A new treefrog from Cordillera del Cóndor with comments on the biogeographic affinity between Cordillera del Cóndor and the Guianan Tepuis (Anura, Hylidae, Hyloscirtus)

Abstract The Hyloscirtuslarinopygion group is a clade of 16 species of large hylids that inhabit cascading Andean streams. They have brown coloration that, in most species, contrasts with bright marks. Herein morphological and genetic evidence is used to describe a new species of the group from Cordillera del Cóndor, a sub-Andean mountain chain that has phytogeographic affinities with the Guianan Tepuis. The new species is characterized by dark-brown coloration with contrasting bright orange flecks and by the presence of an enlarged and curved prepollex protruding as a spine. The new species is closely related to H.tapichalaca and an undescribed species from the southern Andes of Ecuador. The genetic distance between H.hillisisp. n. and its closest relative, H.tapichalaca, is 2.9% (gene 16S mtDNA). Our phylogeny and a review of recently published phylogenies show that amphibians from Cordillera del Cóndor have close relationships with either Andean or Amazonian species. Amphibians do not show the Condor-Guianan Tepuis biogeographic link that has been documented in plants.


Introduction
Hyloscirtus Peters 1882, is a genus of 37 species of treefrogs distributed from Costa Rica to the Andes of Bolivia, Colombia, Ecuador, Peru, and Venezuela (AmphibiaWeb 2018;Frost 2018). They reproduce along streams and share, as a synapomorphy, the presence of wide lateral fringes on fingers and toes (Faivovich et al. 2005 but see Coloma et al. 2012). A well-supported clade within Hyloscirtus is the Hyloscirtus larinopygion species group Coloma et al. 2012;Duellman and Hillis 1990;Rivera-Correa et al. 2016). It is composed of 16 species characterized by large size (SVL < 60 mm) and gray or brown coloration that in many species contrast with bright marks. Species of this group were transferred to the genus Colomascirtus by Duellman et al. (2016). A recent phylogeny showed that the recognition of Colomascirtus rendered Hyloscirtus paraphyletic (Rojas-Runjaic et al. 2018). To maintain taxonomic stability, Colomascirtus was synonymized under Hyloscirtus by Rojas-Runjaic et al. (2018).
The Hyloscirtus larinopygion group is composed of two well-supported clades that replace each other latitudinally with a small area of sympatry in central Ecuador (Almendáriz et al. 2014a). The northern clade is distributed in the Andes of central and northern Ecuador and southern Colombia; the southern clade is distributed in the eastern Andean slopes of central and southern Ecuador and northern Peru (Rivera-Correa et al. 2016). The southern clade is composed of three species: H. condor Almendáriz et al. 2014a, H. tapichalaca (Kizirian et al. 2003), and an undescribed species previously reported as H. lindae (Almendáriz et al. 2014a). Hyloscirtus diabolus Rivera-Correa et al. 2016 is also a putative member of this clade (Rivera-Correa et al. 2016). The four species differ from species in the northern clade by having an enlarged prepollex with the shape of a spine that protrudes below the thumb Rivera-Correa et al. 2016). Recent fieldwork in Cordillera del Cóndor by a field team from the Museum of Zoology, Pontificia Universidad Católica del Ecuador, resulted in the discovery of an undescribed species of the southern clade which also shares a spine-shaped prepollex. Cordillera del Cóndor is a sub-Andean mountain chain with phytogeographic affinities to the Tepuis in the Guiana Region (e.g., Neill 2005). Herein we present morphological and genetic evidence to describe the new species and provide a new phylogeny for the genus Hyloscirtus. We also review recent amphibian phylogenies to explore the existence of biogeographic links between Cordillera del Cóndor and the Guianan Tepuis.

Phylogeny
Phylogenetic relationships were inferred using maximum-likelihood and Bayesian inference. Maximum likelihood analysis were conducted with GARLI 2.0 (Zwickl 2006) using default values, except for the number of generations without topology improvement required for termination (genthreshfortopoterm = 30000) and the maximum number of generations to run and maximum search time (stopgen and stoptime = 5000000). A total of 40 independent searches were run, 20 started from random trees (streefname = random) and 20 from stepwise addition trees (streefname = stepwise). Likelihood values of the 40 searches were within 0.1 likelihood units of each other indicating that all searches converged on similar optimal trees. Support was assessed using 200 bootstrap pseudoreplicates. Bayesian phylogenetic analyses were carried out in MrBayes 3.2.6 (Ronquist et al. 2012). We made four parallel runs of the Metropolis-coupled Monte Carlo Markov for 20 million generations. Each run had five chains, sampled every 1000 generations and with a temperature of 0.1. Convergence into a stationary distribution was measured with software Tracer version 1.4 (Rambaut and Drummond 2007). The search was finished when the average standard deviations of split frequencies was < 0.05 between runs and ESS values were > 200 for all parameters. The consensus tree was generated after discarding 10% of the initial generations as burn-in. Bayesian analyses were carried out at Cipres Science Gateway (available at https//www.phylo.org; Miller et al. 2010).
We also compared qualitative morphological characters between the new species and its closest relatives. Six characters were evaluated: (1) dorsal coloration; (2) ventral coloration; (3) marks on flanks and hidden surfaces of thighs; (4) iris coloration; (5) prepollex condition; and (6) in life, webbing coloration. Life coloration was obtained from color photographs.

Phylogeny and genetic distances
According to PartitionFinder, the best partition strategy consisted of two partitions under model GTR + I + G. Maximum likelihood and Bayesian inference analyses resulted in similar topologies. Four species groups within Hyloscirtus (H. jahni, H. bogotensis, H. armatus, and H. larinopygion group) were recovered with strong support (posterior probability, pp = 1.0 and bootstrap = 100) in both analysis ( Figure 1). However, phylogenetic relationships among these groups were weakly supported (pp < 0.71 and bootstrap < 50), as previously reported Coloma et al. 2012;Guayasamin et al. 2015;Rojas-Runjaic et al. 2018 Table 1 and Appendix 1.
Hyloscirtus condor is sister to a clade conformed by these three species. All together form a strongly supported clade distributed in the eastern slopes of the Andes of central and southern Ecuador and northern Peru (Southern Clade; Figs. 1, 2). The Southern Clade is sister to a clade distributed to the north and confirmed by the remaining species of the Hyloscirtus larinopygion group (Northern Clade; Figs. 1, 3). The Northern and Southern clades have a narrow zone of sympatry in central Ecuador (Figure 2).
Genetic distances between the new species and its closest relatives are characteristic of interspecific distances for the H. larinopygion group. For gene 12S, distances with  Table 2). The genetic divergence between H. hillisi sp. n. and its closest relatives and its unique morphology indicates that it is a new species that we describe below.  Diagnosis. The diagnosis and comparisons are based on one adult female, three adult males, and two subadult females. The new species is diagnosed by the following characters: mean SVL 70.3 mm in adult males (range 66.7-72.3; n = 3), 65.8 mm in one adult female; vomerine odontophores conic-shaped with a gap medially, each process with three to five prominent teeth; supracloacal flap ill-defined; supratympanic fold present; finger webbing formula: I basal II2 --3 -III2½-2IV, toe webbing formula: I2 --2II1 + -2 + III1½-2½IV2½-1 + V; forelimbs hypertrophied in males; enlarged and curved prepollex protruding as a spine in both sexes; fleshy calcar absent; dorsum, flanks, and dorsal areas of limbs dark grayish brown with tiny orange marks varying from abundant to sparse; venter dark grayish brown; iris bronze or yellowish with dark brown reticulation.
Comparisons. Hyloscirtus hillisi is most similar to H. condor, H. diabolus, and H. tapichalaca (Figure 4). They share the presence of an enlarged claw-like prepollex.  Description of the holotype. An adult female (Figs. 5-7), 65.78 mm SVL. Head round in dorsal view, wider than long; snout nearly truncate in lateral and dorsal views; distance from nostril to eye shorter than diameter of eye; canthus rostralis rounded; loreal region slightly concave; internarial region nearly flat; top of head slightly concave; nostrils slightly protruding anterolaterally; lips rounded, not flared; interorbital area slightly convex; eye large, protuberant; diameter of eye 1.85 times diameter of tympanic annulus; supratympanic fold thick, curved, covering posterodorsal edge of tympanum, extending from eye to posterior end of mandible and to shoulder; tympanum rounded; tympanic annulus distinct, rounded, separated from eye by ca. 1.43 times its diameter.
Toes bearing discs broadly expanded, rounded and slightly smaller than those of fingers; relative length of toes I < II < III < V < IV; inner metatarsal tubercle large, oval; outer metatarsal tubercle absent; subarticular tubercles single, round, large, and protuberant; supernumerary tubercles present; toes webbing formula I2 --2II1 + -2 + III1½-2½IV2½-1 + V (Fig. 7). Skin on dorsum, flanks, dorsal surfaces of limbs, throat, chest, dorsal, and inner surfaces of thighs smooth; belly and ventral surfaces of thighs areolate, those of shanks smooth. Cloacal opening directed posteriorly at upper level of thighs, round tubercles below and of vent. Tongue slightly cordiform, widely attached to mouth floor; vomerine odontophores conic-shaped, separated medially, behind level of ovoid choana; each bearing 3-5 vomerine teeth. Additional measurements of the holotype are listed in Table 3.
Color of holotype in preservative. (Figure 6). Dorsal surfaces of head, body, and limbs, including fingers, dark grayish-brown densely stippled with minute, cream flecks. Ventral surfaces of limbs and belly grayish-brown, ventral surfaces of discs, webbing, chest, and throat paler.
Color of holotype in life. ( Figure 5A). Based on digital photographs. Dorsal surfaces same as above except that flecks are bright orange. Ventral surfaces are dark grayishbrown. Ventral pads of digital discs on fingers and toes are gray. Iris is yellowish-cream.
Variation. Dorsal and ventral variation of preserved individuals is depicted in Figure 6. Morphometric variation is shown in Table 3. In preservative, dorsum varies from dark grayish-brown (e.g., QCAZ 68646) in adults to pale grayish-brown (e.g., QCAZ 68647, 68650) or pale gray (e.g., QCAZ 68648) in juveniles and metamorphs. Scattered minutes cream flecks can be present on dorsal surfaces (e.g., QCAZ 68646, 68647). Specimen QCAZ 68647 (juvenile) has cream transverse bars on the dorsal surfaces of the limbs (two to four on the forearm and five to seven on the thigh, shank, and foot). Ventral surfaces vary from pale grayish-brown (e.g., QCAZ 68646) to pale brown or cream (e.g., QCAZ 68648, 68650). Coloration of webbing and discs vary from dark grayish-brown to pale grayish-brown or gray.
In life, (Figure 5), the adult specimens are very similar to the holotype except for the density of bright orange flecks (bright yellow in situ; Figure 11A) on the dorsal surfaces. Background dorsal coloration in juveniles and metamorphs (Figure 8) varies from mottled or uniformly brown (e.g., SC 59268, QCAZ 68650) to light brown (e.g., QCAZ 68648) with or without orange-brown transversal bars on the dorsal surfaces of the limbs. Ventral surfaces vary from dark grayish-brown to cream (e.g., SC 59268). Iris varies from bronze (e.g., SC 59268) to yellowish-cream (e.g., QCAZ 68648).
Tadpole description. The following description is based on a tadpole of series QCAZ 68651 in Stage 25 (Gosner 1960). The specimen was collected in a slow-moving pool along the margins of a stream (Figure 9; 3.5187S, 78.3919W; 1991 m) at the type locality on 7 July 2017. All measurements are in mm. Total length 86.7; body length 29.1 (33.6% of total length). Body ovoid and depressed; width at the level of spiracle 19.2, height at same position 14.7; head width at level of the eyes 17.9; anterior margin of snout uniformly rounded in dorsal view and sloping at level of nares in lateral view; lateral-line system evident with supraorbital, infraorbital, mandibular, angular, postorbital, dorsal body, and ventral body lines. The arrangement of the lateralline system is symmetrical; the supra and infra orbital lines begin at the tip of the snout and join behind the eye, continuing as a single longitudinal line extending along the anterior half of the tail. The dorsal lines extend along the posterior half of the dorsum until reaching the anterior edge of the tail, at the base of the upper fin. The angular line starts behind the orbit and extends longitudinally, contouring the spiracle, to the posterior end of the body, down towards the venter and ending at the base of the vent tube. The postorbital line starts behind the intersection of the supra and infraorbital lines and continues obliquely towards the venter, joining the anteroventral line. The mandibular line originates at the lateral border of the oral disc and runs obliquely until joining the anteroventral line. The posteroventral line forms a V whose vertex is directed towards the midposterior venter ending at the lateral edge of the venter, at the base of the spiracle. The nostrils are ovoid, not protruding and directed anterolaterally, Table 3. Descriptive statistics for measurements of Hyloscirtus hillisi sp. n. Abbreviations: SVL = snoutvent length; FL = foot length; HL = head length; HW = head width; ED = eye diameter; TD = tympanum diameter; TL = tibia length; FEL = femur length. All measurements in mm.  6.8 from tip of snout; internarial distance 8.6. Eyes positioned and directed dorsolaterally; eye length 2.8, eye width 2.5; interorbital distance 9.9. Spiracle sinistral, located at midbody and oriented posterodorsally, inner wall free from body; tube length 2.8, tube width 2.6; spiracular opening directed posterodorsally, diameter 1.6; distance from tip of snout to spiracular opening 22.5. Vent tube medial, opening directed posteriorly; tube length 3.8, tube width 2.6. Tail length 57.5; caudal musculature robust, narrowing gradually until tail terminus. At tail-body junction, tail muscle width 9.6, tail muscle height 11.7; maximum height of tail 17.7. Oral disc located anteroventrally; transverse width 11.6; bordered by two rows of small and rounded papillae; upper jaw sheath forming an arch, unpigmented, transverse width including lateral processes 4.0 (34.4% of transverse width of oral disc); oral apparatus well preserved, showing complete teeth rows. Labial tooth row formula 8(8)/11(1). Only A-8 and P-1 have gaps. Tadpoles were gregarious and fled to the bottom of the pool when disturbed. Color in preservative of tadpoles. In dorsal view, the body is gray, lighter on the tip of snout and towards the base of the tail, grayish cream belly, mouth cream; tail musculature grayish cream with irregular gray spots, upper and lower fins transparent, light gray with irregular dark gray spots.

Adult female (holotype) Adult males (n = 3) Subadult females (n = 2) Juveniles (n = 1)
Color in life of tadpoles. In dorsal view, body brown, including head and snout; in lateral view body dark-brown. Small bronze dots concentrate in the anterior edge of the eye, become diffuse at level of the base of the spiracle. Venter cream, becoming darker medially as result of intestines being dimly visible; oral disc light brown becoming dark brown posteriorly. Iris bronze. Vent tube cream. Muscle tail light brown with gray irregular spots; lower fin transparent cream with a combination of brown and gray irregular spots; upper fin transparent light brown with light brown spots and few scattered dark gray spots. The brown coloration and the pattern of dark gray and brown spots in several individuals is maintained; however, an individual kept in captivity (QCAZ 71182) during 8 months presents an evident change in its coloration, becoming much clearer with a combination of light brown on the back and greenish brown on the flanks; muscles of tail light brown with gray spots; lower fin cream with brown spots, upper fin greenish cream becoming transparent in the distal third with dark brown spots. The differences in coloration after 8 months in captivity may be due to the effects of diet.  Table 4. Seven tadpoles in Stages 25-40 varied in total length, ranging from 57.4 to 101 mm; body length ranged from 20.4 to 34.2 mm; tail length ranged from 37.0 to 67.6 mm. Inter orbital distance from 6.27 to 10.43 mm. Labial tooth row formula varied from 8(8)/11(1) to 7(7)/12(1) (Figure 10).
Etymology. The specific name is a noun in the genitive case and is a patronym for David Hillis, an evolutionary biologist who has made significant contributions to the study of the evolution of amphibians and reptiles. During the 1980s, David Hillis carried out fieldwork in Ecuador that resulted in the discovery of three undescribed species of the H. larinopygion group. In 1990, in collaboration with WE Duellman, he published the first phylogeny for the H. larinopygion group using allozyme data (Duellman and Hillis 1990). Currently he is professor at the University of Texas in Austin.
Distribution and natural history. Hyloscirtus hillisi is only known from two nearby sites (airline distance = 1.7 km) on the slopes of a flattop limestone mountain in the Río Quimi basin, Provincia Zamora Chinchipe, at elevations between 1991 and 2134 m (Figure 2). Biogeographic region is Eastern Montane Forest according to Ron et al. (2018) classification. Vegetation at the type locality ( Figure 11B, C) was dominated by shrubs (1.5 m tall) with sparse trees (10-15 m tall). The ground had cushioned consistency and was covered by roots and bare soil. Mosses and ground-bromeliads were abundant. This type of ground cover is locally known as bamba. Two adults and Table 4. Measurements (in mm) of tadpoles of Hyloscirtus hillisi sp. n. Mean ± SD is given with range in parentheses. Abbreviations: TL (total length), BL (body length), TAL (tail length), TAL/TL (ratio tail length/total length), MHT (Maximum Height of Tail, including dorsal and ventral fins), IOD (inter orbital distance), WOD (transverse width of oral disc), WUJ (transverse width of upper jaw sheath, including lateral processes), WUJ/WOD (ratio width of upper jaw sheath/width of oral disc), TUW (tube transverse width), TUL (tube length spiracle).

Character
Stage 25  one juvenile were found on shrubs next to small streams on the Río Cristalino basin, at an elevation of 2134 m. The tadpoles and juveniles were found in ponds on the margin of Río Cristalino, at an elevation of 1991 m ( Figure 11D). Collections took place in July 2017 and April 2018. The site where the adults were collected is ~500 m from the border between Peru and Ecuador. Therefore, the occurrence of H. hillisi in Peru is almost certain. Conservation status. Hyloscirtus hillisi is only known from two nearby sites in Cordillera del Cóndor. Population size is unknown, but the scant evidence suggests low abundances. In 2017, at the site where the tadpoles and juveniles were found, five hours of nocturnal search by five experienced herpetologists yielded no adults. At the site where the adults were found, ten hours of nocturnal search, for two nights, by two experienced herpetologists, yielded two adults and one subadult. Habitat destruction and fragmentation is evident at a distance of 3.5 km from one of the collection sites (according to Ministerio de Ambiente del Ecuador 2013 map). Cordillera del Cóndor is threatened by large and small-scale mining which has already affected amphibian populations (Valencia et al. 2017). Because of its small known distribution and nearby habitat destruction and mining activities, we suggest to assign H. hillisi to the Critically Endangered category under criteria B1a, b(iii), according to IUCN (2001) guidelines.

Discussion
Our phylogeny is consistent with previous phylogenies of Hyloscirtus (e.g., Coloma et al. 2012;Faivovich et al. 2005;Rojas-Runjaic et al. 2018). The sister clade of the H. larinopygion group appears to be the H. armatus group (e.g., Rojas-Runjaic et al. 2018, Duellman et al. 2016. A close relationship between the H. armatus group and H. larinopygion group is also supported by the shared presence of an enlarged prepollex protruding as a spine in the H. armatus group and in the Southern Clade of the H. larinopygion group. Under Duellman et al. (2016) topology, the absence of the spine in the Northern Clade would result from a secondary loss.
Hyloscirtus hillisi is the second species of the Hyloscirtus larinopygion group to be discovered in Cordillera del Cóndor, a sub-Andean mountain chain with unique geology. While the main Andes are composed of igneous and metamorphic rocks, Cordillera del Cóndor is composed predominantly by sedimentary rocks, specially limestone and sandstone (Neill 2005). Although much younger, Cordillera del Cóndor is geologically similar to the Tepuis in the Guianan region. Remarkably, surveys of the plant communities of Cordillera del Cóndor have recorded at least 10 genera that otherwise are endemic or nearly endemic to the Guianan Tepuis (Ulloa and Neill 2006).
The biogeographic affinity between the biotas of Cordillera del Cóndor and the Guianan Tepuis can be tested with phylogenies. Close relationships between biotas from El Cóndor and the Guianan Tepuis are expected under that biogeographic scenario. However, a review of recently published phylogenies is inconsistent with a Cón-dor-Guianan link. Our phylogeny, for example, shows that both species of Hyloscirtus from el Cóndor are closely related to Andean species from southern Ecuador and northern Peru. Similar results are evident in Pristimantis muranunka (closely related to Pristimantis from the Andes of southern Ecuador; Brito et al. 2017), Pristimantis yantzaza (closely related to Pristimantis from the Andes and adjacent Amazonian lowlands of Peru and Ecuador; Valencia et al. 2017), Excidobates condor (closely related to Excidobates from Cordillera del Cóndor and adjacent Amazonian lowlands; Almendáriz et al. 2012), Centrolene condor (sister to a large clade of Centrolene with species from the Andes of Venezuela, Colombia, Ecuador and Peru; Castroviejo-Fisher et al. 2014), and Chiasmocleis parkeri (closely related to Chiasmocleis from the Amazonian lowlands; Almendáriz et al. 2017). The combined evidence indicates that the biogeographic link between Cordillera del Cóndor and the Tepui region is not discernable in amphibians.
We suspect that the difference in biogeographic pattern observed between plants and amphibians may result from differences in the ecological factors that influence their geographic distribution. In plants, a key factor is soil type (e.g., Clark et al. 1999). The similarity in soil type between Cordillera del Cóndor and the Tepui region (Neill 2005) may explain the biogeographic connection observed in plants. In amphibians, in contrast, edaphic conditions appear to be of minor importance explaining the lack of biogeographic affinity between both regions.
As result of its historic inaccessibility, the organismal diversity of Cordillera del Cóndor is poorly known. During the last two decades, after armed conflicts between Ecuador and Peru ended, roads began to be built and biodiversity surveys became more frequent. These surveys have revealed a large number of unknown species of amphibians, several of which have been recently described (e.g., Almendáriz et al. 2017;Almendáriz et al. 2012;Brito et al. 2017;Brito et al. 2014;Terán-Valdez and Guayasamín 2010;Valencia et al. 2017). Additional expeditions to Cordillera del Cóndor are likely to result in more discoveries since it remains largely unexplored.