To confirm the correct identification of the sequences in our analyses, we examined at least one voucher specimen from most of the clades reported in previous studies (Porter et al. 2007; Larsen et al. 2011; Siles et al. 2013; Siles and Baker 2020). The examined specimens (Suppl. material 1: Table S1) are housed in the following collections: The American Museum of Natural History (AMNH), USA; the Instituto de Ciencias Naturales of Universidad Nacional de Colombia (ICN), Colombia and the Pontificia Universidad Católica del Ecuador (PUCE), Ecuador.

We extracted genomic DNA using the phenol-chloroform method (Sambrook et al. 1989) from fresh samples of liver or muscle tissues preserved in ethanol, and for museum specimens, we extracted punches of tissue from the plagiopatagium. We obtained DNA from individuals of five species, M. hirsuta, M. megalotis, M. microtis, M. minuta, and M. schmidtorum from several localities in Colombia (Suppl. material 2: Fig. S1). We amplified 32 sequences between 700 and 1100 base pairs (bp) of the cytochrome-b gene (Cytb) mitochondrial DNA (mtDNA) and 19 sequences between 500 and 700 bp of the fibrinogen beta chain (Fgb-I7) gene nuclear DNA (nDNA) from the newly obtained specimens. We used the same primers and followed the same laboratory protocols and methods of Porter et al. (2007) for both genes, albeit increasing the annealing temperature from 45–48 °C to 50.5 °C for the Cytb amplification. Additionally, we gathered 101 sequences of Cytb and Fgb-I7 of Micronycteris species from the GenBank and Bold Systems databases and 64 sequences provided in Siles and Baker (2020) and included in the analyses (Suppl. material 1: Table S1).

For the phylogenetic analyses, we included 198 sequences of Cytb (mtDNA) comprising all currently recognizes species of Micronycteris, 150 sequences of Fgb-I7 (nDNA), and 146 individuals with complete data set (Cytb + Fgb-I7) of most of the Micronycteris species except by M. sanborni (list of specimens and sequence numbers in Suppl. material 1: Table S1), covering a wide portion of the geographic distribution of the genus (see Suppl. material 1: Table S1). We performed multiple sequence alignment for both genes using the Clustal W algorithm in BioEdit 7.2.6 software (Hall 1999). We analyzed the Cytb mtDNA, the Fgb-I7 nDNA, and the complete evidence (Cytb + Fgb-I7) data sets, using the following partition scheme: (i) unpartitioned, (ii) partitioned by gene (i.e., each gene fragment treated as a distinct partition) and (iii) maximum partitioning (i.e., each codon of the protein-coding gene Cytb and the Fgb-I7 gene fragment treated as distinct partitions). We assessed the optimal partitioning scheme and best-fit evolutionary models using PartitionFinder v2 and the Bayesian Information Criterion (Lanfear et al. 2016) and selected the maximum partitioning scheme. We applied the following resulting models in a Bayesian analysis (BA) with MrBayes v 3.2.1 (Ronquist et al. 2012): Cytb 1^st codon – K80+G, Cytb 2^nd codon – K81uf+I+G, Cytb 3^rd codon – K80+G and Fgb-I7– HKY+I+G. Those models were incorporated into a single tree search (mixed model partition approach; Nylander et al. 2004), and two parallel runs were carried out using four Markov chains, run for 50 million generations, sampling every 100 generations. We discarded 25% of the resulting trees as burn-in, and 85% of the trees were used for generating a 50% majority-rule consensus. We assessed an acceptable level of the MCMC chains and estimated the effective sample sizes for all parameters using the software Tracer 1.5.4 (Rambaut et al. 2018). Additionally, we performed phylogenetic analyses of the same data sets using the Maximum Likelihood algorithm implemented in RAxML 1.5 beta software (Stamatakis 2014). The default GTR+G model was set across all partitions. Five independent Maximum Likelihood searches were performed with different starting conditions and the rapid bootstrap algorithm to explore the robustness of the branching patterns by comparing the best trees. Afterwards, 1000 non-parametric thorough bootstrap values were computed and plotted against the best tree. For all phylogenetic analyses, we used homologous sequences of other bat species including Desmodus rotundus, Glossophaga soricina, Hsunycteris cadenai, Hsunycteris thomasi, Lampronycteris brachyotis, Lionycteris spurrelli, and Lonchophylla robusta as outgroups (Suppl. material 1: Table S1). We based the description of the resulting phylogenetic hypothesis for the genus Micronycteris on the complete evidence data set analyses.

We evaluated whether populations within the genus Micronycteris corresponded to different independently evolving evolutionary lineages (the General Lineage Species Concept of de Queiroz 1999; 2007) when two or more independent lines of evidence supported their distinctiveness. For that, we used the following different lines of evidence: (1) identification of monophyletic lineages in the mtDNA and mtDNA + nDNA phylogenies, (2) assessment of genetic distances using the mitochondrial gene, (3) identification of mitochondrial monophyletic clades matching unique nDNA haplotypes, and (4) the use of single locus species delimitation methods. Based on the concordance of those lines of evidence we identified whether those lineages corresponded to Confirmed Species or Confirmed Candidate Species (CCS); the later sensu Vieites et al. (2009) defined as: generally deep Cytb lineages (> 3% sequence divergence), differing clearly by morphology or with exclusively nDNA private haplotypes and recognized by at least one species delimitation method.

We started the lineage delimitation by determining the monophyletic clades through the phylogenetic analyses and afterward we searched for mitochondrial lineages with genetic divergences > 3%, using the Species Identifier ‘Cluster’ algorithm in Taxon DNA 1.7 (Meier et al. 2006). We used 3% because it was the percentage of the degree of genetic divergence that often corresponds to species-level units in several phyllostomid bats (see Genetic Species Concept in Bradley and Baker 2001 and Baker and Bradley 2006). We assessed the genetic distances between clades using uncorrected p-distances, implemented in the software Mega v.10.0.5 (Kumar et al. 2018).

Then, we searched for genealogical concordance between mitochondrial and nuclear lineages because such concordance has been considered an essential criterion for species recognition (Avise and Ball 1990). For that, we searched for unique haplotypes in the Fgb-I7 gene corresponding to the mtDNA lineages revealed by the phylogenetic trees and showing genetic distances of > 3%. We trimmed all sequences of Fgb-I7 to equal length and removed the sequences containing ambiguities that could not be interpreted as heterozygotes. We resolved nuclear DNA sequences to haplotypes with the PHASE algorithm (Stephens et al. 2001) implemented in the software DNASp 6 (Rozas et al. 2017). The phased sequences were then used to construct a haplotype network in the software Haplotype Viewer (http://www.cibiv.at/~greg/haploviewer), based on a Neighbour-joining tree from uncorrected p-distances computed with MEGA v.10.0.5 (Kumar et al. 2018).

In addition, we performed the following three different single-locus species delimitations models: (i) The Bayesian implementation of Poisson tree processes (bPTP), (ii) The single-threshold method of the generalized mixed Yule coalescent model (GMYC), and (iii) The multi-rate Poisson tree processes for single locus (mPTP). All analyses were performed using the Exelixis Lab’s web server (bPTP – http://species.h-its.org/ptp/; mPTP – https://mptp.h-its.org/#/tree; GMYC – http://species.h-its.org/gmyc/). For the delimitation analyses, we used unique haplotypes of Cytb across the aligned region to avoid the influence of duplicate haplotypes in the analyses (147 terminals; Suppl. material 1: Table S1). We constructed an ultrametric tree with BEAST v1.7.2 (Drummond et al. 2012) using a lognormal relaxed-clock model, a coalescent constant-size tree prior, and a relative time set with a prior on the ingroup age of one (normal distribution: mean = 1, SD = 0.01). We ran two independent MCMC analyses for 50 million generations, sampling every 1000 generations. We assessed convergence, discarded a fraction of trees as burn-in, and summarized the trees as in MrBayes phylogenetic analyses above. The resulting tree was used for the delimitation analyses. The bPTP delimitation search was performed for 500,000 Markov chain Monte Carlo (MCMC) generations, with thinning set to 100 and a burn-in of 25% of initial samples. The convergence of bPTP analysis was visually checked in the trace plot. Finally, we considered both, the Maximum likelihood and Bayesian solutions for the bPTP delimitation.

The data underpinning the analysis reported in this paper are deposited in the Mendeley Repository at: http://dx.doi.org/10.17632/vyp75f243x.1

For the Cytb gene dataset and the complete evidence, both tree building methods placed all individuals into four well-supported major clades that corresponded to four subgenera of Micronycteris namely: Leuconycteris, Micronycteris, Schizonycteris, and Xenonectes, (Fig. 1). However, the phylogenetic position of M. schmidtorum was placed in the subgenus Leuconycteris (Figs 1, 2). Leuconycteris included the species M. brosseti and M. schmidtorum. Micronycteris included the species M. buriri, M. giovanniae, M. matses, an undescribed species “M. sp.”, and nine clades within M. megalotis complex (including M. microtis, see taxonomic comments in discussion). Schizonycteris included the species M. tresamici, M. sanborni, M. yatesi, M. simmonsae, and seven clades within the M. minuta complex. Finally, Xenonectes contain two clades of M. hirsuta complex (Figs 1, 2).

Figure 1. 

Bayesian phylogram of the genus Micronycteris from the phylogenetic analysis of the Cytb gene, mtDNA. Numbers below the nodes correspond to Bayesian posterior probabilities, and those above correspond to bootstrap support values (percentages). Colors indicate the clades with > 3% of genetic differentiation. Sequences within groups are listed in Suppl. material 1: Table S1 and depicted in Fig. 5. Bootstrap support values are missing in clades not recovered by ML analyses. Micronycteris sanborni was not included in mitochondrial genetic differentiation analyses because it is a chimeric sequence of two individuals.

The first strongly supported major clade matched the subgenus Schizonycteris (Fig. 2). Within this major clade, the first strongly supported subclade corresponded to the species M. tresamici from Honduras. This subclade was recovered basal to the strongly supported species M. yatesi (BA: 1, ML: 90%) from Bolivia. The latter was recovered with robust support (BA: 1, ML: 98%) basal to the species M. simmonsae from Ecuador. These subclades were recovered by both phylogenetic analyses as successive sister taxa of seven subclades comprising the M. minuta complex (denoted with Mi abbreviature; Fig. 2). This complex is represented by two reciprocally monophyletic genetic groups; the first is robustly supported (BA: 1, ML: 99%) and included individuals from eastern Ecuador and Colombia (Mi A), the second showed lower support (not recovered in the ML analyses) included two genetic subgroups. The first subgroup contained an individual from eastern Colombia (Mi B), and its sister robustly supported subgroup (BA: 1, ML: 94%), included individuals from Colombia, Trinidad, and Venezuela (Mi C). The second weakly supported genetic group (BA: 0.9, ML: 38%) included a first strongly supported subgroup (BA: 1, ML: 100%) including individuals from French Guiana and Suriname (Mi E), and a second low supported subgroup made up of three strongly supported subclades including individuals from Bolivia (Mi D; BA: 1, ML: 92%), Guyana and Suriname (Mi G; BA: 1, ML: 100%), and Suriname (Mi F; BA: 1, ML: 100%; Fig. 2).

Figure 2. 

Bayesian phylogram of the genus Micronycteris from the phylogenetic analysis of combined evidence (mtDNA + nDNA). Numbers below the nodes correspond to Bayesian posterior probabilities, and those above correspond to bootstrap support values (percentages). The acronyms Mi, Hi, and Me represent the candidate species for M. minuta, M. hirsuta and M. megalotis, respectively. Sequences per group are listed in Table S1. Inset photos: (Mi A) ICN-24465♀: Colombia, Caquetá, Florencia, Vereda el Venado, Macagual; (Mi B) ICN 23912♂: Colombia, Vichada, Cumaribo, Matavén River; (Hi A) ICN-23867♀: Colombia, Guaviare, San José del Guaviare, Vereda Los Alpes, El Provenir farm; (Hi B) ICN Temporal D3M541♂, Colombia, Santander, Puerto Parra, Bocas del Carare; (Sch) ICN-24479♂: Colombia, Magdalena, Santa Marta, Tayrona National Natural Park; (Me C) ICN-23869♂: Colombia, Guaviare, San José del Guaviare, Vereda Los Alpes, El Provenir farm; (Me E) ICN-23839♂: Colombia, Guaviare, San José del Guaviare, Vereda El Raudal, Angosturas II; (Me G) ICN-24495♀: Colombia, Huila, Gigante, Vereda Matambo, La Ensillada farm; (Me H) ICN-23203♂. Colombia, Meta, La Macarena, Vereda Caño Canoas, High plain of Caño Canoas.

The second strongly supported major clade, matching the subgenus Xenonectes, corresponded to the M. hirsuta complex. This major clade was formed by two strongly supported monophyletic subclades (denoted with Hi abbreviature; Fig. 2). The first corresponded to a strongly supported subclade from eastern Colombia (Hi A; Fig. 2). The second is a strongly supported subclade (BA: 1, ML: 89%) included individuals from Panama, western Colombia, Costa Rica, French Guiana, Suriname, Trinidad, and western Ecuador (Hi B; Fig. 2).

The third major clade corresponded to the subgenus Leuconycteris and included two strongly supported reciprocally monophyletic subclades. The first subclade constituted by M. brosseti from Guyana; and the second subclade by M. schmidtorum, including individuals from French Guiana, western Ecuador, and both western and eastern Colombia (Fig. 2).

Finally, the fourth major clade represented by the subgenus Micronycteris included three inclusive reciprocally monophyletic subclades (Fig. 2). The first robustly supported subclade (BA: 1, ML: 81%) comprised the species M. giovanniae, formed by an individual from western Ecuador, and as the sister taxon to an undescribed species “M. sp.”., from Honduras (Fig. 2). The second moderately supported monophyletic subclade (BA: 1, ML: 84%), comprised a strongly supported lineage corresponding to the species M. matses. This lineage represented by individuals from eastern Peru, and appeared as sister of a weakly supported genetic group (BA: 1, ML: 67%) composed of individuals from French Guyana, Surinam, Guyana, Trinidad, and southern Venezuela (Me A; Fig. 2). Finally, the third weakly supported monophyletic subclade (BA: 0.7, ML: 38%) comprised two genetic groups. The first group was a robustly supported lineage formed by an individual from Brazil (Me B), as the sister taxon of a robustly supported genetic group (BA: 1, ML: 93%) formed by an individual from eastern Colombia (Me C) and a robustly supported lineage (BA: 0.9, ML: 87%) composed by individuals from Argentina and Peru (Me D; Fig. 2). The second genetic group was recovered with weak support (BA: 0.6, ML: 46%), including an individual from eastern Colombia (Me E), as the sister of a lineage composed by two genetic subgroups. The first one corresponded to a strongly supported subgroup (BA: 1, ML: 99%), comprising individuals from Belize, Mexico, Costa Rica, and western Ecuador (Me F; Fig. 2). This genetic subgroup was shown to be the sister of a robustly supported subgroup (BA: 1, ML: 97%) formed by an individual from western Colombia (Me G), as the sister of a robustly supported genetic subgroup comprising a medium supported lineage (BA: 0.7, ML: 86%) including individuals from western Colombia, Panama, eastern Colombia, southern Venezuela (Me H; Fig. 2). The later genetic subgroup was revealed as the sister of a strongly supported subgroup that comprised a supported lineage (BA: 1, ML: 99%) from Trinidad and Tobago (Me I; Fig. 2), being sister of a strongly supported clade of M. buriri form Saint Vincent Island.

The “Cluster” algorithm of TaxonDNA revealed 24 reciprocally monophyletic clades with genetic divergence of more than 3% (Fig. 1, colored clades). Within the subgenus Schizonycteris we found ten lineages, representing to M. yatesi, M. tresamici, M. simmonsae, and seven clades belonging to the M. minuta complex (Fig. 1; Suppl. material 3: Table S2). Within the subgenus Xenonectes, we found two subclades with > 3% of genetic distance (Fig. 1; Suppl. material 3: Table S2). Within the subgenus Leuconycteris, we found two subclades with > 3% of genetic distance, one corresponding to the recognized species M. brosseti, and the second by the species M. schmidtorum (Fig. 1; Suppl. material 3: Table S2). Lastly, within the subgenus Micronycteris we found ten subclades with > 3% genetic distance between them, including (i) M. giovanniae, (ii) an undescribed species (“M. sp.”), and (iii) eight clades forming the M. megalotis complex (Fig. 1; Suppl. material 3: Table S2). The nominal species M. matses and M. buriri failed to have a > 3% of the genetic distance with any other subclade (Suppl. material 3: Table S2; Fig. 1).

Our Fgb-I7 gene TCS haplotype network contained four genetic clusters separated by a minimum of 23 mutational steps (Fig. 3). These clusters corresponded to 130 haplotypes that matched the 24 mitochondrial lineages identified by the TaxonDNA analyses (> 3% genetically of divergence), and M. matses and M. buriri (that have < 3% of genetic distance; Fig. 1), all recovered by the complete evidence phylogeny too (Figs 2, 4). The subgenus Schizonycteris (Cluster 1) was formed by 27 haplotypes, matching the three recognized species (M. tresamici, M. yatesi, and M. simmonsae) and the seven mitochondrial lineages within the M. minuta complex (Mi A–Mi G) separated between one and 12 mutational steps. The subgenus Xenonectes (M. hirsuta complex. Cluster 2) was formed by 19 haplotypes that corresponded to the two mtDNA lineages Hi A and Hi B, separated by a minimum of seven mutational steps. The subgenus Leuconycteris (Cluster 3) included M. schmidtorum with five haplotypes and M. brosseti with two haplotypes, separated by a minimum of 13 mutational steps from the haplotypes of M. schmidtorum (Fig. 3). Finally, the most extensive haplotype cluster matched the subgenus Micronycteris (Cluster 4), formed by haplotypes that diverged between one to 27 mutational steps from each other. Unique haplotypes in this haplogroup matched three species (i.e., M. matses, M. giovanniae, and “M. sp.”) and nine lineages forming the M. megalotis complex (Fig. 3).

Figure 3. 

Haplotype network of the genus Micronycteris for the Fgb-I7 gene nDNA haplotypes based on an alignment of 686 bp. Circle size reflects haplotype frequency and missing haplotypes are represented by small circles. Each line connecting haplotypes corresponds to one mutational step. Colors within each nominal species haplogroup represent the candidate species and Mi, Hi, and Me correspond to their acronyms in M. minuta, M. hirsuta, and M. megalotis respectively.

The single locus species delimitation analyses revealed contrasting results. The bPTP model delimited 55 entities. Nevertheless, only 15 of those clades showed posterior probabilities above 0.95 (Suppl. material 4). The other clades showed between 0.49 and 0.95, which suggests that caution should be considered when interpreting these clades as independently evolving lineages (Suppl. material 4). The single-threshold method of the GMYC model delimited 48 entities. The model was significantly better than the null hypothesis with the likelihood ratio test (LR = 5040.823, LRT results = 0***, threshold time = -0.008590693; Suppl. material 4). Finally, the mPTP model was somewhat more conservative, identifying 33 species (Fig. 4) that matched several lineages delimited by the bPTP and GMYC methods (Suppl. material 4).

Figure 4. 

Bayesian phylogram of the genus Micronycteris from the analysis of Cytb gene mtDNA (BEAST), showing colored bars that represent the different delimitation schemes obtained with > 3% of genetic differentiation, bPTP, GMYC, mPTP, and Fgb-I7 haplotype network. CS: Confirmed species (Red), CCS: Confirmed Candidate Species (Blue). The dashed vertical line indicates the threshold between the Yule and the coalescent process estimated by the likelihood implementation of the GMYC model. Filled circles at internal nodes indicated strong support for Bayesian (BEAST: PP > 0.95), and the size is proportional between 0.95 to 1 PP. The acronyms Mi, Hi, and Me represent the candidate species in M. minuta, M. hirsuta, and M. megalotis, respectively. Sequences within groups are listed in Suppl. material 1: Table S1 and depicted in Fig. 5.

Considering the different lines of evidence, we revealed 27 different genealogical lineages forming the genus (Fig. 4). They belong to eight confirmed species (CS): M. brosseti, M. buriri, M. giovanniae, M. matses, M. schmidtorum, M. simmonsae, M. tresamici, and M. yatesi, and 19 confirmed candidate species (CCS) with either allopatric or parapatric distribution within three species complexes. We present in Fig. 4 the position of each revealed evolutionary lineage in the Cytb tree and in Fig. 5 their hypothesized geographic distribution.

Figure 5. 

Geographic location of the individuals from which the sequences were obtained and included in the phylogenetic analyses. Shades represent a hypothetic estimation of the distribution of each clade based on the locality of the included individuals. The acronyms Mi, Hi, and Me represent the candidate species in M. minuta, M. hirsuta, and M. megalotis, respectively. Sequences within groups are listed in Suppl. material 1: Table S1.

We found seven CCS within the sub genus Schizonycteris, in the M. minuta complex: (1) lineage Mi A formed by individuals from the Amazon of Ecuador and Colombia; (2) lineage Mi B formed by individuals from the north Amazon of Colombia; (3) lineage Mi C formed by individuals from dry-forest of western Colombia, the Orinoco Llanos of Colombia and Venezuela, and the Trinidad Island; (4) lineage Mi D formed by individuals from Bolivia; (5) lineage Mi E formed by individuals from north French Guiana and Suriname; (6) lineage Mi F formed by individuals from south Suriname and (7) lineage Mi G formed by individuals from north Guyana and Suriname. Two more CCS were confirmed within the subgenus Xenonectes in the M. hirsuta group: (1) lineage Hi A, including individuals from eastern Colombia and (2) lineage Hi B, including individuals from Costa Rica, Trinidad, French Guyana, Panama, Suriname, western Colombia, and western Ecuador. Finally, ten CCS within the subgenus Micronycteris: (1) lineage “M. sp.” from Honduras; (2) lineage Me A including individuals from Suriname, French Guyana, easternmost Venezuela, Guyana, and Trinidad; (3) lineage Me B including an individual from Brazil (this lineage could represent the nominal M. megalotis due to its type locality “unknown locality in Brazil”); (4) lineage Me C including individuals from north Amazon of Colombia; (5) lineage Me D comprising individuals from Argentina, Bolivia, and Peru; (6) lineage Me E including an individual from north Amazon of Colombia; (7) lineage Me F including individuals from Belize, Costa Rica, Mexico, and western Ecuador; (8) lineage Me G including individuals from the Magdalena valley and the eastern versant of the Andean Cordillera of Colombia; (9) lineage Me H formed by individuals from Panama, the dry forest of western Colombia and the Orinoco Llanos of Colombia and Venezuela and (10) lineage Me I formed by individuals from Trinidad. Due to the distribution of the Me F and Me H, two available names, Micronycteris mexicana and Micronycteris microtis might be applied to these lineages. Nonetheless, the applications of these names depend on a posterior taxonomic revision. We make no conclusions about the status of M. sanborni because we did not have enough information to include it in the applied framework.

The assessment of species diversity within Micronycteris is a complex task due to the large morphological variation and a lack of precision in the assignation of genetic sequences to specific lineages. Our analyses identified several independently evolving evolutionary lineages that correspond to different species and could be described under an integrative taxonomic approach. Currently, 13 species are accepted (Siles and Baker 2020); but considering the high number of confirmed candidate species (CCS) discussed here, we estimate that the genus comprises at least 27 species. Nevertheless, despite our comprehensive data set there are still geographic gaps that need to be filled in the future, including Central America, Central Amazonia, and the Andean Cordillera.

Several authors have suggested that the genus Micronycteris includes high cryptic diversity (e.g., Porter et al. 2007). However, the assessment of cryptic diversity has been limited by: (1) the incongruence between mitochondrial and nuclear genetic data, (2) the use of the genetic species concept (Bradley and Baker 2001) as the principal line of evidence, and (3) the variability of morphological data. For example, Clare (2011) in a search for cryptic species revealed incongruence between the mitochondrial gene (COI) and a region of paternally inherited Y-chromosome (Dby ^7th intron) in M. megalotis. The incongruence between markers was due to the slower divergence of the Y-chromosome region that could reflect incomplete lineage sorting in that gene (Clare 2011). However, our analyses revealed unique haplotypes in the Fgb-I7, matching most of the clades revealed by the mtDNA data analyses.

In another context, the use of genetic distance for species delimitation within Micronycteris (the genetic species concept) has not been homogeneously employed in the literature. For example, Clare (2011) proposes that M. megalotis is a species with a high intraspecific genetic variation that is the highest reported between recognized bats (COI: mean = 4.2%, range 0–7.7% using Kimura 2 parameter). Nevertheless, some authors have used smaller genetic values (near 2% in the Cytb) to identify and describe new species (e.g., M. buriri by Larsen et al. 2011). The genetic distance as the only criterion for defining species must be used with caution because such a measure is generally based on single mitochondrial genes with relatively high mutational rates like the Cytb (Bradley and Baker 2001). In our study, we used a conservative approach where the genetic distance between our identified genealogical lineages that correspond to undescribed species ranged above 3.0%, except for M. buriri and M. matses that differed between 1.90% and 2.90% respectively from its most closely related lineage.

Finally, Micronycteris has had a complex morphological taxonomy for multiple reasons. Several of the recognized species were described based on morphological characters that, after being described, were re-evaluated or redefined posterior to the revision to a higher number of specimens. That was the case of M. homezorum and M. minuta (Ochoa and Sánchez 2005), M. minuta and M. sanborni (Feijó et al. 2015), and M. microtis and M. megalotis (Martins et al. 2014). Furthermore, the description of several species was based on the morphological revision of a few specimens from a few localities. This was the case of M. matses (eight specimens from only one locality; Simmons et al. 2002), M. giovanniae (one specimen from only one locality; Fonseca et al. 2009), M. yatesi (three specimens from only three localities; Siles et al. 2013) and M. buriri (19 specimens for San Vicente Island; Larsen et al. 2011). This problematic taxonomic scenario weakens the validity of several species of the genus due to the lack of discrete morphological characters, limited knowledge on morphological variation, and the use of the phylogenetic species concept based on a single locus and the small interspecific genetic distances of differentiation (see Bradley and Baker 2001; Zachos et al. 2013). Taking those problems into account and based on our results, in the following section we clarify some aspects concerning essential issues needed for taxonomic stability in the genus Micronycteris, and we propose some lines of research that should be considered for future exploration.