Description and phylogenetic relationships of a new species of treefrog of the Osteocephalus buckleyi species group (Anura: Hylidae)

ABSTRACT The Osteocephalus buckleyi species group is widely distributed in primary and secondary forests of the Amazon Basin and Guiana Region. Based on integrative analysis, including morphological and genetic data, we estimate the phylogenetic relationships and species boundaries among populations of the Osteocephalus buckleyi group from the Ecuadorian Amazon, focusing on the O. verruciger-O. cannatellai species complex. Our results uncovered the existence of one confirmed candidate species from Sangay National Park and one unconfirmed candidate species. Here, we describe the new species which is morphologically and ecologically distinct from other Osteocephalus species. The new species is unusual because it shows quite distinct morphology, but low genetic distances compared to its closest relatives. Zoobank.org registration LSID: urn:zoobank.org:pub:01F86A33-2D07-4D9E-A8A2-FF35622E3DB9


Introduction
The genus Osteocephalus Steindachner [1] is widely distributed in South America, from Venezuela, Guianas, and the Amazon basin to central Bolivia, and from the eastern Andean slopes of Bolivia to Colombia [2]. It is distributed from 0 to 2000 m.a.s.l., in primary and secondary forests [3]. Osteocephalus comprises a total of 27 recognized species [4] classified into five species groups [5]: O. alboguttatus, O. buckleyi, O. leprieurii, O. planiceps, and O. taurinus.
The Osteocephalus buckleyi species group was the first formally defined within Osteocephalus. It was proposed by Cochran and Goin [10]; 1 year later, Duellman and Trueb [3] published the first review of the genus. The most recent reviews of the group carried out by Ron et al. [9] and Jungfer et al. [5] revealed that it still has unresolved taxonomic issues due, to among others, the presence of undescribed species because of its similar morphology and the presence of cryptic species. In this study, we evaluated the species limits within O. verruciger and O. cannatellai species complex, and describe a new species discovered during an expedition to Sangay National Park, in the Amazonian slopes of the Andes of Ecuador. We also present a new phylogeny for Osteocephalus, focusing on the O. buckleyi species group. The species description is based on an integrative analysis of genetic and morphological data.

Morphological analyses
We examined alcohol-preserved specimens of the Osteocephalus buckleyi group (Appendix A) deposited at the herpetological collection of Museo de Zoología at Pontificia Universidad Católica del Ecuador (QCAZ). We included the type material of O. cannatellai (QCAZ 49572) and one syntype of Hyla verrucigera (ZMB 16589) from the Museum für Naturkunde, Leibniz Institute for Evolutionary and Biodiversity Research in Berlin, Germany.
The format for species diagnosis and description follows Duellman and Trueb [3]. Notation for hand and foot webbing is based on Savage and Heyer [18]. Sex and reproductive status were determined by gonad inspection. Morphometric measurements were taken with a digital caliper (to the nearest 0.01 mm). Adult specimens were measured for the following variables [19]: snout-vent length (SVL); head length (HL); head width (HW); tympanum diameter (TD); femur length (FEL); tibia length (TL); foot length (FL); and eye diameter (ED) ( Tables 1 and 2). Color data in life were based on digital photos. We analyzed the SVL for adult males and females to determine size differences CONTACT Marcel A. Caminer marcelcaminer@gmail.com between species with a Student's t-test using the program JMP® 9.01 [20].

DNA extraction, amplification, and sequencing
Genomic DNA was extracted from muscle and liver tissue preserved in 95% ethanol using standard phenol-chloroform extraction protocol [21]. Polymerase chain reaction (PCR) was used to amplify partial mitochondrial genes of 12S rRNA, 16S rRNA, NADH dehydrogenase I (ND1), and cytochrome c oxidase I (CO1  [27], Salducci et al. [28], and Wiens et al. [29]. As outgroup, we included sequences of O. yasuni, O. leprieurii, and O. fuscifacies based on phylogenies of Jungfer et al. [5] and Ron et al. [9]. All newly generated sequences and their GenBank accession numbers are listed in Table 3.
The sequence matrix was imported to Mesquite 3.04 [30] where manual adjustments were made to correct alignment errors.
As there was a possibility that each of our sampled genes or codon positions in protein-coding genes were shaped by different evolutionary processes, sequences were partitioned according to gene and codon position. We used software PartitionFinder v.2.1.1 [31] to simultaneously determine the best evolution model for each partition and the best partition strategy for the complete matrix.
Phylogenetic relationships were inferred under maximum likelihood and Bayesian criteria, using GARLI 2.0 [32] and MrBayes 3.2.2 [33], respectively. Maximum likelihood analyses were carried out using default values, except for the setting maximum program memory usage (availablememory = 10,000), and the number of generations without topology improvement required for termination (genthreshfortopoterm = 500,000). We ran a total of 20 independent searches, 10 of them starting from randomly generated trees (streefname = random) and 10 from stepwise addition trees (streefname = stepwise). We evaluated the exhaustiveness of the global search by comparing the final maximum likelihood value among the 20 replicate searches. We considered that the replicate searches were effective in finding the best tree when more than 90% of them had maximum likelihood values within two units of the best global search. Support was evaluated using 100 bootstrap pseudoreplicates under the same search parameters used to find the best tree.
Bayesian inference analyses consisted of four parallel runs using the Monte Carlo Markov Chain (MCMC) algorithm for 6 × 10 6 generations and sampling every 1,000 generations. Each run had four chains (3 hot and one cold), with a temperature of 0.1. Convergence into a stationary distribution and effective sample sizes (ESS) for all parameters were analyzed using Tracer software version 1.6 [34]. We discarded as burn-in 10% of the sampled generations and combined the four runs to summarize the posterior probabilities of nodes in a maximum clade credibility tree. Pairwise genetic uncorrected p-distances for the 16S rRNA gene were calculated using MEGA 7 [35] software. This gene has been commonly used to compare  48.5 ± 2.3 (44.9-52.0) Female (n = 7) 64.3 ± 3.6 (57. 1-68.9) divergence between amphibian species [36 37,]. We obtained p-distances in the 16S rRNA gene between all species, except for clade G from Colombia, as sequences were not available.

Phylogenetic relationships
The complete DNA sequence matrix contained five mitochondrial genes and up to 4,856 bp for 180 individuals. The software PartitionFinder choose four partitions as the best strategy (best model in parentheses): 12S, 16S, ND1 1st and 3rd position, and CO1 3rd position (GTR + I + G); ND1 2nd position and Cytb 1st position (HKY + I + G); COI 1st position and Cytb 2nd and 3rd position (K80 + I + G); and CO1 2nd position (F81 + I). Topologies of maximum likelihood and Bayesian inference were very similar except for weakly supported nodes (posterior probabilities, pp <0.96 and bootstrap <64). Phylogenetic relationships among species of the Osteocephalus buckleyi group were consistent with those reported by Ron et al. [9] and Jungfer et al. [5].
We Sequence divergence (uncorrected p-distance for gene 16S) within the O. verruciger-O. cannatellai complex (clades A to G) is low and ranges from 0.7% to 1.5% among clades and 0% to 0.1% within clades ( Table 4). The genetic distance between Sangay (clade A) and its closest relative (clade B, O. cannatellai) is only 0.7%. However, the Sangay population shows a quite distinct morphology compared to Osteocephalus cannatellai. Therefore, we conclude that the Sangay population represents a new species that we describe in the following section. The recognition of the new species renders O. cannatellai paraphyletic because some populations of O. cannatellai (clade B) are more closely related to the new species than to other populations of O. cannatellai (clade C; Figure 1). We discuss the status of both clades of O. cannatellai in the taxonomic review section.
We found significant size differences (adult males and females) between most clades ( Figure 3). Males of clade A (O. sangay sp. n.) are smaller than those of clades C -F (all P values for t tests < 0.001), males of clade B are smaller than those of clade C (t = −6.74, df = 29, P < 0.001); clade F has smaller males than clade D (t = −4.05, df = 29, P < 0.001) and E (t = −3.97, df = 29, P < 0.001); and clade C has the largest males among clades A-F. Females of clade A are smaller than those of clades B-F (all P values for t tests < 0.001), females of the other clades (B-F) are similar in size ( Table 2).

Taxonomic review
To assign binomens to clades A-G, we examined the  O. mimeticus, O. mutabor, and O. verruciger. We document those assignments in the following section.

Taxonomic status of Osteocephalus cannatellai
Based on morphology and phylogenetic position of the holotype of O. cannatellai, we assign clade C as O. cannatellai sensu stricto. We suggest that clade B is considered O. cf. cannatellai until a thorough review of both clades is carried out. Differences in size and color suggest that clade B represents a species distinct from O. cannatellai.

Taxonomic status of Osteocephalus verruciger
The description of Hyla verrucigera is based in two syntypes (ZMB 16589 and ZMH A00946). One syntype (ZMH A00946) is an adult specimen and is lost [38]. The type locality is "Ecuador" and the description by Werner [16] is short (Appendix C). It was collected between 1899 and 1900 by Rich Haensch, who traveled through seven provinces in Ecuador (Figure 2) [39]: Bolívar (Balsapamba), Guayas (Palmar), Napo (Ahuano, Archidona, Baeza, Papallacta, and Pucurcu), Orellana (Coca), Pastaza (Canelos, Sarayacu), Pichincha (Santa Inés) and Tungurahua (Baños de Agua Santa). This syntype (Figure 4) is a young male with faded coloration, so it cannot be unambiguously assigned to a specific clade within O. verruciger. However, the itinerary of the collector indicates that the type material should fall within clades D, E, or F (Figure 2). Trueb and Duellman [38] examined syntype ZMB 16589 and considered it "nearly identical" to a juvenile from Cordillera del Dué (Sucumbíos province; KU 123186),  [38] suggests that it has lost color since they examined it in the 1960s. We tentatively follow Trueb and Duellman [38] in considering clade F as O. verruciger sensu stricto, but note that it could also belong to clades D or E.
Werner [16] description of Hyla verrucigera (Appendix C) indicates that the lost syntype is an adult female with heavily tuberculated dorsal areas. This is inconsistent with the morphology of adult females of O. verruciger because heavy tuberculation is only present on adult males. It is likely that the sex of the syntype reported by Werner was incorrect. No other hylid from the foothill forests in the Amazon basin of Ecuador fits Werner [16] description of a heavily tuberculated dorsum. Therefore, we consider highly unlikely that the syntype belongs to a different species of hylid.         tubercles prominent, round to ovoid; single; supernumerary tubercles present; palmar tubercle small and round; prepollical tubercle large, elliptical; prepollex enlarged; fingers webbing formula I basal II2 − -3 + III3-2 1 / 2 IV ( Figure 6). Toes bearing expanded discs, smaller than those of fingers; relative length of toes I < II < V < III < IV; outer metatarsal tubercle present, round; inner metatarsal tubercle present, ovoid, protuberant; supernumerary tubercles present; toe webbing formula I1 2 / 3 -2 1 / 3 II1 + -2III1 + -2 − IV2 − -− V. Skin on dorsum, head and dorsal surfaces of limbs with tubercles; skin on flanks areolate; skin on venter and ventral surfaces of thighs granular; ventral surfaces of shanks smooth. Cloacal opening directed posteriorly at upper level of thighs, round tubercles below vent. Tongue cordiform, widely attached to the mouth floor; dentigerous processes of vomers angular, behind ovoid choanae, each bearing 8 and 10 (left/right) vomerine teeth.
Color of holotype in life. Based on digital photographs (Figure 7(a,b)). Dorsum light to medium brown with irregular dark brown marks; dorsal surfaces of forearms brown with few dark brown marks; dorsal surfaces of thighs, shanks, and feet light to medium brown with dark brown irregular flecks. Head with one big dark brown irregular blotch; sides of head brown with an oblique cream to tan bar below the orbit; tympanum dark brown; flanks cream to tan with dark brown reticulations and irregular dark brown blotches anteriorly; dorsal surfaces of thighs, shanks and forelimbs light to medium brown with dark brown flecks. Venter khaki to brown with dark brown marks; olive green ring around the pupil; bronze iris with black reticulations.
Color of holotype in preservative. Dorsum brown with dark brown irregular blotches; a big dark brown blotch in the head between orbits extends to the flank; dorsal surfaces of forearms and thighs brown with dark brown and light brown flecks; dorsal surfaces of shanks and feet light brown with small brown flecks; flanks light cream with irregular blotches; venter light cream with light brown spots, more abundant in the anterior half of the body; ventral surfaces of hindlimbs and forelimbs light to medium brown; sides of head light brown with a light cream bar below the orbit; iris gray with black irregular reticulations ( Figure 5).
Variation. In life, based on digital photographs (Figure 7), adult male (QCAZ 58840) with dorsal olivegreen coloration and keratinized tubercles, canthal region green with cream subocular mark, tympanum light green, flanks light green to cream with dark brown reticulation, venter tan. In adult females (QCAZ 58825 and 58839) dorsal coloration is the same as in preserved specimens, tympanum light brown, flanks light brown with dark brown reticulation and irregular dark brown blotches posteriorly. Canthal region brown with a cream tan (QCAZ 58839) to light yellow (QCAZ 58825) subocular mark, venter tan with brown reticulation in QCAZ 58839, and cream with a faint light brown reticulation in QCAZ 58825. In preservative ( Figure 5), dorsal background coloration varies from light brown (QCAZ 59125) to dark brown (QCAZ 58825), with irregular dark brown blotches. Young and adult males bear keratinized tubercles (QCAZ 58840 and QCAZ 59124), which are scattered in females (QCAZ 58824). Ventral surfaces vary from light cream (QCAZ 58825 and QCAZ 59125) to tan with dark brown dots (QCAZ 58839). Osteocephalus sangay is most closely related to O. cannatellai (Figure 8(a,b vs. c-f)) from which it differs in having: (1) [15]. Osteocephalus mimeticus differs from O. sangay in having black iris with some golden blotches and adult males with dorsal yellow brown blotches (absent in adult males of O. sangay) [41].
Distribution and ecology. Osteocephalus sangay is known from eastern Ecuador, Morona Santiago province, Sangay National Park, at four nearby localities (maximum distance between localities: 3.7 km), and elevations between 1551 and 1795 m above sea level. Osteocephalus sangay occurs sympatrically with O. verruciger at the type locality ( Figure 2). This region corresponds to Eastern Montane Forest (based on Ron et al. [17] classification). Specimens were found at night in primary terra firme forest, perching on vegetation up to 2 m above the ground.
Etymology. The specific epithet sangay refers to the area where the species was discovered: Sangay National Park, in Morona Santiago province. The park has more than 3220 lakes and three volcanoes: Sangay, Tungurahua and Altar. The word sangay originates from the native-American Shuar word "Samkay" that means to scare, in reference to violent explosions of this volcano. Sangay National Park is one of the most amphibian species-rich national parks in the world with a total of 100 species [42].

Discussion
Our genetic and morphological results document the existence of an undescribed species of Osteocephalus from Sangay National Park in Ecuador. The new phylogeny for Osteocephalus is consistent with phylogenies published by Ron et al. [9] and Jungfer et al. [5]. We found strong support for a clade that closely allies O. verruciger and O. cannatellai. Our results indicate that this clade is a species complex composed of seven clades. We were able to confirm the species status of one of them, Osteocephalus sangay. Clade B, sister to the new species, presents a geographic distribution, morphological and molecular differences that suggest it could be an undescribed species masked within "O. cannatellai". Nevertheless, more data are needed to better assess variation within O. cannatellai and solve the taxonomic status of clade B.
Our study found paraphyly among populations of O. verruciger (also reported by Ron et al. [25] but in contrast with Ron et al. [9], and Jungfer et al. [5]). Ron et al. [25] hypothesized that paraphyly could be a result of incomplete lineage sorting, mitochondrial gene capture or the existence of undescribed species within O. verruciger. In our phylogeny, each of the clades of O. verruciger is well supported (Figure 1), and genetic distances between clades vary from 0.7% to 1.1% (uncorrected p-distance for 16S). More morphological and genetic data from a larger number of populations is needed to solve the status of O. verruciger and describe putative cryptic species.
Within the O. verruciger-O. cannatellai complex we found low genetic divergence between species (0.7% to 1.5% uncorrected p-distance for gene 16S). Previous studies have shown that sister species of hylids are usually separated by higher distances (e.g., Vieites et al. [37]; Caminer and Ron [43]). However, our finding is not completely unprecedented. Coloma et al. [44], for example, found sister species of the Hyloscirtus larinopygion group with distances as low as 1.3%. The occurrence of Osteocephalus sangay and O. verruciger (clade D) in sympatry indicates the existence of reproductive barriers. Sympatric populations showed marked morphological differences without intermediate morphs. Lack of intermediates suggests that hybridization is absent or infrequent between both species.