Introduction

Natural and sexual selection can act on various phenotypic and behavioral traits, leading to differentiation between males and females (Hedrick and Temeles 1989; Shine 1989; Rojas et al. 2018). Difference in body size is the most common sexual dimorphism among vertebrates (Shine 1979; Parker 1992; Anderson 1994; Ralls et al. 2009), and sexual selection has been proposed as its key driving force (Shine 1989; Allen et al. 2011; Rojas et al. 2018). Sexual dimorphism can also manifest in other characteristics such as behavior, physiology, and life history (Cox 2010). In species where females are larger than males, this difference is commonly associated with a high female investment in egg production. In contrast, for species where males are larger, male size is related to male-male competition or female choice (Darwin 1871; Parker 1992). In amphibians is often observed as females are larger than conspecific males (Woolbright 1983); however, in some species, males are similar in size to females or exceed their body size, which appears to be related to the occurrence of physical combats in territorial species (Shine 1979; Katsikaros and Shine 1997; Magalhães et al. 2018).

Territorial behavior, where individuals (usually males) intensively defend resource-limited areas from conspecific intruders (Kaufmann 1983; Maher and Lott 1995), is widespread across different taxa (Brown 1964; Barlow 1974; Wells 1977; Baker 1983; Stamps 1983; Ostfeld 1990). Neotropical poison frogs (Dendrobatidae) are well known not only for their bright coloration and toxicity (Summers and Clough 2001; Santos and Cannatella 2011), but also because males generally defend multipurpose territories (Pröhl 2005) and often engage in physical combats (Shine 1979; Pröhl and Hödl 1999; Méndez-Narváez and Amézquita 2014), which could lead to low levels of sexual size dimorphism (Shine 1979). Resident males perch on calling sites and use advertisement calls to discourage opponents and attract females (Crump 1972; McVey et al. 1981). This territorial behavior could generate a differential use of space by males and females. Males need specific places to establish a territory and perches to call from (Summers 1992, 2000; Roithmair 1992, 1994a, b; Pröhl 1997; Pröhl and Hödl 1999), unlike females, which can use other microhabitats (Pröhl et al. 2019; Fischer et al. 2020). Thus, poison frogs are an excellent system to study sexual differences in space use.

Parental care is essential to offspring survival in several taxa (Clutton-Brock 1991). Dendrobatid frogs exhibit high diversity of parental care behaviors, and male uniparental care is the ancestral (and most common) type of care (Summers and Tumulty 2014). Biparental care has been described in only two species (Caldwell 1997; Brown et al. 2010), and maternal care is present predominantly in the genus Oophaga (Carvajal-Castro et al. 123AD) and Colostethus (Palacios-Rodríguez et al. 2022). Mothers transport the tadpoles to bromeliad axils and then feed them with unfertilized eggs until metamorphosis is complete (Silverstone 1973; Weygoldt 1980; Brust 1993). Because these behaviors entail increased exposure, individuals performing parental care may be more vulnerable to predation (Magnhagen 1991; Reguera and Gomendio 1999; Ghalambor et al. 2013). Thus, parental care behaviors can generate additional pressures on the sex providing care, which could lead to phenotypic sex differentiation (Rojas and Endler 2013).

While warning coloration has been generally studied as an antipredator strategy (Poulton 1890; Ruxton et al. 2004), it can also be under sexual selection via intra-sexual competition or female choice (Summers et al. 1999; Jiggins et al. 2001; Crothers et al. 2011; Crothers and Cummings 2013; Cummings and Crothers 2013; Rojas et al. 2018). Ladybird beetles, Harmonia axyridis, have been shown to maintain polymorphism in the elytra coloration by either seasonal mating variation (Osawa and Nishida 1992) or assortative mating (Awad et al. 2015) whereas wing coloration in Heliconius butterflies has been shaped by sexual selection via mate choice and assortative mating (Jiggins et al. 2001; Merrill et al. 2014; Finkbeiner et al. 2014). In anurans, Oophaga pumilio females have been shown to choose males based on their coloration (Maan and Cummings 2008, 2009), and males seem to behave differently depending on their own colors; males from more conspicuous populations are more active (Pröhl and Ostrowski 2011), vocalize from more exposed sites (Rudh et al. 2011), and tend to be more aggressive and explorative (Rudh et al. 2013) than their duller counterparts. Color variation could be the result of a trade-off between natural and sexual selection, which could in turn facilitate the evolution of sexual dichromatism (Nokelainen et al. 2012; Cummings and Crothers 2013; Rojas and Endler 2013).

While aposematism is known to be important in allowing the evolution and maintenance of complex parental tasks in dendrobatids (Carvajal-Castro et al. 123AD), little is known about how sex differences in behavior interact with the need to signal to predators to influence amphibians’ morphology and space use. Here, we combine phenotypic and ecological data to evaluate whether territorial behavior and parental duties are mediating sex differences in habitat use, coloration, and morphological traits in the critically endangered Lehmann’s poison frog (Oophaga lehmanni). In this aposematic (i.e., warningly colored, chemically defended; (Myers and Daly 1976; Daly et al. 1978; Arenas et al. 2010) frog, males are territorial, and females provide care to their tadpoles by feeding them unfertilized eggs. Considering the territorial defense and vocal behavior in males, we predict that males would be more conspicuous than females. Likewise, given the male-male combats described in this species (Rojas 2002), we expect an absence of sexual size dimorphism as previously described (Myers and Daly 1976; Betancourth-Cundar et al. 2020). Finally, considering differences in parental duties, where females are presumably more exposed to predators while transporting and feeding their offspring, we predict sex differences in morphological traits other than body size that improve transport efficiency or decrease predator detection.

Materials and methods

Study system.

Lehmann’s Poison Frog (Oophaga lehmanni; Fig. 1) is an endemic frog, Critically Endangered and restricted to its type locality in the Colombian Chocó biogeographic region (Myers and Daly 1976; Betancourth-Cundar et al. 2020). In its original description, natural history notes on its abundance and type of habitat were outlined (Myers and Daly 1976), and almost 20 years later, Lötters et al. (1999) evaluated the geographic variation of its advertisement call. In 2009, its conservation status was assessed, finding only 20 individuals (Velásquez et al. 2009), but some behavioral features have been described in captivity (Zimmermann and Zimmermann 1986; Rojas 2002; Lötters et al. 2007). Despite these valuable efforts, very little information (e.g., Vargas-Salinas and Amézquita 2013; Betancourth-Cundar and Palacios-Rodríguez 2018; Gómez-Díaz et al. 2019; Betancourth-Cundar et al. 2020) is available on O. lehmanni’s behavior and ecology in the wild. Within its small distribution range, O. lehmanni exhibits variation in the coloration with red/orange and yellow morphs (Myers and Daly 1976; Lötters 1992) (Fig. 1). While the red morph occurs within protected areas, the yellow morph occurs in unprotected areas and their populations are usually small. Recent studies have shown differences in body size and genetic structure between these morphs (Betancourth-Cundar et al. 2020).

Fig. 1
figure 1

Phenotypic variation in the dorsal color pattern of Lehmann’s Poison Frog: Oophaga lehmanni. The red/orange morph has been found in the Farallones de Cali National Park and these populations are the only ones protected. The yellow morph is extremely difficult to find in the wild, possibly due to its overexploitation for the international wild trade. Photos: M. Betancourth-Cundar

Full size image

Ecological data.

We compiled for the first time a large set of multi-year ecological and phenotypic data in a wild population of O. lehmanni (red morph). Between 2015 and 2018, we conducted five trips to the Farallones de Cali National Park (FCNP), Colombia with 10 to 60 days of sampling. After an exhaustive search, we identified three localities separated between 2.5 and 5 km, where this species is common. FCNP is the only place where this species is sufficiently abundant and accessible to obtain enough data to allow robust statistical comparisons. For each individual observed, we recorded date and time, sex, substrate type on which it was found, and perch height. Sex was determined at the time of collection, based on whether they were calling or not. For individuals that were not calling when captured, we determined the sex by the presence or absence of vocal slits by opening and inspecting their mouths. To test for differential use of the substrate, we used a generalized linear model- GLM with a Poisson error distribution. To test for differences between the sexes in substrate use, we ran a chi-square test for substrates with more than five observations. We recorded perch height, taken as the vertical distance from the soil to the site where the frog was observed, using a measuring tape. We applied the function mvn in the R package MVN (Korkmaz et al. 2014) to check that the variables were normally distributed. To evaluate sexual differences in perch height, we conducted an Analysis of Variance (ANOVA). All statistical analyses were conducted in R (R Core Team 2013).

Phenotypic data.

We measured the body mass of each individual to the nearest 0.01 g with a digital balance approximately 20–30 min after capture. Dorsal and ventral photographs were taken with a digital camera (Canon PowerShot SX170, Tokyo-Japan), using a size scale and a color standard (Kodak Q-13). Photographs were taken with the same equipment and camera settings, positioning the camera parallel to the frog’s body (Approx. 10 cm) and holding it by its thighs. Using ImageJ software (Schneider et al. 2012), we measured eight morphological variables, to the nearest 0.1 mm, from the photographs: body length (SVL); head, body, groin, and thigh width; and forearm, arm, and thigh length. To avoid redundancy between the morphological variables, we used a Principal Component Analysis (PCA), followed by a Linear Discriminant Analysis (LDA) (McLachlan 2004) to identify whether there were sex differences in morphology. To evaluate differences between the sexes in the principal components and the discriminant function, we conducted an ANOVA. We used as response variables the first three principal components and the discriminant function score, respectively.

To evaluate differences in colour pattern and reflectance between the sexes, we first analyzed the red vs. black shape of the pattern of the frogs’ dorsal area using JPEG photographs. Because the photographs were taken in the field with different light conditions, we standardized their brightness using the automatic white balance command of the GNU Image Manipulation Program, GIMP (The GIMP Team 2014). With this software, we also exported the photographs as 8-bit uncompressed files (.tiff) and then cut out the dorsal area of the frog’s body. The cut-outs were standardized to the same layer size of 300 × 600 pixels, and the background color was set to white. Later, the standardized images were proessed with the ‘patternize’ R package (Van Belleghem et al. 2018), which quantifies variation in color patterns obtained from 2D image data. We quantified the variation in the color patterns using the ‘patPCA’ function. We represented this variation with a PCA, which shows the main variation in color pattern boundaries among groups and predict color pattern changes along the PC axis (Van Belleghem et al. 2018). Positive values indicate a higher predicted presence of a particular pattern element, and negative values show its absence. Elements of the color patterns that are similar in all samples have a predicted value of zero, i.e., these pixels do not contribute variance among images (Van Belleghem et al. 2020). Also, we estimated the relative size of the red and black color patterns using the ‘patArea’ function. This function computes the relative area in which a color pattern is observed in each sample (Van Belleghem et al. 2018), i.e., the proportion of the total body area in which the red or black pattern is expressed. We represented the relative area of the color pattern as a density plot for each color (red and black), and by sex. We compared the percentage of the red-colored area and the black-colored area between sexes using ANOVA and within each sex using a paired Student’s t-test.

For the reflectance analysis, we measured the light reflectance in six parts of the frog’s body: red and black colors in the dorsal, lateral, and ventral areas. All reflectance measurements (λ = 300–700 nm) performed using a 600 μm bifurcated fiber-optic (QR600-7-UV/VIS) coupled to a spectrometer (Ocean Optics Inc. USB4000, Dunedin, FL, USA) and a pulsed xenon light lamp (PX-2) as a light source. Before each measurement, the lamp was calibrated with a white standard (WS-1-SL) and a probe with a rubber back cover was used to exclude ambient light. Measurements were taken at 2 mm and perpendicular (90º) to the surface. The analysis of reflectance was conducted using the “pavo” package (Maia et al. 2013) implemented in R (R Core Team 2013). Reflectance data were graphically represented by a plot of percent reflectance of red color vs. wavelength. Differences between males and females were evaluated considering the overlapping 95% confidence intervals.

Results

We observed 207 individuals, including 199 adults (137 males, 40 females, and 22 unknown sex), four juveniles (small-sized frogs), and four tadpoles. The sample size was different for each analysis and depended on whether the frogs were collected or only observed. For ecological analysis, we used 118 males and 40 females. For morphological variation analysis, we employed 35 males and 22 females, and for body size and weight 117 males and 34 females. For coloration pattern analysis, we used 32 males and 31 females, and for reflectance analysis 39 males and 20 females.

Frogs were found on diverse substrates such as fallen trunks, leaf litter, green leaves, roots, rocks, and soil, but trunks were significantly more used than other substrates (Z = 5.504, df = 11, P < 0.001). Furthermore, males were more likely to be found on trunks (Chi-sq = 32.190, df = 1, P < 0.001) and green leaves (Chi-sq = = 5.260, df = 1, P = 0.022), than females. Because of the low number of observations, we did not statistically compare the use of the least frequently used substrates between the sexes (soil: 2 males, 2 females; rocks: 2 males; Fig. 2a). Males were also found to occupy higher perches (about 30 cm higher) than females (F = 5.462, df = 156, β = 26.339, P = 0.021; Fig. 2b).

Fig. 2
figure 2

Habitat use of males (n = 118) and females (n = 40) of O. lehmanni in the Farallones de Cali National Park. Light gray color indicates females and dark gray indicates males. (a). Types of substrates used by males and females. This species differentially uses fallen trunks, and males were found more frequently than females on trunks and green leaves. Values in the boxes indicate the number of individuals of each sex per substrate. *** in black denotes statistically significant deviations from the null expectation by sex, and +++ by type of substrate. (b). O. lehmanni males occupy higher perches than females. Black points represent raw data, the horizontal bar shows the mean, the shaded diagram is a smoothed density curve showing the full data distribution, and the rectangle represents the uncertainty around the mean using a 95% Bayesian highest density interval

Full size image

We found a positive relationship between body length and body mass across all frogs (F = 38.700, df = 135, P < 0.001; Fig. 3a). However, we found no sexual dimorphism in body mass (mean ± sd, males: 2.965 ± 0.272 g, n = 103; females: 2.888 ± 0.319 g, n = 34; F = 1.861, df = 1, 135, P = 0.175) or body length (F = 3.154, df = 1, 149, P = 0.078), although males were slightly larger (SVL = 35.559 ± 2.073 mm, n = 117) than females (SVL = 34.857 ± 1.854 mm, n = 34). Our PCA using morphological data indicated that three principal components explain 79% of the variation in morphology. PC1 explained 52% of the variation and was positively associated to SVL, body and head width, and forearm length. PC2 explained an additional 16% variation and was positively related to thigh width and thigh length, and PC3 explained 11% of the variation and was positively associated with arm length (Fig. 3b). We found no significant sex differences in PC1 (F = 0.020, df = 1, 55, P = 0.889) and PC2 (F = 1.456, df = 1, 55, P = 0.233,), but we found that females had longer arms than males, as suggested by differences in PC3 (F = 6.583, df = 1, 55, P = 0.013). This was corroborated by the LDA (LD coef. =-0.711), which indicated that males and females were morphologically differentiated using one dimension (P < 0.001, F = 33.2, df = 1, 55; Fig. 3c), i.e., arm length, which was larger in females and was negatively associated with LD1.

Fig. 3
figure 3

Differences in the morphology of females and males of Lehmann’s Poison Frog. (a) Body size (Snout-Ventral Length-SVL) and body mass are positively correlated for males and females of O. lehmanni. We found no sex differences in body size. Light gray indicates females (n = 34) and dark gray indicates males (n = 117). (b) The relative contribution of each morphological variable to the first five principal components was obtained in the PCA analysis. The absolute value of each contribution is represented according to the size of the circle, while blue and red colors show positive and negative contributions, respectively. SVL, with snout-vent length, body width, forearm length, and head width are positively associated with PC1, while thigh width and thigh length are positively related with PC2, and arm length is positively associated with PC3. (c) LD1 indicates sexual dimorphism in morphology in this species, mainly associated with the arm length, which is larger in females. This trait is negatively associated with LD1. Light gray indicates females (n = 22) and dark gray indicates males (n = 35). Black points represent raw data, the horizontal bar shows the mean, the shaded diagram is a smoothed density curve showing the full data distribution, and the rectangle represents the uncertainty around the mean using a 95% Bayesian highest density interval

Full size image

Although the pattern of coloration varied widely among individuals and was used for individual identification, no differences between the sexes were found. Considering our PCA on dorsal coloration, there were no significant sex differences for both red and black colors. The first two components explained 16% of the variation in the black color and 20% in the red color (Fig. 4a). Similar results were found for the percentage of the red-colored and black-colored areas, which did not differ between sexes (Fig. 5b). Overall, the percentage of black-colored area (mean = 57.499 ± 8.244%, n = 63) was greater than that of red (mean = 28.599 ± 7.322%, n = 63) for males (t-test = 10.549, df = 31, P < 0.001) and females (t-test = 11.718, df = 30, P < 0.001) (Fig. 5b). Finally, we found no significant sex differences in the reflectance spectra of any of the red-colored areas measured (Fig. 4). In short, there were no detectable sex differences in variables associated with warning coloration.

Fig. 4
figure 4

Analysis of the color patterns of males and females of O. lehmanni indicates that there is no sexual dimorphism in coloration for this species. Light gray indicates females (n = 31) and dark gray indicates males (n = 32). Right plots show data related with black color and left plots with red color. (a) Principal components analysis of the dorsal pattern coloration for males and females of O. lehmanni. We found no sex differences in coloration patterns. Ellipses indicate the 95% confidence level of data. (b) The relative area of the color patterns black (right) and red (left), expressed as a percentage, of the dorsal body surface for males (dark gray) and females (light gray)

Full size image
Fig. 5
figure 5

Percentage of reflectance for males and females of O. lehmanni for the red coloration in ventral, dorsal, and lateral views. Colors indicate the sex, gray for males (n = 39) and red for females (n = 20). We did not find sex differences in this variable for all individuals analyzed. The shaded area shows the 95% confidence interval

Full size image

Discussion

Our study shows sex differences in behavioral, ecological, and morphological traits, but not in coloration, in the Lehmann’s Poison Frog, which suggests that selection could affect males and females differently.

Oophaga lehmanni males were found more often on trunks and green leaves, and on higher perches than females. This differential habitat use may be associated with territorial behavior, whereby males call from elevated perches to announce their presence to male conspecifics and to attract females (Pröhl 2005). Calling-perch height has a positive influence on sound propagation (Rodríguez et al. 2020), as calling from higher perches allows the signal to propagate over larger distances increasing the probability of the stimulus reaching potential receivers (Brenowitz et al. 1984; Kime et al. 2000; Parris 2002; Schwartz et al. 2016). Also, in some anuran species, perch height is directly related to mating success (Greer and Wells 1980; Pröhl and Hödl 1999). O. lehmanni males defend territories throughout the year and for several years (at least three), whereas females live and forage around male territories; after mating, females live within their mate’s territory (MBC, pers. obs.). Thus, possession and defense of territories is a characteristic that distinguishes males from females.

Sex differences in microhabitat use (Donnelly 1991; Rojas and Pašukonis 2019) might reflect differences in foraging activity and feeding rates, but also in diet composition (Katsikaros and Shine 1997). Such variation may lead to differences in the accumulation of diet-derived alkaloids in the skin, and sex differences in alkaloids have been previously documented in other amphibian species (Saporito et al. 2010; Stynoski et al. 2014; Brunetti et al. 2019). Further research should investigate sex differences in the chemical composition of alkaloids in aposematic species to determine if such variation is associated with differential space use.

We found no evidence of sexual dimorphism in body size, adding support to previous studies (Myers and Daly 1976; Betancourth-Cundar and Palacios-Rodríguez 2018). Although amphibian females tend to be larger than males (Howard 1981; Woolbright 1983; Lee 2001; Lowe and Hero 2012), in some dendrobatids this pattern appears to be diminished or reversed. In this group, males appear to be as large or larger than females, which has been associated with physical combat behavior (Shine 1979). Our results are in line with this hypothesis, as male and female O. lehmanni have similar body sizes, and males have been reported to engage in physical combats (Rojas 2002). Interestingly, frogs of the genus Oophaga exhibit maternal care (Weygoldt 1980; Summers and Tumulty 2014), which implies a higher reproductive investment by females (Pröhl and Hödl 1999). However, when we were conducting a study on territoriality, we observed that tadpoles are deposited on bromeliads within the male’s territory, and the territory owner actively calls near these plants while the female feeds the larvae with unfertilized eggs. We also observed that females rest near the male within his territory. These behaviors may be indicative of pair bonding, as has been shown in other poison frogs (Caldwell 1997; Brown et al. 2010). Male parental investment may be greater than previously thought if the male’s behavior directly or indirectly benefits his offspring (Tumulty et al. 2013). For example, if males participate in the tadpole transport, or if females need the male’s presence and calling to orchestrate the egg provisioning. Possibly, a substantial parental investment of males in addition to costs of advertising and maintaining territories may be associated with the absence of size sexual dimorphism (Woolbright 1983). Field observations targeted to directly examine parental investment are necessary to corroborate the occurrence of pair bonding and genetic monogamy in O. lehmanni, and to understand the adaptive value of male care in dendrobatids.

Arm length was the trait that contributed the most to morphological differentiation between sexes. In anurans, the accumulation of muscle mass in the arms, especially in the forearm, is often sexually dimorphic and predictive of mating success (Howard and Kluge 1985; Duellman and Trueb 1986; Lee 2001). Males usually have bigger arms than females, which confers an advantage both in male-male physical combats (to resist attempted takeovers by competing males) (Magalhães et al. 2018) and during amplexus (to retain their grip on a female), mainly for explosive-breeding anurans (Lee 2001). In dendrobatids, there are no accounts of axillary amplexus in aposematic species (Grant et al. 2006; Castillo-Trenn and Coloma 2008; Carvajal-Castro et al. 2020), but cephalic amplexus has been observed in the genus Phyllobates (R. Marquez, per. Obs.) and several species of the Colostethinae subfamily (Grant et al. 2006; Castillo-Trenn and Coloma 2008). Considering that male-male combats are common (Pröhl and Hödl 1999; Méndez-Narváez and Amézquita 2014), we would expect males to have larger arms. Our findings, however, contradict this expectation, which is somewhat surprising. While we are unaware of any functionality of longer arms in females, we propose that longer arms may be useful to climb trees or to lower into or get out of the bromeliad funnel when feeding the tadpoles. In O. lehmanni and other dendrobatids, larvae are deposited individually within the male’s territory in bromeliads located at about 4–8 m from the ground (MBC, pers. Obs.; Schulte et al. 2010; Rojas 2014; Fouilloux et al. 2021). O. lehmanni females feed the tadpoles every three to four days with unfertilized eggs until they are fully developed (Brust 1993; Crump 1996; MBC, pers. Obs.), making climbing requirements strikingly different between sexes. This could, in turn, result in sex differences in arm morphology (Moen 2019), as shown in macroevolutionary studies where microhabitat use has been found to dominate frog morphological evolution and locomotor performance (Moen et al. 2016; Citadini et al. 2018). In Dendrobates tinctorius, for example, males transport the tadpoles, and they have wider discs than females (Rojas and Endler 2013). In Anolis lizards, habitat use, and behavior appear to be reflected in the morphology and functional capabilities of the organisms, suggesting an adaptive component in morphology variation (Pounds 1988; Losos 1990; Velasco and Herrel 2007). This same principle of evolution of morphology associated with ecology and behavior may be operating in O. lehmanni females.

We found no sexual dimorphism in any aspect of O. lehmanni’s coloration, suggesting that this trait is not subject to sexual selection. It is widely recognized that the benefits of aposematism increase as a function of the frequency/density of a given signal (Endler and Mappes 2004; Mappes et al. 2005); therefore, by having similar coloration males and females would probably share the costs of predator learning. In D. tinctorius males have a higher proportion of yellow in their dorsal area than females, which is linked to their presumably extended exposure to predators due to tadpole transport (Rojas and Endler 2013). In O. lehmanni calling males defending their territories could potentially be at similar levels of exposure than caring mothers. So, natural selection might give equal advantages in avoiding predation for both sexes while carrying out their parental duties.

Sexual selection has been suggested to play a crucial role in shaping amphibian coloration (reviewed in Rojas 2016). For example, in some populations of O. pumilio males are significantly more brightly colored than females (Solarte Island), and females prefer brighter males (In three populations) (Maan and Cummings 2009). Our data do not support the hypothesis that sexual selection drives the evolution of color patterns in O. lehmanni because males and females show similar colouration. However sexual selection may act on other sexually dimorphic traits in O. lehmanni such as the presence of white spots on their limbs. This frog exhibits an elaborate courtship in which males display foot flagging and arm waving, similar to other anurans who display colored parts in reproductive contexts (Hödl and Amezquita 2001; Rojas 2016). It would also be interesting to examine the toe-pad size, which is a sexually dimorphic trait in D. tinctorius (Rojas and Endler 2013).

Here, we document different aspects of the morphology, ecology, and behavior of a wild population of an endemic, endangered species with a restricted distribution. Identifying sex differences in habitat use is fundamental for understanding the ecological requirements of this species in the field and, in turn, for the design of programs and management of ex-situ populations. While we found no sex differences in coloration patterns, we found that the large inter-individual variation in color patterns allows reliable identification at the individual level, which is very useful for establishing population monitoring programs. Furthermore, unlike traditional capture-recapture marking methods (e.g., toe-clipping), this method is non-invasive, and has been successfully implemented and recommended in other vertebrate species (Donnelly et al. 1994; Jonas et al. 2011; Chase et al. 2015; Ringler et al. 2015; Rojas and Pasukonis 2019; Landeo-Yauri et al. 2020). We found more males (137) than females (40). Knowledge of population sex ratios is important for conservation management because females are often the limiting sex when it comes to reproduction, and in this species, females are under higher demand in the illegal pet market (Betancourth-Cundar et al. 2020). The numbers sex ratio presented here may be biased due to the greater conspicuousness of males during calling. However, due to the importance of sex ratio in conservation, this deserves more attention. Improving methods for unbiased sampling of the sexes should be a priority for future research. Finally, we hope that our natural history findings will be useful not only for further studies on the behavior and evolutionary ecology of poison frogs, but also for the conservation and monitoring of this endemic and critically endangered poison frog.