The genus Molossops includes two monotypic species of insectivore bats distributed in South America: Molossops neglectus and Molossops temminckii. Both can be differentiated, based on sizes, M. temminckii being smaller (forearm less than 33 mm). Despite being monotypic, at least two additional subspecies have been described for M. temminckii, of which M. temminckii griseiventer from the inter-Andean Valley of the Magdalena River in Colombia might represent a valid taxon. To test the taxonomic status of M. t. griseiventer, we reviewed specimens of M. temminckii from cis- and trans-Andean localities in Colombia. We used Cytochrome-b and Cytochrome Oxidase I comparisons to test the phylogenetic position of cis- and trans-Andean samples and compared qualitative morphology, morphometric and bioacoustics. Our results show that M. t. griseiventer is differentiated from cis-Andean specimens, providing further evidence of its validity at the species level. Furthermore, M. temminckii (sensu stricto) is also distributed in Colombia, but both M. griseiventer and M. temminckii are allopatric, with the Andes acting as a barrier. The specific identity of the specimens from the Caribbean Region of Colombia needs a new evaluation, but our results clearly show that the diversity of Molossops is underestimated.

Andes, biogeography, endemism, Mammalia, nomenclature, subspecies, taxonomy

The genus Molossops Peters, 1866, comprises two species distributed in South America: the Rufous Dog-faced bat, Molossops neglectus Williams & Genoways, 1980 and the Dwarf Dog-faced Bat, M. temminckii Burmeister, 1854 (Díaz et al. 2021; Mammal Diversity Database 2022). The genus is monophyletic, based on morphological data (Gregorin and Cirranello 2016), but only M. temminckii has been included in the molecular phylogenetic reconstructions of the family using one mitochondrial and three nuclear genes of a single specimen from Paraguay (Ammerman et al. 2012).

Species of Molossops are characterised by their small size (forearm less than 39 mm), with relatively large tragus, a short and wide antitragus and long and pointed ears that have a flexible fold where they are attached to the head (Eger 2008). Molossops neglectus and M. temminckii are currently considered monotypic (Eger 2008). Molossops neglectus is the rarest species and is only known from a few localities in Colombia, Venezuela, Guyana, Suriname, French Guiana, Peru, Brazil and Argentina. In contrast, M. temminckii has a broader distribution, with records in Colombia, Venezuela, Guyana, northern Ecuador, Peru, Brazil (type locality at “Lagoa Santa,” Minas Gerais), Bolivia, Paraguay, Argentina and Uruguay (Gamboa Alurralde and Díaz 2019; Mammal Diversity Database 2022) with a huge gap between northern South American populations and those from the south of the Amazonia.

Despite Eger (2008) treating the species as monotypic, there are two additional described subspecies: M. t. sylvia Thomas, 1924, with type locality at “Goya”, Corrientes, Argentina; and M. t. griseiventer Sanborn, 1941, with type locality “Espinal, west of Magdalena River on the plains of Tolima, Colombia”, tentatively endemic to the inter-Andean Valley of the Magdalena River of Colombia. Molossops t. griseiventer exhibits morphometric differences compared to the nominal population, although, due to the limited samples, its subspecific status was not fully supported (Eger 2008; Gamboa Alurralde and Díaz 2019). In recent years, Chacón-Pacheco et al. (2021) confirmed that individuals of M. t. griseiventer are larger than specimens from the Caribbean and Orinoco Llanos Regions in Colombia and M. griseiventer was elevated to species level, based on karyotype descriptions of specimens collected in cis-Andean localities in Venezuela (Volleth et al. 2023), far from and separated from the type locality for the Oriental (Eastern) Cordillera of Colombia. However, no other morphological and molecular data have been used to validate the specific status. Here, we assess the cryptic diversity and the validity of M. t. griseiventer reflected by the morphometric variation found in previous works (Eger 2008; Chacón-Pacheco et al. 2021) using mitochondrial, morphological and acoustic data. We also evaluated if the morphologic variation is associated with geographical barriers in northern South America.

To explore the cryptic diversity within M. temminckii, we collected 18 specimens in both sides of the Andes in Colombia in contrasting areas: the inter-Andean Valley of the Magdalena River (close to the type locality of M. t. griseiventer) and the Orinoco Llanos Region (M. temminckii). The specimens were deposited at the Instituto de Ciencias Naturales, Universidad Nacional de Colombia (ICN) and the Colección de Mamíferos, Museo de Historia Natural of the Universidad de Caldass (MHN-UCa). We identified the specimens as belonging to Molossops, based on the following traits: tips of ears elongated and pointed with inner margins not arising from the same point, but joined to head by a flexible fold; no granulations on the forearm; the skull flattened; one upper premolar; upper incisors in contact with each other and projecting forward; palate distinctly domed and anterior palate not emarginate; basisphenoid pits clearly developed; lacrimal processes well developed; lacrimal width of rostrum considerably greater than postorbital constriction; one lower incisor in each ramus; third commissure of M3 well developed, about 1/2 the length of the second; the second phalanx of digits III and IV equal to or longer than first (Eger 2008). We also explored discrete traits suggested in literature as unique for M. t. griseiventer, such as colouration and size of the patch of the throat and colouration of the venter (Sanborn 1941).

We obtained nine cranial and external measurements including the length of the forearm (FA), greatest length of the skull (GLS), condylo-basal length (CBL), breadth of the braincase (BB), postorbital constriction (POC), width across the last upper molars (M-M), zygomatic breadth (ZB), maxillary tooth length (C-M³) and mastoid breadth (MB). To explore the morphometric variation within cis- and trans-Andean populations in Colombia, we performed Principal Component Analyses (PCA) using only seven cranial measurements (GLS, CBL, POC, BB, C-M³, MB, M-M). We computed the analyses using the covariance matrix to preserve the information about the relative scale amongst variables using the statistical package PAST version 2.2 (Hammer et al. 2001). We also obtained discriminant analyses and correct classification rates for pairwise comparisons amongst M. t. griseiventer and specimens of M. temminckii from eastern Colombia and localities in the Southern Cone of South America.

To compare genetic affinities of the cis- and trans-Andean specimens of M. temminckii, we extracted genomic DNA from muscle tissues preserved in 96% ethanol of five specimens. Following the manufacturer’s protocol, DNA was extracted with a Wizard Genomic DNA Purification kit (Promega Corporation). We performed amplification of a fragment of the mitochondrial gene Cytochrome-b (Cyt-b) using the primers LGL765F (Bickham et al. 1995) and LGL766R (Bickham et al. 2004) and the reaction volumes and the amplification conditions detailed in Ramírez-Chaves et al. (2021a). We checked for discontinuities and stop codons for all the obtained sequences using the ExPASy translate web tool (https://web.expasy.org/translate/). We gathered homologous sequences of Neotropical species of Molossops from GenBank including seven Cyt-b sequences of M. temminckii. We also analysed the genetic affinities in the mitochondrial gene Cytochrome Oxidase I (COI). For that gene, we retrieved sequences deposited in GenBank and Bold Systems platforms, including a sequence of M. t. griseiventer from Caucasia in the inter-Andean Valley of the Cauca River in the Department of Antioquia, Colombia and 23 sequences of M. neglectus (16 sequences) and M. temminckii (7 sequences) species. GenBank and Bold Systems accessions, museum vouchers and localities of the analysed sequences are shown in Appendix 1.

We inferred phylogenetic trees for each gene using Bayesian Inference in MrBayes 3.2.6 (Ronquist et al. 2012). We ran two independent replicates of the Metropolis coupled chain Monte Carlo analysis for 10×10⁶ generations with trees sampled every 1,000 generations. Convergence was inspected in the programme Tracer 1.6 (Rambaut et al. 2014). We discarded 25% of the samples in each run as burn-in and combined the remaining samples to estimate tree topology, the mean likelihood and posterior probabilities. We used different outgroup sets for each gene because of the different availability of sequences. The species Cynomops planirostris, Eumops auripendulus, Nyctinomops laticaudatus and Tadarida teniotis were used as outgroups for the COI gene and Cynomops abrasus, E. auripendulus, Nyctinomops aurispinosus and T. teniotis for Cyt-b analyses. Besides, we estimated genetic distance values for Cyt-b and COI using the p-distance method in the software MEGA X (Kumar et al. 2018). We analysed the haplotype network for Cyt-b and COI genes to assess non-bifurcated similarities. We trimmed all sequences to equal lengths and removed the sequences containing ambiguities (Cyt-b = 1071 bp, COI = 650 bp). Posteriorly, we built each haplotype network using the TCS parsimony algorithm (Clement et al. 2002) as implemented in PopArt (Population Analysis with Reticulate Trees) 1.7 software.

Finally, we examined acoustic variables in the echolocation calls of M. temminckii from southern South America and M. t. griseiventer from the Magdalena River Basin to explore their differences. Recordings of M. t. griseiventer were obtained from two different localities in the Magdalena River Basin using transects (AnaBat Walkabout, Titley Scientific) and passive acoustic monitoring (AnaBat Swift, Titley Scientific) methods. Our data were compared with available information on the species from Brazil (Oliveira et al. 2018). Spectrograms and oscillograms of echolocation call sequences were displayed simultaneously using Raven Pro 1.6 software (K. Lisa Yang Center for Conservation Bioacoustics 2023) from 512 consecutive fast Fourier transforms with an 85% overlap and visualised on a Hamming-type window. For each pulse in the sequences analysed, we manually measured the following parameters: 1) call duration (measured in ms from the start to the end of the pulse), 2) peak frequency (corresponding to maximal intensity in the power spectrum), 3) maximum, 4) and minimum frequency and, finally, 5) the bandwidth (measured as the absolute difference between Fmax and Fmin). Recordings of M. t. griseiventer are available at the Environmental Sound Collection – ‘Mauricio Álvarez-Rebolledo’ (Colección de Sonidos Ambientales IAvH-CSA) at the Instituto Humboldt in Colombia.

Our analyses clearly showed the validity at the species level of M. griseiventer, based on morphologic and genetic evidence supporting previous claims of morphometric differentiation of this taxon when compared with M. temminckii (Eger 2008). While differences in all acoustic variables for both UFM and DFM signals were observed, further analyses are necessary to determine the statistical significance of these differences. Notably, variations were evident across all recording contexts used in Brazil (Oliveira et al. 2018). However, particular emphasis should be placed on the recordings captured during manual release, as this context closely resembles free-flying conditions.

Cryptic diversity within Molossops was previously suggested, based on morphometric comparisons (Eger 2008; Chacón-Pacheco et al. 2021) and the diversity within the genus may still be under-described. For example, Volleth et al. (2023) argued that differences between the karyotypes of specimens from cis-Andean Colombia and Venezuela and those from Brazil and published morphometric data (i.e. Chacón-Pacheco et al. 2021) are enough to elevate the subspecies M. t. griseiventer to species rank. However, although we provide evidence of the validity of M. griseiventer as a full species here, the conclusion of Volleth et al. (2023) is incorrect, but supports the presence of possible additional cryptic diversity. Volleth et al. (2023) linked the karyomorphs from the Orinoco Region with M. t. griseiventer even though these are distant from the type locality and are separated by the Cordillera Oriental of Colombia. The Orinoco Llanos´ karyotypes correspond to the populations of M. temminckii of northern South America and the cranial and external measurements provided for these specimens fall within the range of the specimens from eastern Colombia (Table 4), being smaller than those of M. griseiventer. Consequently, the specimens of M. temminckii from eastern Colombia and western Venezuela for which the karyomorph is 2n = 42, FNa = 56–60 (Gardner 1977; Volleth et al. 2023) could represent a different species. We found that cis-Andean specimens of Colombia and Ecuador formed monophyletic groups in both COI and Cyt-b genes. Nevertheless, we did not explore morphometric, morphologic or acoustic differences and the values of genetic distances are lower than 3% in both genes. Therefore, although these populations could represent a distinct undescribed taxon, we do not have the evidence to describe them. We also highlight the presence of at least two distinct species within M. neglectus, one from Brazil and one from Suriname and Guyana recovered as two different monophyletic groups in our genetic analysis.

Our review also increases the number of Molossops species to three and highlights the role of the Andes in the diversification of the genus, as has been suggested for other bat species (Patterson et al. 2012). With these findings, the number of molossid species in Colombia reaches 30 species (Ramírez-Chaves et al. 2021b), a richness like other highly diverse countries in South America, such as Brazil (32 species; Garbino et al. (2022)) and Peru (31 species; Velazco (2021)). Furthermore, the number of bat species in Colombia reaches 220 (see Ramírez-Chaves et al. (2021b); Esquivel et al. (2022)).

The distribution limits of M. griseiventer are still unclear. So far, M. griseiventer is restricted to Colombia between Antioquia and Huila Departments, containing the first genus records for the Department of Caldas (Ramírez-Chaves et al. 2021c). However, the identity of specimens from the Caribbean Region of Colombia (Chacón-Pacheco et al. 2021), from Cucuta (Royal Ontario Museum ROM 84999) and the Biogeographic Chocó (Bahía Solano: ROM 69533) are still in need of a taxonomic evaluation, these potentially being extensions of the distribution of the species. To contribute to the specific differentiation of M. griseiventer, we present amended systematic information for the species:

Order Chiroptera

Family Molossidae P. Gervais, 1856

Genus Molossops W. Peters, 1866

Molossops griseiventer Sanborn, 1941. New rank.

Molossops temminckii griseiventer Sanborn, 1941:385. Type locality “Espinal, west of Magdalena River on the plains of Tolima, Colombia.”

Molossops temminckii: Chacón-Pacheco et al. 2021. Part. Not Molossops temminckii (Burmeister 1854: 72).

Molossops griseiventer: Volleth et al. 2023:3. Part. First use of the current combination.

Type material. Holotype: Field Museum of Natural History (FMNH 51727), adult female in alcohol with skull removed. Collected on 21 September 1940 by Brother Nicéforo María (Sanborn 1941).

Type locality. “Espinal, west of Magdalena River on plains of Tolima, Colombia.” The type locality is currently located in the Department of Tolima.

Emended diagnosis. A small molossid bat (FA: 30.7–31.9 mm; weight: 7 g) with brownish dorsal colouration. Ventrally, it exhibits a white patch around 9 × 7 mm on the throat. Externally, the tips of ears are elongated and pointed with inner margins not arising from the same, but joined to the head by a flexible fold; lips smooth, lacking evident facial papillae (Fig. 5). The nose, ears and membranes are blackish (Fig. 6). The skull is small (GLS: 14.04–15.34 mm) and flattened (Fig. 7). The inner incisors are long, reaching almost half of the length of the canines. The postorbital constriction is well marked (3.97–4.34 mm); the sagittal crest is present, but low.

Figure 5. 

Details specimens of dwarf dog-faced bat (Molossops) A Molossops temminckii B Molossops griseiventer. Note the elongated and pointed tips of ears and the smooth lips in both specimens, the differences in ears and rostrum colouration being darker in M. griseiventer, the white gular patch in M. griseiventer that extends into the chest and the differences in the antitragus being more robust and rounded in M. griseiventer.

Figure 6. 

Details of the skin of males A Molossops griseiventer and B Molossops temminckii, from Colombia. Note the darker colouration of the membranes in M. griseiventer. Red arrow shows the lighter ventral patch in the throat.

Figure 7. 

Details of the skull of A Molossops temminckii and B Molossops griseiventer, from Colombia. Red arrow shows the marked nasal septum in M. griseiventer. Note the differences in the size of the upper incisors and the shape of the skull in lateral view. Scale bar: 10 mm.

Distribution. M. griseiventer is distributed in the inter-Andean Valleys of the lower Cauca and the Magdalena River Basin. All the locality records are trans-Andean and under 900 m a.s.l. (Fig. 8). The species is only confirmed from localities in the Departments of Antioquia, Cundinamarca, Huila, Caldas, Santander and Tolima, but additional records from the Magdalena Valley need confirmation.

Figure 8. 

Distribution of Molossops griseiventer and Molossops temminckii in South America. We detailed the allopatric distribution of two cis-Andean populations of Molossops temminckii from “North” and “South” South America. The star represents the type locality of Molossops griseiventer in the inter-Andean Valley of the Magdalena River. Trans-Andean records are attributed to M. griseiventer.

Comparisons. M. neglectus is larger (forearm > 36 mm) than M. griseiventer and M. temminckii. Molossops griseiventer is similar to M. temminckii (Table 4), but can be differentiated by a set of characters. Externally, M. griseiventer has darker colouration of the ears, nose and membranes than M. temminckii (Fig. 7). General colouration varies in M. temminckii. The largest series revised from the cis-Andean lowlands of the Orinoco Region of Colombia shows variation in the dorsal fur colouration ranging from light to yellowish-brown to brown and a clear monotonous venter. M. griseiventer has a dark brown dorsal colouration, but the venter shows a large white patch on the throat (Fig. 6). The antitragus of live individuals is more elongated in M. temminckii than in M. griseiventer, which is more rounded; that characteristic is lost in dry specimens. The lips of both M. griseiventer and M. temminckii from Colombia lack facial papillae (present in M. temminckii from Argentina and Brazil; Gregorin and Cirranello (2016); Gamboa Alurralde and Díaz (2019)). M. griseiventer appears, in general, larger in external and cranial measurements (Table 4), such as the forearm (31.3–32.9 mm versus 28.9–32.5 mm) and the post-orbital constriction (3.97–4.34 mm versus 3.4–4.1 mm). The skull is more flattened and larger, with a more defined nasal septum in M. griseiventer (Fig. 7) than in M. temminckii from Colombia (but similar in shape to the skull of M. temminckii from Argentina, see: Gamboa Alurralde and Díaz (2019)).

DMM thanks Ministerio de Ciencia Tecnología e Innovación and the Fulbright-Colombia commission for supporting the doctoral studies through the Fulbright-MinCiencias 2022 scholarship. HERC thanks the Fulbright Colombia: Investigador Visitante Colombiano cohort 2023 scholarship and Bruce D. Patterson and The Science and Scholarship Committee of the Field Museum of Natural History, Chicago, for support and allowing the review of specimens under their care. We also thank Sofía Terán Sánchez, Alison Guerrero Díaz and Jose Henao-Osorio for helping during fieldwork and for the photographs of specimens in collections. Sergio Solari, DeeAnn Reeder and one anonymous reviewer provide useful comments that improved the manuscript.