Study Area and Geological Background

The Ioiô-Impossível (12°23′13.56″S, 41°32′57.48″W) is a large cave system located in the Iraquara karst region (Laureano and Cruz, 2002; Laureano, 1998; Cruz, 1998) (fig. 1), a subunit of the Precambrian carbonate rocks of the Ulna Group (Auler and Farrant, 1996) situated in the geologically well-known Irecê sedimentary basin (Souza et al., 1993). Paleontological explorations carried out in this cave have been concentrated on the main ducts, which are partially filled with water. These ducts can be accessed via the Impossível main entrance, which is situated at the bottom of a collapsed sinkhole that is oval in shape and approximately 100 m in diameter and 50 m in height (fig. 2). The cave ducts are partially filled by sediments that contain concentrations of fossil vertebrate bones in some locations, mostly medium-sized animals such as carnivores (Felidae, Procyonidae, Mustelidae) and xenarthrans (Dasypodidae, Myrmecophagidae).

FIGURE 1.

Map illustrating northeastern Brazil. Light gray shading indicates the extent of the Caatinga Biome, dark gray the Serra do Espinhaço mountains, and medium gray shading shows the extent of the Serras da Diamantina plateau. Numbers indicate locations of caves from which fossil bats have been collected: 1, Toca da Boa Vista; 2, Toca da Barriguda; 3, Gruta dos Brejões; 4, Toca dos Ossos; and 5, the new fossil site Ioiô-Impossível.

FIGURE 2.

Photograph showing one of the openings of the Ioiô-Impossível cave system as viewed from the top edge of the sinkhole. The main opening to the “Nothrotherium duct” is indicated by a dashed arrow.

Many of the medium-sized and larger specimens found in the bone concentrations are disarticulated, but nearly all are spatially arranged in such way that they seem to have maintained, at least to some extent, their anatomical relationships. This suggests that very little bone displacement occurred after the death of these animals, which in turn implies that the flux of water and sediments has been very slow and without sufficient energy to disperse larger bones along the cave ducts. A complete analysis of the taphonomy of these fossils will be the subject of another work, but the most likely hypothesis is that these nonvolant animals fell into the cave system via skylights, small holes, and fissures (all present in the area) that can function as pitfalls, and that they were preserved near where they died or their carcasses came to rest (Hubbe et al., 2011). In contrast, the isolated microvertebrate remains recovered in the cave system do not present evidence of the conditions described above, seeming instead to have been accumulated as a mix of disarticulated bones in certain areas in the cave. Some coincide with the sites of accumulation of medium-sized mammals, while others are dispersed along the ducts of the cave system. The micro vertebrate fauna thus preserved includes bats, rodents, and marsupials, as well as lizards and frogs. Regardless of how they entered the cave, many of the microvertebrates apparently were disarticulated and their bones transported after decomposition.

FIGURE 3.

Cross section of the “Nothrotherium duct” with schematics showing the airlift system used to collect fossils of microvertebrates including bats: (A) air-compression system placed at the cave entrance; (B) sieve system; and (C) at the Nothrotherium grid, a diver operating the pipe end with valve control for the air injection, which sucks water and sediments through the main pipe to the sieve system. Photographs below show images of the “Nothrotheriutn duct”: (D) dry area approximately 50 m from the entrance; (E) waterlogged portion approximately 150 m from the entrance; and (F) underwater view of the Nothrotherium grid.

Table 1.

Measurements of humeri of fossil bats from the Impossível-Ioiô cave system. Measurements follow standards described by Czaplewski and Cartelle (1998) and are in mm.

Abbreviations are as follows: MN, Museu Nacional (Rio de Janeiro) catalog number; ch, complete humerus; ps, proximal end of humerus with shaft; p, proximal end only; ds, distal end of humerus with shaft; d, distal end only; TL, total length; SD, shaft diameter; WP, width of proximal end of humurus; WD, width of distal end of humerus; ^*, measurement approximate due to damage.

Two areas with concentrations of bones of small mammals were selected for sampling (fig. 3), and the one that was most intensively sampled was the site where a skeleton of †Nothrotherium sp. (Xenarthra, Nothrotheriidae) was found. The latter specimen represents the only source of age reference for these fossils thus far. At Beta Analytic Inc. ( http://www.radiocarbon.com) significant amounts of collagen were successfully extracted from bone samples for AMS radiocarbon dating of the †Nothrotherium specimen, with a resulting date of 8700 ± 50 years BP (Perini et al., in prep.).

The Nothrotherium site covers an area of approximately 10 m² on the bottom of one of ducts that connects the main entrance of the Impossível cave, approximately 280 m from the entrance. A cross section of the so-called Nothrotherium duct is illustrated in figure 3 along with a sequence of photos that document the major passages of the duct. The Nothrotherium site is oval in shape, submerged in a water column of approximately 8 m (depth apparently not significantly varying between dry and rainy seasons) with lateral walls separated by about 5 m. In one of the walls, laminar carbonate concretions are present at different depths, some close to the bottom level, which is covered by silt, clay, and sand grains. In order to collect micro - vertebrate remains associated with these submerged sediments, some of the established quadrants were collected by using a pressurized air-suction device or airlift (Salles et al., 1999). Sediments containing fossil fragments were sucked through a pipe that ended at a sequence of sieves of different square mesh sizes varying from 0.2 to 1.2 cm. A sequence of schematic illustrations (fig. 3A–C) shows how the airlift system was used at the Impossível-Ioiô cave system. Small samples of fossils were also collected by hand with the use of surgical gloves and plastic zip bags. All remains representing the order Chiroptera were manually sorted from mixed packages of microvertebrate bones.

FIGURE 4.

The terminology used to describe humerus morphology (modified from Czaplewski et al., 2008) with proximal ends at the top and distal ends at the bottom of the figure. Letters are indicated on a posterior view (left) and anterior view (right) of a left humerus of Noctilio leporinus (MN 47199): a, head; b, greater tuberosity (= trochiter); c, lesser tuberosity (= trochin); d, pectoral ridge; e, medial ridge; f, epitrochlea, the region located medial to the trochlea and including both the medial process and spinous process; g, medial process; h, spinous process; i, radial fossa; j, lateral ridge of capitulum; k, medial ridge of capitulum; 1, trochlea.

Table 2.

Quaternary bat records from the Chapada Diamantina region compared with the extant records from the region and Quaternary records from elsewhere in the Neotropics.

Abbreviations and data sources as follows: CD, Chapada Diamantina Plateau, Bahia (Quaternary: Czaplewski and Cartelle, 1998; extant: Sbragia, 2012); LS, Lagoa Santa, Minas Gerais (Lund, 1840; Winge, 1892; Paula Couto, 1946); SM, Serra da Mesa, Goiás (Fracasso and Salles, 2005); RRV, Ribeira River Valley, São Paulo (Trajano and de Vivo, 1991; Ameghino, 1907); Ga/Sa, Garivaldino and Sangão sites, Rio Grande do Sul (Hadler et al., 2010); Caa, bat records from the Caatinga Biome that occurs in the Chapada Diamantina Plateau (Sá-Neto and Marinho-Filho, 2013); SA, South America, excepting Brazil (Morgan et al., 1988; Pardiñas and Tonni, 2000; Czaplewski et al., 2005); CA, Central America and Caribbean (Reynolds et al., 1953; Choate and Birney, 1968; Silva-Taboada, 1974, 1979; Olson et al., 1990; Arroyo-Cabrales and Polaco, 2008; Olson and Nieves-Rivera, 2010; Orihuela, 2012; Velazco et al., 2013); NA, North America (Mexico only) sites (Arroyo-Cabrales and Polaco, 2008); ^*, data from current work; ^**, first record for area.

Specimen Identification

The fossils collected were deposited at the mammal section of the Museu Nacional—Universidade Federal do Rio de Janeiro. Identifications were based on literature descriptions and comparisons with extant specimens in the mammal collections of the Museu Nacional, Universidade Federal do Rio de Janeiro (MN); American Museum of Natural History, New York (AMNH); and Instituto Nacional de Antropología e Historia, México City (INAH) (appendix 1).

FIGURE 5.

Posterior and anterior views of a distal portion of the right humerus of Furipterus horrens (MN 57429). Scale bar = 1mm.

Standard measurements for the fossil humeri collected are presented in table 1. All specimens whose measurements are provided here were considered adults based on epiphyseal fusion. The terminology used for humerus morphology was standardized based on a modified version of that proposed by Czaplewski et al. (2008) (fig. 4). A comparative study of humeral morphology of Noctilio is provided in an attempt to reveal patterns of variation within the genus that could be used to identify fossils without relying exclusively on size.

Family Furipteridae Gray, 1866

A distal end of a left humerus from Impossível-Ioiô is identified as Furipterus horrens (fig. 5), which represents the second fossil record for the species and the third for the family (Czaplewski and Cartelle, 1998; Morgan and Czaplewski, 1999). The first known fossil record of F. horrens (a left mandible and four humeri fragments) is also from the Quaternary of the Chapada Diamantina (Toca da Boa Vista; Czaplewski and Cartelle, 1998). The Chapada Diamantina records are consistent with the present distribution of the species, which ranges from Costa Rica to northeastern Brazil (Simmons, 2005; Peracchi et al., 2011), although it is considered a rare species and is difficult to collect with mist nets (Simmons and Voss, 1998; Bredt et al., 1999).

The humerus fragment from Impossível-Ioiô preserves only a small portion of the shaft with a damaged distal end. The dimensions of this specimen (table 1) fall well within the range of variation previously reported for Furipterus horrens (Czaplewski and Cartelle, 1998). Distal morphology of the humerus of F. horrens is distinct from that of other bats of comparable size, including thyropterids, natalids, emballonurids, and vespertilionids. The distal surface of the epitrochlea in F. horrens has a unique rounded and concave shape, with the anterolateral border extending over the anterior portion of the trochlea. The humerus in natalids such as Natalus stramineus resembles to a certain degree the general pattern found in Furipterus. However, Furipterus has a considerably larger and thicker trapezoidal spinous process, located directly adjacent to the posterior surface of the trochlea (fig. 6A), differing this way from natalids, but somewhat similar patterns are also observed among vespertilionids. This similarity in position with respect to the medial-lateral axis contrasts with significant differences in shape and direction of growth of the spinous process compared to that in vespertilionids, which display a markedly more posterior projection. The typical natalid medial extension of the posterior surface of the epitrochlea is entirely absent in F. horrens (b′ in fig. 6).

FIGURE 6.

Differences in distal humeral morphology between Furipterus horrens (left) and Natalus stramineus (right) as seen in anterior view with lateral toward the right side of the figure; see text for discussion: (a/a′) shape and development of the spinous process (larger and more trapezoidal in Furipterus); (b/b′) development of the epitrochlea (less expanded mediolaterally in Furipterus); and (c/c′) development of the posterior surface of the lateral ridge of the capitulum (posterior surface expanded over the lateral border of a well-developed radial fossa in Furipterus).

Extension of the posterior surface of the lateral ridge of the capitulum over the lateral border of a well-developed radial fossa (preserved in the fossil) is a character that unequivocally distinguishes Furipterus horrens from both natalids (c/c′ in fig. 6) and thyropterids, although this condition is also present in other bat lineages, such as emballonurids and noctilionids. These two lineages, however, have a very different overall pattern of distal morphology, mostly in shape and degree of development of the trochlea and the spinous process, which is relatively elongated and detached from the trochlea in both families.

FIGURE 7.

Anterior, posterior, lateral, and medial views of a right humerus of Chrotopterus auritus (MN 57430). Scale bar = 1mm.

FIGURE 8.

Differences in distal humeral morphology between Chrotopterus auritus (left) and Vampyrum spectrum (right) as seen in anterior view with lateral toward the right side of the figure; see text for discussion: (a/a′) shape and development of the spinous process (smaller and more trapezoidal in Chrotopterus); (b/b′) anteroposterior development (width) of the epitrochlea (narrower in chrotopterus); and (c/c′) degree of the development of the trochlea (narrower and less expanded over the epitrochlea in Chrotopterus).

Family Phyllostomidae Gray, 1825

The family Phyllostomidae is represented by humeral remains at Impossível-Ioiô by only one species, Chrotopterus auritus (fig. 7). The Pleistocene/Holocene records of this species are limited to reports from the Yucatán Peninsula of México (Arroyo-Cabrales and Alvarez, 1990), from Sierra de Perijá in Venezuela (Rincón, 1999), and from the following localities in Brazil: Lagoa Santa, Minas Gerais (Winge, 1892); Iporanga, São Paulo (Ameghino, 1907); Chapada Diamantina, Bahia (Cartelle and Abuhid, 1994; Czaplewski and Cartelle, 1998); Serra da Mesa, Goiás (Fracasso and Salles, 2005); and Brochier, Rio Grande do Sul (Hadler et al., 2010). These records include cranial and postcranial remains and are biogeographically consistent with the known extant distribution pattern of the species (Simmons, 2005).

The new record of Chrotopterus auritus from Chapada Diamantina consists of a single well-preserved right humerus that lacks only portions of the pectoral ridge and the tip of the spinous process (fig. 7). This specimen is the size and shape typical for the species. The shaft is arched with a concave curvature toward the middistal portion (viewed from the lateral side), and also exhibits torsion of this portion toward the medial side. This morphology also occurs in Vampyrum spectrum as well as in other phyllostomines such as Trachops cirrhosus and Tonatia saurophila. The proximal end of the humerus in Chrotopterus auritus and Vampyrum spectrum are very similar. In both species the humerus has a nearly spherical humeral head and a lesser tuberosity that is either similar in size or a bit larger than the greater tuberosity (Chrotopterus) or a lesser tuberosity that is distinctly larger than the greater tuberosity (Vampyrum) (fig. 8). Chrotopterus has a humerus with a distal end that is slightly less robust than in Vampyrum, with a less well-developed radial fossa and an overall smaller epitrochlea. The epitrochlea has an elongated shape, narrowing toward the medial end with a somewhat triangular distal surface (associated with the median process), a condition shared by Vampyrum and other phyllostomids. Nevertheless, it is anteroposteriorly wider in Vampyrum compared to the condition observed in Chrotopterus (b/b′ in fig. 8). The spinous process in Chrotopterus is small, located near the medial end of the epitrochlea, and has a trapezoidal shape quite different from the lower and wider process seen in Vampyrum (a/a′ in fig. 8). The degree of development of the trochlea also distinguishes these two species, being considerably wider and expanded over the epitrochlea surface in Vampyrum (c/c′ in fig. 8). The sister-taxon relationship found between Chrotopterus and Vampyrum in recent phylogenies (Wetterer et al., 2000; Baker et al., 2003; Dávalos et al., 2012) seems unquestionably supported by the patterns of variation in the humerus morphology seen in the family Phyllostomidae.

Family Mormoopidae Saussure, 1860

The family Mormoopidae is represented by a set of 27 humeral fragments identified as Mormoops cf. megalophylla (fig. 9), Pteronotus gymnonotus (fig. 11), and P. parnellii (fig. 12). The record of Mormoops is particularly noteworthy because these bats do not occur in the region today, instead having a distribution that is primarily limited to Central America and the Caribbean, with a very restricted range in northwestern South America (Simmons, 2005). The presence of the genus in the Quaternary of Chapada Diamantina was previously reported by Cartelle and coauthors (Cartelle and Abuhid, 1994; Czaplewski and Cartelle, 1998) based on craniodental remains as well as a number of humeri (although only measurements of a subset of 54 were provided). Those authors tentatively identified all of these fossils as Mormoops megalophylla, although they also suggested that, based on their large size, they might represent †M. magna, an extinct species exclusively known from humeri remains from Cuba and the Dominican Republic (Silva Taboada, 1974; Jiménez Vázquez et al., 2005; Velazco et al., 2013).

The new records of Mormoops recovered from the Impossível-Ioiô cave system consist of nine humeral fragments, four of which are complete and all of which we assign to Mormoops cf. megalophylla. These are unquestionably Mormoops, but biogeographic considerations and lack of comprehensive analyses of variation within the genus suggests that caution should be used in assigning them to M. megalophylla. The Chapada Diamantina fossils of Mormoops have a morphology very much like that of M. megalophylla, but they are slightly more robust than typical extant specimens, as pointed out by Czaplewski and Cartelle (1998). M. megalophylla differs from M. blainvillei in its significantly larger size (Silva Taboada, 1974; Rezsutek and Cameron, 1993). The morphology of the proximal humerus is very similar in these taxa, but distally the species differ in the development of the trochlea with respect to its posterior extension over the anterior surface of the spinous process. In M. megalophylla this surface is relatively well developed, reaching the base of the spinous process at its most posterior extent. In contrast, M. blainvillei has the trochlea clearly less developed and ending at a more anterior plane (a/a′ in fig. 10). As can also be seen in figure 10, the development and shape of the spinous process (b/b′) and of the posterolateral crest (c/c′) are distinct in these two species. The Mormoops samples from the Quaternary of Chapada Diamantina resemble M. megalophylla but have a slightly larger posterolateral crest (a structure on the posterior surface of the distal epiphysis extending over the shaft, laterally following the posterior ulna-trochlear facet) than that seen in our comparative sample of extant Mormoops. However, the significance of this trait, as well as the slight size differences observed, remain unclear. Accordingly, we treat the fossils of Mormoops from Chapada Diamantina as M. cf. megalophylla.

FIGURE 9.

Anterior, posterior, medial, and lateral views of a left humerus of Mormoops cf. megalophylla (MN 57402). Scale bar = 1mm.

FIGURE 10.

Differences in distal humeral morphology between Mormoops megalophylla (left) and M. blainvillei (right) as seen in posterior view with lateral toward the left side of the figure; see text for discussion: (a/a′) development and posterior extension of the trochlea (trochlea more strongly developed and extending further posteriorly in M. megalophylla); (b/b′) differences in shape and size of the spinous process (more bulbous distally and with strongly developed crestlike form posteriorly in M. megalophylla); and (c/c′) degrees of the development and shape of the posterolateral crest (more strongly developed in M. megalophylla).

Pteronotus specimens are relatively abundant in the Chapada Diamantina collections, with 18 humeral fragments representing two species, P. parnellii and P. gymnonotus. Of the seven specimens representing P. gymnonotus, two are very well preserved, with one nearly intact and the other only slightly damaged proximally. Similarly, of the 11 humeri of P. parnellii in the collection, three are complete and three others have the shaft with both ends preserved but with some damage to various surfaces and processes. The morphology of the humerus in Pteronotus is quite different from that seen in Mormoops. The humerus of Pteronotus seemingly retains more of the basal noctilionoid morphology, mostly notably in the distal end. The main diagnostic features that distinguish the genus Pteronotus from other noctilionoids include the prominent anterior position of the humeral head relative to the greater tuberosity (the latter structure being more posteriorly positioned in Pteronotus than in other noctilionoids), and the marked posterolateral displacement of the spinal process, which is partially dissociated from the medial process.

FIGURE 11.

Anterior, posterior, lateral, and medial views of a right humerus of Pteronotus gymnonotus (MN 57413). Scale bar = 1mm.

FIGURE 12.

Anterior, posterior, medial, and lateral views of a left humerus of Pteronotus parnellii (MN 57423). Scale bar = 1mm.

Within Pteronotus a number of features of both the proximal and distal ends of the humerus allow unequivocal species identifıcations, with the distal end more drastically modified. Differences among species include variation in shape, angle, and degree of development of the lesser tuberosity at the proximal end, and at the distal end variation in trochlear shape, position, size, and in the relationship between the trochlea and the spinous process. The humeri of P. parnellii and P. gymnonotus stand clearly as the two largest in the genus, with P. parnellii the larger. Similar size distinctions are also observed among the fossils, although they are overall somewhat smaller than the extant specimens examined (a topic to be addressed further in future contributions). The morphology of the proximal end of the humerus of P. parnellii differs from that of P. gymnonotus in a number of details. The most conspicuous difference involves the development of the midbasal portion of the anterior surface of the greater tuberosity, which is wider in P. gymnonotus, extending over the anterior extremity of the supraglenoid fossa (a somewhat similar condition is also observed in P. macleayii and P. quadridens, and to a certain extent also in P. personatus and P. davyi). In contrast, in P. parnellii the basal portion of the greater tuberosity is relatively narrow and the anterior extremity of the supraglenoid fossa is more open, a condition shared by other noctilionoids. The most obvious distinction between the distal ends of the humeri of Pteronotus parnellii and P. gymnonotus concerns the position and angle of development of the spinous process. Pteronotus parnellii exhibits a unique condition in which the spinous process is located near a midline between the anterior and posterior plane of the shaft and projects distally, whereas in P. gymnonotus the spinous process is placed much more posteriorly from that midline, projecting more posteriorly and with its tip extending over the limits of the posterior surface of the shaft (a/a′ in fıg. 13). The latter condition is shared by all species of Pteronotus (including the extinct species †P. pristinus) except for P. parnellii. Pteronotus species are also distinguished by different shapes of the distal trochlea when seen in lateral view, with this flange in P. parnellii displaying a circular outline as opposed to a more rectangular outline in P. gymnonotus (b/b′ in fig. 13). The fossils of the two species from Chapada Diamantina have a distal humeral morphology that conforms well to the known patterns of P. parnellii and P. gymnonotus as described above.

Pteronotus gymnonotus was documented as an extant taxon from Chapada Diamantina (Sbragia, 2012), but fossils of this species have not been previously reported from the region (Cartelle and Abuhid, 1994; Czaplewski and Cartelle, 1998). This species is known from the fossil record of South America only from a single locality, Cueva Toromo, Venezuela (Rincón, 1999). Material identified as Pteronotus aff. gymnonotus by Cartelle and Abuhid (1994) was reexamined by Czaplewski and Cartelle (1998), who reidentifıed it as P. parnellii, a species not known from the extant fauna of the Chapada Diamantina.

Family Noctilionidae Gray, 1821

The evolutionary history of the bulldog bats (genus Noctilio) extends at least to the late Tertiary of South America. A series of upper and lower molars nearly indistinguishable from those of modern Noctilio were reported by Czaplewski (1997) and Czaplewski et al. (2003) from the Middle Miocene (11.8–12.5 mya) of La Venta, Colombia. Based on size and morphology, these authors identified the La Venta fossils as Noctilio albiventris. Another fragmentary Noctilio fossil from the Miocene—a lower m1 from Rio Acre, 30 km W of Iñapari, Department of Madre de Dios, Peru—dates to the Huayquerian South American Land Mammal Age (9.5 mya). Because this specimen does not appear to fit within the range of variation known for N. albiventris, Czaplewski (1996) named it as a new species, †N. lacrimaelunaris. Merino et al. (2007) described an upper incisor referable to Noctilio sp. from deposits dated approximately 0.4 mya (Middle Pleistocene) from Las Grutas, Necochea, Province of Buenos Aires, Argentina. This discovery extended the known range for the genus at least 6° S latitude from the present distribution. However, the highly incomplete nature of the specimen precluded identification to species level.

Fossils assigned to the larger species of Noctilio, N. leporinus, have been found in Late Pleistocene and Holocene cave deposits of Barbuda, Cuba, and Puerto Rico (Reynolds et al., 1953; Choate and Birney, 1968; Silva Taboada, 1979; Pregill et al., 1994). However, no such fossils have been previously reported from mainland sites despite the fact that the species today ranges from Sinaloa, Mexico, south through Central and South America as far as northern Argentina and Paraguay (Simmons, 2005; Gardner, 2008a). New remains of Noctilio recovered from Chapada Diamantina stand as the first confirmed Quaternary record of the species N. leporinus (fig. 14) for South America. The material consists of two well-preserved humeral fragments and a few craniodental remains (the latter to be described in a subsequent publication). Noctilio leporinus has been reported from the extant fauna of the Chapada Diamantina (Oliveira and Pessôa, 2005), although no voucher specimen is known.

The morphology of the humerus of Noctilio is quite distinctive from that seen in other noctilionoids. The shaft is nearly straight and both proximal and distal articular surfaces lie in line with the shaft (fig. 14). Proximally, the humerus has an elliptical head, a feature not observed in any other noctilionoid except Mormoops. Distally, the humerus exhibits a unique medially extended epitrochlea, with the spinous process placed at its extremity.

The most obvious difference in humeral morphology between Noctilio leporinus and N. albiventris concerns size, with the humerus of N. leporinus much larger and more robust than that of N. albiventris. The cross-sectional shape of the shaft near the distal epiphysis also varies between these taxa: in N. albiventris the shaft tends to be more cylindrical in cross section, as opposed to a more rectangular cross-sectional shape in N. leporinus (c/c′ in fig. 15). Proximally, N. albiventris has a humeral head with a constriction toward the anterior end (associated with a prominent laterally curved proximal edge), which contrasts with a more smoothly rounded condition observed in N. leporinus. Distally, among some specimens of N. albiventris, the proximal border of the radial fossa has a prominent curved form (apparently correlated with the shaft having a cylindrical cross section near the epiphysis). This contrasts with the condition of N. leporinus in which the proximal border of the radial fossa is not as well developed over the anterior surface of the shaft (b′ in fig. 15). The epitrochlea is proportionally somewhat more mediolaterally expanded in N. leporinus than in N. albiventris. Noctilio leporinus has a distinct medial process that projects slightly over the anterior border of the epitrochlea, whereas this process is not developed in N. albiventris (e/e′ in fig. 15). The spinous process in N. leporinus has a relatively conspicuous tip that is recurved laterally and somewhat posteriorly oriented, whereas in N. albiventris the spinous process is relatively poorly developed, not recurved, and is partially fused with the lateral end of the epitrochlea (f/f′ in fig. 15). In N. leporinus the anteroproximal rim of the lateral ridge of the capitulum is narrow and forms a pointed tip that is curved toward the posterior side, whereas in N. albiventris this process is more rounded (a/a′ in fig. 15). The Noctilio humerus fossils from Chapada Diamantina are clearly referable to N. leporinus based on both size and morphology.

FIGURE 13.

Differences in distal humerus morphology between Pteronotus parnellii (left) and P. gymnonotus (right) as seen in medial view with posterior toward the left of the figure; see text for discussion: (a/a′) differences in degree of development and orientation of the epitrochlea and spinous process (epitrochlea less developed and spinous process shorter and directed more distally in P. parnellii); and (b/b′) differences in the outline of the trochlea (more circular in medial view in P. parnellii).

The genus Noctilio is in need of review and patterns of morphological variation within species need to be reevaluated with larger sample sizes from more localities to better control for intraspecifıc variation. Phylogeographic studies of Lewis-Oritt et al. (2001) and Pavan et al. (2013) have suggested that N. albiventris as currently recognized is composed by more than one taxon (see below). However, there is no evidence that N. leporinus represents of more than one species despite its broad geographic distribution (Lewis-Oritt et al., 2001; Pavan et al., 2013).