Systematics and Taxonomy of Tonatia saurophila Koopman & Williams, 1951 (Chiroptera, Phyllostomidae)

Abstract The Stripe-headed Round-eared bat, Tonatia saurophila, includes three subspecies: Tonatia saurophila saurophila (known only from subfossil records in Jamaica), Tonatia saurophila bakeri (distributed from southeastern Mexico to northern Colombia, Venezuela west and north of the Cordillera de Mérida, and northwestern Ecuador), and Tonatia saurophila maresi (distributed in Venezuela east and south of the Cordillera de Mérida, the Guianas, Trinidad and Tobago, northeastern Brazil, and along the upper Amazon basin in Colombia, Ecuador, Peru, and Bolivia). The last two subspecies are an attractive example to test predictions about the historical role of the Andes in mammalian diversification. Based on morphological descriptions, morphometric analyses, and phylogenetic reconstruction using the mitochondrial gene Cyt-b and the nuclear exon RAG2, this study evaluates the intraspecific relationships within Tonatia saurophila and the taxonomic status of the taxon. The three subspecies of T. saurophila are recognizable as full species: Tonatia bakeri, Tonatia maresi, and Tonatia saurophila. The latter is restricted to its type locality and possibly is extinct. Tonatia bakeri, in addition to being larger than T. maresi, is morphologically distinguishable by possessing an acute apex at the posterior edge of the skull, a well-developed clinoid process, and relatively robust mandibular condyles, and by lacking a diastema between the canine and the first lower premolar. The genetic distance between T. bakeri and T. maresi is 7.65%.


Introduction
The Neotropical bat genus Tonatia Gray, 1827 (Phyllostomidae, Phyllostominae) includes two species: T. saurophila and T. bidens. This genus is widely distributed from the southern Mexico to northern Argentina and Paraguay (Williams and Genoways 2008). Members of the genus Tonatia are easily differentiated from other phyllostomines by a combination of the following characteristics: a single pair of lower incisors; three pairs of lower premolars; a tail that extends to the middle of the uropatagium; short and wide ears that are not joined by a band of skin on the forehead; absence of a notch on the lower margin of the pinna; hairy face, ears, forearms, legs, and feet; and postorbital constriction greater than 5 mm (Williams et al. 1995;Williams and Genoways 2008).
Since its description in 1823 the taxonomic history of Tonatia has been controversial. Spix (1823) described Vampyrus bidens based on a single specimen from Bahia, Brazil. Four years later, Gray (1827) described the genus Tonatia and considered V. bidens as its type species. A decade later, Gray (1838) described Phyllostoma childreni, a species with similar characteristics to V. bidens, based on a single specimen without a precise location in South America. Concurrently, d 'Orbigny (1836) described the genus Lophostoma based on a specimen of L. silvicola collected in Bolivia, the only species he included in the genus (d 'Orbigny 1836;d'Orbigny and Gervais 1847). Dobson (1878) listed three species for Lophostoma: L. bidens (Spix 1823), L. brasiliense Peters, 1867, and L. amblyotis Peters, 1867; he also listed L. silvicolum as a junior synonym of L. amblyotis. Additionality, Dobson (1878) synonymized V. bidens and P. childreni under the name Lophostoma bidens. Based on Dobson's taxonomic arrangement and because L. bidens was assigned to the genus Tonatia Gray, 1827, Palmer (1898 treated Lophostoma as a junior synonym of Tonatia Gray, 1827. More than a century later, Koopman and Williams (1951) described Tonatia saurophila, based on fragmentary subfossil material found by H. E. Anthony (1919)(1920) in two caves in Jamaica (Wallingford and Dairy). Later, Koopman (1976) considered T. saurophila as a subspecies of Tonatia bidens. The synonymy of Lophostoma under Tonatia was not challenged until studies of G-banded karyotypes, allozymes, and albumin immunology (Patton and Baker 1978;Baker and Bickham 1980;Arnold et al. 1983;Honeycutt and Sarich 1987) showed that the T. bidens complex was divergent from the other species of the genus. Lee et al. (2002) examined DNA sequence variation in the mitochondrial ribosomal genes and found significant differentiation and paraphyly within the genus Tonatia. Based on these results, Lee et al. (2002) recommended restricting Tonatia to T. bidens and T. saurophila and transferring brasiliense, carrikeri, evotis, schulzi, and silvicola to Lophostoma. Based on morphological and morphometric differences, Williams et al. (1995) elevated T. bidens saurophila from subspecies to species level, resulting in Tonatia to include two species: Tonatia bidens and T. saurophila. Furthermore, Williams et al. (1995) described two subspecies in T. saurophila: T. s. bakeri and T. s. maresi.
The Stripe-headed Round-eared bat, Tonatia saurophila, is characterized by presenting a secondary process in the mastoid that partially obscures the base of the mastoid bulla. It also presents a conspicuous space between the cingula of the lower canines, lower premolars relatively uncrowded with the first lower premolar slightly overlapping on the second premolar, and a clear line of short fur on top of the head stretching laterally between the eyes and crown (Williams et al. 1995;Tirira 2017). Currently three subspecies are recognized: T. saurophila saurophila (only subfossil records; type locality: Balaclava, Jamaica); T. saurophila bakeri, distributed from southern Mexico southward into South America to northern and western Colombia, northwestern Ecuador, and Venezuela (west and north of the Cordillera de Mérida); and T. saurophila maresi distributed in Venezuela (east and south of the Cordillera de Mérida), the Guianas, northeastern Brazil, and along upper Amazon basin of Colombia, Ecuador, Peru, Bolivia, and Brazil (Williams et al. 1995;Williams and Genoways 2008).
Tonatia saurophila is in need of a taxonomic and systematic revision. Its wide geographic distribution in Central and South America, which includes populations on both sides of the Andes, raises questions regarding the role of this mountain range as a potential barrier to gene flow and a promoter of diversification. The phylogenetic relationships between the species of Tonatia, and within the subspecies of T. saurophila, have not been investigated. Herein, we aim to evaluate the taxonomic status of the subspecies of T. saurophila based on morphological, morphometric, and molecular data. We discuss the role of the Andes in the diversification and the taxonomic substructure of this taxon in light of the results.

Sampling and measurements
We examined 137 adult specimens of Tonatia saurophila (68 females, 67 males, and two specimens of undetermined sex), of which 31 were collected in localities west of the Andes, and 106 from localities east of the Andes (Appendix 1). We evaluated 21 craniodental and external measurements based on the external and osteological characteristics defined by Williams et al. (1995) and Velazco and Gardner (2012). The measurements of the body and skull were taken from the left side and of the jaw from the right side. All measurements were made using digital calipers with an accuracy of 0.01 mm. The craniodental and external measurements used in this study were: FA, forearm length; METIII, metacarpal III length; METIV, metacarpal IV length; METV, metacarpal V length; TiL, tibia length; HF, hind-foot length; GLS, greatest length of skull; CB, condylobasal length; CCL, condylocanine length; CIL, condyloincisive length; BB, braincase breadth; PB, postorbital constriction breadth; MPW, mastoid process width; ZB, zygomatic breadth; BC, breadth across upper canines; PL, palatal length; DENL, dentary length; COH, coronoid height; MANDL, mandibular toothrow length; MTRL, maxillary toothrow length; and M2M2, width at M2s.

Morphological and morphometric analyses
In order to identify morphological differences among T. saurophila populations across their range, 23 specimens from localities west of the Andes and 77 from localities east of the Andes were examined. From these observations, each individual was examined and variation patterns of various qualitative and discrete characteristics were described, following Velazco (2005). Descriptive statistics (mean, standard deviation, and minimum and maximum values) were obtained from all the measured specimens. Data transformation was performed through a standardization process. Differences between sexes and between subspecies were evaluated by Principal Component Analysis (PCA) and Discriminant Function Analysis (DFA). Factorial points of these multivariate tests were graphically plotted in morphospace to show relationships between subspecies. All tests were done using the statistical software package PAST v. 1.0 (Hammer et al. 2001).

Molecular analyses
Tissue samples from 15 specimens stored in the QCAZ mammal collection were used for the molecular analysis. DNA was obtained from liver, muscle, or tail preserved in 95% ethanol and stored at −80 °C, as well as from dried skin fragments from specimens in fluid. The DNA was extracted using the salt protocol (Bilton and Jaarola 1996), modified in the use of 300 μl of ethanol for the washes instead of 1000 μl. The concentration and quality of DNA was measured using the NanoDropTM1000 v. 3.7 spectrophotometer (Thermo Scientific). From the stock solution, aliquots of 20 ng/μl of DNA concentration were prepared to be used in PCR reactions. Sequences of the Cytochrome b (Cyt-b) and the Recombination Activating Gene (RAG2) were amplified and sequenced for this study. The following primer pairs were used for the Cyt-b gene: forward primer glo7L and reverse primer glo6H, and for RAG2: forward primer RAG2-F1 and reverse primer RAG2-R1. Thermal profile for the Cyt-b PCR followed Hoffmann and Baker (2001), and for RAG2, PCR followed Baker et al. (2000). The amplicons were visually evaluated with gel electrophoresis and subsequently purified with ExoSap-IT (GE Healthcare, Chalfont St. Giles, UK). Amplicons were then sequenced by Sanger method at Macrogen Inc. (Seoul, South Korea).
Sequences were edited using Geneious R11 (https://www.geneious.com), and aligned using the ClustalW tool. We calculated interspecific and intraspecific genetic distances using software MEGA v. 7.0 (Kumar et al. 2015); to get corrected distances we used the Kimura 2 Parameters. The best partition strategies along with corresponding models of evolution were obtained in PartitionFinder v. 1.1 (Lanfear et al. 2012). For the Bayesian Inference Analysis (BI) the best substitution models for Cyt-b were: first position K80+G, second position HKY+I and third position GTR+I. For RAG2 they were: first and second position HKY+G and for third position K80+G, while for the Maximum Likelihood Analysis (ML) substitution model used was GTR. The ML analysis was conducted using RAXML (Stamatakis 2014). Nodal support was determined by 1000 bootstrap replicates. The BI analysis was conducted using MrBayes v. 3.2.2 (Ronquist et al. 2012). Four Markov chains were run twice for 10,000,000 generations. Trees were sampled every 1,000 generations resulting in 20,000 trees saved per analysis. Adequacy of chain mixing was assessed by examining effective sample sizes (ESS) in Tracer, with ESS > 200 considered as satisfactory and plotting the -ln L per generation. After analyzing convergence, chain mixing, and sampling, the first 1000 trees sampled were discarded as "burn-in". The remaining trees were used to obtain a consensus tree by 50% majority rule. To evaluate the monophyly and phylogenetic relationships of our Tonatia samples, several phyllostomines were selected as outgroup (Table 1), which have been used in previous works (Lee et al. 2002;Velazco and Cadenillas 2011;Camacho et al. 2016). The outgroup sequences were obtained from GenBank (www.ncbi.nlm.nih.gov/Genbank). Table 1. Specimens used for phylogenetic analyses. Species, museum and tissue ID numbers, and Gen-Bank accession numbers are given for the Tonatia and outgroup samples used in the phylogenetic analyses presented in this study. *Misidentified specimens.

Morphometric analyses
The statistical analyses performed on data obtained from measurements of the entire sample set of Tonatia saurophila found no sexual dimorphism within groups for the analyzed variables (contrary to that observed in Tonatia bidens and in some species of the genus Lophostoma; Davis and Carter 1978;Williams et al. 1995). In addition, we found that specimens from Central America and the western foothills of the Andes in Venezuela, Colombia, and Ecuador (Tonatia saurophila bakeri) share similar morphometric characteristics that separate them from specimens from Trinidad and Tobago, Guyana, French Guiana, Brazil and individuals from eastern foothills of the mountain range in Colombia, Ecuador, and Peru (Tonatia saurophila maresi), which form morphometrically independent groups (Figs 1, 2). The subspecies T. saurophila bakeri presents larger craniodental and external measures than T. saurophila maresi (PCA, percentage of variation: 71.3%; DFA, percentage of variation: 86.84%). CCL, GLS, CIL, MTRL, MANDL, and ZB are variables that contribute the most in discriminating these two subspecies (Table 2).

Morphological analyses
Four patterns of cranial and morphological variation were recognized: (1) from a dorsal view of the skull, the posterior edge of the cranial cavity in T. saurophila maresi presents a blunt vertex due to presence of a poorly developed sagittal process, whereas T. saurophila bakeri presents an acute apex due to presence of a well-developed sagittal process ( Fig. 3a, b); (2) in an occlusal view of jaw, the mandibular condyle is feeble in T. saurophila maresi and robust in T. saurophila bakeri (Fig. 3c, d); (3) in lateral view of mandible, the diastema between canine and first premolar is small in T. saurophila bakeri, while in T. saurophila maresi the diastema is larger (Fig. 4a, b); and, (4) in view through the foramen magnum (basioccipital view), the clinoid process is well developed in T. saurophila bakeri and poorly developed or absent in T. saurophila maresi (Fig. 4c, d).
Externally, the coloration of the nose leaf, warts of lower lip, and skin surrounding the mouth is lighter in T. saurophila bakeri, whereas the skin color in those areas in T. saurophila maresi is darker. In addition, T. saurophila maresi has darker pelage that present patches of hair with reddish tips, while the pelage in T. saurophila bakeri is lighter and uniform in color (Fig. 5).

Phylogenetic analyses
Maximum Likelihood and BI analyses of the two genes analyzed independently (Cytb and RAG2; Fig. 6) recovered the genus Tonatia as monophyletic and as sister to a  clade that included representatives of Phyllostomus, Phylloderma, Gardnerycteris, and Lophostoma (Fig. 6). Within Tonatia, the monophyly of T. bidens was strongly supported. Within T. saurophila, two well-supported clades were recovered. One clade Table 2. Measurements (in mm) of Tonatia bakeri and Tonatia maresi. Measurements are given for the holotypes and for the specimens included in this study (mean and observed range). Holotype data taken from Williams et al. (1995). Both clades exhibit high support values (Fig. 6). The Cyt-b gene topology showed two clades within T. saurophila maresi. Samples from the Guiana Shield, Brazil, and Peru formed a poorly supported clade, while samples from eastern Ecuador formed a clade with high support (Fig. 6). We recovered the sample DQ903829 identified as T. bidens (Brazil) nested within the clade of T. s. maresi. The exon RAG2 topology for T. s. maresi did not record these two clades. The sample FN641681 identified as P. discolor (Costa Rica) is located within the clade of T. s. bakeri (Fig. 6). In the case of the FN641681 sample, Dávalos et al. (2012) confirmed that the specimen was misidentified. The DQ903829 sample may have the     same identification error due to the morphological complexity of the genus Tonatia and due to the use of few genetic samples from T. saurophila and T. bidens. This same misidentification could have occurred in previous works, but, in our case, when using more sequences of T. saurophila, the error was notorious. The average Cyt-b pairwise distance between clades from west of the Andes (T. s. bakeri) and east of the Andes (T. s. maresi) is 7.65% ± 0.65. The clades of T. s. bakeri and T. bidens exhibit a genetic differentiation of 13.66% ± 1.12, and the clades of T. s. maresi and T. bidens differ by 13.52% ± 1.02 (Table 3). Levels of intraspecific variation within T. s. bakeri and T. s. maresi are 1.73% ± 0.29 and 4.01% ± 0.3, respectively (Table 3). Within the T. s. maresi clade, the specimens grouped by the country of origin (e.g., Brazil, Ecuador, Peru, and the Guiana Shield) exhibit values between 3.51% and 5.88% (Table 4).

Tonatia bakeri Tonatia bakeri Tonatia maresi Tonatia maresi
A century has passed since Harold E. Anthony recovered the subfossil material that was used by Koopman and Williams to describe T. saurophila. Despite numerous bat surveys since then throughout the West Indies, no additional records of this bat either alive or as subfossil remains have been confirmed. In spite of being known from fragmentary remains, morphological and morphometric differences have been found between the subfossil samples of T. saurophila and specimens of T. bakeri and T. maresi. Based on the aforementioned information we support the recognition of †Tonatia saurophila as an extinct full species, with a distribution restricted to Jamaica. Distribution. The only record of this species is based on the subfossil remains found by H. E. Anthony in the aforementioned caves in Jamaica (Koopman and Williams 1951 ; Fig. 7).

Taxonomy
Diagnosis. Similar to Tonatia bidens, but differing in having: the axis of the talonid of m3 straight in an anteroposterior sense, instead of running obliquely in a lingual-labial direction; slightly lower coronoid; more bulbous forehead; a well-developed labial posterior lobe of the last upper premolar; overall size smaller (Koopman and Williams 1951).
Description. The holotype is a partial mandible. The mandible is comprised of the entire dentary bone except for the end of the angular process. All three molars are  Williams and Genoways (2008), and based on the localities of the specimens included in this study. present along with the last premolar. Complete dental formula of the mandible can be determined from alveoli. There are two small roots, a large canine root and a single small incisor root in front of the last premolar. Also, the coronoid is moderately high. The last premolar is anteriorly squared and therefore the middle premolar is relatively larger (Koopman and Williams 1951). The subfossil fragment of the specimen AMNH 147205 (as shown in the description) includes the entire rostrum, except the extreme anterior end, almost the entire hard palate, and the roots of the teeth except the incisors. Also, the anterior border of the orbit rises obliquely to join the dorsal border. This subfossil also has a slightly bulbous forehead and the presence of an anterior lobe on the last premolar.
Measurements of the holotype taken by Koopman and Williams (1951) are: mandibular toothrow length, 9.8 mm, and depth of ramus behind last molar, 3.1mm. Also, measurements of the paratype taken from Koopman and Williams (1951) are: mandibular toothrow length, 9.5 mm; depth of ramus behind last molar, 2.9 mm; and coronoid height, 6.3 mm.
Comparisons. Tonatia saurophila is smaller than any other species within the genus. Tonatia saurophila differs from Tonatia bidens in having the axis of the talonid of m3 running not obliquely in a lingual-labial direction but straight anteroposteriorly, in having a somewhat more bulbous forehead, and in possessing a well-developed posterior lobe on the last premolar (Koopman and Williams 1951). On the other hand, T. saurophila differs from T. bakeri and T. maresi by having smaller craniodental measurements. Mandibular toothrow length in the holotype of T. saurophila is 9.8 mm, while in specimens of T. bakeri and T. maresi, analyzed in this study, the mandibular toothrow length averaged 11.37 mm and 10.51 mm, respectively. In addition, the coronoid height in the paratype of T. saurophila is 6.3 mm, while in specimens of T. bakeri and T. maresi, analyzed in this study, the coronoid height averaged 7.61 mm and 7.07 mm, respectively. Also, the axis of the talonid of the last molar running not obliquely in a lingual-labial direction, while tends to be oblique in T. bakeri and T. maresi.
Etymology. The etymology of the term Tonatia is unknown (Medellín and Arita 1989). The name saurophila is the union of the Latin terms saurus, lizard and philus, loving. This is because T. saurophila was described based on some subfossil material from the "Lizard" stratum of the Wallingford Roadside cave (Williams and Genoways 2008).
Remarks. In 1951, Koopman and Williams considered the fragmentary subfossil material found in the Jamaican caves as a new species. Then, Williams et al. (1995) recognized the Jamaican taxon as the subspecies Tonatia saurophila saurophila, along with two other subspecies (T. saurophila bakeri and T. saurophila maresi). Herein, we recognize T. bakeri and T. maresi as full species, and support that T. saurophila be considered as an extinct monotypic entity, as only subfossil specimens have been recorded in 1920, and there have been no new records since then.

Tonatia bakeri Williams, Willig & Reid, 1995
Tonatia saurophila bakeri Williams et al. 1995: 622. Holotype. Adult male, deposited at the Museum of Texas Tech University (TTU 39120), collected on 31 January 1983 by R. J. Baker (original field number 1195), 6 km SW of Cana, Darién, Panama, 1200 m. Prepared as skin and skull by M. S. Hafner. Karyotype reference number TK22573. No paratypes were designated by Williams et al. (1995), but several specimens from Mexico, Belize, Costa Rica, Guatemala, Honduras, Nicaragua, Panama, and Venezuela were listed and used in the description.
Distribution. Tonatia bakeri is distributed from southeastern Mexico southward into South America to northwestern Ecuador, northern Colombia, and northern Ven-ezuela (west and north of Cordillera de Mérida). The southernmost records of the species are located in the Province of Esmeraldas, Ecuador (Fig. 7).
Diagnosis. Tonatia bakeri is distinguished from T. bidens and T. maresi mainly by craniodental and external measurements. Tonatia bakeri is larger than any other species within the genus (Williams et al. 1995; Table 2). The measurements that explain most of the variability are postorbital constriction length, mastoid width, and upper canine width. The skin around the mouth, nose leaf, and warts of the lower lip presents a light coloration. The posterior edge of the cranium presents an acute apex, the mandibular condyles are robust, the diastema between the canine and first lower premolar is absent or not evident, and clinoid process is well developed (Figs 3a, c, 4a, c). Measurements of the holotype taken from Williams et al. (1995), as well as averages of the external craniodental and external measurements of the specimens analyzed in this study, are presented in Table 2.
Description. The holotype has bicolored gray-brown dorsal fur with dark tips; hairs behind ears and neck are bicolored with white bases, and are overall slightly lighter than the rest of the dorsum. Hairs on shoulders are gray-brown, presenting white bases and tips (tricolored). The hairs on the top of head have white tips forming a pale stripe between the ears. Ventral pelage is paler than dorsal pelage; hairs are fawn with lighter tips, but around the throat the fur has a lighter and more uniform color. Dorsal and ventral pelage is dense. Dorsal hairs are longer (12.0 mm) than ventral hairs (5.0 mm). The proximal third of the forearm is covered by short hair (the ventral surface is more densely furred than dorsal), as well as base of the thumbs and proximal side of the feet. The proximal ventral margins of the uropatagium and wing membranes have sparse short hair. The skull of the holotype is complete and in perfect condition, presenting a well-defined sagittal crest.
Comparisons. Tonatia bakeri and T. maresi differ from T. bidens in the presence of a clear line of short fur on the top of the head between ears, a secondary process in the mastoid that partially obscures the base of the bulla (unique within subfamily Phyllostominae), a larger gap between lower canines, a lower crowded appearance of the premolars, and a narrower postorbital constriction. Tonatia bakeri differs from T. maresi in its narrower breadth across the lower incisors (Williams et al. 1995), presence of an acute apex in the posterior edge of the braincase due to the presence of a well-developed sagittal process, robust mandibular condyles, a reduced or absent space between the canine and the first lower premolar, observable in the mandible body (lateral view), and a well-developed clinoid process (Figs 3a, c, 4a, c). Additionally, T. bakeri specimens are larger than those of T. maresi (Table 2).
Etymology. The name bakeri was coined in recognition to Robert J. Baker for his contributions to the genetics and systematics of the family Phyllostomidae (Williams et al. 1995).
Remarks. Little is known on the natural history of the genus Tonatia. It has been reported that it uses hollow trees as day roosts, within which forms monospecific groups, or multispecific groups with other bat species (Williams and Genoways 2008). The diet of Tonatia includes arthropods, fruit, and small vertebrates (Tirira 2017).

Tonatia maresi Williams, Willig & Reid, 1995
Tonatia saurophila maresi Williams et al. 1995: 623. Tonatia saurophilla Falcão et al. 2005; incorrect subsequent spelling of T. saurophila Koopman & Williams, 1951. Holotype. Adult female, deposited at the Museum of Texas Tech University (TTU 9774), collected on 12 July 1969 by R. J. Baker (original field number 318) in Blanchisseuse, Trinidad and Tobago. Prepared as skin, skull, and partial postcranial skeleton by S. L. Williams. No paratypes were designated by Williams et al. (1995), but several specimens from Colombia, Venezuela, Guyana, Suriname, French Guiana, Trinidad and Tobago, Brazil, Peru, and Ecuador were listed and used in the description. Distribution. Tonatia maresi is restricted to South America. It occurs in Venezuela (east and south of Cordillera de Mérida), the Guianas, northeastern Brazil, and along the upper Amazon basin of Colombia, Ecuador, Peru, and Bolivia, as well as in the South American islands of Trinidad and Tobago (Fig. 7).
Diagnosis. Tonatia maresi is distinguished from other extant species of Tonatia by its smaller craniodental and external measurements. The measurements that explain most of the variability are postorbital constriction length, mastoid width, and upper canine width. The skin around the mouth, nose leaf, and warts of the lower lip present a dark coloration. The posterior edge of the cranium presents a blunt vertex due to the poorly developed sagittal process, mandibular condyles are gracile, the canine and the first lower premolar are separated by a diastema, and the clinoid process is poorly developed or absent (Figs 3b,d,4b,d). The measurements of the holotype, taken from Williams et al. (1995), as well as the averages of the external craniodental and external measurements of the specimens analyzed in this study are presented in Table 2.
Description. The holotype has dark gray-brown dorsal fur with patches of hair having reddish tips (bicolored). The hairs on the shoulder have white tips and, like the hairs behind the ears and around the base of the neck, present a white base (tricolored). The hairs on the top of head have white tips forming a pale stripe between the ears. Ventral pelage is grayer, and paler, than the dorsal hair and has white tips. The throat region has a uniformly colored hair. The body is densely furred, with the dorsal hairs longer (12.0 mm) than the ventral hairs (5.5 mm). The forearm presents shorter hairs on the proximal half of its length, with the ventral surface being more densely furred. Short, sparse hairs occur on the inner margins of the ventral surfaces of the uropatagium and the wing membranes; short hairs also occur on the thumbs and feet. The skull of holotype is complete, and in perfect condition.
Comparisons. Specimens of Tonatia maresi are smaller than those of T. bakeri and T. bidens. Additionally, T. maresi can be distinguished from T. bakeri by its wider breadth across the lower incisors (Williams et al. 1995), by the presence of a blunt vertex in the posterior edge of the braincase due to a poorly developed sagittal process, delicate mandibular condyles, the presence of a space between the canine and the first lower molar, observable in the mandible body (lateral view), and the lacking or poorly developed clinoid process (Figs 3b,d,4b,d). The coloration of the dorsal pelage on T. maresi is dark and usually presents patches of hair with reddish tips; while in T. bakeri it is light and uniform in color. The warts on the lower lip are darker in T. maresi than in T. bakeri.
Etymology. The name maresi was coined in recognition to Michael A. Mares for his contributions to the systematics, ecology, and zoogeography of South American mammals (Williams et al. 1995).
Remarks. Little is known about its natural history. The diet of Tonatia includes various arthropods such as crickets, cicadas, and spiders. Additionally, they consume fruit, and small vertebrates such as lizards and birds (Tirira 2017). The species roosts in hollow trees, forming monospecific groups, or multispecific groups with other bat species (Williams and Genoways 2008). Recently, three species of ectoparasites (Mastoptera minuta, Pseudostrebla greenwelli, and Strebla tonatiae: family Streblidae) were found in specimens of Tonatia in the Reserva Natural La Palmita, Department of Casanare, in the Colombian Llanos. This locality occurs within the range of T. maresi (Liévano-Romero et al. 2019).

Discussion and conclusion
Our study shows that, within extant Tonatia saurophila populations, there are two clearly differentiated genetic lineages, namely, the lineage of Tonatia maresi, which includes samples from eastern Venezuela, Colombia, Ecuador, Peru, Trinidad and Tobago, northeastern Brazil, and the Guiana Shield, and the lineage of Tonatia bakeri, which includes samples from Costa Rica, Colombia, Venezuela, and western Ecuador (Fig. 6). These lineages were formerly considered to be subspecies of a single species (T. saurophila maresi, T. saurophila bakeri; Williams et al. 1995). For decades, the recognition of three subspecies in Tonatia saurophila was commonly accepted; however, the integration of morphological, morphometric, and molecular evidence indicates that T. bakeri and T. maresi are two well-supported and distinguishable taxa. In addition to being dissimilar in size and morphology, they differ genetically at the 7.65% level, a percentage within the 3.3%-14.7% range of genetic distances known to separate sister species of mammals when the Cyt-b gene is considered (Baker and Bradley 2006). More recent works that resolve the taxonomic status of species in Phyllostomidae have shown similar genetic differences ranges to those of this study, such as between Sturnira burtonlimi and S. adrianae: 3.93% + 0.25, Sturnira hondurensis and S. ludovici: 5.74% + 0.46 (Molinari et al. 2017), and, specifically in the subfamily Phyllostominae, Lophostoma silvicolum and L. evotis: 5.02% + 0.49; Lophostoma carrikeri and L. brasiliense: 12.78% + 0.97 (Camacho et al. 2016); Gardnerycteris crenulatum and G. koepckeae: 11.2% + 1.0 (Hurtado and D'Elia 2018).
According to simultaneous phylogenetic analyses, it is estimated that diversification between bat genera of the subfamily Phyllostominae occurred during beginning and mid-Miocene (23-16.9 Ma; Hoffmann et al. 2008). Specifically the split between Tonatia and a clade that included Artibeus (Stenodermatinae) and Anoura (Glossophaginae) is estimated to have occurred 22 Ma ago (Teeling et al. 2005). At the end of this period, there were probably several ancestral migration processes of Tonatia throughout northern South America (as also occurred with Carollia), especially in the Northern Andes . The divergence between T. bakeri and T. maresi may have occurred due to a process of allopatric speciation, in which an ancestral population split into two separate lineages as a result of the uplift of the Andes, although this separation had to have happened very recently, along with the rising of the Northern Andes. The Central Andes of Peru and Bolivia and the Northern Andes of Ecuador, Colombia, and Venezuela showed complex landscape transformations during the Miocene and Pliocene. However, the Central Andes experienced its most surface uplift in the late Miocene-Pliocene between 25-14 Ma ago, and the Northern Andes, with a different tectonic history, experienced rapid elevations between 2 and 5 Ma, reaching its modern elevations by around 2.7 Ma (Gregory-Wodzicki 2000).
We speculate that at the beginning of the Pliocene, Tonatia may have taken advantage of the newly formed Isthmus of Panama and all subsequent biogeographic processes (i.e. forests expansions) to colonize Central America and some Caribbean islands, including Jamaica Arita et al. 2014;Leigh et al. 2014). Finally, glaciations and tectonic activity in the Andes, during the Pliocene and Early Pleistocene, could have facilitated vicarious speciation within Tonatia saurophila (Chesser 2000). However, with the Andes uplift, a reduction in temperature and a shortage of resources at high altitudes possibly were impediments for migratory processes and gene flow to occur between western and eastern populations (Graham 1990). This kind of speciation process has been proposed for other bat species, such as Artibeus jamaicensis (Larsen et al. 2007), Carollia castanea (Vaca 2016), Uroderma bilobatum , and Gardnerycteris crenulatum and G. keenani (Hurtado and D'Elia 2018). In the case of Tonatia maresi, relationships between the Amazon and the Guiana Shield samples are not clear yet, since the mitochondrial (Cyt-b, Fig. 6) and nuclear (RAG2, Fig. 6) gene trees are not congruent. The relatively high level of genetic differentiation may indicate the existence of more than one species.
Despite numerous bat surveys throughout the West Indies in recent years, no new records of Tonatia saurophila have been confirmed. Tonatia specimens, recorded in Trinidad and Tobago (AMNH 180261-180264 and 182923) have been identified as T. maresi on the basis of their morphometric characteristics. Morphometric differences were found between subfossil samples of T. saurophila and specimens of T. bakeri and T. maresi (also considering the individuals collected in Trinidad and Tobago). For example, in T. saurophila the mandibular toothrow length and the coronoid height are smaller than in T. bakeri and T. maresi. These facts support the recognition of Tonatia saurophila (formerly classified as T. saurophila saurophila) as a full extinct species, with a distribution restricted to Jamaica.
In the Neotropics, more studies on the richness, distribution, and conservation status of the species are urgently needed. Diversity of better-known groups should be studied continuously and consistently, given increasing rates of habitat loss and global climate change. In bats, some of the recently described species were formerly recog-nized and treated as synonyms or subspecies until extensive mammal collections reviews showed that they were indeed different species (Solari and Martínez 2014). The conservation of biodiversity requires accurate and up to date studies of the taxonomy, distribution, and habitat preferences of species in order to effectively manage and protect them.

Taxonomic perspectives
With the elevation of T. bakeri and T. maresi to the species category, the genus Tonatia now includes three extant species, including T. bidens, and one extinct species.