The diary of Ross McFarland, one of the members of the IHAEC, was requested from the Ross A. McFarland Collection in Aerospace Medicine and Human Factors Engineering at the Wright State University Archives. The diary is listed as “Box 63, Folder 5: Ross McFarland’s Diary (May 1935–September 1935)” in the collection’s inventory (Hoffman and Ritchie 1987: 29). From the same collection, we obtained the video footage recorded by McFarland during the expedition (Items 2213, 2217 and 2218; Hoffman 1987: 113), which shows a concrete swimming pool at the source of Loa River. Individual frames were extracted from the video and panoramic views of the four different positions of the cameraman were generated, using the open-source software HUGIN – Panorama photo stitcher (version 2019.2.0).

On 31 October 2020, a field trip to the site called Miño (21°12'S, 68°40'W; 3900 m elevation; Calama Commune, El Loa Province, Antofagasta Region, Chile) was performed to locate the frog population that was described as T. halli (Correa 2021). The historical reference for this search was based on the swimming pool and other features of the landscape that appear in the recordings made by McFarland.

As biosecurity measures to prevent the spreading of chytridiomycosis and other infectious diseases, we disinfected car tires, boots, and utensils with F10 Super Concentrate Disinfectant (Health and Hygiene Pty.) at a concentration of 1:250 (Webb et al. 2007).

We made a general description of the study area, considering the topography of the landscape and more specific conditions at microhabitat level. We measured the stream dimensions at various points and took air and water temperatures at different times of the day. The composition of the adjacent vegetation along the stream was ascertained and a nocturnal survey was undertaken to detect possible sympatric amphibians.

Both, adults and larvae, identified as Telmatobius, were collected during the daytime from the stream using a hand net. The sampling site was ~ 300 m upstream from the pool identified as the historical place where T. halli was collected (see details in Results). The animals were measured, photographed, and finally released back to the capture site. Each individual was handled separately with an unused pair of disposable nitrile gloves (Thomas et al. 2020). To avoid possible toxic effects, the gloves were rinsed and the rinse water was discarded away from the watercourse (Cashins et al. 2008).

In order to obtain bioacoustic data, an AudioMoth recording unit (Hill et al. 2019) was placed beside the stream, at a spot where adult individuals had been sighted during sampling. The device recorded continuously between 8 p.m. and 7 a.m., but we did not obtain vocalizations that could be unquestionably attributed to Telmatobius.

During the night, the AudioMoth took a measurement of the air temperature every 15 minutes, but the sensor only has an accuracy of ± 3 °C (Open Acoustic Devices 2020). Water temperature was taken with a generic digital thermometer.

Seven morphometric features were measured on 11 adult specimens (Watters et al. 2016): snout-vent length (SVL), head width (HW), head length (HL), inter-orbital distance (IOD), inter-nostril distance (IND), foot length (FL) and tibia length (TL). FL and TL were assessed on the right hindlimb. In the case of the tadpoles (n = 9), body length (BL) and total length (TTL) were measured (Altig 2007) and the development stages (Gosner 1960) were determined. All measurements were taken using a vernier caliper to the nearest 0.05 mm.

Three tadpoles (Gosner stages 36–37) were anesthetized by immersing them in a buffered solution of MS-222 (0.2%) (Mitchell 2009), and a small portion of the membrane was cut from the caudal fin. After recovery from the anesthesia, they were released at the collection site. The tissue samples were stored in 96% ethanol until DNA extraction.

The DNA was extracted with a commercial kit (Promega ReliaPrepTM gDNA Tissue Miniprep System, Madison, WI) following the manufacturer’s instructions. We obtained fragments of two mitochondrial genes, 16S rRNA and cytochrome b (cytb), the same fragments that were used in the phylogenetic analyses of Sáez et al. (2014). The reagent mixtures, reaction conditions, and primers used in the PCRs are detailed in Sáez et al. (2014) and references therein. Electropherograms were edited with the program Bioedit v7.1.3 (Hall 1999). Substitu­tion saturation of the sequences was assessed with DAMBE7 (Xia 2018). Sequences were deposited in GenBank (accession numbers OL412556–OL412561).

The sequences of both fragments were aligned with MUSCLE (Edgar 2004) and the alignments were then inspected by eye. Bayesian phylogenetic analyses were performed with the program MrBayes v3.2.7 (Ronquist et al. 2012), in which all Telmatobius species from Chile and all sampled populations of the genus geographically close to Miño were included (Appendix 1). Both gene fragments were concatenated, but a reversible-jump Markov Chain Monte Carlo method for exploring the space of all General Time Reversible sub-models, plus gamma and proportion of invariable sites parameters, was applied independently to each fragment (an additional analysis was carried out considering the different positions of the codons of the cytb as distinct partitions). Both analyses consisted of two groups of four Markov chains that run independently for 20 million generations, sampling every 1,000 gener­ations. The first 25% of generations was conservatively discarded as burn-in after observing the stationarity of ln-likelihoods of trees in Tracer v1.7.1 (Rambaut et al. 2018). Convergence and mixing of chains were assessed by examining values of average standard deviation of split frequencies (ASDSF) and expected sampling sizes (ESS) and Potential Scale Reduction Factor (PSRF) for all parameters. Sampled trees were rooted with one specimen of Telmatobius sibiricus De la Riva & Harvey, 2003, a representative of the T. bolivianus species group (Sáez et al. 2014; Barrionuevo 2017) which constitutes the sister clade of the three species groups present in Chile (Sáez et al. 2014).

We collected all available information on the morphology of T. halli and the two populations to which the same name was assigned before being formally described as different species (T. dankoi and T. vilamensis) to compare their diagnostic characters as well as the proposed differences between them. The morphological details were obtained from the literature as follows: T. halli (Noble 1938; Veloso et al. 1982; Formas et al. 1999, 2003), T. dankoi (Veloso et al. 1982; Formas et al. 1999, 2003; Barrionuevo 2017), T. vilamensis (Benavides et al. 2002; Formas et al. 2003; Barrionuevo 2017). We further contrasted our observations of adults from Miño with the published data and added some minor comments regarding morphological traits observed in the populations from Las Cascadas (T. dankoi) and Vilama River (T. vilamensis).

As pointed out in Correa (2021), according to the chronicles of the IHAEC by Keys (1936a, b) and Dill (1979, 1980), the collection site of the type series of T. halli was the surroundings of a concrete swimming pool filled with warm water at the source of the Loa River (Figs 2, 3). Here, we provide the additional historical evidence extracted from the diary and the video recording of Ross McFarland that allowed us to identify the exact spot of the type locality. In the diary entry for Sunday, 23 June 1935, he wrote: “Trip in cars to hot springs at source of Rio Loa with Mr. Bell, Watson & Packard. Swimming & walk in green valley.” (McFarland 1935) (Fig. 4). The diary also confirms that the date of the departure of the IHAEC from Collahuasi (railway station Montt) (Fig. 1B) back to Ollagüe was Tuesday, 25 June 1935. As stated in Fibla et al. (2018), this means that the original collection date was June 23 and not June 25, as specified by Noble (1938).

Figure 2. 

Historic and current panoramic view of the area surrounding the concrete swimming pool in Miño A panorama extracted from video footage from the IHAEC, 1935. Yellow arrows indicate rock formations that are easily recognizable B current state of the habitat. Red rectangle = location of the concrete pool. The mountain in the left background is Miño Volcano.

Figure 3. 

Historic and current view of the concrete swimming pool in Miño A panorama extracted from video footage from the IHAEC, 1935. Yellow arrows indicate the upper and lower pool walls B same view in 2020.

Figure 4. 

Extract from the diary Ross McFarland wrote in 1935 during the IHAEC.

Regarding McFarland’s video material (Suppl. material 1), the mountain in the background of the video takes can easily be identified, even using Google Earth’s perspective view, as Miño Volcano, because of a characteristic bulge in its profile. Although strong erosion events have reshaped part of the landmarks, multiple rock formations of the canyon walls still remain identical and corroborate the congruence of the place with respect to the one depicted in the video (Fig. 2).

As expected, the remains of the mentioned concrete swimming pool were found at 21°12'01"S, 68°40'09"W (3,900 m) (Fig. 3). Even though the stream broke through the lower end of the pool’s wall and the bottom is filled with sand, most of the boundaries are still in place and it is evident that the structure corresponds to the one shown in the recording. The pool is rectangular, approximately 6.5 m wide, 20 m long and between 1.5 and 2 m deep. The side walls are made of stones, joined together with concrete, while the upper and lower walls comprise massive concrete blocks. There are other more recent concrete structures inside the stream, one immediately above the pool and another one ~ 300 m upstream.

The Loa River originates mainly from meltwater from throughout its upper drainage basin, where snow accumulates during austral winter. Several temporal ravines also gather the characteristic precipitations during the austral summer months (December to March), known as Altiplanic winter (Berenguer and Cáceres 2008; Delsouc et al. 2020). Lower down and descending from the east, there also are some important permanent affluents fed by aquifers.

For the first few kilometers, the riverbed is a broad and dry wadi named Miño River. Only ~ 4 km north of Miño, the arid riverbed gradually turns greener and ends in a small bog with grass tussocks, covering an area of ~ 5 ha. No significant water flow was registered during this time of the year (late October). At Miño, there are some well-preserved ruins of an old mining camp from the 18^th and 19^th centuries at both sides of the bog (Berenguer and Cáceres 2008), serving as an easily recognizable landmark (Fig. 1C).

From this point on, the river bears the name Loa, as it receives its first permanent tributary, the “Estero Nacimiento” creek (Berenguer and Cáceres 2008). This spring emerges at the head of a small ravine of ~ 1.3 km in length, a place called Ojos del Miño (21°11'43"S, 68°39'40"W) (Flores 2001) (Fig. 1C).

Below the confluence, the river suddenly turns into a pronounced canyon with vertical cliffs. The concrete pool is located precisely at the beginning of the canyon. Soon after, the river gets a little broader, forming larger natural ponds and sections with rapids. The canyon goes on in a similar manner for almost 100 km, until reaching the Conchi water reservoir.

At the sampling point, the current of the Nacimiento Creek flows rapidly, though the terrain is not very steep. The stream is between 2.5 and 5 m broad and 25–50 cm deep. The water is clear and the bottom is mostly sandy with some stones and scarce detritus at the bends. The margins are almost entirely covered with vegetation, mainly Festuca chrysophylla Phil. and a few bushes of Parastrephia lucida (Meyen) Cabrera. The overhanging grass cushions are ideal refugia for the frogs, forming at times gallery-like cavities along the riverbank. At some points inside the stream, patches of Myriophyllum aquaticum (Vell.) Verdc. can be found, alternating with mats of undetermined filamentous green algae.

At the pool site, the bottom is also sandy; however, there is a little more mud and detritus, probably coming from the bog and consequently a more abounding aquatic vegetation. The stream at the exit of the pool measures ~ 4 m in width and 50 cm in depth. Downstream from the pool, the vegetation coverage at the banks decreases a bit, which leaves fewer shelters for the frogs. In fact, a lower population density was detected there.

Adults of T. halli were found mainly under the tussocks, where they shared their refugia with other adults and larvae. On one occasion, 11 adults and one tadpole were captured from below the same plant. Tadpoles also exhibit gregarious habits, but somehow seem to prefer to shelter inside the aquatic vegetation, at the bottom of the stream. Still, they are not absent under the cushions at the riverbank. Most of the observed larvae were at approximately the same development stage (Gosner stage 36–37); however, two specimens were younger (Gosner stages 27 and 33). Directly inside the pool, there were very few Telmatobius tadpoles and only one adult was found a few meters below the outlet.

During the daytime, two adults of Rhinella spinulosa were found under the riparian vegetation in the pool and after nightfall, numerous individuals of these toads were observed outside the water along the stream. A small ravine, adjacent to the pool, was occupied by hundreds of Rhinella larvae in semi-lentic, shallow puddles, which are ideal for their development. Additionally, one specimen of Pleurodema marmoratum (Duméril & Bibron, 1840) was found walking around at night; hence, all three potential anuran species were present in the area. Since no case of syntopy between the Chilean Telmatobius has been reported, no other congener is expected to be encountered in Miño.

In the afternoon (05:00 p.m.), the air temperature was 21.8 °C, almost equal to the water temperature at the outlet of the pool (21.4 °C). In contrast, in the morning (8:00 a.m.) the air temperature was -2.4 °C, while water temperatures at the pool and the sampling site were 19.0 °C and 20.7 °C, respectively. After sunset, the air temperature dropped quickly to around -11.0 °C (00:30 a.m.) and remained alike until dawn. The minimum value was -13.1 °C at 03:30 a.m. The water temperature, which is generally higher than that of other localities of the genus (Lobos and Rojas 2020) and which remains more or less constant (19–21.4 °C), is consistent with the description of the original capture site (“a warm spring”; Noble 1938).

Overall, T. halli is a medium-sized frog (Table 2), with a depressed body, thin forelimbs and anterodorsolaterally orientated eyes (Fig. 5). In dorsal view, the head is slightly broader than long (HL/HW = 0.96), but narrower than the body. On average the head length is 29.65% of SVL. The snout tends to be long but truncated in dorsal view, although it can be rather elliptical in some individuals. In lateral view, the snout profile is quite variable, as it can be flat with a rounded tip or short and acuminate. Telmatobius halli presents a smooth skin with minuscule granules, which in some specimens are almost absent on the dorsum. In other cases, they can be more evident on the limbs, flanks, or even covering the ventral surface. These granules become most prominent on the posterior and ventral parts of the thighs. Mature males have very small spines associated with the granules, in addition to conspicuous, black nuptial pads on their thumbs. The coloration of dorsum and extremities can be described as a broad spectrum of brown, olive and yellowish speckles that alternate with dark, almost black spots or marks. Some frogs have fewer dark spots and the brown colors predominate, others show extensive dark areas (Fig. 5A–D). The ventral coloration is lighter, with shades of cream or pink, mixed with yellow areas or white dots (Fig. 5E). A noteworthy character is the light, yellow annulus around the eyes of some specimens (Fig. 6A), a trait that is shared with T. dankoi and T. vilamensis (JvT, pers. obs.), but seemingly not with other Chilean congeners. Loose skin folds at the posterior part of the thighs can be more or less developed, but seem more frequent in corpulent individuals and mature males. Another highly variable character is the extent of the interdigital membrane. All examined animals had fully webbed toes, but while in some cases the webbing was barely distinguishable towards the tips of the phalanges, others presented very prominent lateral fringes. The tadpoles are large and robust (97.27 mm at Gosner stages 36–37) (Table 3), with a thick, pointed tail (tail length = 1.52× BL; stages 36–37) and show approximately the same pigmentation patterns as adults, but with entirely smooth skin (Fig. 5F).

Figure 5. 

Selected specimens of Telmatobius halli from Miño A–D dorsal views of adult specimens, showing variation in coloration patterns E ventral view of the specimen from C F tadpole; scale bar: 1 cm (A–F).

Figure 6. 

Adults from the three known populations of Telmatobius halli, as recognized in this study, showing the similarity in their external appearance A Miño B Las Cascadas and C Vilama River. The inlay in the upper right corner of C shows a detail of the keratinous spines. Photograph credits for the Vilama River specimen: Felipe Rabanal. Scale bars: 1 cm.

We obtained final alignments of 568 nucleotide sites for the fragment 16S and of 975 for the cytb. However, both alignments were incomplete because the sequences of several specimens included from previous studies are shorter, particularly some fragments of the cytb of the T. marmoratus group from De la Riva et al. (2010). The topologies obtained in the Bayesian consensus trees (50% majority-rule) of the analyses with two or four (considering the different codon positions of the cytb) partitions were virtually identical; only slight differences were observed in branch lengths and in a few posterior probability values. The relationships recovered in both analyses were in agreement with those obtained by Sáez et al. (2014) and Fibla et al. (2017), recovering the monophyly of the three species groups (T. marmoratus, T. hintoni, and T. pefauri) present in Chile, although the last one with low support (posterior probability, pp < 0.95) (Fig. 7). Also, the relationships among species and populations within groups are consistent with those studies; for example, the close relationship among populations of Ascotán and Carcote salt flats + T. philippii + T. fronteriensis and between Telmatobius pefauri Veloso & Trueb, 1976 and the clade made up of T. dankoi + T. vilamensis (although in this case with low support, pp = 0.75). In our analyses, the three samples of T. halli group with T. dankoi and T. vilamensis with the maximum support (pp = 1). All the specimens of T. dankoi (n = 4) and T. vilamensis (n = 5) make up a polytomy with two of the tadpoles of T. halli (L2 and L3), which constitutes the sister group of the third tadpole (L1) (Fig. 7). The polytomy results from the fact that the sequences of all these specimens are identical in their entirety (the 1,543 sites of both fragments), while the separation of the haplotype of tadpole L1 is due to two differences in the cytb fragment.

Figure 7. 

Bayesian consensus tree (50% majority-rule; mitochondrial genes concatenated, treated as two separated partitions), showing the relationships among Chilean Telmatobius and the species groups recovered by Sáez et al. (2014). The specimens of the species and populations of the extreme south of the distribution of the genus in Chile are highlighted with the same colors of the map in Fig. 1A. The values next to the nodes correspond to posterior probabilities and the scale bar below the tree represents the expected substitutions per site along the branches. Identification of populations of Copaquire, Quebrada Chiclla, Quebrada Choja, and Aguas Calientes follows the taxonomy prior to Fibla et al. (2018) and Cuevas et al. (2020). The red box indicates the taxonomic changes proposed in this study.

Sáez et al. (2014) were the first to include T. dankoi and T. vilamensis in a molecular phylogenetic analysis. They obtained identical mitochondrial sequences (genes 16S and cytochrome b) from several specimens of both species and based on their morphological similarity (including some diagnostic characters, see Table 4) suggested that they corresponded to the same species. Fabres et al. (2018) tested 29 microsatellites in several Telmatobius species and found the same allele size ranges at various loci in T. dankoi and T. vilamensis. They note that this is observed only between these two species, supporting the suggestion of Sáez et el. (2014) that both correspond to the same taxon. Here, we obtained mitochondrial sequences from two individuals of T. halli that are identical to those of T. dankoi and T. vilamensis (a third individual differs by only two bases), suggesting a possible synonymy of these three species. The descriptions and diagnoses of T. halli (Noble 1938; Formas et al. 2003), T. dankoi (Formas et al. 1999) and T. vilamensis (Formas et al. 2003) are mainly based on external and osteological features, so considering the identity of the mitochondrial sequences among these three species, it is important to reevaluate the morphological differences that have been described between them. Table 4 compares the traits that have been included in the diagnoses of the three species as they appear in different sources. Below, for each trait, we highlight possible instances of polymorphism as well as the discrepancies that emerge when comparing the different sources and incorporating new observations.

The rudimentary nature of the maxillary teeth of T. halli was one of the features that motivated the description of the species. Since we did not examine the dentition of the frogs from Miño, an evaluation of this issue remains pending. At first glance this point seems decisive, adding the fact that the absence of teeth is also listed as an important trait in the diagnoses of T. dankoi and T. vilamensis. Nevertheless, Barrionuevo (2017) pointed out that the presence or absence of teeth can vary intraspecifically in some species of Telmatobius and even cited T. vilamensis as an example where both conditions have been observed.

Telmatobius dankoi was distinguished by having small keratinous spines on the head, flanks, posterior third of the dorsum and extremities in both sexes (Formas et al. 1999), while the skin of T. halli and T. vilamensis was described as smooth (Noble 1938; Formas et al. 2003). Some of the individuals from Miño had indeed smooth skin, others presented small granules in different densities. We also observed mature males with small black spines mainly on the flanks, extremities, and the posterior part of the dorsum, but in some cases even on the chest and venter. Formas et al. (2003) differentiate T. vilamensis from T. dankoi alluding to the skin being smooth in the former and spiny in the latter, yet they also state that the holotype of T. vilamensis has numerous, minute, transparent or white spines on the venter and the ventral surface of the extremities. We show an individual of T. vilamensis that presents small black spines on flanks, extremities, and posterior dorsum (Figure 6C; Felipe Rabanal, pers. comm.). It is important to note that Veloso et al. (1982) indicate that the adults of T. halli from the Loa River at Calama (later described as T. dankoi by Formas et al. 1999, type locality Las Cascadas) have “smooth dorsal and ventral skin”, in clear contrast to what appears in the description of T. dankoi. Furthermore, they do not mention any difference in skin texture between the population of Calama and that of Vilama River (T. vilamensis), which they also consider T. halli. In any case, interpopulational variation of the skin texture is not a novelty in the genus, as it has been reported for T. rubigo (Barrionuevo and Baldo 2009).

Another feature on which emphasis was made in the descriptions of T. dankoi and T. vilamensis is the presence of postfemoral folds. Both species differ from their congeners by presenting well-developed folds, although it was reported that these are smaller and thinner in T. vilamensis (Formas et al. 2003). For T. halli, Noble (1938) does not mention anything about this trait in the original description, but Formas et al. (2003), in the redescription of the species, explicitly indicate the lack of these folds. However, in the photographs of the holotype (Fibla et al. 2018: fig. 6A; Cuevas et al. 2020: fig. 3E, F) this trait seems to be present. The paratype depicted in Fibla et al. (2018: fig. 6B) does not have folds, which suggests that this would also be an intraspecific polymorphism. All of the adults that we observed in Miño, both males and females, presented postfemoral folds, although they seem more developed in males.

In the case of T. vilamensis, the shape of the snout was stated as an outstanding character and is described as being “strongly depressed” (Formas et al. 2003). Noble (1938) used the term “flat” for the snout of the holotype of T. halli, while Formas et al. (2003) described the snout of the same specimen as truncated in dorsal and short in lateral view. On the other hand, Veloso et al. (1982) mentioned a pointed and depressed snout for the frogs from Calama (as T. halli), while Formas et al. (2003) clearly categorized the snouts of the animals from the same population as “not depressed” (as T. dankoi). Veloso et al. (1982) do not mention differences in the shape of the snout (pointed, depressed) between the populations of Calama and Vilama River, which is contradicted by Formas et al. (2003), who include the shape of the snout among the few traits that distinguish T. dankoi of T. vilamensis (not depressed versus strongly depressed, respectively). In Miño, we observed variable snout lengths and forms, and some degree of variation in this feature is to be expected as well in the Vilama River and Las Cascadas.

Almost half of the diagnostic characters of T. vilamensis are cranial bone structures, which contrasts with the diagnoses of T. halli and T. dankoi, where few osteological characters were included. Therefore, from an osteological point of view, there are not many elements to compare the three populations. Some of the aforementioned osteological features have been attributed to immature stages of post-metamorphic development in T. dankoi and T. vilamensis, in comparison to other species of the genus (Barrionuevo 2013). Barrionuevo (2017) presents two possible explanations for this interspecific variation: i) Formas et al. (2003) used immature individuals for their osteological analysis, or ii) the differences in the analyzed attributes are produced by phenotypic plasticity. Since the specimens used for the descriptions of the skeletons of T. dankoi (Formas et al. 1999) and T. vilamensis (Formas et al. 2003) were explicitly stated to be adults, the second explanation would be more likely. However, neither of these cases can assure a reliable species delimitation. Regarding the different degrees of cranial ossification in T. dankoi and T. vilamensis (Formas et al. 2003), it is necessary to consider that it might correspond to intraspecific variation, as described for T. oxycephalus (Barrionuevo 2013).

The development of webbing and fringes on the toes has been included in the diagnoses of T. halli and T. vilamensis (Formas et al. 2003), but the differences described for these traits are only of degree, are very subtle, or vary within the population of Miño. Another feature that appears in diagnoses is the shape of the tongue. In fact, it is one of the characteristics with which Formas et al. (2003) differentiate T. vilamensis from T. dankoi. However, the description of the form of the tongue of T. dankoi varies between sources, being the description of Veloso et al. (1982) very similar to that of T. vilamensis (Table 4). The absence of the tympanum and tympanic annulus has also been included in the diagnoses of T. halli and T. vilamensis, but in this case all three species lack these structures.

With respect to the tadpoles, Formas et al. (1999) mentioned that those of T. dankoi have the tip of the tail black; however, when Veloso et al. (1982) described the appearance of larvae from the same population, they mentioned a “uniformly pigmented tail” and did not say anything about such a conspicuous mark (Table 4). We also endorse that at least some tadpoles from Las Cascadas do not exhibit dark tail tips (JvT, pers. obs.). Between the two sources listed above, there is also a discrepancy for the same species in the description of the shape of the tip of the tail.

One particular character was not included in the diagnosis of any of the three species, but was used to differentiate T. dankoi from its Chilean congeners (Formas et al. 1999) and in the dichotomous key to adults of the Telmatobius species from Chile (Formas et al. 2003). It refers to the condition of the tibio-tarsal joint not reaching the eye when bent forward in T. dankoi. Interestingly, in the key, the same species was categorized as having a tibio-tarsal joint that “reaches or exceeds the posterior border of [the] eye” (Table 4). This could mean that this attribute is polymorphic or, otherwise, that the authors might have overlooked this detail.

The usefulness of three additional characters that appear in the diagnoses can be discarded. Body size was included in the diagnosis of T. dankoi (SVL = 49.7–51.7 mm, Formas et al. 1999), but this range is within the size limits of the T. halli type series (42–57 mm, Noble 1938) and overlaps with the size range of T. vilamensis (38.36–50.81 mm, Formas et al. 2003). All of these ranges of values fall within the size limits measured by us in the Miño individuals (Table 2). Another feature included in the diagnosis of T. vilamensis is the number and shape of the chromosomes: 26 bi-armed chromosomes (Formas et al. 2003). However, these same authors point out that the karyotype of Telmatobius species is uniform, with 26 chromosomes and a fundamental number of 52, and that in all described cases the secondary constriction always is found in the short arm of pair 6. Finally, the coloration of the dorsum was included in the diagnoses of T. vilamensis and T. halli (Formas et al. 2003). In general, the coloration of the frogs from the three focal localities (Miño, Las Cascadas, and Vilama River) is very similar (Fig. 6A–C). The size and number of darker spots vary between individuals of the same population, but this variation is subtle if compared to the differences that can be found in other Telmatobius species. Just to give examples of this in species and populations from Chile, intraspecific heterogeneity in body coloration is described for T. marmoratus (Veloso et al. 1982) and well-illustrated for T. marmoratus (Sáez et al. 2014: fig. 3), T. pefauri (Fibla et al. 2017: fig. 4 A–E) and T. chusmisensis (Fibla et al. 2018: fig. 5A–C).

In summary, considering all the available information, in the literature there are two contrasting views on morphological variation among T. halli, T. dankoi, and T. vilamensis. On the one hand, there are two studies where the limits of T. halli are broadened: one that adds the population of Arroyo Vilama (Cei 1962) and another that includes this same population and that of Calama (Veloso et al. 1982). In both studies, no morphological differences between these populations were described. On the other hand, there are two studies that focus on dissimilarities observed in frogs from these localities and split the populations into separate species (Formas et al. 1999, 2003). Combining the information from the different sources, including novel observations made here, differences and discrepancies arise in the descriptions of many traits for the same species (Table 4). In the cases where the information is specified, authors examined different collection specimens from the same populations (Veloso et al. 1982; Formas et al. 1999, 2003), so the differences observed among them can be interpreted as polymorphisms, which in many cases are shared with the other populations. This applies to most of the characters that were described as diagnostic for one or more species: skin texture (presence/absence of spines), snout shape, webbing and fringes of the toes, tongue shape, shape, and pigmentation of the end of the tadpole’s tail, and extension of the tibio-tarsal joint when extends forward. The presence of postfemoral folds and the absence of the tympanum and tympanic annulus, apply to all three species, while the presence/absence of teeth is a polymorphism in T. vilamensis (Barrionuevo 2017) that could be shared with the other species. This leaves as diagnostic differences only the degree of ossification of the skull (between T. dankoi and T. vilamensis) and a slight distinction in the shape of the choanae, but these osteological observations are based on a limited number of specimens (two of T. dankoi and three of T. vilamensis in the case of the skull). Furthermore, as mentioned above, variation in osteological characters has been described within some Telmatobius species, including the degree of ossification (Barrionuevo 2013 and references therein). Regardless of the position adopted (morphological uniformity among populations or widespread polymorphism), neither of the two supports the distinction of species.

Therefore, the external, osteological, and ecological characteristics as a whole do not allow to clearly distinguish T. halli, T. dankoi, and T. vilamensis, and the described variation of some morphological characters can be interpreted as intra- and interpopulation polymorphisms between the three known populations. Bearing in mind also their indistinguishable mitochondrial sequences and their high genetic affinity detected with microsatellite markers (Fabres et al. 2018), we herein propose to consider T. dankoi and T. vilamensis, by nomenclatural precedence, junior synonyms of T. halli. We further suggest adopting the vernacular name of T. dankoi (Loa Water Frog), as it has gained popularity (Lobos and Rojas 2020) and would represent the species appropriately.