The primary study site is Cerro Chucantí that includes the highest peak (1439 m a.s.l. at 8.8046°N, 78.4595°W; Fig. 1) in the Maje Mountains, an isolated massif in Darién, Panama. It is a sky island with a small cloud forest around its peak of < 5 km² width. The nearest comparable cloud forests are at least 100 km away on Cerro Pirre and Cerro Pechito Parao. The higher elevations of Cerro Chucantí are part of the Eastern Panamanian Montane Forests ecoregion (Fund 2014). Annual precipitation varies between 3,000 mm and 4,000 mm and occurs mainly from April to December (Rio Maje Meteorological Station, 70 m a.s.l. http://www.hidromet.com.pa/, accessed on 19 September 2015). The average temperatures on Cerro Chucantí, measured with data loggers in 2018 and 2019, decreased with elevation from 23.5 °C at 770 m a.s.l., to 21.1 °C at 1025 m a.s.l., and 19.1 °C at 1269 m a.s.l., yet was with 22.1 °C again higher on the top at > 1400 m a.s.l., possibly as a result of increased solar radiation due to reduced canopy cover in the cloud forest. The following vegetation zones occur on Cerro Chucantí at different elevations: Lowland Moist Forest (0–500 m a.s.l.), Premontane Moist Forest (500–1000 m a.s.l.) and a small area of Premontane Wet Forest (≈ cloud forest) higher than 1000 m a.s.l. (Holdridge 1966). All geographical coordinates were recorded in the WGS 1984 datum and presented in decimal degrees. The distributional map was created using QGIS (QGIS 2018) with an Open Street Map (OSM) layer (OSM 2015).

Figure 1. 

Map showing locations of Pristimantis gretathunbergae sp. nov. and Pristimantis cruentus in Panama. Numbers in the map correspond to localities mentioned in methods. Internal divisions in the map correspond to provinces in black letters.

Additional material for molecular and/or morphological analysis was obtained from specimens collected in eastern Panama (location names in bold as used in article): 1) Maje Ambroya, Maje Mountains, Panama Province; 2) Cerro Chucantí, Province Darién; 3) Río Tuquesa, Cerro Pechito Parao at Bajo Pequeño, Darien Mountains, Embera-Wounaan Comarca (= indigenous autonomous region). Third party material was collected in Central Panama from: 4) Cerro Brewster, Piedras-Pacora Mountains of Chagres National Park, Province Panama; 5) only photographic vouchers from Cerro Bruja, Chagres National Park, Province Colon; 6) a single DNA sequence from El Cope National Park at Rio Blanco, Penonomé Mountains, Province Cocle; and 7) Altos del Maria, vicinity of Gaita Hills, Province Panama Oeste (Fig. 1).

For molecular analyses of Panamanian samples, DNA was extracted from fresh liver tissue. The 16S mtDNA extraction and sequencing follow previously described routines (Batista et al. 2016a). The COI fragments were sequenced in the Southern China DNA Barcoding Center. The mtDNA sequences obtained were compared and related specimens from Colombia and Ecuador published in GenBank, with those retained for the analysis that were informative per region (i.e., only one sequence/taxon/location). The sequences were aligned with CLUSTAL W (Larkin et al. 2007) and edited by eye using Geneious version 4.8.5 (Kearse et al. 2012). A list of specimens included in the genetic analysis with corresponding GenBank accession numbers is appended in the Suppl. material 1. GenBank sequences of Craugastor sagui and C. crassidigitus were used as outgroups. The final 16S alignment comprised 97 sequences obtained from this study and from GenBank consisting of 500 bp, of which 339 sites were variable, 187 parsimony-informative, and 50 singletons. For COI gene analyses Diasporus diastema, C. longirostris and C. crassidigitus were used as outgroup. The final alignment comprised 8 sequences from our material and 30 ones from GenBank, consisting of 567 bp, of which 293 sites were variable, 282 parsimony-informative, and 11 singletons. The final alignment for the COI and 16S genes together comprised 49 sequences consisting of 1053 bp, of which 412 sites were variable, 359 parsimony-informative, and 52 singletons.

A Maximum Likelihood analysis (MA) was conducted for both genetic markers using IQ-TREE (Nguyen et al. 2015; Trifinopoulos et al. 2016). To estimate support, 1000 replicates of ultrafast bootstrapping (Minh et al. 2013) were performed. A nodal or branch support with SH-aLRT values ≥ 80% is considered reliable for a clade (Guindon et al. 2010). A substitution model using JModeltest 0.1.1 (Posada 2008) was selected for the Bayesian Inference (BI) analysis under the corrected Akaike Information Criterion (AICc; Akaike 1974), for the 16S gene. However, the resulting TIM2+I+G model was replaced by the GTR model (Lecocq et al. 2013). The 3-parameter model with rate heterogeneity, TIM2+I+G (Kimura 1981) was implemented for the combined gene data set. We ran a Bayesian phylogenetic analysis in MrBayes 3.1.2 (Huelsenbeck and Ronquist 2001) for 10,000,000 generations with four default chains, sampling every 1000 generations and for the final consensus tree discarding the first 25% as burn-in. To test species delimitation among the Pristimantis species included in this study, the Automatic Barcode Gap Discovery (ABGD) algorithm (Puillandre et al. 2012) was applied for the 16S gene under the following settings: steps = 20, distance = Kimura 2-parameter model with transversion/transition ratio of 2.0. The setting for the minimum relative gap width (X) was set to 0.5.

Collecting permits for 2009 (SC/A-8-09, SC/A-28-09), 2011 (SC/A-37-11), 2012 (SC/A-33-12), 2016 (SE/A-60-16) and 2018 (SE/A-33-18) as well as export permits for 2012 (SC/A-33-12) and 2013 (SEX/A-7-13) were provided by UNARGEN-Ministerio de Ambiente, Panama. Finally, we applied the traditional Environmental Vulnerability Score (EVS) methodology by Wilson and McCranie (2004) to calculate the conservation status of this species. This method assigns increasing values to higher conservation priorities based on geographic and habitat distribution, and reproductive mode; in anurans from 1–17, or up to 20 in the revised version by Johnson et al. (2015).

Specimens were collected by hand, photographed alive, euthanized with the Solution Tanax T-61, fixed with a preservative solution of 5 mL formalin (36%) in 1 L ethanol (94%), and subsequently stored in ethanol (70%) following the protocol of Batista et al. (2016a). Preserved specimens were later analyzed at the Zoological Laboratory of the Universidad Autónoma de Chiriquí. All figures were assembled and some improved using Adobe Photoshop CS6. Specimens are deposited at the Museo Herpetológico de Chiriquí (MHCH, Universidad Autónoma de Chiriquí, David) in Panama, and at the Senckenberg Forschungsinstitut Frankfurt (SMF) in Germany. The abbreviations for museum collections follow Sabaj (2016), with field numbers AB for initials of Abel Batista and MG of Macario Gonzalez. Morphological nomenclature, measurements and standardized diagnosis characters follow Duellman and Lehr (2009). Some comparative morphological data of similar Pristimantis species in Colombia were extracted from the respective original descriptions, as well as a few online photo repositories (see online Suppl. material 2). For color descriptions, we applied the code of Köhler (2012).

Sex of specimens was determined by morphometric characters and presence of eggs in Panamanian samples. Measurements were taken to the nearest 0.1 mm, using a stereomicroscope and precision digital calipers. Following variables were measured according to Batista et al. (2016a) and Duellman and Lehr (2009):

SVL Snout-Vent Length

HW Head Width, measured between posterior end of jaws

HL Head Length, measured between posterior end of jaws and tip of snout

InD Internarial Distance as shortest line between inner edges of narial openings

IoD Interorbital Distance as shortest distance between visible eyes, reflecting size of braincase

EW Eyelid Width, perpendicular distance to the outer edge of the upper eyelid

ED Eye Diameter as length of exposed eye

EN Eye-Nostril distance as shortest distance between anterior corner of eye and posterior margin of nostril

TY Tympanum Diameter

TL Tibial Length from knee to distal end of tibia

FL Foot Length between proximal edge of inner metatarsal tubercle to tip of fourth toe

FAL Forearm Length between elbow and hand

HAL Hand Length between proximal edge of palmar tubercle to tip of third finger

BW Body Width as largest width on trunk

AGD Axilla-Groin Distance as length between hind and front limbs along the trunk

TrL Trunk Length as SVL minus HL

3FW Width of 3^rd Finger at penultimate phalanx just anterior to disc

3FD Disk Width of 3^rd Finger

3TW Width of 3^rd Toe at penultimate phalanx just anterior to disc

3TD Disk Width of 3^rd Toe

4TW Width of 4^th Toe at penultimate phalanx just anterior to the disc

4TD Disk Width of 4^th Toe

Interspecific differences among Pristimantis spp. and related species are known to be relative lengths of heads, hind limbs, and feet (Duellman and Lehr 2009). Consequently, multivariate analyses were conducted to investigate morphometric differences between sympatric P. cruentus and the new species. To reduce the impact of ontogenetic and gender differences on measures of all body parts, 15 meaningful ratios of our initially measured variables were applied. To account for potential head shape differences, the measured distances along the head were put into relation to head length, i.e., InD/HL, IoD/HL, ED/HL, TY/HL, EN/HL, EW/HL, and proportionally larger head size was reflected by TrL/HL, whereas stockiness is investigated by BW/SVL. Similar, sizes relating to limb length were put into relationship with the approximate trunk length, i.e., FL/TrL and TL/TrL, whereas hand and foot length were measured against forearm HAL/FAL, respectively shank FL/TL. Relative size of disk width to digits of finger and toes were represented by 3FD/3FW, 3TD/3TW, and 4TD/4TW. We applied a Principal Component Analysis (PCA) for variable selection and therefore removed redundant (highly correlated) ones. A Linear Discriminant Function Analysis (LDFA), or simply Linear Discriminant Analysis (LDA), with the remaining morphometric variables (ratios) followed to elucidate the potential differences of body proportions between the two sympatric Pristimantis species.

We conducted a Multiple Correspondence Analysis (MCA) in R (Version 4.0.3), using the FactoMineR package, on categorical variables to compare the presence/absence of certain color pattern and tubercle characters between sympatric Pristimantis gretathunbergae sp. nov. and P. cruentus. We assessed the character state of six variables from photographic material of 26 P. gretathunbergae sp. nov. and 17 P. cruentus, whereby several specimens were collected and inspected by us, and their taxonomic allocation confirmed through molecular means. The value 0 was assigned to the morphological state typical for P. gretathunbergae sp. nov., whereas the value 2 is typical for P. cruentus, and the value 1 represents an intermediate expression. Follow do the descriptive states for these six variables: A) iris coloration: 0 = blackish to very dark red; 2 = whitish, golden, or light reddish, B) iris reticulation: 0 = no pattern, some with golden red sparkling; 1 = some dark red, small patches; 2 = reticulation, C) upper eyelid tubercle: 0 = single conical to spine-like; 1 = short singular, but spine-like, higher than other subtriangular humps; 2 = not singular or none at all, D) upper lip coloration: 0 = light colored and sharply demarcated to darker snout coloration; 1 = light color with some dark patches ingressing vertically from the snout, but upper border of light colored parts on the lip still contrastingly sharp bordered; 2 = no light color or very diffuse, no upper dark border, E) groin coloration: 0 = colors are relatively uniform white, yellow, light olive, or red, but some show a flecking pattern of these colors; 2 = dark brown to black flecks or patches on a light ground, F) ventral coloration: 0 = unicolored or fine spotting on white, yellow or orange: 2 = heavily dark mottled.

In the following, we present information on both genes (16S and COI) separately, as well as their combined results. However, we focus more on 16S for the presentation on closely related taxa, as 16S is more widely used and thus comparable with many Neotropical anurans (Fouquet et al. 2007; Vacher et al. 2020). Furthermore, 16S tends to be more appropriate when using a few phylogenetically and/or geographically close taxa. Limitations with COI are the lack of a universal primer for the PCR amplifications across numerous different species and high rates of intraspecific genetic variations (Vences et al. 2005, 2012; Grosejan et al. 2015; Estupinan et al. 2016).

The ABGD analysis generated ten groups of species by initial partition with prior maximal distance P = 1.45^e-02 (Distance K80 Kimura MinSlope = 0.5) and a relative width of barcoding gap of 0.05 X-value (Fig. 2). Genetic divergence values among groups for 16S and COI genes combined are shown in Table 1, whereas the respective values of each gene alone are shown in the Suppl. material 1: Tables S1, S2). Group 1 received a high SH-aLRT support of 96% (bootstrap support of 95%) and includes all Panamanian specimens that have originally been labeled as P. aff. latidiscus on GenBank, but which show a large genetic divergence of > 11% at 16S to the original P. latidiscus from Ecuador, South America. All other samples were grouped with high bootstrap support in their corresponding known species, with the lowest, yet still good support of 89.% for Group 3 (P. erythropleura) and 86.8% for Group 2 that consisted of single specimens originally labeled as P. cruentus (SMF 97539), P. paisa (AJC 1344) and P. viejas (EMM 247), see also Reyes-Puig et al. (2020). However, these latter three specimens actually represent P. penelopus from northwestern Colombia, errors that were already addressed/corrected by Restrepo et al. (2017). The third specimen of Group 2 (P. cruentus SMF 97539) is a close relative collected by us from a 200 km distant, montane site near the Pacific coast of Panama. Morphologically it resembles P. cruentus but was provisionally labeled as P. aff. sanguineus/penelopus due to genetic results. Group 5 represents P. cruentus with a perfect SH-aLRT support of 100% but a comparatively lower, yet still moderately well-supported bootstrap value of 75%, that possibly is the result of a large genetic variation and indicates an unresolved species complex (Crawford et al. 2012; Estupinan et al. 2016), which is also shown by the large genetic distance of 6% at 16S between the distinct groups cruentus and aff. cruentus (see Suppl. material 1: Table S1). Furthermore, the cruentus clade contains the specimen CH 6456 from Cana, Darien Province, Panama, originally labeled as P. aff. latidiscus (Crawford et al. 2012). This specimen was relabeled as P. cruentus, hence, P. latidiscus is removed from the list of Panamanian Pristimantis species, because all other originally labeled P. latidiscus (members of Group 1) actually represent the new, undescribed species. The shortest genetic distance (16S and COI combined) of this new rainfrog to any other Pristimantis species is 9.6% to P. penelopus and 11% to P. erythropleura (Table 1).

Figure 2. 

Phylogenetic tree of Pristimantis spp. based on mtDNA genes 16S and COI performed by a Shimodaira-Hasegawa approximate likelihood ratio test (SH-aLRT test). Numbers on nodes indicate estimated SH-aLRT support/bootstrap support with SH-aLRT values ≥ 80% are considered reliable for a clade (Guindon et al. 2010). The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. Colored bars and G-numbers represent the groups generated by ABGD analysis (see results of phylogenetic analysis and Suppl. material 1: Table S1 for more details).

On a wider perspective, the phylogenetic inference based on combined 16S and COI sequences placed the Pristimantis erythropleura-penelopus clade, P. cruentus, P. cisnerosi (data of cisnerosi available only for 16S, see also Suppl. material 1: Table S1 and Suppl. material 2: Figs S1, S3), and the new species into one larger clade with moderate support, possibly reflecting still unresolved species complexes. However, the new species clearly formed a distinct lineage (Fig. 2). With the phylogeny analyzed by the approximative likelihood test (SH-aLRT test), the South American species were grouped in a clade separated from P. gretathunbergae sp. nov. and P. cruentus, both from Panama. Applying only 16S, a ML analysis placed P. cisnerosi as sister species to the new species (not shown). Pristimantis cisnerosi is a rainfrog of lower elevations, 70–600 m a.s.l., from the Choco forests of southwestern Colombia (see Reyes-Puig et al. 2020, and specimens depicted in Suppl. material 2: Fig. S16), whereas P. gretathunbergae sp. nov. occupies elevations higher than 700 m a.s.l. at sites of more than 400 km further north in Panama. In contrast to ML, results of 16S from a BI analysis placed P. cisnerosi as sister to the P. erythropleura-penelopus clade, with P. erythropleura occurring at elevations primarily > 900 m and P. penelopus inhabiting lower to higher elevations around 150–200 km south of the new species (Suppl. material 2: Fig. S3). Biologically and geographically, the BI tree with 16S alone or combined with COI is in accordance with the current distributional patterns of the species included in our phylogenetic analyses (Fig. 2).

Pristimantis gretathunbergae sp. nov. reveals a genetic variation of < 3% for 16S between our focus populations from Maje Mountains (Cerro Chucantí and Maje Ambroya) to related populations from Rio Tuquesa (divergence of 1.5–2.8%) and Cerro Brewster (1.9–2.9%), but also to a single sequence from farther west at El Cope, Central Panama (2.3%). This corresponds to the suggested and applied minimum sequence divergence of 3% between Neotropical frog species (Fouquet et al. 2007; Vacher et al. 2020). Additional genetic variations by species for single genes 16S and COI are displayed in Suppl. material 1: Tables S1, S2.

The shortest genetic distance for 16S mtDNA between the new Pristimantis species and any other Pristimantis sample was 4.4% and pertains to two specimens of allopatric Colombian relatives, one P. erythropleura (minimum of 14.6% at COI to the new species) and one P. penelopus (min. of 15.5% at COI to the new species). The mean difference at 16S in these Colombian subsamples to P. gretathunbergae sp. nov. is 4.8%, that increased with the addition of a few samples from other sites in the same general region to 5.9% (P. erythropleura), respectively 6% (P. penelopus, see also Suppl. material 1: Table S1). While a slightly shorter mean sequence divergence to the new species is also reflected at COI (16.0% in P. erythropleura vs. 16.3% in P. penelopus; Suppl. material 1: Table S2), the combination of both genes reversed that order, as P. penelopus exhibits a shorter distance to the new species (16S and COI combined: 9.9% in P. penelopus vs. 11.0% in P. erythropleura, Table 1). Yet, their very close relationship is displayed in the SH-aLRT phylogenetic tree (Fig. 2). Genetic divergence of the new Pristimantis species is similarly low towards a single specimen of P. viejas with a mean value of 5.5% at 16S, however, with considerable higher values at COI (19.7%) and COI with 16S combined (13.7%). Furthermore, P. viejas was placed in a clade with P. cerasinus (Fig. 2), corroborating the results in Amezquita et al. (2019). In contrast to allopatric Colombian species, sympatric Pristimantis spp. in Panama are more distant, as P. cruentus is the closest relative with a divergence at 16S of > 9.6% to the new species (19.0% at COI; and 14.9% at 16S and COI combined; Table 1, Fig. 2).

Results of morphometric measurements of adult specimens of Pristimantis gretathunbergae sp. nov. are presented in Table 2. It generally resembles the sympatric P. cruentus, yet differs from it, as well as all other known Pristimantis spp. occurring in Panama (see Comparative diagnosis and Figs 3–6) by having poorly defined tympanic membrane, absence of vocal slits, and absence of nuptial pads (illustrative examples in Figs 4, 5 and Suppl. material 2: Figs S8–S11). Other qualitative variables (color pattern, tubercles) and parametric variables (body proportions) have been analyzed statistically.

A PCA revealed the following relative variables to contribute mostly to the principal components: TrL/HL, 3TD/3TW, 3FD/3FW, 4TD/4TW with strong loadings and IoD/HL, ED/HL, EW/HL with medium loadings (Suppl. material 2: Fig. S4 and Suppl. material 1: Table S3 with loadings). In the first PCA axis (67.64%), P. gretathunbergae sp. nov. and P. cruentus display no differences (Mann-Whitney-U-Test, W = 301, p = 0.834), whereas the second PCA axis (14.04%) reveals significant differences between the two species (Welch Two Sample t-test, t = 6.74, df = 15.473, p < 0.001; Suppl. material 2: Fig. S5).

A subsequent Linear Discriminant Analysis LDA correctly separated P. gretathunbergae sp. nov. (n = 9) from P. cruentus from eastern Panama (n = 27) and western Panama (n = 37). Pristimantis cruentus had to be split into two separate geographic groups based on LDA-conditions (LDA graph in Suppl. material 2: Fig. S6). On average, 79.4% of the specimens from all three groups were classified correctly according to our a priori groupings. Pristimantis gretathunbergae sp. nov. was classified correctly by 77.78%. The four morphometric variables that contributed the most to the LDA groupings in order of relevance were: 1) IoD/HL, 2) EW/HL, 3) ED/HL, 4) 4TD/4TW, followed by TrL/HL and characters of finger disk proportions (coefficients of LDA in Suppl. material 1: Table S4). These results indicate a principal difference between the three groups in head morphology, eye size (eyelid width EW and eye-diameter ED likely relate similarly to eye size), and toe characters.

A final univariate analysis corroborates that in four morphometric variables used in the LDA P. gretathunbergae sp. nov. differ significantly from P. cruentus, for which Eastern and Western populations were combined (unlike in the LDA): P. cruentus exhibits a relatively larger eye (mean ED/HL = 0.414; mean P. gretathunbergae sp. nov. = 0.322; Welch Two Sample t-test, t = 6.297, df = 25.65, p < 0.001), and eyelid width (mean sp. nov. EW/HL = 4.50, mean P. gretathunbergae sp. nov. = 0.376; Welch Two Sample t-test, t = 4.667, df = 25.97, p < 0.001); a longer trunk (mean TrL/HL = 2.109; mean P. gretathunbergae sp. nov. = 1.529; Mann‑Whitney‑U‑Test, W = 548, p < 0.001), and a wider head (mean IoD/HL = 0.368; mean P. gretathunbergae sp. nov. = 0.259; Welch Two Sample t-test, t = 7.591, df = 18.8, p < 0.001). These results indicate that the head morphology relates primarily to the separation of P. gretathunbergae sp. nov. and all P. cruentus. Toe characters, important in the LDA, are only significantly different between P. cruentus from eastern Panama and western Panama (4TD/4TW: Mann‑Whitney‑U‑Test, W = 686, p = 0.011), whereas they are marginal to not different between P. gretathunbergae sp. nov. and P. cruentus from eastern Panama (4TD/4TW: Mann‑Whitney‑U‑Test, W = 73, p = 0.079) or from western Panama (4TD/4TW: Mann‑Whitney‑U‑Test, W =151, p = 0.683), respectively.

A Multiple Correspondence Analysis (MCA) of six categorical variables of color pattern and tubercle properties results in a clear distinction between the new Pristimantis gretathunbergae sp. nov. and its closest relative in sympatry, P. cruentus. The most frequent and/or typical expression of these variables in Pristimantis gretathunbergae sp. nov. (with the comparative expression of P. cruentus in parenthesis) are: 1) blackish eyes or iris (light colored iris in P. cruentus), 2) no iris reticulation (reticulated), 3) a single conical tubercle on the upper eyelid (rarely so, generally more variable from subtriangular to spine-like, and from none at all to several ones), 4) light upper lip contrastingly bordered to dark coloration on snout above (none, diffusely colored lips, or light, but not demarcated), 5) coloration of groin, as well as 6) venter is unicolored whitish, yellow or reddish, sometimes with fine spotting (heavily black and white to dark and light mottled, see methods for a more detailed and expanded species-specific variable definition and quality scoring). A photographic example of a Pristimantis gretathunbergae sp. nov. and a P. cruentus in a face-off position is depicted in Fig. 4C, while more explicit photographic material for comparison between these two species can be viewed in Suppl. material 2: Figs S10, S11.

The first and second dimension of the MCA describe 73.89% of the total variance, allowing a conclusive two-dimensional display of the scores (Fig. 3), with further graphic variables representation and their weighing in Suppl. material 2: Fig. S7. All variables correlate strongly with Dimension 1, with the iris coloration and reticulation having the highest correlation ratio, 0.937 and 0.933, respectively. Dimension 2 is mainly correlated with the iris reticulation, with 0.637 producing the only correlation ratio > 0.5.

Figure 3. 

Map of the Multiple Correspondence Analysis (MCA) of P. gretathunbergae sp. nov. (red dots) and P. cruentus (black dots): Number labels for individual frog with lines pointing to specimen location on the map. Following correlation ratio (Dimension 1/Dimension 2) resulted from the MCA: iris coloration 0.937/0.001; iris reticulation 0.933/0.637; upper eyelid tubercle 0.751/0.331; upper lip coloration 0.735/0.326; groin coloration 0.852/0; ventral coloration 0.810/0.001. The qualitative scoring of the variables and its species-specific expression is explained in the methods.

Two distinct clusters appear in the MCA that clearly represent the two sympatric species P. gretathunbergae sp. nov. and P. cruentus (Fig. 3). The distinction of these species in the first and second dimension of the MCA is highly significant (Dimension 1: Mann‑Whitney

U‑Test, W = 133, p < 0.001, and Dimension 2: Mann-Whitney U-Test, W = 321, p = 0.026). These results strongly separate the two Pristimantis species on qualitative morphological characters, with the distinctive eye color and pattern being a particular easy and obviously useful character to separate P. gretathunbergae sp. nov. from P. cruentus (see Suppl. material 2: Figs S10, S11). The distinctive black eyes without reticulation of P. gretathunbergae sp. nov. also separates it from the even closer related, but allopatric, P. erythropleura-penelopus clade from Colombia, and likewise from Colombian and Ecuadorian P. cisnerosi, P. viejas, and P. paisa (consult respective species-specific photographic panels in the Suppl. material 2: Figs S12–S16).

Based on molecular divergence and morphological consensus, we assign the undescribed Pristimantis sp. with the type material from Cerro Chucantí, Maje Mountains as a new species to science. It belongs to the Pristimantis ridens species group (sensu Reyes-Puig et al. 2020), defined by having large digital disks, finger I shorter than finger II, toe III shorter than toe V, tympanum concealed, vomerine odontophores oblique, no toe webbing and vocal slits absent. It is most closely related to the allopatric P. erythropleura-penelopus group, which inhabits similar montane forests along the Andean slopes of western and central Colombia. Following is the formal description of the new species of Pristimantis.