < IV < III; palmar tubercle bilobed, thenar tubercle oval (Fig. 2C); subarticular tubercles round, prominent; supernumerary palmar tubercles present at the base of all fingers; disc cover finger I barely expanded, those of fingers III and IV extensively expanded (Fig. 2C), outer discs of fingers as wide as those of toes; discs covered with elliptical ventral pads defined by circummarginal grooves. Hind limbs slender; tibia length 48% of SVL; foot length 94% of tibia length; tarsal fold absent, tarsal tubercles absent; heel lacking tubercles; toes without lateral fringes; subarticular tubercles round, prominent; inner metatarsal tubercle oval, about 2.4 times the size of subconical outer tubercle (Fig. 2D); supernumerary plantar tubercles low at the base of all toes; discs covers slightly expanded; toes with defined pads; discs pads nearly elliptical; relative length of toes I < II < III < V < IV; tip of toe V reaching proximal border of distal subarticular tubercle IV; tip of toe II reaching proximal border of distal subarticular tubercle of Toe III (Fig. 2D).
Coloration in life (Fig. 3A, B). Dorsum, flanks, and dorsal surface of limbs are yellowish-brown with dark brown transversal bands and reddish-orange tubercles; dorsal surface of head bearing a dark brown interorbital bar; canthus rostralis yellowish-brown, loreal region dark brown, labial bars dark brown, and white minute flecks present on the upper labial region. Throat, chest, belly, ventral surface of limbs, hands and feet dark brown with white flecks; groins, posterior surface of thighs, and posterior surface of shanks dark brown. Anterior surface of thighs is dark brown. Iris copper yellow with a vertical black line and dark reticulations, black pupil surrounded by a copper orange ring.
Coloration in preservative ( Fig. 2A, B). Despite the skin on dorsum suffering damage, the coloration is similar to coloration in life, except that dark brown coloration turned yellowish-brown. The venter, limbs and ventral surfaces of hands and feet are yellowish-brown with creamy-white flecks, discs on hands and feet dark grey; iris gray.
Intraspecific variation. Dorsal coloration from yellowish brown to dark brown in juvenile CORBIDI 13952 ( Figure 3G, H) and olive brown in female CORBIDI 14429 ( Figure 3E, F). Moreover, CORBIDI 13952 has a pattern consisting of a middorsal creamy white line going from the tip of the nose to the cloaca, a pale interorbital bar, and a thin creamy white mid-throat line going from the chin to the distal edge of the throat. Furthermore, male CORBIDI 14433 ( Figure 3C, D) has a yellowish-brown throat, a greyish-white belly, and yellowish-brown coloration on the ventral and posterior surface of thighs. Table 1 reports variation in measurements and proportions.  Etymology. The species epithet "sira" is a noun in apposition, referencing El Sira Communal Reserve, a protected area established in 2001, containing the type locality of the new species. El Sira also protects one of the last large extensions of primary mountain forests in Central Peru.

Discussion
Previous genetic analyses have shown that a group of Pristimantis frogs characterized by cruciform eyes are closely related, and include P. antisuyu, P. cruciocularis, and P erythroinguinis (Catenazzi and Lehr 2018). Our study suggests that P. sira sp. nov. belongs to the same group. Likewise, these species are forming a clade into the larger P. platydactylus -P. llojsintuta complex (Padial et al. 2009) (see Figure 1). We propose to name the clade "Pristimantis cruciocularis group" following Article 61 (Principle of typification) of the International Code of Zoological Nomenclature 1999, which recommends using the oldest bearing-type species belonging within the defined taxonomic boundaries of the group. Additionally, the cruciform eye is a character present in two other species that are not closely related to the P. cruciocularis group: P. ashaninka and P. altamazonicus. Pristimantis ashaninka inhabits mountain forests, whereas P. altamazonicus is an Amazon lowlands dweller. Our phylogenetic analyses suggest that P. ashaninka is not closely related to P. sira (but only 16S rRNA fragments are available for P. ashaninka). Further analyses and gene sampling are necessary to confirm the relationships of montane species with cruciform eye.
We compared measurements and proportions of P. antisuyu, P. cruciocularis, and P. erythroinguinis, all montane species morphologically similar to P. sira (Table 1), concluding that they appear very similar in measurements. We also realised that proportions published in the table 3 of the original description of P. cruciocularis (Lehr et al. 2006) seem to be incorrect: males proportions correspond to females and viceversa. Thus, in order to help future research and publish confident morphological data, we present a new measurement dataset of P. cruciocularis, including populations formerly assigned to P. flavobracatus Lehr, Lundberg, Aguilar & von May, 2006, a junior synonym of P. cruciocularis sensu Catenazzi and Lehr (2018).
We have observed illegal gold mining activity near the type locality of P. sira sp. nov. Gold mining involves the establishment of mining camps and the subsequent clearing of big areas, which would affect the habitat of the new species. Following the IUCN criteria, and considering the aforementioned threat and that P. sira has been recorded in only two localities with a possible EOO not wider than 6000 km 2 (which is the entire extension of the El Sira Communal Reserve, SERNANP 2020), we suggest this species to be placed into the Vulnerable category of the IUCN red list.