A new cryptic species of the Pristimantis lacrimosus group (Anura, Strabomantidae) from the eastern slopes of the Ecuadorian Andes

With 566 species, the neotropical genus Pristimantis is the most speciose vertebrate genus. As a result of its striking diversity, taxonomic reviews remain a challenge. Herein, we present an updated phylogeny of the Pristimantis lacrimosus group and describe a new species from Llanganates and Sangay National Parks. We also report, for the first time, the phylogenetic position of Pristimantis degener, P. eugeniae, P. katoptroides, and P. petersi. Based on our phylogeny, we add two species to the Pristimantis lacrimosus group. Through the integration of molecular and bioacoustic evidence, we describe a new species which was hidden under “Pristimantis petersi”. Pristimantis petersioides sp. nov. is most closely related to Pristimantis petersi and an undescribed species from Peru. It can be distinguished from P. petersi by its advertisement call and large genetic differences (uncorrected p-genetic distances 7.9% to 8.4% for gene 16S). Moreover, the new species and P. petersi are not sister species. We suggest assigning the new species to the Endangered Red List category because it has a small distribution range with deforestation as result of agriculture and other anthropogenic influences.


Introduction
The genus Pristimantis Jiménez de la Espada, 1870 has bewildered scientists for its striking diversity. Comprising 566 Neotropical species it is the most speciose vertebrate genus (Hedges et al. 2008;Frost 2021). In Ecuador, this genus encompasses more than one third of all anuran species, with 233 out of 640 species .
The astounding species richness of Pristimantis has been attributed to terrestrial breeding -direct embryonic development without a tadpole stage (Padial et al. 2014) -and the appearance of geographic barriers as result of the Andean uplift (Lynch and Duellman 1997;Mendoza et al. 2015). However, other sympatric Andean genera with similar reproductive mode (e.g., Strabomantis Peters, 1863) and age are not as diverse suggesting that terrestrial breeding and Andean uplift are not the only factors explaining the high diversity of Pristimantis.
The number of described species of Pristimantis is increasing rapidly as result of the use of DNA sequences allowing the discovery of a large number of cryptic species. DNA sequences helps to achieve better informed taxonomic decisions and speed up species discovery with more than 40 species of Pristimantis (e.g. Ortega et al. 2015;Guayasamin et al. 2017;Páez and Ron 2019) described in Ecuador in the last five years. In some clades, the number of undescribed species outnumbers described species (e.g., Ortega et al. 2015;Páez and Ron 2019) suggesting that there could be hundreds of undescribed species of Pristimantis.
Pristimantis petersi holotype (KU 143508) is from 16.5 km NNE of Santa Rosa, Napo Province, 1700 m. However, most specimens used in the species description by Lynch and Duellman (1980) were from other populations including localities in the central Andes of Colombia (Huila and Putumayo Departments), and central Ecuador (Napo and Pastaza provinces). As currently defined, Pristimantis petersi is considered to have a wide distribution from the central Andes of Colombia (Lynch and Duellman 1980;Mueses-Cisneros 2005;Stuart et al. 2008) to the eastern slopes of the Ecuadorian Andes, from Sucumbíos to Morona Santiago Provinces at altitudes ranging between 1400-2000 m a.s.l Ron et al. 2019). Lynch and Duellman (1980) mentioned that Pristimantis petersi exhibits body size variation throughout its distribution range and, remarkably, realized that individuals from the upper Pastaza trench were larger than individuals from the type locality, Ecuador, and suspected that populations from Pastaza may represent another species.
The wide geographic distribution of Pristimantis petersi suggest that it may be a species complex. Recent reviews of Andean Pristimantis suggest that most species have highly restricted distributions. For example, in the subgenus Huicundomantis Páez & Ron, 2019 all species had distribution ranges below 5000 km 2 and the seemingly large distribution of "P. phoxocephalus" and "P. riveti" were an artifact of the combined distribution of several cryptic species (Páez and Ron 2019). Similarly, "P. calcarulatus" was considered to be distributed in the Andes from central Ecuador to southern Colombia (Hutter and Guayasamin 2015). However, genetic and morphological evidence demonstrated that "P. calcarulatus" was a complex of three different species, each with a small distribution range. Similar results have been found within "P. orestes", "P. ornatissimus" (Guayasamin et al. 2017;Urgiles et al. 2019), and P. ventrimarmoratus (Moravec et al. 2020). It is unclear if species of Andean Pristimantis with large distributions (> 10000 km 2 ) actually exist. We are unaware of any species of Pristimantis with such distribution withstanding a taxonomic review based on genetic and phenotypic characters. The available evidence suggests that species with seemingly large distributions, like P. petersi, may represent species complexes.
The existence of cryptic diversity within P. petersi is also suggested by reports of body size differences among populations of P. petersi (Lynch and Duellman 1980;Brito et al. 2017). Recent fieldwork conducted by field staff of the QCAZmuseum from Pontificia Universidad Católica del Ecuador resulted in collections of P. petersi near its type locality and in the discovery of populations of a species similar to P. petersi in Sangay National Park and Llanganates National Park. Through the integration of genetic and bioacoustic data with an exhaustive population sampling, we demonstrate that those populations are distinct from P. petersi. We describe the new species and review the content and phylogenetic relationships of the Pristimantis lacrimosus group.

DNA extraction, amplification and sequencing
We inferred the phylogenetic relationships of the new species and closely related taxa based on DNA sequences of one nuclear gene: Recombination activating gene 1 (RAG-1) and three mitochondrial genes: 12S rRNA (12S), 16S rRNA (16S), NADH dehydrogenase subunit 1 (ND1) and their flanking tRNAs. DNA was extracted from muscle or liver tissue preserved in 95% ethanol using standard Guanidine thiocyanate extraction protocols. We used polymerase chain reaction (PCR) to amplify DNA fragments. Primers used for amplification of 12S were t-Phe-frog and t-Val-frog (Wiens et al. 2005), 12SZ-L and 12SK-H (Goebel et al. 1999), for 16S, primers were 12sL13 (Feller and Hedges 1998), 16L19 and 16H36E (Heinicke et al. 2007), for ND1, primers were WL379, WL384, t-Met-frog and 16S-frog (Moen and Wiens 2009), for RAG1, primers were R182, R270, Rag1FF2, Rag1FR2 (Heinicke et al. 2007). PCR amplification was performed under standard protocols and sequenced in both directions by the Macrogen Sequencing Team (Macrogen Inc., Seoul, Korea). All sequences were assembled in Geneious 7.1.7. and then exported to Mesquite version 3.40 where each genomic region was aligned separately using default parameters in Muscle (Edgar 2004). Unambiguous alignment errors were corrected manually in Mesquite (Maddison and Maddison 2018). The aligned matrix is available in https://doi.org/10.5281/ zenodo.3785738. To calculate the uncorrected pairwise genetic p-distances of 16S we used MEGA7 on a fragment of 653 pb (Kumar et al. 2016).

Phylogeny
The phylogeny was inferred using Maximum Likelihood as optimality criterion. To choose the substitution models that best adjusted to our sequences, we used Model Finder under the command MFP+MERGE (Kalyaanamoorthy et al. 2017;Chernomor et al. 2016) as implemented in IQ-TREE 1.6.8 (Nguyen et al. 2015). We partitioned the sequences by gene and by codon position in coding genes. For the ML search we used IQ-TREE 1.6.8 (Nguyen et al. 2015) under default values. To assess branch support we obtained ultrafast bootstrap values from 2000 pseudoreplicates and 10000 iterations as maximum number to stop (commands -bb 2000 and -nm 10000 in IQ-TREE) and SH-like approximate likelihood ratio test (SH-aLRT ) with 1000 replicates (-alrt 1000 command, Guindon et al. 2010). We considered that branches with bootstrap values > 95 and SH-aLRT values > 80 had strong support. Additionally, we inferred phylogenies from mithocondrial DNA and the nuclear gene RAG1 separately to compare the topology of the phylogenetic tree derived from DNA regions with independent segregation.

Morphology
Diagnostic characters and comparisons are based on preserved specimens from Museo de Zoología at Pontificia Universidad Católica del Ecuador, Quito (QCAZ) and, when available digital photographs. Examined specimens are listed as Suppl. material 1. Character definitions and terminology follow Duellman and Lehr (2009). For subarticular tubercle terminology we follow Ron et al. (2020). Sex was determined by presence of nuptial pads or vocal slits, and direct inspection of gonads. Descriptions of coloration and variation in life are based on digital photographs. We examined the following qualitative characters: dorsal and ventral skin texture, presence of tympanic membrane and annulus, snout shape, presence of rostral papilla, presence of vomerine odontophores, presence of vocal slits and gular sac in males, relative length of fingers and toes, disc shape, presence of dorsolateral, discoidal and supratympanic folds, presence of lateral fringes on fingers and toes, presence of palmar, ulnar, tarsal, metatarsal, subarticular, supernumerary, knee, heel, and eyelid tubercles, and webbing on fingers and toes. We follow the name "hyperdistal tubercle" proposed in Ospina-Sarria and Duellman (2019) to refer to the most distal tubercle in Fingers and Toes.
To explore morphometric differentiation between species, we applied a Principal Components Analysis (PCA).
To remove the effect of size covariation, we carried out linear regressions between the morphometric variables and SVL. The PCA was applied to the residuals of the regressions. Morphometric variables associated with eyes (i.e., eye diameter, interorbital distance, eyelid width, internarial distance and eye-nostril distance) had weak correlation with snout-vent length. Low correlation appears to be a result of the difficulty of defining the eye edge on preserved specimens. Therefore, were removed those variables from the analysis. Prior to the PCA, we ran a MANOVA on the residuals to test for morphometric sexual dimorphism independent of size differences . Because the MANOVA was non-significant, we pooled the measurements of both sexes on a single PCA.

Bioacoustics
To assess species limits between the new species and the closely related P. petersi, we analyzed calls from three males of the new species: QCAZ58940, SVL = 19.3 mm, from Refuge 1, Sardinayacu, Sangay National Park, Morona Santiago Province (2.0983°S, 78.1555°W, 1406 m) collected on 21 January 2015, air temperature 19 °C, recorded in situ by Daniel Rivadeneira, and QCAZ59466, SVL = 19.1 mm, from the ravines of Yurugyacu river, Zarentza community, Llanganates National Park, Pastaza Province, (1.3524°S, 78.0597°W, 1419 m) collected on February 24 2015 and recorded in captivity on 6 March 2015 by Santiago R. Ron and one male not collected from Sardinayacu, Sangay National Park, Morona Santiago Province, recorded by Diego Batallas. Advertisement calls of P. petersi were analyzed from two adult males (not collected) from near its type locality, Cocodrilos, Napo Province, (0.66812°S, 77.7975°W, 1725 m) recorded on 22 June 2016 by Santiago R. Ron. We did not have size data for the recorded individuals of P. petersi. However, we collected other adult males on the same night and chorus and used the size of those individuals to assess interpopulation size differences. For two of the recorded males of the new species, SVL was 19.1 and 19.3 mm; for Pristimantis petersi average SVL was 17.2 mm (16.5-17.8 mm; n = 3). Recordings were made in WAV format, with a sample rate of 44100 Hz and 16-bits. Call variables were measured with RAVEN PRO 1.5 (Charif et al. 2010), under a Hanning function, 2048 DFT, sample rate of 46 kHz and a grid spacing of 20 kHz.
Most of our recordings lacked temperature information. However, the variables that allowed us to differentiate the new species from the closely related P. petersi, where call duration and call frequency which are static and are not strongly influenced by ambient temperature (Köhler et al. 2017). Moreover, equatorial Andean forests have low seasonality and with low temperature variation, so it is unlikely that our results are biased by temperature differences between localities or seasons. At Zarentza, for example, in 2015, the monthly average of the minimum daily temperature varied between 14.7 (January) and 15.9 °C (May) while at Cocodrilos (where recordings for P. petersi were made) it had a range between 15.1 (January) and 16.0 °C (November). Temperature data were obtained from the WorldClim database (https:// www.worldclim.org/).
For call measurements and terminology, we followed the call-centered approach by Köhler et al. (2017). We followed the step-by-step guide of Köhler et al. (2017) for measuring variables: we measured temporal variables such as call duration, call rate, call interval, call rise time and amplitude modulation in the oscillogram and the spectral variables such as frequency band, fundamental frequency, and dominant frequency in the spectrogram. Due to call structure, parameters such as notes, notes rates, note series, pulses and pulse rates were absent and only eight acoustic parameters (modified from Köhler et al. 2017) were measured: (1) Call duration = time from beginning to end of the call, measured from oscillogram; (2) Call rate = number of calls per minute; (3) Call interval = time from end of call to beginning of next call;(4) Call rise time = time from beginning of call to point of maximum amplitude; (5) Amplitude modulation = change in the amplitude level of a sound wave over time; (6) Frequency band = difference between upper and lower frequencies measured visually along the entire call (7) Fundamental frequency = frequency with highest energy on 1 st harmonic in the call; (8) Dominant frequency = frequency with highest energy along entire call. Recordings are deposited in the Sound Archive of Museo de Zoología QCAZof Pontificia Universidad Católica del Ecuador and are available at the Anfibios del Ecuador website, https:// bioweb.bio/faunaweb/amphibiaweb/).
The ML tree from mitochondrial DNA shows similar topology to the ML tree from all genes. The best-fit models of DNA evolution for each partition are available as Suppl. material 2.
The mtDNA tree shows strong support for the Pristimantis lacrimosus group (bootstrap = 99), for Pristimantis petersi (bootstrap = 100) and for the new species (bootstrap = 99). The ML tree inferred from RAG1 shows lower support values but is congruent in showing a monophyletic Pristimantis lacrimosus group (bootstrap = 91) and in confirming a close relationship between P. petersi and the new species. The mtDNA and RAG1 phylogenies do not show strongly supported incongruences. Mitochondrial DNA and RAG1 phylogenetic trees are available as Suppl. material 3.
The new species is the sister to P. petersi and an undescribed species from Cordillera Escalera, Peru. The uncorrected pairwise p-genetic distances for 16S between P. petersi and the new species range from 7.9% to 8.4%. The clade comprising these three species is sister to an Figure 1. Maximum likelihood tree of the Pristimantis lacrimosus group inferred from a partitioned analysis of 4026 aligned sites of DNA sequences of the mitochondrial genes 12S, 16S, and ND-1 and the nuclear gene RAG-1. Bold characters highlight individuals included for the first time in a phylogeny, red taxa highlight previously misidentified species. Pristimantis petersioides sp. nov. is highlighted in blue. SH-aLRT support (above branch) and ultrafast bootstrap support (below) are shown as percentages; asterisks denote 100% for both measures. Outgroups are not shown. Number for voucher museum specimens are shown to the left of the species name; locality is shown to the right and country abbreviation at the end, as follows: ECU Ecuador, PER Peru, COL Colombia, VEN Venezuela, GUY Guyana, PAN Panama. undescribed species from Bombuscaro, Podocarpus National Park, Zamora Province, Ecuador. Samples of the new species separate in two sister clades, a northern clade in Zarentza, Pastaza Province and a southern clade from Sardinayacu and Bosque Protector Abanico, Morona Santiago Province; the uncorrected pairwise p-genetic distances between these clades range from 1.2% to 1.4%.

Morphometric analysis
MANOVA results showed no sexual dimorphism (after removing size effects) on the new species and P. petersi. The PCA show broad overlap in morphometric space between both species (Fig. 2). PC I (49.5%) had high load-ings on head width, head length, and tibia length while PC II (21.5%) had high loading on tympanum diameter. Both principal components explained 71.0% of the morphometric variation ( Table 2).

Systematic account
The differences in advertisement calls (see Comparisons with other Species section), the branch lengths in the phylogeny, and genetic distances indicate that the new species, in fact, represents a different species from P. petersi (see below). In the following section, we update the species content of the Pristimantis lacrimosus group and describe the new species.

Pristimantis lacrimosus species group
Content. We include all the descendants from the most recent common ancestor of P. eremitus and P. lacrimosus according to Ron et al. 2020. We exclude Pristimantis eugeniae from this group because it belongs to the sister clade of P. lacrimosus species group. We also include P. degener which is sister to P. subsigillatus and the new species.
Distribution. The Pristimantis lacrimosus group is distributed in Central America, the Guianan Shield, Pacific Basin of Ecuador, and the Amazon Basin. Its species richness peaks in the Ecuadorian Andes (n = 19) and Amazon basin of Ecuador and Peru (n = 14).
Remarks. We refrain from assigning Pristimantis sneiderni (Ospina-Sarria and Duellman 2019) to the Pristimantis lacrimosus group due to the lack of molecular evidence and following Ron et al. (2020).
Description of the holotype. Adult female (QCAZ58939). Measurements (in mm): SVL 22.02; tibia length 12.07; foot length 10.72; head length 8.82; head width 9.09; eye diameter 2.96; tympanum diameter 1.35; interorbital distance 2.52; upper eyelid width 2.44; internarial distance 1.59; eye-nostril distance 2.59; tympanum-eye distance 0.71. Body slender; head slightly wider than long, wider than body; snout rounded to truncate with rostral papilla in dorsal view, truncate in lateral profile; canthus rostralis distinct, slightly curved in dorsal view; loreal region concave; interorbital space flat, no cranial crests; eye large, protuberant; upper eyelid about 97% of interorbital distance, bearing one subconical tubercle. Tympanic membrane and annulus distinct, rounded, with inconspicuous supratympanic fold, partially obscuring anterodorsal edge; horizontal diameter of tympanum about 13% of head length, separated from eye by a distance about one half tympanum length; choanae large, rounded, not concealed by palatal shelf of maxillary arc; dentigerous processes of vomers prominent, oblique, bearing a transverse row of five teeth; tongue big, elliptical, posterior border slightly notched, 40% of the anterior surface adherent to floor of mouth. Skin on dorsum smooth to shagreen; dorsolateral folds absent; skin on upper flanks bearing scattered low tubercles; skin on belly weakly areolate; skin on throat and chest smooth; discoidal fold ill-defined; skin in upper cloacal region shagreen. Forearms slender bearing low antebrachial tubercle and one subconical ulnar tubercle at the distal half of the forearm; fingers large and slender, all with broadly expanded pads, all fingers with discs; fingers bearing narrow lateral fringes; relative lengths of fingers I < II < IV < III; three subarticular tubercles on finger III (Fig. 4B), the most distal we refer as hyperdistal, all the tubercles well defined, round in ventral and lateral view; several supernumerary tubercles present, prominent at the base of the fingers and lower, indistinct at the palmar surface; palmar tubercle bifid, heart-shaped, about the same length and twice the width of elliptical thenar tubercle (Fig. 4B).
Hindlimbs slender; tibia length about 55% of SVL; upper surfaces of hindlimbs smooth; foot length about 48% of SVL, posterior surfaces of thighs smooth, ventral surfaces of thighs slightly areolate; knee and heel lacking tubercles; outer surface of tarsus bearing three low, inconspicuous tubercles, equally distributed along tarsus; toes bearing narrow lateral fringes; webbing between toes absent; discs on toes broadly expanded as those on fingers, rounded; relative lengths of toes: I < II < III < V < IV; Toe V much longer than Toe III (disc on Toe III reaches proximal edge of penultimate subarticular tubercle on Toe IV, disc on Toe V exceeds the distal edge of penultimate subarticular tubercle on Toe IV), subarticular tubercles rounded, simple, elevated; plantar surface with low supernumerary tubercles, bearing four subarticular tubercles (Fig. 4A), inner metatarsal tubercle prominent, elliptical, approximately 3x size of oval and conical outer metatarsal tubercle (Fig. 4A).
Color of holotype in preservative. (Fig. 3C, D) Background color pale grayish cream with scattered, irregular dark brown chevrons, head bearing dark brown supratympanic and canthal stripe, upper lip bearing ill-defined stripe formed by irregular dark brown dots; upper flanks bearing dark brown, irregular flecks and blotches densely distributed; venter, ventral surfaces of forearms and hindlimbs pale creamy white, chest and throat with diminutive dark brown dots uniformly distributed (visible under magnification); ventral surfaces of hands and foot with dense minute dark brown dots, posterior surfaces of thighs pale cream to dark brown; iris reddish coppery with fine, dense, black reticulation.
Color of holotype in life. (Fig. 3A, B) Dorsal surfaces yellowish green with scattered, irregular dark brown chevrons; canthal stripe and supratympanic fold black, upper flanks pale cream with dark brown irregular flecks and blotches; venter creamy white; axils pinkish white; ventral surfaces of limbs, thighs yellowish green; iris reddish copper with dark bronze faint horizontal streak and thin irregular black reticulations.
Variation in life. (Fig. 7). Tuberculation pattern varies from dorsum completely smooth (e.g., QCAZ58943, 58951) to dorsum shagreen (e.g., QCAZ58938, 58939), some individuals bear scattered small tubercles on anterior half of dorsum (e.g., QCAZ58880) or have the dorsum densely tuberculated (e.g., QCAZ59463). When dorsum is tuberculated, flanks and limbs usually bear scattered tubercles more conspicuous than those in the dorsum. Similarly, the interorbital tubercle and upper eyelid tubercles are more prominent when the dorsum is tuberculated. There is extensive variation in dorsal coloration (Fig. 7). Dorsum varies from dark greenish brown (e.g., QCAZ59471), bright orange (e.g., QCAZ58943), olive green (e.g., QCAZ58938), to pale yellowish green (e.g., QCAZ58941). Dark marks on dorsum vary from scattered dark brown flecks to irregular brown chevrons that form a triangle that extends from the ilium to the scapula, to ill-defined, dark brown flecks and spots (e.g., QCAZ58948, 59455). Some individuals bear bright orange blotches limited by dark brown contours (e.g., QCAZ58951, 59458), a bright orange middorsal bar that extends from the snout to the cloaca (e.g., QCAZ59456), black dorsolateral stripes suffused with supratympanic stripes (e.g., QCAZ58943). Bright orange to yellow with a darker contour interorbital stripe or bar may be present (e.g., QCAZ59455,59458,59462) or absent (e.g., QCAZ58943, 59456). Snout varies from dark green-   Table 3. The call is a metallic click with an average duration of 0.25 s (0.19-0.32 s; n = 3; Fig. 8). The amplitude peak occurs at 20-30 ms and then decreases gradually towards the end (Fig. 8). The calls are repeated at a mean rate of 19.89 calls per minute (11.26-25.78; n = 3). Three or four harmonics are visible, but most of the energy is located on the first one. The dominant frequency (= fundamental frequency) is  Hz; n = 3).  Distribution and natural history. Pristimantis petersioides sp. nov. is known from six localities in the eastern Andean slopes of central Ecuador between 1221-2300 m (Fig. 9). It inhabits the Eastern Andean Foothills Forest and Eastern Montane Forest natural regions (as defined by Ron et al. 2019). It has been recorded in primary forest and, less frequently, in secondary forest. Individuals were found during nocturnal surveys, usually perching on ferns, herbs, or Heliconia leaves, branches, or inside bromeliads up to 350 cm above the ground, usually near water bodies. Three amplectant pairs were found on January and February 2015 in Sardinayacu and Zarentza.
Etymology. The specific epithet is a masculine noun in apposition. The suffix oides is derived from the Greek eidos meaning similar. The name makes reference to the   Lynch and Duellman (1980), Mueses-Cisneros (2005), and Stuart et al. (2008). similarity between the new species and its sister species, Pristimantis petersi.
Conservation status. Four out of six known localities are inside National Parks (Sardinayacu in Parque Nacional Sangay and Ankaku, Zarentza and Salcedo-Tena road in Parque Nacional Llanganates); nonetheless, based on a vegetation cover map (Ministerio del Ambiente 2018a) and a deforestation map 2016-2018 (Ministerio del Ambiente 2018b), Zarentza is < 1 km from deforested areas for agriculture. At the year of collection (2009) the locality at Salcedo-Tena highway was in a forested region with small, deforested patches at distances > 2.5 km (based on a 2008 deforestation map by Ministerio de Ambiente). Sardinayacu, refuge 3 occur > 6 km from pastures, while Sardinayacu, refuge 1 is < 0.5 km from deforested areas for agriculture.
In Sardinayacu, this species was one of the most common during surveys (24 individuals found in 9 days by 13 people) which suggest it can be locally abundant.  also reported abundant populations in the upper basin of the Upano river, Sangay National Park, Morona Santiago Province (referred both as "Pristimantis petersi" and also "P. aff. petersi"). Its extent of occurrence is 1402 km 2 (based on a minimum convex polygon). Despite being locally abundant, we consider Pristimantis petersioides sp. nov. to be in the Red List category Vulnerable (VU) following B1, B2ab(iii) IUCN criteria because: (i) it is only known from six localities (sensu IUCN 2017), (ii) its Extent of Ocurrence is less than 5000 km 2 (1433 km 2 ); and approximately 9% of its Extent of Ocurrence has been affected by deforestation, human settlements and agriculture (Fig. 10).

Discussion
On the identity of Pristimantis petersi Pristimantis petersi was considered to have a wide distribution from the central Andes of Colombia in Caquetá, Huila, and Putumayo (Lynch and Duellman 1980;Mueses-Cisneros 2005;Stuart et al. 2008), to the eastern slopes of the Ecuadorian Andes, from Sucumbíos to Morona Santiago Provinces Ron et al. 2019). Herein, we show that it was composed of two species which appear to be allopatric, south and north of the Quilindaña paramos in Napo Province. Lynch and Duellman (1980) remark of size differences between populations from the north and south of "P. petersi" was not supported in our data but their suspicion of the distinctiveness of the populations from the Pastaza trench was correct. Based in our review, we tentatively consider Pristimantis petersi as distributed from the central Andes of Colombia to Napo Province (Fig. 9). We recommend verifying the identity of Colombian populations using genetic data. Recent reviews of Andean Pristimantis indicate that species usually have a restricted distribution (e.g., Páez and Ron 2019). The geographic distance of Colombian populations (up to 320 km from the type locality) suggest that, at least some of them, could represent a separate species. Guayasamin and Funk (2009) reported an abundant population of "Pristimantis cf. petersi" at Yanayacu Biological Station. Examination of voucher specimens deposited at the QCAZcollection indicate that they are not conspecific with P. petersi nor P. petersioides sp. nov.
Our results and those of previous systematic reviews (show that eastern montane forests still harbor many undescribed species of Pristimantis. Similar findings have been previously reported by Ortega et al. (2015), Páez and Ron (2019), and Ron et al. (2020). As in previous reviews (e.g., Restrepo et al. 2017;Páez and Ron 2019), we also found broad intraspecific and intrapopulation variation in dorsal color within P. petersioides sp. nov. (Figs 6, 7) and P. petersi (Fig. 11). This large intraspecific and intrapopulation variation hinders the use of dorsal coloration for diagnosis between both species. Most individuals have greenish dorsal color which is characteristic of several species of the P. lacrimosus group. We did not find diagnostic morphological characters to distinguish the new species from Pristimantis petersi, which highlights the importance of including molecular and bioacoustic data to clarify species identity.

Use of bioacoustics for species delimitation
Similar to Páez and Ron (2019), our morphometric analysis was of little help to distinguish closely related species of Pristimantis. In contrast, advertisement calls and genetic data indicate that P. petersioides sp. nov. represents a lineage independent from P. petersi. We found differences with little or no overlap in two static call traits, call duration and dominant frequency (Köhler et al. 2017). Moreover, differences in call frequency are likely an underestimate because the calling males of P. petersioides sp. nov. were larger than those from the recorded chorus of P. petersi. Because there is an inverse relationship between body size and call frequency (Gerhardt and Huber 2002), the higher frequency of the call of P. petersioides sp. nov. would be likely higher after a size correction.
Bioacoustic comparisons are of importance for taxonomy because advertisement calls mediate species recognition and mate choice (e.g., Ryan and Rand 1995). It has been widely accepted that calls are among the most useful characters differentiating closely related anuran species (Vences and Wake 2007). However, calls have been of limited use in the taxonomy of Pristimantis (Duellman and Lehr 2009). Our study and some recent works (Hutter and Guayasamin 2015;Páez and Ron 2019;Székely et al. 2020) highlight the usefulness of bioacoustic characters in Pristimantis taxonomy. Future taxonomic reviews will benefit from a more comprehensive knowledge of advertisement calls in Pristimantis.