Fifty years after: A taxonomic revision of the amphibian species from the Ecuadorian biodiversity hotspot Abra de Zamora, with description of two new Pristimantis species

RESEARCH ARTICLE Fifty years after: A taxonomic revision of the amphibian species from the Ecuadorian biodiversity hotspot Abra de Zamora, with description of two new Pristimantis species Paul Sze´ kelyID1,2*, Juan Sebastia´n EguigurenID3, Leonardo Ordo´ ñez-DelgadoID2,4, Diego Armijos-Ojeda2,4, Diana Sze´kely2,5 1 Museo de Zoologı´a, Universidad Te´ cnica Particular de Loja, Loja, Ecuador, 2 Laboratorio de Ecologı´a Tropical y Servicios Ecosiste´ micos - EcoSs Lab, Departamento de Ciencias Biolo´ gicas, Universidad Te´cnica Particular de Loja, Loja, Ecuador, 3 Biotechnology Center (BIOTEC), Center for Molecular and Cellular Bioengineering (CMCB), Technische Universita¨ t Dresden (TUD), Dresden, Germany, 4 Programa de Doctorado en Conservacio´ n de Recursos Naturales, Escuela Internacional de Doctorado, Universidad Rey Juan Carlos, Mo´ stoles, Madrid, Spain, 5 Faculty of Natural and Agricultural Sciences, Ovidius University Constanța, Constanța, Romania * jpszekely@utpl.edu.ec Abstract Abra de Zamora is an important biodiversity hotspot in southern Ecuador. Between 1938 and 2010, eleven species of frogs were described from here: Lynchius flavomaculatus, Gas- trotheca psychrophila, Pristimantis balionotus, P. colodactylus, P. cryptomelas, P. percul- tus, P. versicolor, P. vidua, Telmatobius cirrhacelis, P. andinognomus, and Atelopus podocarpus. Unfortunately, many of these species were not re-encountered after their origi- nal description, and for the majority DNA samples were not available, making their phyloge-

netic position unknown. In this study, we assess the current state of the amphibians from Abra de Zamora, by: i. redescribing the species which were first reported from the area, by contributing genetic delimitation (for L. flavomaculatus, P. balionotus, P. colodactylus, P. percultus, and P. vidua), release call (L. flavomaculatus) and advertisement call descriptions (for P. balionotus, P. vidua and P. versicolor); ii. presenting an updated amphibian species list of Abra de Zamora, with the description of two additional Pristimantis species; iii. updat- ing the distribution of these species, including data collected in similar montane habitats from surrounding areas; and iv. amending recommendations regarding their conservation status. Introduction Despite its small size, Ecuador is one of the most biodiverse countries [1], showing the highest density of species (number of species per area unit) in the case of amphibians [2] and reptiles [3], as well as a remarkable proportion of endemic species [4]. Several areas in Ecuador are recognized as genuine hotspots of amphibian diversity from where numerous species were PLOS ONE PLOS ONE | https://doi.org/10.1371/journal.pone.0238306 September 10, 2020 1 / 59 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Sze´kely P, Eguiguren JS, Ordo´ñez- Delgado L, Armijos-Ojeda D, Sze´kely D (2020) Fifty years after: A taxonomic revision of the amphibian species from the Ecuadorian biodiversity hotspot Abra de Zamora, with description of two new Pristimantis species. PLoS ONE 15(9): e0238306.

https://doi.org/10.1371/journal.pone.0238306 Editor: Stefan Lo¨tters, Universitat Trier, GERMANY Received: May 20, 2020 Accepted: August 12, 2020 Published: September 10, 2020 Copyright: © 2020 Sze´kely et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: DNA matrices are available in Zenodo: https://doi.org/10.5281/ zenodo.3952997. Funding: This study was funded by the Critical Ecosystem Partnership Fund (CEPF), grant CEPF- 108984, through the “Amphibian Conservation in the Abra de Zamora Key Biodiversity Area of Ecuador” project. Fieldwork and research of marsupial frogs was supported by the Mohamed bin Zayed Species Conservation Fund, through project 172515484 "Conservation of the Ridge

described, such as Parque Nacional Yasunı´ in western Amazon [5] or Parque Nacional Sangay [6] and the upper basin of the Pastaza River [7] in Tropical Andes. One such a biodiverse hot- spot, with a comparatively much smaller extension, is Abra de Zamora. The name “Abra de Zamora” was used by William E. Duellman and John D. Lynch in their ’70s publications [8, 9], to designate the “ridge (or crest) between Loja and Zamora Chinchipe provinces in the Cordillera Oriental in southern Ecuador” or just the area “13–15 kilometers east (by road) of Loja City, Ecuador”. The term Abra is used in orography to refer to the depression in a line of mountain peaks, which generally, for logistical reasons, is used to build the roads that cross the mountain range [10]; this is also the case of the road connecting two Ecuadorian cities from the tropical Andean region, Loja and Zamora. However, this name was not used by the local residents, who refer to the area between the two provinces closer to Loja as “El Tiro” (the Tiro Pass), and to the lower parts, close to El Tambo sector as “Los Trigales” (the Wheat Fields). Duellman and Lynch conducted their pioneer expeditions in search of amphibians on the old road that connected Loja with Zamora at that time (or Vı´a Antigua Loja-Zamora); this road was used only until the mid ’80s, when a new road (the E50 highway) was constructed on the other side of the valley, and is still in use today (Figs 1 and 2A). Between 1938 and 2010, eleven species of frogs were described from Abra de Zamora. The

first species, Lynchius flavomaculatus, was described in 1938 by Hampton Wildman Parker [11], with specimens collected by the Ecuadorian naturalist, Clodoveo Carrio´n Mora. In 1974, William E. Duellman described Gastrotheca psychrophila [8], based on specimens collected in 1968 and 1971. John D. Lynch described six Pristimantis species from Abra de Zamora in 1979 [9], with specimens collected in 1968, 1971 and 1975: P. balionotus, P. colodactylus, P. crypto- melas, P. percultus, P. versicolor and P. vidua. Linda Trueb described Telmatobius cirrhacelis in 1979, using specimens collected in 1975 [12]. Paratypes collected from Abra de Zamora were used for the description of an additional two species: P. andinognomus [13], with specimens collected in 2003, and Atelopus podocarpus [14], with specimens collected in 1968 by Lynch and 1975 by Duellman. Due to its location (Huancabamba depression, Andes), Abra de Zamora has been a center of endemism and probably a center of amphibian diversification [15]. From this list of eleven species originally described from this area, six species are strictly endemic to the area. An addi- tional five species are currently undescribed, and similarly have not yet been encountered outside the area. This information emphasizes the importance of the region as a center of endemism, and suggests its role as a center of amphibian diversification, particularly for some groups of Terrarana. The advancement of molecular techniques allows precise species identifi-

cation, which is especially important in areas of diversification and for cryptic species [16]. Unfortunately, many of these species were not encountered after their original description, and DNA samples were available only for a couple of them (G. psychrophila, P. andinognomus, P. cryptomelas and P. versicolor) which made their phylogenetic relationships impossible to assess. After almost 50 years of the important work of Duellman and Lynch we evaluated the presence and conservation status of the amphibian populations from this important hotspot. Through an intense fieldwork encompassing Abra de Zamora and several other potentially similar habitats in southern Ecuador, our study targets the following aspects: i. redescribe the species which were described from Abra de Zamora, including their genetic delimitation and advertisement call; ii. present the updated amphibian species list of Abra de Zamora with the description of two additional Pristimantis species; iii. search for their presence in similar mon- tane habitats in surrounding areas from southern Ecuador, in an attempt to more precisely evaluate their distribution; and iv. update recommendations regarding their conservation status. PLOS ONE Taxonomic revision of the amphibian species from Abra de Zamora PLOS ONE | https://doi.org/10.1371/journal.pone.0238306 September 10, 2020 2 / 59 Marsupial Frog (Gastrotheca psychrophila) in southern Ecuador". DS benefited from a research grant from Universidad Te´cnica Particular de Loja (Convocatoria Estancias Cortas de Investigacio´n 2019).

Competing interests: The authors have declared that no competing interests exist.

Materials and methods Ethics statement This study was carried out in strict accordance with the guidelines for use of live amphibians and reptiles in field research compiled by the American Society of Ichthyologists and Herpe- tologists, the Herpetologists’ League and the Society for the Study of Amphibians and Reptiles. Research permit was issued by the Ecuadorian Ministry of Environment (MAE-DNB-CM- 2015-0016). This study was evaluated and approved by the Ethics Committee of Universidad Te´cnica Particular de Loja (UTPL-CBEA-2016-001). Specimen collection and study site We focused our study on all the species described from Abra de Zamora. Even though there are no exact limits to the area, as a starting point, we used Abra de Zamora as referenced by Duellman and Lynch—the area between Loja and El Tambo longitudinally crossed by Vı´a Antigua Loja-Zamora and the E50 highway, which connects Loja and Zamora cities (Fig 1). Based on that, we defined our study site as an approximately 50 km2 area which includes all points of collection for amphibian species described from Abra de Zamora, along with Fig 1. Abra de Zamora and nearby study sites. 1. Huacapamba, 2. Reserva Biolo´gica San Francisco, 3. Reserva Madrigal del Podocarpus, 4. Parque Nacional Podocarpus (Cajanuma sector), 5. Reserva El Cristal. https://doi.org/10.1371/journal.pone.0238306.g001 PLOS ONE Taxonomic revision of the amphibian species from Abra de Zamora PLOS ONE | https://doi.org/10.1371/journal.pone.0238306 September 10, 2020 3 / 59

neighboring favorable habitats. Almost half of this area is inside the Parque Nacional Podocar- pus, basically delimiting the park’s northern border. Parque Nacional Podocarpus was declared officially in 1982 and comprises an area of 146,289 ha. The national park has a very irregular topography covering altitudes from 900 to 3650 m, with large areas of diverse natural habitats [17] and it is an important part of the Tropical Andes hotspots, the richest in biodiver- sity on the planet [18, 19]. The study site has an altitudinal range between 1700 and 3000 m a.s. l. and consists of evergreen upper montane forest (on the western slope, near Loja, up to 2700 m), subpáramo or evergreen elfin forest (on the crest, between 2700 and 3000 m), evergreen upper montane forest (on the eastern slope, between 2700 and 2100 m), and evergreen lower montane forest (on the eastern slope, near El Tambo, bellow 2100 m [20]) (Fig 2). In the study area, the average annual precipitation is around 4865 mm [21] and the average annual temper- ature around 7.3 ˚C [22]. Fieldwork was carried out between January 2016 and March 2020 (a total of 97 days) in Abra de Zamora and in several neighboring areas: Huacapamba (about 10 km to the north; 3.9061˚ S, 79.1909˚ W; datum WGS84), Reserva Madrigal del Podocarpus (about 9 km to the south; 4.0595˚ S, 79.1632˚ W) and the Cajanuma sector of Parque Nacional Podocarpus (about 13 km south to Abra de Zamora; 4.1163˚ S, 79.1719˚ W) (Fig 1). Additional specimens were collected from the wetland complex of Oña, Nabo´n, Saraguro and Yacuambi (3.5810˚ S,

79.0939˚ W), Bosque Protector Washapamba (3.6615˚ S, 79.2738˚ W), Ramos Urcu (at about 5 km west to San Lucas; 3.7111˚ S, 79.3019˚), Reserva Tapichalaca (4.4859˚ S, 79.1477˚ W), and Reserva Cerro Plateado (4.6565˚ S, 78.8699˚ W) in Loja and Zamora Chinchipe provinces. Fig 2. Abra de Zamora. A. General view of the higher part of Abra de Zamora (2800 m a.s.l.) with subpa´ramo ecosystem in the foreground. These terrestrial bromeliads are the type of microhabitat where Gastrotheca psychrophila specimens were collected in the ‘60s and ’70s. In the background the old road that connected Loja with Zamora (Vı´a Antigua Loja-Zamora) is visible. B. Subpa´ramo ecosystem of the higher part (2750 m a.s.l.), habitat of Pristimantis balionotus, P. colodactylus, P. matildae sp. nov., P. samaniegoi sp. nov., and P. versicolor. C. Evergreen upper montane forest ecosystem (2600 m a.s.l.), habitat of P. andinognomus, P. atratus, P. cryptomelas, P. samaniegoi sp. nov., and P. versicolor. D. Evergreen lower montane forest ecosystem (1850 m a.s.l.), habitat of Bolitoglossa sp., Rhinella margaritifera, Gastrotheca testudinea, P. galdi and an undescribed Pristimantis species. https://doi.org/10.1371/journal.pone.0238306.g002 PLOS ONE Taxonomic revision of the amphibian species from Abra de Zamora PLOS ONE | https://doi.org/10.1371/journal.pone.0238306 September 10, 2020 4 / 59

We made intensive visual encounter surveys and auditory surveys both during the day and during the night (12h00–02h00). Due to the heterogeneity of the surveyed ecosystems and sampling methods, we could not calculate the relative abundance or the survey effort. However, we use the following terms in ranking the relative abundance (adapted from [23, 24]): common: when the species presence was detected (seen or heard), in the proper habitat, in large or moderate numbers, on 50– 100% of the sampling days/nights; uncommon: when the species presence was detected in moderate or small numbers, on 25–50% of the sampling days/nights; rare: when the species presence was detected in small numbers on less than 25% of the sampling days/nights. Collected specimens were photographed alive, after which they were euthanized using 20% benzocaine, fixed in 10% formalin, and stored in 70% ethanol. Tissue samples for genetic anal- yses were preserved in 96% ethanol. Examined specimens (listed in the type-series and S1 Appendix) are housed in Museo de Zoologı´a, Universidad Te´cnica Particular de Loja, Loja, Ecuador (MUTPL), Museo de Zoologı´a de la Universidad del Azuay, Cuenca, Ecuador (MZUA), Museo de Zoologı´a, Pontificia Universidad Cato´lica del Ecuador, Quito, Ecuador (QCAZ) and Museo de Zoologı´a, Universidad Tecnolo´gica Indoame´rica, Quito, Ecuador (MZUTI). Morphology The description of qualitative and quantitative morphological characters, as well as the format of the description follows Duellman and Lehr [25]. Sex was determined by the presence of

vocal slits and/or by gonadal inspection. Color data in life were based on field notes and digital photos. The specimens were weighted (body mass: BM) before euthanasia using a My Weigh Triton T3 portable scale with 0.01 g precision. Measurements were taken under a stereo micro- scope, with a Vernier caliper, and rounded to the nearest 0.1 mm. Specimens were measured for the following morphometric variables: (1) snout-vent length (SVL), distance from the tip of snout to posterior margin of vent; (2) head width (HW), widest portion of the head, mea- sured at level of jaw articulation; (3) head length (HL), distance from the tip of snout to poste- rior angle of jaw articulation; (4) interorbital distance (IOD), distance between the inner margins of the orbits; (5) internarial distance (IND), distance between the inner edges of the narial openings; (6) upper eyelid width (EW), the perpendicular distance to the outer edge of the eyelid; (7) eye diameter (ED), distance between anterior and posterior borders of eye; (8) eye-nostril distance (EN), distance from posterior margin of nostril to anterior margin of eye; (9) tympanum diameter (TD), horizontal distance between peripheral borders of tympanic annulus; (10) femur length (FL), length of femur from vent to knee; (11) tibia length (TL), length of flexed leg from knee to heel; (12) foot length (FoL), distance from proximal margin of inner metatarsal tubercle to tip of Toe IV; (13) hand length (HaL), distance from proximal edge of palmar tubercle to the tip of Finger III. Measurements are given as mean ± SD.

DNA extraction, amplification and sequencing DNA extraction was performed directly from 96% ethanol-preserved liver tissue, using the Extract-N-Amp™ Tissue PCR Kit (Sigma-Aldrich, Merck KGaA, Darmstadt, Germany), fol- lowed by PCR reactions under the manufacturer’s protocol. Additionally, in the case of hard to identify and potentially endangered groups (especially Gastrotheca spp.), species identity was confirmed by non-destructive technique, using buccal swabs for DNA extraction [26]. Individuals were afterwards released at capture site. The 12S and 16S rRNA mitochondrial genes, and the nuclear RAG-1 gene were amplified using the primers and annealing tempera- tures given in S1 Table. Success of PCR amplification was tested by gel electrophoresis, using a PLOS ONE Taxonomic revision of the amphibian species from Abra de Zamora PLOS ONE | https://doi.org/10.1371/journal.pone.0238306 September 10, 2020 5 / 59

1% agarose gel, stained with GelRed™ Nucleic Acid Gel Stain (Biotium Inc., Landing Parkway, CA). PCR products were purified using 1.1 volumes of PEG (20% Polyethylene glycol 8000, 2.5 M NaCl), followed by incubation at 37˚C for 15 min. Samples were then centrifuged at 14,000 rpm at room temperature for 15 min, and afterwards the supernatant was discarded. The DNA pellet was washed twice with ice cold 80% ethanol, spinning the samples at maxi- mum speed in the centrifuge for 2 min between ethanol washings. The ethanol was discarded, and the DNA pellet was dried at room temperature for 5 min or until no trace of ethanol was visible in the microcentrifuge tube. DNA was then resuspended in 15 μL sterile ddH2O. Amplicons were sent for sequencing at Macrogen Sequencing Service (Macrogen Inc., Seoul, Korea), using the corresponding primers for each gene. The newly generated DNA sequences were deposited in GenBank (S2 Table). DNA sequence analyses Molecular data were analyzed using sequences of two mitochondrial genes (12S and 16S rRNA) and one nuclear gene (RAG-1) from 123 individuals of 56 species from 55 different localities from Ecuador and Peru (S2 Table). We used all the GenBank-available sequences for Lynchius [27, 28], Pristimantis orestes group [16, 29], Huicundomantis subgenus of Pristiman- tis [30], and 129 new sequences (for 13 species) generated by our study (S2 Table). As out- groups we used Oreobates amarakaeri, O. cruralis, Phrynopus bracki, P. juninensis, Niceforonia brunnea, N. dolops, Bryophryne cophites, Barycholos pulcher, Noblella heyeri, and Strabomantis

anomalus (for the Lynchius tree), Pristimantis unistrigatus, P. ceuthospilus, P. imitatrix, P. dia- dematus, P. rhodoplichus, P. melanogaster, P. wiensi, and P. galdi (for the P. orestes group) and the same Pristimantis species + P. simonsii, P. orestes and P. colodactylus for the Huicundoman- tis subgenus. The sequences were edited, assembled and aligned [MAFFT algorithm, 31] using the program Geneious Prime (Biomatters Ltd.). The edited alignments of 12S, 16S and RAG-1 sequences were concatenated into a single matrix, which was then used for the phylogenetic analyses. The phylogenetic analyses were based on a 2925 bp dataset for Lynchius (1028 bp 12S, 1285 bp 16S and 612 bp RAG-1), a 2683 bp dataset for Pristimantis orestes group (964 bp 12S, 1077 bp 16S and 642 bp RAG-1) and a 2463 bp dataset for Huicundomantis (905 bp 12S, 915 bp 16S and 643 bp RAG-1). The aligned matrices are available at https://doi.org/10.5281/ zenodo.3952997. Molecular phylogenetic relationships were inferred using Maximum Likelihood (ML) and Bayesian Inference (BI). We used PartitionFinder v. 2.1.1 [32] to select the best-fit models of sequence evolution and the best partition scheme with the AICc (for ML) and BIC (for BI) models of selection. ML analyses were conducted in GARLI v. 2.1 [33] performing four inde- pendent searches (two with the “streefname” set to random and two set to stepwise), with 250 replicates each, and with the “genthreshfortopoterm” set to 100,000. Node support was assessed with non-parametric bootstrapping [34] with 1,000 pseudoreplicates. The 50% major-

ity rule consensus for the bootstrap trees was obtained with Geneious Prime (Biomatters Ltd.). BI analyses were conducted with MrBayes 3.2.6 [35], the Markov chain Monte Carlo runs being performed twice, independently, for 50 million generations, with a sampling frequency of 1000. Convergence of the runs was assessed from the average split frequency of standard deviations (p < 0.001) and by checking the potential scale reduction factors (PSRF ~ 1.0) for all model parameters. The first 25% of the trees were discarded as burn-in and the remaining ones were used to generate a 50% majority rule consensus tree, as well as to estimate the Bayes- ian posterior probabilities. Throughout the text, we considered that a node has “strong sup- port” when its bootstrap value was > 70 and its Bayesian posterior probability was > 0.95, PLOS ONE Taxonomic revision of the amphibian species from Abra de Zamora PLOS ONE | https://doi.org/10.1371/journal.pone.0238306 September 10, 2020 6 / 59

“moderate support” for 50–70 and 0.90–9.95 and “weak support” or non-resolved for lower values of 50 or 0.90, respectively. Uncorrected p-genetic distances for gene 16S were estimated with software MEGA6 [36] (S3 Table). Call recordings and analysis We analyzed 16 calls of 15 individuals belonging to five species: release call of L. flavomacula- tus and advertisement calls of P. vidua, P. samaniegoi sp. nov., P. balionotus and P. versicolor. The calls were recorded in the field or in laboratory (L. flavomaculatus) using an Olympus LS- 11 Linear PCM Recorder and a RØDE NTG2 condenser shotgun microphone at 44.1 kHz sampling frequency and 16-bit resolution, in WAV file format. Air temperature and humidity were measured with a data logger (Lascar Electronics, model EL-USB-2-LCD, accuracy: ± 0.5˚C; ± 5%). All analyzed call recordings are deposited in original form, full length at Fono- teca UTPL (record IDs are provided in S4 Table). Acoustic analysis was conducted using Raven Pro 1.6 (Center for Conservation Bioacoustics 2019). We measured the temporal parameters from the oscillograms and the spectral parameters from spectrograms obtained with the Hanning window function, DFT: 512 samples, 3 dB filter bandwidth: 124 Hz, and a 50% overlap. The terminology and procedures for measuring call parameters follow Cocroft and Ryan [37], Toledo et al. [38] and Ko¨hler et al. [39] and a call-centered approach was used to dis- tinguish between a call and a note [sensu 39]. The following temporal and spectral parame-

ters were measured and analyzed: (1) call duration: time from the beginning to the end of a call; (2) inter-call interval: the interval between two consecutive calls, measured from the end of one call to the beginning of the consecutive call; (3) call rate: number of calls per min- ute, measured as the time between the beginning of the first call and the beginning of the last call; (4) note duration: the duration of a single note within a call, measured from begin- ning to the end of the note; (5) inter-note interval: the interval between two consecutive notes within the same call, measured from the end of one note to the beginning of the conse- cutive note; (6) note rate: number of notes per second, measured as the time between the beginning of the first note and the beginning of the last note; (7) number of pulses: the num- ber of pulses per note; (8) pulse rate: number of pulses per second, measured as the time between the beginning of the first pulse and the beginning of the last pulse; (9) dominant frequency: the frequency containing the highest sound energy, measured along the entire call; and (10) the 90% bandwidth, reported as frequency 5% and frequency 95%, or the mini- mum and maximum frequencies, excluding the 5% below and above the total energy in the selected call. Nomenclatural acts The electronic edition of this article conforms to the requirements of the amended Interna- tional Code of Zoological Nomenclature, and hence the new names contained herein are avail- able under that Code from the electronic edition of this article. 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) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix "http://zoobank.org/". The LSID for this publication is: urn:lsid:zoobank.org:pub:32448A- CE-C45F-4D84-9C2A-670E0C29B389. The electronic edition of this work was published in a journal with an ISSN, and has been archived and is available from the following digital reposi- tories: PubMed Central and LOCKSS. PLOS ONE Taxonomic revision of the amphibian species from Abra de Zamora PLOS ONE | https://doi.org/10.1371/journal.pone.0238306 September 10, 2020 7 / 59

Results Class Amphibia Linnaeus, 1758 Order Anura Fischer von Waldheim, 1813 Family Bufonidae Gray, 1825 Genus Atelopus Dume´ril and Bibron, 1841 Atelopus podocarpus Coloma, Duellman, Almenda´riz, Ron, Tera´n-Valdez, and Guayasa- min, 2010 Common English name. Podocarpus Harlequin Frog Common Spanish name. Jambato de Podocarpus Etymology. The specific name is a noun in apposition and refers to the Podocarpus National Park. Holotype. QCAZ 6801, adult female, from Lagunas del Compadre (ca. 3300 m), Parque Nacional Podocarpus, Provincia Loja, Ecuador, collected on 1 December 1994 by Jenny Rudston. Paratypes. EPN 1223–26, from Lagunas del Compadre (3400 m), Parque Nacional Podo- carpus, Provincia Loja, Ecuador, no collection date stated on field notes, but possibly collected on September 1981, by Luis H. Albuja V.; EPN 2143–45, from Cerro Toledo (2900 m), Provin- cia Loja, Ecuador, collected on 11 September 1981 by Luis H. Albuja V.; EPN 3235, 3237, Mir- ador path nearby the Centro Administrativo Cajanuma of the Parque Nacional Podocarpus (2850–2990 m), Provincia Loja, Ecuador, obtained on 16–18 November 1986 by Ana Almen- da´riz; KU 120389, from 15 km east of Loja (2700 m), Provincia Zamora Chinchipe, Ecuador, obtained on 10 June 1968 by John D. Lynch; KU 165004–5, from 13 km east of Loja, Abra de Zamora, Provincia Loja, Ecuador, obtained by William E. Duellman; KU 178418, from Provin- cia Zamora Chinchipe: 13.7 km east of Loja, Provincia Loja, Ecuador; KU 201115–7, from 3.7 km south of Saraguro (2800 m), collected on 08 March 1984 by J. E. Simmons; QCAZ 2838

(cleared and double stained skeleton), from Centro Administrativo Cajanuma (2800 m), Par- que Nacional Podocarpus, obtained on 25 June 1987 by Luis A. Coloma, Mario Garcı´a Saltos, and Renato Leo´n. Remarks. Despite numerous searching efforts carried out by different teams from various institutions (including us), in the last several years, in Abra de Zamora and nearby areas, the species was not re-encountered. The last sighting of Atelopus podocarpus dates from 1994, from Parque Nacional Podocarpus (Lagunas del Compadre), and the species is considered probably extinct. Family Hemiphractidae Peters, 1862 Genus Gastrotheca Fitzinger, 1843 Gastrotheca psychrophila Duellman, 1974 Gastrotheca (Duellmania) psychrophila: Dubois, 1987 Gastrotheca (Gastrotheca) psychrophila: Duellman, 2015 Common English name. Ridge Marsupial Frog Common Spanish name. Rana marsupial del Abra de Zamora Etymology. The specific name is derived from the Greek psychros, meaning “cold,” and the Greek philos, meaning “having an affinity for”; the name was used in allusion to the cold cli- mate at the type locality [40]. Holotype. KU 120760, an adult female, from the ridge between Loja and Zamora, 2850 m, 13– 14 km E (by road) of Loja, Provincia Loja, Ecuador; obtained on 10 June 1968 by John D. Lynch. Paratypes. Collected at the type locality: KU 120761, 10 June 1968, John D. Lynch; KU 141586, 21 May 1971, Richard R. Montanucci; KU 142631–7, 21–23 July 1971, William E. Duellman. PLOS ONE Taxonomic revision of the amphibian species from Abra de Zamora

PLOS ONE | https://doi.org/10.1371/journal.pone.0238306 September 10, 2020 8 / 59

Remarks. Despite intensive searches carried out since 2016, we were not able to find this species. We managed to survey an area of almost 100 km2 around the type locality, but unfor- tunately, we were not able to find any G. psychrophila individuals. The results are even more worrisome as our intensive searches in the type locality failed to find any traces of marsupial frog presence (no tadpoles or calls), although if the ecosystem remained practically unchanged since the species was found in 1968 (Fig 2A). Two other Gastrotheca species were encountered, relatively close to the type locality: G. elicioi (on the western slope) and G. testudinea (on the eastern slope), but both well below G. psychrophila’s altitudinal range of 2700–2800 m. The identity of the encountered Gastrotheca species was confirmed through non-destructive DNA tissue sampling (using buccal swabs). In the light of these current findings, we are considering G. psychrophila probably extinct. Superfamily Brachycephaloidea Gu¨nther, 1858 Family Strabomantidae Hedges, Duellman, and Heinicke, 2008 Subfamily Pristimantinae Pyron and Wiens, 2011 Genus Lynchius Hedges, Duellman, and Heinicke, 2008 Phylogeny For Lynchius, PartitionFinder under AICc (for ML) identified five partition schemes as the best strategy (best model in parentheses): 12S (GTR+I+G), 16S (GTR+I+G), RAG-1 1st posi- tion (TRNEF+G), RAG-1 2nd position (K81UF+G) and RAG-1 3rd position (GTR+G). Under BIC (for BI) PartitionFinder identified three partitions: 12S and 16S (GTR+I+G), RAG-1 1st

position (K80+G) and RAG-1 2nd and 3rd positions (HKY+G). The phylogenetic trees con- structed by Bayesian inference and Maximum likelihood showed the same topology, with the only difference being the position of L. nebulanastes, which remained non-resolved in both trees (Fig 3). The phylogram of our analysis showed exactly the same topology with the one constructed by Sanchez-Nivicela et al. [28]. The L. flavomaculatus sequence (KU 218210) used in all published studies (e.g. [27, 28, 41, 42]), was obtained from a specimen collected in 1990, from 19.4 km south of Yangana, Loja province, which is located more than 60 km south from the type locality, Abra de Zamora. In June 2018 we encountered an adult female (MUTPL 342) in Parque Nacional Podocarpus (Cajanuma entrance), from about 13 km south of the type locality, and in the same type of eco- system as of the type locality (shrub pa´ramo). The uncorrected p-genetic distance for the gene 16S between the two samples is 3.1% (S3 Table). Based on the conservative 3% threshold of the genetic distances in the 16S gene for species delineation [43], together with observed morpho- logical differences between several similar species from the nearby areas (pers. obs.) we con- sider the specimen from Yangana as belonging to a different, currently undescribed species. Thus, it seems that the previously known L. flavomaculatus is in fact a complex of cryptic spe- cies, composed of at least four different species in Ecuador: one species from Yangana, L. flavo-

maculatus sensu stricto from the type locality (Abra de Zamora), one species from San Lucas and Saraguro (we are preparing its description) and one species nearby Plan de Milagro, Mor- ona Santiago province. In addition, very likely the specimen collected in Peru [25] is a differ- ent, currently undescribed species. Lynchius flavomaculatus (Parker, 1938) (Fig 4) Eleutherodactylus flavomaculatus Parker, 1938 Niceforonia flavomaculata: Lynch, 1969 Phrynopus flavomaculatus: Lynch, 1975 Eleutherodactylus flavomaculatus: Lehr, 2006 PLOS ONE Taxonomic revision of the amphibian species from Abra de Zamora PLOS ONE | https://doi.org/10.1371/journal.pone.0238306 September 10, 2020 9 / 59

‘‘Eleutherodactylus” flavomaculatus: Heinicke, Duellman, and Hedges, 2007 Lynchius flavomaculatus: Hedges, Duellman, and Heinicke, 2008 Common English name. Yellow-spotted Andes Frog Common Spanish name. Rana andina de manchas amarillas Etymology. The specific name is derived from the Latin adjective flavus, meaning “yellow” and the Latin noun maculatus, meaning “spots” and refers to the yellow spots on the venter [25]. Fig 3. Maximum likelihood phylogram of the Lynchius species based on 2925 base pairs of concatenated DNA from 12S, 16S, and RAG-1 gene fragments. Bootstrap values (%) and Bayesian posterior probabilities (decimal) are shown except when they are below 50 (bootstrap) or 0.5 (posterior probability); asterisks indicate support values of 100% or 1 (posterior probabilities). Outgroup is not shown; the tree was routed with Strabomantis anomalus. With green are marked the newly generated sequences for the present study. The collection number, species name, province and short locality name (for the Ecuadorian samples) or country (for the Peruvian samples) of the samples are shown next to each terminal (associated data listed in S2 Table). https://doi.org/10.1371/journal.pone.0238306.g003 Fig 4. Lynchius flavomaculatus (MUTPL 342, adult female) in life. A. Dorsolateral view; B. Ventral view; C. Profile view of the head. https://doi.org/10.1371/journal.pone.0238306.g004 PLOS ONE Taxonomic revision of the amphibian species from Abra de Zamora PLOS ONE | https://doi.org/10.1371/journal.pone.0238306 September 10, 2020 10 / 59

Holotype. BMNH 1947.2.16.11 (formerly 1935.11.3.16), adult female from 15 km East of Loja City, Ecuador, 3000 m, collected by C. Carrio´n. Paratypes. BMNH 1947.2.16.12–14 (formerly 1935.11.3.17–20), a male and three immature females from the type locality. Diagnosis. Lynchius flavomaculatus is a large species distinguished by the following combi- nation of traits: (1) skin on dorsum shagreen (trait more visible in life) and of the flanks smooth; dorsolateral folds absent; low middorsal fold present (trait more visible in life); skin on venter smooth; discoidal and thoracic folds absent; (2) tympanic membrane and annulus distinct; large supratympanic fold obscuring upper and posterodorsal edges of tympanum; one large, elongated postrictal tubercle present; (3) snout short, rounded in dorsal view, slightly inclined anteroventrally in profile; canthus rostralis weakly concave in dorsal view, angular in profile; (4) upper eyelid without tubercles; low, cranial crests present; (5) dentigerous processes of vomers prominent, triangular, with 4 to 5 teeth; (6) males with vocal slits on each side of the tongue and no nuptial pads; (7) fingers long and slender; Finger I longer than Finger II; tips of digits narrow, not expanded, rounded; circumferential grooves absent; fingers lacking lateral fringes; (8) subarticular tubercles prominent, round; supernumerary palmar tubercles large, rounded, smaller than subarticular tubercles; palmar tubercle partially divided into a much larger (inner) and a smaller (outer) tubercles; thenar tubercle prominent, elliptical; (9) ulnar

tubercles absent; (10) heel lacking tubercles; outer edge of tarsus lacking tubercles; inner edge of tarsus lacking tubercles or fold; (11) inner metatarsal tubercle elliptical, about 2x round outer metatarsal tubercle; subarticular tubercles prominent, subconical; supernumerary plan- tar tubercles absent; (12) toes long and slender; Toe V slightly shorter than Toe III; tips of digits narrow, not expanded, rounded; circumferential grooves absent; toes without lateral fringes; webbing absent; (13) in life, dorsum mottled tan, reddish or dark brown, dull greenish gray or yellow; flanks yellow with or without brown spots; venter mostly yellow, with or with- out gray on the throat; throat light gray with yellow blotches; yellow blotches on groin and anterior surfaces of thighs; iris pale green, with a brown median horizontal streak, and fine black reticulations; (14) SVL 34.6–44.2 mm in adult females (n = 31) and 24.0–35.2 mm in adult males (n = 27) [27] (data regarding the SVL contains individuals from several locations, and probably include specimens belonging to other, cryptic, currently undescribed species than L. flavomaculatus). Advertisement call. Unknown. Release call. When caught and manipulated, the female MUTPL 342 emitted low, short, screeching calls (Fig 5), which we interpreted as release calls (calls usually emitted during amplexus or after being touched by other males, when the amplexus is not desired by the amplected individual, [38]), and not distress calls (calls emitted during subjugation by a poten-

tial predator). We base this observation on the fact that the distress call of a similar, unde- scribed, Lynchius species is very different (long, loud shrieking sounds; pers. obs.). We recorded several of these calls in the laboratory, from a 20 cm distance, while holding the female by the sides of its body, in the axillary and inguinal regions, simulating an amplexus (S4 Table). The calls are single-pulsed notes (tonal sounds) characterized by a duration of (range and mean ± SD in parenthesis) 0.026–0.041 s (0.033 ± 0.006, n = 6), a dominant frequency of 2239.5–3100.8 Hz (2555.3 ± 343.088, n = 6) and a 90% bandwidth range of 2153.3–2842.4 Hz (2426.1 ± 295.4, n = 6) to 2325.6–3186.9 Hz (2641.4 ± 343.088, n = 6). The fundamental fre- quency and harmonics are not recognizable. Distribution. Currently, the presence of L. flavomaculatus is confirmed (with DNA sam- ples) only from Parque Nacional Podocarpus (Cajanuma entrance), in a shrub pa´ramo eco- system [44], from about 13 km south to the type locality, Abra de Zamora, in Loja province (Fig 6). Despite intensive searches, we did not find any individuals in the type locality. Based PLOS ONE Taxonomic revision of the amphibian species from Abra de Zamora PLOS ONE | https://doi.org/10.1371/journal.pone.0238306 September 10, 2020 11 / 59

on the available data we conclude that the specimen KU 218210 collected at 19.4 km south of Yangana, Loja province, is most likely a different, undescribed species. Also, the specimens collected at 8–9 km north to San Lucas or 10 km south of Saraguro, in Loja province [27], actually belong to a currently undescribed new species. As for the specimens from Morona Santiago province and Peru [25, 27] they represent very probable different, undescribed species, taking in consideration the large distances from the type locality. The species was encountered at an altitudinal range between 3000 and 3250 m a.s.l. in shrub pa´ramo ecosystems. Natural history. This species is considered terrestrial, usually found beneath rocks or logs in pa´ramo and subpa´ramo habitats [45]. We found the female MUTPL 342 emerging from a thick moss bed, which is likely a preferred microhabitat for Lynchius species. From the type locality, there were collected 44 specimens between 1968 and 1984 (22 specimens in 1968, 7 in 1971, 14 in 1975, and 1 in 1984) by J.D. Lynch, R.R. Montanucci and W.E. Duellman, accord- ing to the Kansas University Herpetology Collections database (https://biodiversity.ku.edu/ herpetology/collections). However, our ongoing fieldwork since 2016 in the type locality did not yield any specimens. It is possible that the population is experiencing a decline, but Fig 5. Release call of Lynchius flavomaculatus (MUTPL 342, FUTPL-A 246). A. Oscillogram of a single call; B. Spectrogram of a single call; C. Power spectrum of a single call. https://doi.org/10.1371/journal.pone.0238306.g005

conclusive conclusions are hard to reach due to the species low detectability. The female MUTPL 342 had 31 large unpigmented eggs (SVL 42.5 mm, BM 8.21 g). Conservation status. Lynchius flavomaculatus is currently listed as Data Deficient in view of the absence of recent information on its extent of occurrence, population status and ecologi- cal requirements [46]. Based on the available (revised) information we consider L. flavomacu- latus to be Endangered following B1ab(iv)c(iii)+2ab(iv)c(iii) IUCN criteria [47] because: (1) its Extent of occurrence (EOO) and Area of occupancy (AOO) are estimated to be less than 50 km2; (2) it is known from only two close locations; and (3) we suspect a decline in the number of locations or subpopulations. Remarks. The diagnosis provided herein is based on the thorough revision of Motta et al. [27] and one adult female specimen, MUTPL 342, collected from about 13 km south to the type locality. Our diagnosis concurs with most of the morphological features described; how- ever we identified some differences, which we attribute to the fact that the specimens examined by the authors [27] belong to at least four different cryptic species. Also, the opportunity to examine a live specimen helped to complete the diagnosis by adding descriptions of characters which are evident in life, but lost in preserved specimens. The most important difference is the Fig 6. Distribution of Lynchius flavomaculatus and of the species from the Pristimantis orestes species group described from Abra de Zamora. Records are

based on specimens (and DNA sequences) deposited at the Museo de Zoologı´a, Universidad Te´cnica Particular de Loja (MUTPL). 1. Bosque Protector Washapamba, 2. Reserva Madrigal del Podocarpus, 3. Parque Nacional Podocarpus (Cajanuma sector). https://doi.org/10.1371/journal.pone.0238306.g006 PLOS ONE Taxonomic revision of the amphibian species from Abra de Zamora PLOS ONE | https://doi.org/10.1371/journal.pone.0238306 September 10, 2020 13 / 59

lack of discoidal and thoracic folds in the specimen collected from Cajanuma. Duellman and Lehr [25] list the discoidal fold as absent, but Motta et al. [27] consider that discoidal and tho- racic folds are present. Specimen MUTPL 342 lacked both folds (in life as well as in preserva- tive; Fig 4B). On the other hand, all the specimens collected by us nearby Saraguro and San Lucas have evident discoidal and thoracic folds, so it might be possible that this feature is char- acteristic for the individuals of this currently undescribed species. Similarly, we observed the skin on dorsum as shagreen, with a more granular texture on the back and dorsal surfaces of hindlimbs, but without tubercles (Fig 4A), and thus, in accordance to Duellman and Lehr [25], we did not observed any tubercles on the upper eyelids, either in life or in preservative (Fig 4C). Genus Pristimantis Jime´nez de la Espada, 1870 Pristimantis orestes species group [sensu 6, 16, 48]. Three species of this species group were described from Abra de Zamora. We add our description of an additional two new spe- cies further below. Phylogeny For the Pristimantis orestes species group, PartitionFinder under AICc (for ML) identified four partition schemes as the best strategy (best model in parentheses): 12S (GTR+I+G), 16S (GTR+I+G), RAG-1 1st and 2nd positions (F81+G), and RAG-1 3rd position (TVM+G). Under BIC (for BI) PartitionFinder identified three partitions: 12S and 16S (GTR+I+G), RAG-1 1st and 2nd positions (F81+G), and RAG-1 3rd position (K80+G). The phylogenetic trees con-

structed by Bayesian inference and Maximum likelihood showed exactly the same topology, but with overall stronger support in the case of the Bayesian inference (Fig 7). Once again, we recovered the P. orestes species group as monophyletic, with strong support (bootstrap values = 100%; posterior probabilities = 1) in both ML and Bayesian analyses (Fig 7). Our analysis confirmed the membership of P. vidua, P. colodactylus and two new species, i.e. P. samaniegoi sp. nov. and P. matildae sp. nov., to the P. orestes species group. In general, the phylogenetic tree in our analysis showed the same topology with the one constructed by Urgilles et al. [16], but the addition of the new species to the group changed the kinship between some of the species. Also, differences appeared in the position of some unresolved branches such as P. tiktik or P. saturninoi (Fig 7). Our phylogenetic tree reveals three, strongly supported, large subgroups of species: the P. orestes subgroup (with P. andinognomus, P. orestes, P. vidua, an undescribed species and P. cajanuma), the closely related P. simonbolivari subgroup (with P. tiktik, P. saturninoi, P. simon- bolivari, P. bambu, P. mazar, P. quintanai, P. samaniegoi sp. nov and two undescribed species) and the morphologically distinct, bromeliad specialist, P. colodactylus subgroup (with P. matil- dae sp. nov., P. colodactylus, P. muranunka and an undescribed species). Pristimantis andinognomus Lehr and Coloma, 2008 Common English name. Andean Dwarf Robber Frog Common Spanish name. Cutı´n nomo

Etymology. The specific name is derived from the New Latin noun Andinus meaning “Andes” and the New Latin noun gnomus meaning “dwarf” and refers to the small size of the species, which at the time of description was the smallest species of the genus [13]. Holotype. QCAZ 16695, an adult male, from Reserva Tapichalaca, on the road between Yangana and Valladolid (2667 m), Provincia Zamora-Chinchipe, Ecuador, collected by Italo G. Tapia and Queti M. Tapia on 17 September 2001. Paratypes. 20 males: QCAZ 16669, 16672–74, 16685–88, 16690–93, 26326, 26329, 26331, 26343, 26347, 26352, 26964–65, 19 females: QCAZ 16670–71, 16675–78, 16680–84, 16689, PLOS ONE Taxonomic revision of the amphibian species from Abra de Zamora PLOS ONE | https://doi.org/10.1371/journal.pone.0238306 September 10, 2020 14 / 59

Fig 7. Maximum likelihood phylogram of the Pristimantis orestes species group based on 2683 base pairs of concatenated DNA from 12S, 16S, and RAG-1 gene fragments. Bootstrap values (%) and Bayesian posterior probabilities (decimal) are shown except when they are below 50 (bootstrap) or 0.5 (posterior probability); asterisks indicate support values of 100% or 1 (posterior probabilities). Outgroup is not shown; the tree was routed with Pristimantis galdi. With green are marked the newly generated sequences for the present study; with blue and red are marked the newly described species. The collection number, species name, province and short locality name of the samples are shown next to each terminal; all samples are from Ecuador (associated data listed in S2 Table). https://doi.org/10.1371/journal.pone.0238306.g007 PLOS ONE Taxonomic revision of the amphibian species from Abra de Zamora PLOS ONE | https://doi.org/10.1371/journal.pone.0238306 September 10, 2020 15 / 59

16694, 16697, 26325, 26330, 26351, 27033, 29248, 24 subadults: QCAZ 16670–72, 16674–78, 16680–94, 16697 collected at the type locality along with the holotype by Italo G. Tapia and Queti M. Tapia on 17 September 2001; QCAZ 26325–26, 26329–31, 26343, 26347, 26351–52, from Abra de Zamora (2800 m), boundary between Provincias Loja and Zamora Chinchipe, collected by Italo G. Tapia, Erik R. Wild, Andre´s Merino-Viteri, and Luis A. Coloma on 16 August 2003; QCAZ 26964–65, 27033, from Reserva Tapichalaca (2400 m), on road between Yangana and Valladolid, collected by David Kizirian and Juan C. Santos on 30 October 2000; QCAZ 29248, collected by Carlos Carrio´n-Bonilla and Paolo Piedrahita-Piedrahita on 18 Feb- ruary 2001. Remarks. It seems that the population of Pristimantis andinognomus from Abra de Zamora is a different species from the one from the type locality, Reserva Tapichalaca, and it needs a detailed revision. This species is common in the higher parts of Abra de Zamora, and, together with P. versicolor, represents the most abundant frog species in the area. The calls of the males were registered during each night of fieldwork at the site, regardless of weather conditions, and sometimes even during the day. More data is needed to clarify this species phylogenetic relationships and current distribution. Pristimantis vidua (Lynch, 1979) (Fig 8) Eleutherodactylus vidua Lynch, 1979 Eleutherodactylus (Eleutherodactylus) vidua: Lynch, 1996; Lynch and Duellman, 1997 Pristimantis vidua: Heinicke, Duellman, and Hedges, 2007

Pristimantis (Pristimantis) vidua: Hedges, Duellman, and Heinicke, 2008 Common English name. Mountain Crest Robber Frog Common Spanish name. Cutı´n viuda Fig 8. Color variation of Pristimantis vidua in life. A–F females: A. MUTPL 312, B. MUTPL 164, C. MUTPL 586, D. MUTPL 590, E. MUTPL 153, F. MUTPL 150; G–I males: G. MUTPL 495, H. MUTPL 496, I. MUTPL 148. Abra de Zamora (A, G, H), Bosque Protector Washapamba (B, E, F, I), Parque Nacional Podocarpus—Cajanuma (C, D). Most common coloration of this species is presented in A, G and H. https://doi.org/10.1371/journal.pone.0238306.g008 PLOS ONE Taxonomic revision of the amphibian species from Abra de Zamora PLOS ONE | https://doi.org/10.1371/journal.pone.0238306 September 10, 2020 16 / 59

Etymology. Latin, “a widow”, used as a noun in apposition. The choice of this name reflected the belief of the author that this species might be all-female [9]. Holotype. KU 120082, adult female obtained 15 km E Ciudad Loja, Provincia Zamora- Chinchipe, Ecuador, 2800 m, on 10 June 1968 by John D. Lynch. Paratypes. KU 120083–88, KU 120090–91, topotypes; KU 141994, 15 km E Loja, Provincia Zamora-Chinchipe, Ecuador, 2710 m; KU 120089, 13–14 km E Loja, Provincia Loja, Ecuador, 2850 m; KU 120092, 8–9 km N San Lucas, Provincia Loja, Ecuador, 3000–3100 m. Diagnosis. Pristimantis vidua is a small species distinguished by the following combination of traits: (1) skin on dorsum finely tuberculated (in life the tuberculated texture of the skin is more evident); skin on venter areolate; discoidal fold visible; dorsolateral folds present; low middorsal fold present; (2) tympanic membrane absent; ventral part of tympanic annulus visi- ble, concealed by skin; its length about 43% of the length of eye; supratympanic fold present; (3) snout short, rounded in dorsal view and in profile; canthus rostralis weakly concave in dor- sal view, rounded in profile; (4) upper eyelid bearing several small tubercles, similar in size and shape with the ones from the dorsum (trait more visible in life), about 61% IOD in females and 69% IOD in males; cranial crests absent; (5) dentigerous processes of vomers prominent, oblique, round or slightly ovoid, separated medially by distance lower than width of processes;