A new species of Andean toad (Bufonidae, Osornophryne) discovered using molecular and morphological data, with a taxonomic key for the genus

Abstract Combining a molecular phylogeny and morphological data, we discovered a new species of Osornophryne from the Amazonian slope of the Ecuadorian Andes. Morphologically, the new taxon is distinguished from all others species in Osornophryne by having the Toes IV and V longer than Toes I–III, a short and rounded snout with a small rostral papilla, and conical pustules on flanks. The new species previously was confused with Osornophryne guacamayo. A taxonomic key is provided for all known species of Osornophryne.

Given the complex topography of the Andes and the opportunity for allopatric speciation in areas with similar climatic conditions, it is possible that morphologically similar populations are evolving independently. Herein, we report and describe a new species of Osornophryne, previously confused with O. guacamayo.

Material and methods
Morphology. We examined alcohol-preserved specimens from the herpetological collections at Museo de Zoología of the Pontificia Universidad Católica del Ecuador (QCAZ), Escuela Politécnica Nacional (EPN), and Museo Ecuatoriano de Ciencias Naturales (DH-MECN), all based in Quito, Ecuador. Specimens examined are listed in Appendix I. Fingers are numbered preaxially to postaxially from I-IV to facilitate comparison with previous literature dealing with anurans; however, we stress that in an evolutionary perspective anuran fingers should be numbered from II-V, consistent with the hypothesis that Digit I was lost in anurans (Shubin and Alberch 1986;Fabrezi and Alberch 1996). Morphological measurements were taken with digital calipers to the nearest 0.1 mm and are, as follow: (1) snout-vent length (SVL = distance from tip of snout [excluding the proboscis] to posterior margin of vent); (2) tibia length (TIB = length of flexed hind leg from knee to heel); (3) foot length (FL = distance from base of inner metatarsal tubercle to tip of Toe IV); (4) head length (HL = distance from tip of snout to articulation of jaw); (5) head width (HW = greatest width of head measured between jaw articulations); (6) interorbital distance (IOD = shortest distance between medial margins of upper eyelids); (7) upper eyelid width (EW = greatest width of eyelid measured perpendicular to medial axis of skull); (8) internarinal distance (IND = distance between internal borders of nostrils); (9) eye-nostril distance (EN = distance from anterior corner of eye to posterior border of nostril); (10) snout-eye distance (SE = distance from anterior corner of the eye to the tip of the rostrum); (11) eye diameter (ED = distance between anterior and posterior corners of eye); (12) Finger-III length (FIIIL = distance from proximal border of Finger I to distal end of Finger III); (13) Finger-IV length (distance from proximal border of Finger I to distal end of Finger IV); (14) Toe-IV length (TIVL = distance from proximal edge of Toe I to distal tip of Toe IV); (15) Toe-V length (TVL = distance from proximal border edge of Toe I to distal tip of Toe V). Sexual maturity was determinate by the presence of nuptial pads in adult males and convoluted oviducts in adult females. Techniques for clearing and double-staining specimens with Alcian Blue and Alizarin Red were those of Taylor and Van Dyke (1985). Illustrations were made with the aid of a Wild M3B Heerbrugg stereo dissecting microscope equipped with a camera lucida. Osteological terminology is that of Duellman and Trueb (1994), Fabrezi (1992Fabrezi ( , 1993, and Trueb (1973Trueb ( , 1993; bufonid osteological character states are illustrated in Pramuk (2006). Molecular data. Fresh liver samples were preserved in 90% alcohol, and stored at -80°C. We used salt-precipitation protocols to extract genomic DNA from ethanolpreserved tissues (M. Fujita, unpubl. data). To amplify the mitochondrial gene 12S, we used the primers MVZ59 and tRNA-val, developed by Graybeal (1997) and Goebel et al. (1999), respectively; Polymerase Chain Reaction (PCR) amplification protocol was, as follows: 1 cycle of denaturation 2 min at 94°C, annealing for 30 sec at 42°C, extension for 1 min at 72°C, 5 cycles of denaturation 30 sec at 94°C, annealing for 30 sec at 42°C, extension for 1 min at 72°C, 22 cycles of denaturation 30 sec at 94°C, annealing for 30 sec at 50°C, extension for 1 min at 72°C, final extension at 72°C was conducted for 5 min. PCR products were visualized in 0.7% agarose gel, and unincorporated primers and dNTPs were removed from PCR products using ExoSap-it purification. Cycle sequencing reactions were conducted by the commercial company Macrogen Inc. Data from heavy and light stands were compared to generate a consensus sequence for each DNA fragment with Sequencer Ver. 4.8. We obtained sequences of 71 speci-mens, including all the species in Osornophryne, except O. talipes, and three outgroup taxa. In addition, sequences were downloaded from GenBank (NCBI). Sequences were initially aligned in Clustal X (Larkin et al. 2007) and adjusted in Mesquite 2.71 (Maddison and Maddison 2009). Best-fit model of molecular evolution was selected in jModeltest 1.1 (Posada 2008) under the Akaike Information Criterion (AIC). Model parameters estimated from jModelTest were used in Bayesian analyses.
Phylogenetics. Analyses were conducted using Maximum Parsimony (MP), Maximum Likelihood (ML), and Bayesian Analyses (BA). Parsimony analyses were performed in PAUP (Swofford 2009) using heuristic searches (10,000 stepwise random additions with TBR branch-swapping) and clade support was estimated via 1000 bootstraps with 10 random additions. Maximum likelihood was run in GARLI 0.951 (Zwickl 2006), which uses a stochastic genetic algorithm-like approach to find the topology, branch lengths, and substitution model parameters that maximize the loglikelihood simultaneously (Zwickl 2006). We performed a total of 50 runs to reduce the probability of inferring a suboptimal likelihood solution. Node support was assessed via 1000 bootstrap replicates. For Bayesian analyses, we implemented the model of nucleotide substitution selected as the best fit for the particular dataset according to the Akaike Information Criterion (AIC) in jModeltest 1.1 (Posada 2008). Bayesian analysis of the mitochondrial dataset was performed in Mr Bayes 3.1 (Ronquist and Huelsenbeck 2003). The analysis consisted of 10 million generations and two Markov chains with default heating values. The prior used for the rate matrix was a uniform Dirichlet and no prior information on topology was incorporated. Trees were sampled every 1000 generations; stationarity was assessed by examining the standard deviation of split frequencies and by plotting the -lnL per generation using Tracer v1.4 (Rambaut and Drummond 2005), and trees generated before stationary were discarded as ''burn-in." Bootstrap values p > 70% are considered to indicate strong support (Hillis and Bull 1993, with their caveats). Clades with posterior probabilities p > 0.95 are considered strongly supported, but we caution that relatively high posterior probabilities for short internodes (particularly those with low bootstrap values) may be overestimates of confidence (Alfaro et al. 2003;Erixon et al. 2003).

Results
For most of the species and population of Osonophryne and three species of Atelopus, we obtained a total of 800 bp from the mitochondrial marker 12S rRNA (Table 1). Parameter value estimates for best-fit models for 12S gene generated by jModeltest 1.1 are TIM2 + I (0.001) + G (0.4700). The only taxon for which we could not obtain molecular information was O. talipes, a species that, in Ecuador, is only know from a specimen collected on 02 August 1970 (Cannatella 1986). The different analyses (MP, ML, and AB) are congruent (Fig. 1). The topology resolves most of the relationships among species in Osornophryne, and reveals the presence of the previously unrecognized taxon described below. Species Description. Ten adult males and one adult female. Females of medium size (SVL = 33.0 mm, n = 1); males small (SVL = 17.6-26.1 mm; mean = 21.1 ± 2.40, n = 10; Table 2). Head length 77.2-95.1% head width; male head width 34.9-40.8% SVL; female head width 37.3% SVL; width of head greater at level of posterior margin of mouth; snout short, rounded, with rostral papilla in dorsal and lateral views; nostrils slightly swollen; each nostril oblique, oval, directed laterally; internarial area concave in males and slightly concave in female; interorbital region with skin co-osified with underlying bone, which has few low tubercles; occipital region mostly flat, but with few bony tubercles and cranial crests in males and females; upper eyelids finely tuber-culate in females, with conical tubercles in males; interorbital region wider than the upper eyelid (upper eyelid 73.0-87.5% of interorbital distance in males, n = 9; 64.4% in female); outer edge of the eyelid delineated by a continuous row of warts, which are more conical in males than in female; canthus rostralis straight; loreal region slightly concave, with small warts in males and female; pale brown lips; eyes with oval horizontally pupil; infraorbital and postorbital regions with some prominent tubercles of variable size in males and females. Skin of dorsum highly tuberculate, with discontinuous row of conical tubercles starting at level of posterolateral edge of cranium and ending at level of sacrum in males and females; in males, ventral skin with several small pustules and few conical tubercles on gular region and toward the flanks, pustules much denser on chest and abdomen and less conical; in females, ventral skin smooth, with small, non-conical isolated pustules, pustules more numerous on abdomen.
Forelimb long, slender, finely granular, with several larger tubercles extending along inner and outer edges of fingers in males; in females, tubercles smaller than in males. Hand of moderate length, representing 25.0-30.4% (n = 10) of SVL in males and 28.5% in female; extensive webbing between fingers (Fig. 4); lengths of fingers in order of increasing length : I < II < IV < III; palms with numerous tubercles; subarticular tubercles not distinguishable; palmar tubercle rounded, thenar tubercle almost undistinguishable.    Hind limbs long and slender; well-defined pustules present on inner and outer edges of fingers in males, females with less pronounced pustules than those in males; tibia and foot, respectively, 32.8-36.9% and 34.9-42.5% of male SVL, and 33.9% and 41.9% of female SVL; webbing between Toes I-III more extensive that webbing between Toes IV-V (Fig. 4); lengths of Toes: I < II < III < V < IV; Toe V much longer than Toe III, soles with numerous tubercles; subarticular tubercles indistinguishable; inner metatarsal tubercle oval. Choanae slightly rounded; adult males lacking vocal sacs; vocal slits absent; nuptial pads on proximal surfaces of Toes I and II, not pigmented; cloacal opening medial to thighs.
Coloration in alcohol. Dorsum, head, forearms, and hind limbs brown to dark brown, with some orange patches; tubercles on upper eyelid, proboscis, and flanks pale yellow. Throat pale yellow; venter cream with brown tubercles.
Coloration in life. Dorsum, head, forearms, and hind limbs dark brown to light brown with some lighter patches; tubercles on upper eyelid, proboscis and flanks orange to yellow. Throat cream yellow, with small dark marks; venter orange-brown.
Osteology. The following osteological description of Osornophryne simpsoni is based on a cleared-and-double stained adult male (QCAZ 45899, SVL = 19.5 mm). The osteological description of females was not possible because only one female is known.
Cranium. Shape and proportions. The skull is widest posterior to the orbit at the level of the articulation of the maxilla with the quadratojugal. The braincase is broad; at the level of the midorbit, the width of the braincase is about 41.2% of the greatest width of the skull and 26.2% of the medial skull length.
Neurocranium. The neurocranium is formed by five bones-the sphenethmoid, and the paired prootics and exoccipitals. Anteriorly, the neurocranium is completely ossified. A minute septomaxilla is embedded in the anterior nasal capsule cartilage. In dorsal aspect, the cartilaginous planum antorbitale has a perpendicular orientation in relation to the longitudinal axis of the skull. In lateral and ventral views, a broad cartilaginous separation between bony sphenethmoid and prootic is evident. The frontoparietal fontanelle is partially exposed medially between the frontoparietals. Distally, the otic capsules are cartilaginous. Medially, the exoccipitals are slightly separated from one another. The dorsal surface of each prootic is smooth. The epiotic eminences are prominent.
Auditory apparatus. The stapes and tympanic annulus are absent. The operculum is oval and cartilaginous.
Dermal investing bones. Dorsal investing bones are well developed. The nasals are separated from one another and cover most of the nasal capsules dorsally. The maxillary process of the nasal overlaps the pars fascialis of the maxilla to form a bony anterior margin of the orbit. The frontoparietals are well developed and have a narrow separation between one another along its longitudinal axis. The posteriormedial margin of each frontoparietal contacts the exoccipital, but is not fused to it. Posterolateraly, each frontoparietal bears a bony extension that reaches the epiotic eminence. Each frontoparietal has a lamina perpendicularis that is narrow anteriorly and greatly expanded posteriorly (Fig. 5C). The dorsal surface of each frontoparietal bears small, bony tubercles that are visible externally; the tubercles seem to be co-ossified with the overlying skin.
Ventral investing and palatal bones. The parasphenoid has the shape of an inverted T. The broad cultriform process extends anteriorly to about the mid-level of the orbit, where it is narrowly separated from the posterior border of the sphenethmoid. The cultriform process reaches its maximum width at a level that is coincident with the posterior margin of the optic fenestra. The parasphenoid alae are robust, investing the cartilaginous floor of the otic capsule anterior to the exoccipitals; the length of each ala is 61.8% the length of the cultriform process. A broadly acuminate posteromedial process of the parasphenoid terminates just anterior to the margin of the foramen magnum. The vomers are small, arcuate, broadly separated bones that support the medial margins of the choanae; the bones are unornamented, edentate, and lack dentigerous processe; the prechoanal ramus of the vomer is especially short. The neopalatine is short and narrow; medially, it reaches the anterolateral margin of the sphenethmoid; medially, the neopalatine does not contact the maxilla (Fig. 5B). Maxillary arcade. The premaxillae and maxillae lack teeth. The arcade is complete and has a tenuous articulation with the short quadratojugals. The pars palatinae of the premaxillae are broad. The premaxilla bears two palatine processes, a narrow longer medial and a broad lateral process. There is a simple, juxtaposed articulation between the anterior end of the maxilla and the premaxilla. The pars facialis of the maxilla is well-developed anteriorly, covering the the posterior region of the olfactory capsule; also, the pars facialis has a well-developed preorbital process, which covers most of the planum antorbitale (Fig. 5B, C).
Suspensory apparatus. The tridiate pterygoid bears a slightly curved anterior ramus that is orientated anterolaterally toward the maxilla, with which it articulates. The pterygoid is in close proximity to the maxilla and the narrow space between them is filled by the pterygoid cartilage. The medial and posterior rami of the pterygoid are about equal in length; however, the medial ramus is more robust than the posterior. The lateral end of the medial ramus overlaps the lateral edge of the prootic. The squamosal has the shape of an inverted L; the zygomatic ramus is almost absent, whereas the otic ramus is long and almost reaches the posterior end of the skull. The otic ramus overlaps the lateral margin of the crista parotica slightly. The ventral ramus invests the lateral surface of the palatoquadrate, and articulates with the quadratojugal (Fig. 5A, C); along its anterior margin, the ventral ramus has a conspicuos flange, which extends along the upper border of the otic ramus (Fig. 5C).
Hyoid. The width of the cartilaginous hyoid corpus is narrower than its medial length (width 63.1% of length). The anterolateral and posterolateral processes of the hyoid are absent. The bony posteromedial processes are slightly expanded proximally; each process has a bony flange along the posteromedial margin. The hypoglossal sinus is broadly U-shaped. The hyalia are simple and lack any processes (Fig. 5D).
Postcranium. Vertebral column. There are six prepresacral vertebrae. Presacrals I and II are not fused and are notably shorter than Presacrals III-VI. The vertebral profile in decreasing order of overall width of bony parts is: Sacrum > III > IV > V > VI > II > I. Presacral I, or the atlas, lacks transverse processes. All presacrals are non-imbricate. The transverse processes of Presacral II have a anterolateral orientation, Presacrals III-V have a slightly posterolateral orientation, and Presacral VI is approximately perpendicular to the longitudinal axis of the body. The bony sacral diapophyses are broadly expanded; posteriorly, the sacrum is broadly fused with the urostyle, which is greatly expanded laterally. The urostyle bears a well-developed dorsal crest throughout most of its length (Fig. 6).
Pectoral girdle. The clavicles have a slight orientation, with the medial tips distinctly separated from one another and located at about the same level of the anterolateral end of the clavicle, which articulates with the pars acromialis of the scapula (Fig. 7). The coracoid is notably stout, with the sternal end having a moderate expansion and the sternal end being heavily expanded (sternal end 45% of glenoid end); the inner edge of coracoid has an angle of about 45˚, wheras the external edge is straight (no angle). The pectoral fenestra is has a tringular shape, in which the base is anteriorly convex. The scapula is moderately long with a prominent pars acromialis that is separated from the pars glenoidalis; the leading and posterior edges of the scapula are slightly concave. The suprascapula is mostly cartilaginous, but it is mineralized at both ends, with the ossified cleithrum apparent as a slender bone along the leading edge of the suprascapular blade and with a proximal end that is wider than its distal end. The sternum is small and completely cartilaginous; it contacts the epicoracoid cartilage, which is extensive, and the posterior margin of the coracoid. The omosternum is absent.
Pelvic girdle. The long, slightly concave, and slender ilial shafts bear small dorsal crests, which extend from the anterior third to the posterior end of the shafts (Fig. 6). The ilial prominence is broad and low; the pubes is highly mineralized.
Manus and pes. The phalangeal formulae for the hand and foot are standardi.e., 2-2-3-3 and 2-2-3-4-3, respectively; however, the distal phalange of Finger I, Toe I, and Toe III are greatly reduced and formed mostly by cartilage (Fig. 8). Relative length of fingers, in increasing order, is: I-II-IV-III, and of the foot is: I-II-III-V-IV. The carpus is composed of a radiale, ulnare, Element Y, Carpal 2, and a large postaxial assumed to represent a fusion of Carpals 3-5. Element Y is about 3 times the size of Carpal 2, and the prepollex is an elongated cartilage. The terminal phalanges are acuminate, except Finger III that is slightly T-shaped. The tarsus is composed of two tarsal elements, presumably Tarsal 1 and Tarsal 2 + 3. The prehallux is presented by a proximal mineralized cartilage element associated with a small bony element.
Etymology. The specific name simpsoni is a patronym for Dr. Nigel Simpson in recognition for his continual efforts in protecting the Andean cloud forests of Ecuador. Dr. Simpson is a collaborator of two of the most important conservation NGOs in Ecuador, EcoMinga Foundation (www.ecominga.net) and Jocotoco Foundation (www. fjocotoco.org). As the common name of the species, we suggest "Simpson's Plumb Toad." In Spanish, we suggest the name "Osornosapo de Simpson."

Distribution and conservation.
Osornoprhyne simpsoni is only known from the type locality and surrounding areas, Reserva Zuñac (1°20'58"S, 78°09'31"W) and Reserve Ankaku-Zona (1°16'35"S, 78°4'21"W; Fig. 9). These localities are included in the Bosque de Niebla Montano (Montane Cloud Forest) according to the classification proposed by Valencia et al. (1999). Vegetation is dominated by Clusia spp. trees. All individuals of O. simpsoni have been found on leaves of bromeliads and ferns during the night. Simpatric anurans include Pristimantis altamis, P. bicantus, P. imcomptus and P. galdi. Following the IUCN (2001) criteria, we consider O. simpsoni as Data Deficient; however, it is likely that O. simpsoni has a restricted distribution, as observed in other Osornophryne species. Discussion. It has become increasingly evident that lineage independence is not always accompanied by morphological change when ecological conditions remain similar (Wiens 2004). Therefore, combining different sources of data in the process of species discovery increases the probabilities of revealing evolutionary species (sensu Simpson 1961;Wiley 1978;Padial et al. 2009). The discovery of Osornophryne simpsoni represents a good example of such approach. The phylogeny presented in Figure 1 shows some interesting issues that will need further research. For example, given the current gene sampling, there is no genetic differentiation between O. angel and O. bufoniformis; similarly, O. antisana and O. puruanta are not reciprocally monophyletic, although they have conspicuous morphological differences (e.g., body size). Last, within O. guacamayo, there are two genetically distinctive populations that might represent evolutionary species.