Species of Thyroptera are insectivorous foliage-roosting bats that inhabit lowland moist forests (including gallery formations in savanna landscapes) from Mexico to southeastern Brazil. Although four species are currently recognized, only one or two species were previously known to occur at most localities. Recent inventory work in northeastern Peru has documented the local cooccurrence of four species of Thyroptera, one of which is here described as new. The new species (T. wynneae), which also occurs in Brazil, can easily be recognized by a combination of diagnostic morphological traits. The latter include small size, tricolored ventral pelage, long and woolly hairs between the shoulders, a uropatagium with the proximal half densely covered by long hairs, wing tips sparsely covered by long hairs, a calcar with two lappets and five tiny skin projections between the foot disk and the proximal lappet, a rostrum considerably shorter than the braincase, third lower incisors that are subequal in height to the first and second lower incisors, and third lower incisors with two well-developed accessory cusps. We illustrate the crania of all five known species of Thyroptera and provide a key based on craniodental and external characters. Unexpectedly high local diversity of these elusive bats poses a challenge for future inventory research and raises interesting questions about ecological-niche partitioning in Neotropical bat communities and the evolutionary history of thyropterids.
INTRODUCTION
Disk-winged bats (Thyropteridae: Thyroptera) comprise a group of foliage-roosting, insectivorous species that occur in lowland moist forests from southern Mexico to southeastern Brazil (Wilson, 2008). Thyropterids are easily distinguished from all other Neotropical bats by the presence of a circular adhesive disk on the sole of the foot, an oval or circular disk attached by a short pedicle to the base of the thumb, and by fusion of the soft tissues of digits III and IV of the hind foot (Simmons, 1998). In the possession of adhesive disks thyropterids resemble Old World sucker-footed bats of the family Myzopodidae, but anatomical and evolutionary analyses concur that the adhesive disks in these two groups evolved convergently (Schliemann, 1970, 1971; Simmons, 1998). Thyropterids are additionally characterized by their small size (forearm length = 31–41 mm); an elongate, slender muzzle; circular and well-separated nares; and funnel-shaped ears. The tail extends one to a few millimeters beyond the distal margin of the long interfemoral membrane. The toes have only two phalanges and, as noted above, the third and fourth toes (including claws) are fused. Wing digit II is reduced to an incomplete metacarpal, and the third phalanx of digit III is ossified. The manubrium of the sternum is small and keeled, and its width is about one-half the combined length of the manubrium and mesosternum. The mesosternum is broad, flat, and lacks a keel. The first and second thoracic vertebrae are fused. The trochiter of the humerus is larger than the tro chin and articulates with the scapula. The skull has a rounded braincase elevated above the slender rostrum, and has complete premaxillae, but lacks postorbital processes (Wilson, 2008).
Four species of living thyropterids are currently recognized. Thyroptera discifera (Lichtenstein and Peters, 1854) is known from Nicaragua, northeastern Costa Rica, and central Panama in Central America, and from northern South America (Colombia, northern Venezuela, and the Guianas) to northwestern Ecuador, eastern Peru, northern Bolivia, and southeastern Brazil (Gregorin et al., 2006: fig. 5). Thyroptera tricolor Spix, 1823, is patchily distributed in Central America from southern Mexico (Veracruz) to Panama (it has not been recorded in El Salvador or Nicaragua); in South America it has a similar distribution to that of T. discifera, but it seems to be more widely distributed in southeastern Brazil (Wilson, 2008: map 222). Thyroptera lavali Pine, 1993, is known from eastern Ecuador and Peru to northeastern Venezuela and the central Brazilian Amazon (Solari et al., 2004). Thyroptera devivoi Gregorin et al., 2006, is only known from southwestern Guyana and northeastern Brazil, but the considerable distance (ca. 2000 km) spanning documented collection localities (Gregorin et al., 2006: fig. 5) suggests that the species is widely distributed.
Recent inventory work in the Peruvian Amazon and southeastern Brazil has resulted in the collection of three specimens of Thyroptera that do not fit the diagnosis of any previously described species in the genus. In this report we describe these specimens as a new species, and we discuss the remarkably high local diversity of disc-winged bats in northeastern Peru, where four species are now known to occur between the Yavarí and Ucayali rivers.
METHODS
The external and osteological characters we examined include (but are not limited to) those defined by Gregorin et al. (2006) and Solari et al. (2004). Premolar nomenclature follows O'Leary et al. (2013) and Dávalos et al. (in press): the three upper premolars are P1, P4, and P5, and the three lower premolars are p1, p4, and p5. We examined 80 specimens of adult Thyroptera (36 males and 44 females; appendix 1) for comparative purposes. All measurements reported herein are from adult individuals with closed epiphyses. The first four measurements listed below were taken from skin tags or other records made by the original collector; all other measurements were taken by us using digital calipers and were recorded to the nearest 0.01 mm. Linear measurements are given in millimeters (mm), and weights are reported in grams (g). Descriptive statistics (mean and observed range) were calculated for all samples. Measurements are defined as follows:
Total length (TL): Distance from the tip of the snout to the tip of the last caudal vertebra.
Length of tail (LT): Measured from the point of dorsal flexure of the tail with the sacrum to the tip of the last caudal vertebra.
Hind foot length (HF): Measured from the anterior edge of the base of the calcar to the tip of the claw of the longest toe.
Ear length (Ear): Measured from the ear notch to the fleshy tip of the pinna.
Free tail length (FTL): Length of the free portion of the tail, measured from the tip of the last caudal vertebra to the distal border of the interfemoral membrane (uropatagium).
Forearm length (FA): Distance from the elbow (tip of the olecranon process) to the wrist (including the carpals). This measurement is made with the wing at least partially folded.
Greatest length of skull (GLS): Distance from the posteriormost point on the occiput to the anteriormost point on the premaxilla (excluding the incisors).
Condyloincisive length (CIL): Distance between the anteriormost point on the upper incisor and a line connecting the posteriormost margins of the occipital condyles.
Braincase breadth (BB): Greatest breadth of the globular part of the braincase, excluding mastoid and paraoccipital processes.
Rostral length (ROL): Distance from the alveolar process of the premaxilla above the first upper incisor to the ipsilateral postorbital constriction.
Zygomatic breadth (ZB): Greatest breadth across the zygomatic arches.
Postorbital breadth (PB): Least breadth at the postorbital constriction.
Maxillary toothrow length (MTRL): Distance from the anteriormost surface of the upper canine to the posteriormost surface of the crown of M3.
Width at M3 (M3—M3): Greatest width of palate across labial margins of the M3s.
Length of mandible (LMA): Distance from the anteriormost point on the first lower incisor to the posteriormost point on the ipsilateral coronoid process.
Mandibular toothrow length (MANDL): Distance from the anteriormost surface of the lower canine to the posteriormost surface of m3.
Eight craniodental measurements (GLS, CIL, BB, PB, MTRL, M3—M3, LMA, and MANDL) and one external measurement (FA) of male specimens were log-transformed for multivariate statistical analysis, and we used principal components analysis to assess species differences in these data. A few measurements (TL, LT, HF, Ear, FTL, ROL, and ZB) were omitted because they were unavailable from more than a few specimens. Principal components were extracted from the variance-covariance matrix using the program PAST (Hammer et al., 2001). A dichotomous identification key for all the species of Thyroptera was constructed based on morphological traits identified during the course of the study.
The specimens discussed this report belong to the following collections:
Glossary
AMNH
American Museum of Natural History, New York.
CEBIOMAS
Centro de Ecología y Biodiversidad, Lima, Peru.
CMUFLA
Coleção de Mamíferos da Universidade Federal de Lavras, Lavras, Minas Gerais, Brazil.
FMNH
Field Museum of Natural History, Chicago, Illinois.
MUSM
Museo de Historia Natural de la Universidad Nacional Mayor de San Marcos, Lima, Peru.
MZUSP
Museo de Zoologia da Universidade de São Paulo, São Paulo, Brazil.
ROM
Royal Ontario Museum, Toronto, Ontario, Canada.
SYSTEMATICS
Family Thyropteridae Miller, 1907
Genus Thyroptera Spix, 1823
Thyroptera wynneae
, new species
Patricia's Disk-winged Bat
Holotype: The holotype (CEBIOMAS 237; figs. 1, 2, 4E, 5E, 6E), an adult male specimen preserved in alcohol with the skull removed and cleaned, was collected by Robert S. Voss and subsequently prepared by Paúl M. Velazco (original number: PMV 2469) on 29 January 2012 at the Centro de Investigaciones Jenaro Herrera (45°05′S, 73°40′W), a forestry research station located about 2.5 km inland from the right bank of Río Ucayali in the Peruvian department of Loreto (fig. 3, locality 27). Frozen tissues are deposited at the American Museum of Natural History (AMNH 278486).
Paratypes: Two additional specimens (CMUFLA 694 and CMUFLA 1160), both adult males preserved in alcohol with the skulls removed and cleaned, were collected at a site called Campolina (ca. 19°42′S, 42°30′W; fig. 3), about 25 km E of Marliéria in the Parque Estadual do Rio Doce, Minas Gerais, Brazil. Campolina is shown on a map of bat sampling localities by Tavares et al. (2007: fig. 1), who also worked at this locality. One paratype (CMUFLA 1160) was collected by Renato Gregorin on 11 January 2013 and the other (CMUFLA 694) was collected by Karina Lob˜o Vasconcellos and Bruno Bret Gil (original field number: PERD 310) on 11 March 2013. CMUFLA 694 has a damaged skull (the braincase is broken in the parietal region and the left auditory bulla is missing), but the other specimen is in good condition.
Distribution: Thyroptera wynneae is currently known only from the type locality in northeastern Peru and from the Parque Estadual do Rio Doce in southeastern Brazil (fig. 3; appendix 1).
Diagnosis and Description: Thyroptera wynneae is a small thyropterid with a short, slender rostrum and a globular braincase (figs. 1–2, 4–6). The dorsal fur is light brown with a slightly grayish overtone. Dorsal hairs are unicolored for most of their length, but a tiny basal portion of each hair is lighter than the rest of the shaft (fig. 1 A, 2A). Ventral hairs are tricolored with the basal third of each hair whitish in color, the middle segment light brown, and the tip darker. The fur is long and woolly, approximately 11 mm long between the shoulders and 7 mm on the chest. The proximal portion of the forearm (roughly 12 mm of the shaft just distal to the elbow) is densely furred with long hairs. Most of the dorsal surface of the pinna is densely covered with long hairs, but the distal portion is naked (fig. 1A).
The wing membranes of Thyroptera wynneae are dark brown. The plagiopatagium and uropatagium (fig. 1C, 2A) are marked by transverse lines of elastine fibers punctuated by regularly spaced wartlike somatosensory domes (sensu Zook, 2007). The dorsal surface of the uropatagium is densely covered with long hairs proximally, and sparsely covered with long hairs distally. The tip of the wing is also sparsely covered with long hairs, and the uropatagium, plagiopatagium, and dactilopatagium have hairs distributed sparsely along their edges. The adhesive disks of the thumb and foot are oblong (fig. 1B, 1C). The thumb disks vary somewhat in size among individuals, ranging from 2.7–4.0 mm in length and 1.8–3.1 mm in width (length being measured along the longest axis of the disk). Similarly, foot disks range from 1.6–2.5 mm in length and 1.4–2.2 mm in width. Some of this variation, however, might be due to differences in preservation. The calcar is long (8.5–12.0 mm) and has two lappets (small cartilaginous flanges) that project posterolaterally from the shaft of the calcar. In addition to these skeletal processes, five tiny skin projections are also present between the foot disk and the proximal lappet (fig. 1C).
The skull of Thyroptera wynneae has a smoothly rounded braincase that lacks a sagittal crest (figs. 4E, 6E). The angle between the rostrum and braincase is approximately 130° (fig. 6E) and the rostrum is considerably shorter than the braincase (table 1). There is a large foramen ovale which is 0.5 mm in length. A transverse ridge is present between the right and left hamular processes of the pterygoid. This ridge is intersected anteriorly by a midventral bladelike crest, dividing the mesopterygoid fossa into two separate concavities (fig. 5E; Gregorin et al, 2006, fig. 2A).
Like all other thyropterids, Thyroptera wynneae has a dental formula of I2/3, C1/1, P3/3. M3/3 = 38 teeth. The upper incisors are bicuspidate. The second upper incisor (I2) has a mesial cusp that is larger than the distal cusp. Both cusps on 12 are perpendicularly arranged relative to the long axis of the toothrow. The first and second upper incisors are in contact with one other and are separated from the canine by a small diastema. P1 is rounded in occlusal view. Anteroposterior length of P1 is less than that of P4, and crown height of P1 is slightly less than that of P4. P1 and P4 both have narrow lingual and labial cingula. P4 and P5 are subequal in anteroposterior length but crown height of P4 is slightly less than that of P5. P5 has a narrow labial cingulum and a very large lingual cingulum that forms a distinct lobe, the latter extending as far lingually as the protocone of M1. A well-developed anterolingual cusp is present on the labial cingulum of P5. M1 and M2 are subequal in size. A parastyle is present and the protocone is well developed on all upper molars including M3. A metastyle is present and the paracone and metacone are subequal in size on M1 and M2. In contrast, a metastyle is absent and the paracone is taller than the metacone on M3. Lingual cingula are present on all three upper molars. The preparacrista is shorter than the postparacrista on M1 and M2, but the preparacrista is longer than the postparacrista on M3. The premetacrista is shorter than the postmetacrista on M1 and M2, but the postmetacrista is absent on M3.
The first and second lower incisors (i1 and i2) are tricuspidate. The third lower incisor (i3) has two small accessory cusps on each side (mesostyles and disto styles). The i3 is subequal in height to i1 and i2, and is 1.5 times the buccolingual width of i1 and i2. The p1 is triangular in occlusal view. The p4 and p5 are buccally flattened and ovate in occlusal view. All three lower premolars are subequal in height. The lower molars are similar in shape. The m3 is slightly shorter in anteroposterior length than m1 and m2. The protoconids are taller than the hypoconids on all three lower molars. All of the cristids on all lower molars are clearly notched.
Comparisons: External and craniodental measurements for Thyroptera wynneae and its congeners are provided in tables 1 and 2. T. wynneae can be easily distinguished from T. devi- voi, T. lavali, and T. tricolor by its shorter forearm and smaller greatest length of skull (tables 1–2). T. wynneae and T. discifera overlap somewhat in size (tables 1–2) but can be unambiguously distinguished based on pelage characteristics.
Externally Thyroptera wynneae can be distinguished from all the other species in the genus by its tricolored ventral pelage, which contrasts with the unicolored ventral pelage of T. discifera and T. tricolor, and with the bicolored ventral pelage of T. devivoi and T. lavali. The overall appearance of the ventral pelage is light brown in T. wynneae, whereas it is grayish brown in T. devivoi, yellowish brown in T. discifera, medium brown in T. lavali, and white or whitish in T. tricolor. Dorsally the hairs between the shoulders are long (11 mm) and woolly in T. wynneae (fig. 2A) while they are long (8.0–12.0 mm) and not woolly in T. discifera and T. tricolor, and short (3.5–6.0 mm) and woolly in T. devivoi and T. lavali. The proximal portion of the forearm is well furred in T. wynneae and T. discifera, whereas it is only sparsely haired in T. devivoi, T. lavali, and T. tricolor.
The dorsal surface of the uropatagium is densely covered with long hairs proximally, and sparsely covered distally, in T. wynneae and T. discifera. In contrast, the entire dorsal surface of the uropatagium is sparsely covered with short hairs in T. lavali and T. tricolor. The tip of the wing is covered sparsely by long hairs (> 1.8 mm) in T. wynneae and T. tricolor, whereas these hairs are short (< 1.2 mm) in T. discifera and T. lavali (we were not able to examine T. devivoi for this characteristic). The adhesive disks of the thumb are oblong in T. wynneae (fig. 1B), T. devivoi, and T. lavali, whereas they are circular in T. discifera and T. tricolor. The calcar has two lappets that project posterolaterally from the shaft of the calcar in T. wynneae (fig. 1C) and T. tricolor. In contrast, they are absent or only one faintly developed lappet is present in T. devivoi, and only a single lappet is present in T. discifera and T. lavali. Five tiny skin projections are present between the foot disk and the proximal lappet in T. wynneae (fig. 1C) and T. tricolor, whereas these structures are entirely absent in T. devivoi, T. discifera, and T. lavali.
Cranially, Thyroptera wynneae can be distinguished from all other species of Thyroptera by having a rostrum that is considerably shorter than the braincase. The rostrum is of equal length or longer than the braincase in T. devivoi, T. discifera, T. lavali, and T. tricolor. The foramen ovale is large (0.50–0.65 mm) in T. wynneae, T. devivoi, T. lavali, and T. tricolor, whereas it is smaller (< 0.45 mm) in T. discifera. The transverse ridge present between the hamular processes of the pterygoid is anteriorly intersected by a midventral bladelike crest, dividing the mesopterygoid fossa in two separate concavities in T. wynneae, T. devivoi, T. lavali, and T. tricolor. The midventral bladelike crest does not reach the transverse ridge in T. discifera (fig. 5).
Dental characters can also be used to distinguish species of Thyroptera. The mesial cusp is larger than the distal cusp on I2 in T. wynneae, T. discifera, and T. tricolor, whereas these cusps are subequal in T. devivoi and T. lavali. Both cusps on I2 are perpendicularly arranged relative to the long axis of the toothrow in T. wynneae, T. devivoi, and T. lavali, whereas they are obliquely arranged in T. discifera and T. tricolor. The first and second upper incisors are in contact with each other in T. wynneae, T. devivoi, T. lavali, and T. tricolor but are separated by a small diastema in T. discifera. P1 is rounded in occlusal view in T. wynneae, T. devivoi, T. lavali, and T. tricolor, whereas it is elongated in T. discifera. All three lower incisors are subequal in height in T. wynneae, whereas the i3 is taller than il and i2 in T. devivoi, T. discifera, T. lavali, and T. tricolor. The third lower incisor has two small accessory cusps on each side (mesostyle and distostyle) in T. wynneae and T. discifera, but these accessory cusps are only faintly developed in T. devivoi and T. tricolor and are either faintly developed or entirely absent in T. lavali. The i3 is 1.5 times the buccolingual width of il and i2 in T. wynneae, T. discifera, and T. tricolor, whereas the i3 is twice the buccolingual width of il and i2 in T. devivoi and T. lavali. The p4 and p5 are buccally flattened and ovate in occlusal view in T. wynneae, T. discifera, whereas they are flattened and squarish in T. tricolor, and wide and squarish in T. devivoi and T. lavali.
Multivariate Analysis: Our multivariate statistical analysis included only male specimens because females are unknown for Thyroptera wynneae. We conducted a principal component analysis of measurements of 21 specimens of Thyroptera, including 3 each of T. wynneae, T. devivoi, and T. discifera, 2 of T. lavali, and 10 of T. tricolor (appendix 1). The first three principal components accounted for 88.5% of the total variance in the log-transformed measurements of this material (table 3). PC 1 accounted for the highest percentage (70.1%), and this vector has uniformly positive loadings, suggesting that it is a size factor (with a notably high loading for forearm length). Correspondingly, specimens of small-bodied species (T. wynneae and T. discifera) have lower scores along this axis, large-bodied T. lavali has high scores, and medium-sized species (T. tricolor and T. devivoi) have intermediate values (fig. 7). Although T. wynneae and T. discifera have overlapping scores on PC2 (fig. 7A), these species are resolved as distinct clusters on PC3 (fig. 7B). Factor loadings on PC3 suggest that this separation is primarily attributable to taxonomie differences in forearm length and width at M3.
Natural History: The Centro de Investigaciones Jenaro Herrera is a forestry research station located about 2.5 km inland from the right bank of the Río Ucayali in the Peruvian department of Loreto. The station is on a terrace above the river floodplain and is not subject to annual flooding by river water. The local vegetation consists of primary and secondary lowland rainforest (“high terrace forest” sensu López-Parodi and Freitas, 1990) on well-drained soils, but a distinctive riparian formation (“creek forest” sensu López-Parodi and Freitas, 1990; “bosque de quebrada” sensu Kahn and Mejia, 1990) occurs along streams, and palm swamps dominated by Mauritia flexuosa (locally known as “aguajales”) occupy depressions with permanently waterlogged soils. Small plantations (of yuca, pineapple, etc.) are also scattered throughout the area. The holotype of Thyroptera wynneae was found roosting with another individual (which escaped capture) in the late morning (about 11:00 hrs). Both bats were concealed within a dead Cecropia leaf that was hanging from understory vegetation by its petiole about 2 m above the ground in secondary forest next to a trail about 50 m from the main buildings of the biological station (fig. 8).
The Parque Estadual do Rio Doce protects some 36,000 hectares of semideciduous forest, most of which is thought to be secondary growth (Lopes et al., 2002). The park includes hills and low mountains (Stallings, 1988; Lopes et al., 2002), but Campolina is located in primary forest with tall (20–30 m) trees and a relatively open understory on level terrain. Both paratypes were captured in ground-level mist nets.
Etymology: We are pleased to name this species in honor of Patricia J. Wynne (fig. 9), longtime artist-in-residence in the AMNH Department of Mammalogy, who has drawn mammals for scientific and educational publications for over 40 years. Exceptional for clarity, elegant rendering, and meticulous attention to relevant detail, Patricia's work is immediately recognizable and inimitable (e.g., figs. 3–6).
TABLE 1.
Measurements (mm) and Weights (g) of the Type Series of Thyroptera wynneae.
![t01_01.gif](ContentImages/Journals/novi/2014/3795/3795.1/graphic/WebImages/t01_01.gif)
TABLE 2.
Measurements (mm) and Weights (g) of Four Species of Thyroptera.
![t02_01.gif](ContentImages/Journals/novi/2014/3795/3795.1/graphic/WebImages/t02_01.gif)
FIG. 1.
Right oblique view of the head (A), right wrist with adhesive disk (B), and left hind limb (C) of the holotype of Thyroptera wynneae (CEBIOMAS 237). Note that due to the angle from which image (B) was taken, the adhesive disk on the thumb looks circular, but it is in fact oblong vertically. Photos: Burton Lim.
![f01_01.jpg](ContentImages/Journals/novi/2014/3795/3795.1/graphic/WebImages/f01_01.jpg)
FIG. 2.
Dorsal (A) and ventral (B) views of the holotype of Thyroptera wynneae (CEBIOMAS 237). Photos: Burton Lim.
![f02_01.jpg](ContentImages/Journals/novi/2014/3795/3795.1/graphic/WebImages/f02_01.jpg)
FIG. 3.
The Yavarí-Ucayali interfluvial region (boundaries highlighted in grey) showing the type locality of Thyroptera wynneae (arrow) and adjacent localities where other thyropterid species have been collected. The inset shows where the paratypes were collected in southeastern Brazil. Numbers are keyed to entries in the gazetteer (appendix 2).
![f03_01.jpg](ContentImages/Journals/novi/2014/3795/3795.1/graphic/WebImages/f03_01.jpg)
FIG. 4.
Dorsal views of the skulls of A, Thyroptera devivoi (ROM 35588♂); B, Thyroptera discifera (AMNH 16686♀); C, Thyroptera lavali (ROM 104026♀); D, Thyroptera tricolor (AMNH 273160♀); E, Thyroptera wynneae (CEBIOMAS 237♂, holotype). Scale bar = 5 mm.
![f04_01.jpg](ContentImages/Journals/novi/2014/3795/3795.1/graphic/WebImages/f04_01.jpg)
FIG. 5.
Ventral views of the skulls of A, Thyroptera devivoi (ROM 35588♂); B, Thyroptera discifera (AMNH 16686♀); C, Thyroptera lavali (ROM 104026?); D, Thyroptera tricolor (AMNH 273160♀); E, Thyroptera wynneae (CEBIOMAS 237♂, holotype). Scale bar = 5 mm.
![f05_01.jpg](ContentImages/Journals/novi/2014/3795/3795.1/graphic/WebImages/f05_01.jpg)
FIG. 6.
Lateral views of the skulls and lower jaw of A, Thyroptera devivoi (ROM 35588♂); B, Thyroptera discifera (AMNH 16686♀); C, Thyroptera lavali (ROM 104026♀); D, Thyroptera tricolor (AMNH 273160♀); E, Thyroptera wynneae (CEBIOMAS 237♂, holotype). Scale bar = 5 mm.
![f06_01.jpg](ContentImages/Journals/novi/2014/3795/3795.1/graphic/WebImages/f06_01.jpg)
TABLE 3.
Factor Loadings for the First Three Factors Extracted from a Principal Component (PC) Analysis of Nine Variables for All Species of Thyroptera.
![t03_01.gif](ContentImages/Journals/novi/2014/3795/3795.1/graphic/WebImages/t03_01.gif)
DISCUSSION
The discovery of a new species of Thyroptera at Jenaro Herrera was unexpected because two congeners (T. discifera and T. lavali; Solari et al., 2004) were already known to occur there. Furthermore this locality had previously been sampled intensively for bats by Ascorra et al. (1993), who recorded almost 2500 captures in the course of their multiyear inventory study. Additionally, Jenaro Herrera (fig. 3: locality 27) lies within the Yavarí-Ucayali interfluve, where a fourth species (T. tricolor) has been collected at Nuevo San Juan (fig. 3: locality 22). Although Jenaro Herrera and Nuevo San Juan are almost 70 km apart, they are otherwise separated only by the Río Gálvez, a small river that dwindles to a narrow stream in its adjacent headwater region (>300 m above sea level) and is unlikely to be an effective faunal barrier. Therefore, it seems plausible that as many as four species of Thyroptera might be sympatric at some sites within this interfluvial region.
FIG. 7.
Results of principal components analysis, illustrating the dispersion of specimen scores for male Thyroptera devivoi (filled circles), Thyroptera discifera (open circles), Thyroptera lavali (open triangles), Thyroptera tricolor (asterisk), and Thyroptera wynneae (filled squares). See text for explanation and table 3 for factor loadings and other results.
![f07_01.jpg](ContentImages/Journals/novi/2014/3795/3795.1/graphic/WebImages/f07_01.jpg)
TABLE 4.
Records of Sympatry among Species of Thyroptera.
![t04_01.gif](ContentImages/Journals/novi/2014/3795/3795.1/graphic/WebImages/t04_01.gif)
Such diversity poses a challenge for bat inventory research, because disk-winged bats are hard to capture in mist nets, even where they are known to be locally common (Vonhof and Fenton, 2004). For example, Simmons and Voss (1998) captured 37 specimens of T. tricolor by searching for roosts, but the same researchers captured only four individuals of this species in 112 nights of mist-netting at their French Guianan study site. If other congeneric species usually roost in less accessible situations than T. tricolor (e.g., in the canopy or subcanopy; Pine, 1993), then it may seldom be possible to know whether local thryopterid diversity has been completely inventoried. The recent discovery of T. discifera at La Selva, Costa Rica (Tschapka et al., 2000), one of the most thoroughly inventoried rainforest sites anywhere in the New World tropics (Voss and Emmons, 1996: appendix 2) underscores the elusive nature of these bats.
Given the broadly overlapping geographic distributions of thyropterids, four or more species might be sympatric throughout much of cis-Andean South America, despite the fact that only one or two are known to cooccur at sites other than Jenaro Herrera. To date, sympatry has been documented for six of the 10 possible pairwise combinations of Thyroptera species (table 4). The obvious ecological outlier in this context is T. devivoi, which occupies edaphically moist gallery formations in savanna landscapes (Gregorin et al., 2006). Among the four species known to inhabit continuous tracts of climatically moist (evergreen or semideciduous) forest, five of the six possible combinations of pairwise sympatry have been observed. What ecological differences among these species might facilitate such coexistence? Divergent roosting behavior is one possibility.
Well-developed adhesive disks on the wing and foot occur in only two families of bats: the Neotropical family Thyropteridae and the Old World family Myzopodidae, which is endemic to Madagascar (Goodman et al., 2007; Wilson, 2008).3 Such disks allow these bats to adhere to smooth surfaces when roosting, and their occurrence in two distantly related families is thought to represent an extraordinary example of convergent evolution (Eisenberg, 1981; Riskin and Racey, 2010; Schliemann and Goodman, 2011; Schliemann and Maas, 1978). However, the mechanisms of sub strate-disk adhesion used in each case are different: whereas thyropterid disks use suction to adhere to surfaces, myzopodid disks operate by wet adhesion (Goodman et al., 2007; Riskin and Fenton, 2001; Riskin and Racey, 2010; Schliemann and Goodman, 2011). Although bats in both families are said to use large, furled leaves for roost sites (Kunz and Lumsden, 2003), most of what is actually known about thyropterid roosts is derived from observations of a single species, Thyroptera tricolor.
FIG. 8.
Roost of Thyroptera wynneae in secondary growth at the Centro de Investigaciones Jenaro Herrera, Loreto, Peru. Two bats occupied the dark interior of this dead Cecropia leaf (arrows), which was hanging in understory vegetation by its petiole about 2 m above the ground. One specimen was captured in a butterfly net placed underneath the leaf, but the other bat escaped. Cecropia (Cecropiaceae) is a speciose genus of trees commonly found in secondary vegetation throughout the Neotropics, where hanging dead leaves like this one are abundant in the sub canopy and understory.
![f08_01.jpg](ContentImages/Journals/novi/2014/3795/3795.1/graphic/WebImages/f08_01.jpg)
By far the most commonly collected thyropterid, Thyroptera tricolor has often been reported as roosting in the half-unrolled new leaves of large understory mono cots such as Heliconia (Heliconiaceae), Musa (Musaceae), Calathea (Marantaceae), and Phenako sper mum (Strelitziaceae). In general, leaves used as roosts by T. tricolor form upright, tubular (or conical) shelters with openings 40–100 mm in diameter; they are located <4 m above the ground; and they are not in direct contact with other vegetation (Goodwin and Greenhall, 1961; Wimsatt and VillaR., 1970; Findley and Wilson, 1974; Simmons and Voss, 1998; Vonhof and Fenton, 2004). Recently, Chaverri and Gillam (2013) reported that besides shelter, T. tricolor uses the furled leaves as acoustic horns that amplify incoming and outgoing social calls among group members. Such shelters are ephemeral because unfurling new leaves soon lose the small-diameter tubular/conical form preferred by bats; therefore, colonies of T. tricolor must change roosts every night and seem to favor habitats with dense concentrations of suitable host plants (Findley and Wilson, 1974; Vonhof and Fenton, 2004; Chaverri and Kunz, 2011). Because the ventral surfaces of roosting bats are pressed against the roosting substrate, and because young leaves are paler green than mature leaves (fig. 10), the whitish ventral coloration of T. tricolor might be an adaptation for roosting in translucent shelters, where dark ventral fur might often betray the presence of resident bats to visually oriented predators.
FIG. 9.
Patricia J. Wynne at her microscope in the AMNH Department of Mammalogy. Patriciaapos;s first mammalogical illustration appeared almost 40 years ago (in Hooper, 1975), and she has lost track of how many she has drawn since then. In addition to her technical work for museum researchers, Patricia has illustrated museum exhibition labels, numerous educational publications, and dozens of popular science books. She is now busier than ever in semiretirement. Photo: Denis Finnin.
![f09_01.jpg](ContentImages/Journals/novi/2014/3795/3795.1/graphic/WebImages/f09_01.jpg)
To the best of our knowledge, no other Thyroptera species has ever been collected in shelters similar to those normally occupied by T. tricolor, and from this fact alone it may be supposed that they use different diurnal refugia. Apparently, only three day roosts of T. discifera have been described, all of them under (Robinson and Lyon, 1901) or inside (Torres et al., 1988) dead banana leaves. Dead banana leaves are brown, opaque, and hang suspended from collapsed petioles, so dead-leaf roosts are dark and open downward, quite unlike the shelters normally occupied by T. tricolor. The only described roost of T. devivoi was characterized by the collector as ’under an eaté palm leaf” (Gregorin et al, 2006: 239),4 and one specimen of T. lavali is also thought to have also been collected from a palm (Solari et al., 1999). Apparently, these five observations, together with our discovery of T. wynneae roosting in a dead Cecropia leaf, comprise all that is known of the roosting habits of other species of Thyroptera.
FIG. 10.
Roost of Thyroptera tricolor in the half-unrolled new leaf of a small Heliconia in primary forest at Paracou, French Guiana. The adhesive disks of the roosting bats are visible as dark spots through the translucent tissue of the leaf. The bats themselves (an adult male and three adult females) form a dark mass within their tubular shelter (arrow).
![f10_01.jpg](ContentImages/Journals/novi/2014/3795/3795.1/graphic/WebImages/f10_01.jpg)
If populations of disk-winged bats are limited by roost availability (Findley and Wilson, 1974; but see Vonhof and Fenton, 2004), then there may have been strong selection for divergent roosting habits in the course of thyropterid phylogeny, possibly resulting in species that routinely occupy dead leaves or palm fronds.5 Morphological products of such selection might include taxonomie differences in size, ventral pelage coloration, and disk morphology. Smaller bats, for example, can probably occupy smaller shelters (e.g., Cecropia leaves) than larger bats, bats with dark ventral fur might inhabit darker refugia (e.g., dead or opaque-green leaves) than pale-ventered bats, and bats with oval adhesive disks might be better equipped to roost in leaves (such as palm fronds) with vanes divided by well-developed ribs than bats with rounded disks (if these routinely use smoother leaves as shelters). In the current absence of reliable information about the roosting habits of most Thyroptera species, these are mere conjectures, but they could be tested by actively searching for different kinds of roosts in areas where multiple species could be expected to occur.
Interestingly, not all of the diurnal shelters known to be used by Thyroptera afford substrates suitable for disk adhesion. In particular, the finely corrugated surface of dried Cecropia leaves seem unsuitable for suction-based attachment. Instead, bats inhabiting these (and perhaps other) rough-surfaced shelters may use their claws instead (Riskin and Fenton, 2001), and in downward-opening roosts it is possible that they hang head-down (like other bats), rather than the heads-up posture normally observed in T. tricolor.
Reconstructing the evolutionary history of roosting habits among disk-winged bats will obviously require much more information about the behavior of extant species, as well as knowledge about their phylogenetic relationships. In particular, dated phylogenies will be crucial for correlating relevant character transformations with historical events in the evolution of Neotropical plant communities, and the thyropterid fossil record is obviously important in this context. Although fossil thyropterids are known from the Miocene of South America, they are poorly preserved and taxonomically problematic. Czaplewski (1997) described fossils from the Miocene La Venta deposits in Colombia as a new species, Thyroptera robusta, but he subsequently synonymized robusta with lavali (Czaplewski, 1996; Czaplewski et al., 2003).6 The La Venta material currently referred to T. lavali consists only of isolated fragments: a broken dentary with ml and m2; another dentary fragment preserving only the talonid of m3; two upper canines, one lacking enamel; an upper Ml and M2; and fragments of an upper and lower molar (Czaplewski, 1996, 1997; Czaplewski et al., 2003). Although we agree that these fossil fragments may not be distinguishable from recent material of T. lavali, in our opinion they simply indicate that a lavali-sized thyropterid existed in the Miocene. Czaplewski (1997: 420) also described from La Venta a series of isolated teeth, probably from one skull, that he referred to “Thyroptera cf. T. tricolor.” Noting that differences between these teeth and those of tricolor were minor and might not exceed the expected limits of individual variation in tricolor, Czaplewski (1997: 422) concluded:
Although with more and better specimens it may eventually be determined that these teeth represent an undescribed species of Thyroptera, I cannot find sufficient reason to name a new species based on the fossils, despite the fact that they are separated by some 12.5 million years from their living relatives. These specimens indicate a significant period of evolutionary stasis for disk-winged bats consistent with their present lack of diversity.
Two additional extant species of Thyroptera have since been discovered, and our review indicates that few dental characters distinguish the five living species from one another. We suggest that the Miocene fossils described by Czaplewski (1996, 1997) and Czaplewski et al. (2003) probably represent taxa related to, but not conspecific with, extant species. More complete fossils and more comprehensive analyses will be needed before the relationships of Miocene thyropterids to extant lineages can be determined. Rather than providing evidence that T. lavali and T. tricolor diverged from T. discifera at least 12 Mya (as suggested by Solari et al., 2004), we would argue that the most we can say based on the fossil record is that the genus Thyroptera was clearly distinct by about 12.5 Mya.
Molecular-clock dating and biogeographic analyses indicate that the lineage leading to Thyroptera diverged from that of other noctilionoids approximately 36–45 Mya, and that this divergence occurred in the Neotropics (Miller-Butterworth et al., 2007). Therefore, thyropterids were probably present in South America for approximately 23 to 33 million years before the La Venta fauna existed. Given that we do not know the relationships of these Miocene fossils to extant species, it is premature to speculate about the timing of the origin of the unique anatomical specializations and unusual roosting habits seen in modern members of this group.
Key to the Species of Thyroptera
1. Disk at base of thumb clearly oval in outline (not round); ventral pelage bicolored or tricolored; both cusps of the second upper incisor perpendicularly arranged relative to the long axis of the toothrow 2
1′. Disk at base of thumb nearly circular in outline; ventral pelage unicolored; both cusps on the second upper incisor obliquely arranged relative to the long axis of the tooth row. 4
2. Forearm ≥ 35.0 mm; GLS ≥ 14.0 mm; ventral pelage bicolored; proximal portion of forearm sparsely haired; rostrum of equal length or longer than the braincase; accessory cusps on the third lower incisor absent or only faintly developed 3
2′. Forearm ≤ 34.5 mm; GLS ≤ 13.5 mm; ventral pelage tricolored; proximal portion of the forearm densely haired; rostrum considerably shorter than the braincase; accessory cusps well developed on the third lower incisor Thyroptera wynneae
3. Ventral pelage distinctly frosted, individual hairs dark brown to blackish at the base with grayish-brown to whitish tips; calcar lacks lappets or only one faintly developed lappet is present; GLS ≤15.1 mm and ZB ≤7.7 mm Thyroptera devivoi
3′. Ventral pelage bicolored, not distinctly frosted; calcar has one well-developed lappet that project posterolaterally from the shaft of the calcar; GLS ≥ 15.2 mm and ZB ≥ 8.0 mm Thyroptera lavali
4. Ventral pelage unicolored white or pale gray; proximal portion of forearm sparsely haired; large foramen ovale (0.50–0.65 mm) Thyroptera tricolor
4′. Ventral pelage unicolored yellowish-brown (not white); proximal portion of forearm densely haired; small foramen ovale (< 0.45 mm) Thyroptera discifera
ACKNOWLEDGMENTS
We thank all the participants of the “International Bat Course 2012” organized by CEBIO (Centro de Ecología y Biodiversidad) for their camaraderie at Jenaro Herrera, and we thank course organizer Erika Paliza for inviting P.M.V. to attend this event and for obtaining the necessary permits that enabled P.M.V. and R.S.V. to collect bats there after the course ended. The following curators and collection staff graciously provided access to specimens under their care: Erika Paliza (Centro de Ecología y Biodiversidad, Lima), Bruce D. Patterson and John Phelps (Field Museum of Natural History, Chicago), Víctor Pacheco (Museo de Historia Natural de la Universidad Nacional Mayor de San Marcos, Lima), and Burton Lim (Royal Ontario Museum, Toronto). We are also grateful to Burton Lim for his live photographs of Thyroptera wynneae and to Wendy Calderón for assistance in the field. Brock Fenton and Burton Lim provided helpful comments on the submitted draft of this manuscript. Partial funding for R.G.'s fieldwork in Brazil was provided by the Fundaç˜o de Amparo à Pesquisa do Estado de Minas Gerais.
REFERENCES
Appendices
APPENDIX 1
Specimens Examined
The following list includes all the specimens used in this study, with their respective localities. See Material and Methods for abbreviations. Coordinates are provided in appendix 2. Specimens used in the morphometric analyses are marked with an asterisk.
Thyroptera devivoi (N = 3). BRAZIL—Piauí: Estação Ecológica Uruçui-Una, Bom Jesus (MZUSP 30073*). Tocantins: Estaçao Ecológica Serra Gerai do Tocantins (Jalapao) (MZUSP 32345*). GUYANA— Upper Takutu-Upper Essequibo: Tamton (ROM 35588*).
Thyroptera discifera (N = 10). BRAZIL—Bahia: probably near Salvador (MZUSP 16395, 16396). Pará: Caxiricatuba (MZUSP 5628); Santarém, Tapajós River, Tauari (AMNH 94549). COLOMBIA—Córdoba: Socorre, Upper Río Sinú (FMNH 69509*). Valle del Cauca: La Costa, El Tambo (FMNH 140791). ECUADOR—Pichincha (AMNH 46687). PERU—Loreto: Requena, Jenaro Herrera, Centro de Investigaciones Jenaro Herrera (MUSM 5546*). Ucayali: Atalaya, Cumaría (FMNH 46160*). VENEZUELA—Distrito Federal: La Guaira, San Julián (AMNH 16686).
Thyroptera lavali (N = 7). ECUADOR—Napo: Yasuní National Park, 42 km S, 1 km E Pompeya Sur (ROM 104026, 105215). PERU—Loreto: Mariscal Ramón Castilla, Río Yavarí Mirím, Quebrada Esperanza (FMNH 89118*, 89120, 89121); Requena, Jenaro Herrera, Centro de Investigaciones Jenaro Herrera (MUSM 5944*).
Thyroptera tricolor (N = 57). BRAZIL—Amapá: Rio Amapari (MZUSP 16393). Pará: Santarém, Tapajós River, Piquiatuba (AMNH 94550*); Abaetetuba (MZUSP 8583–8586); Rio Tocantins, Ihla do Taiuna (AMNH 97129, 97130, 97131*, 97132–97134, 97136, 97137). Rio de Janeiro: Angra dos Reis (MZUSP 6585). São Paulo: Iguape (MZUSP 5702). COLOMBIA—Valle del Cauca: Anchicaya river (AMNH 239084-239086, 239088); Dagua, Cali-Buenaventura highway, near Dagua River (AMNH 239077, 239078*, 239080, 239081*). COSTA RICA—Puntarenas: Esparza (FMNH 43980*); Parque Nacional Carara (FMNH 180726). ECUADOR—Tungurahua: Palmira (AMNH 67592*). FRENCH GUIANA—Paracou, near Sinnamary (AMNH 267216, 267217,266353*). PERU—Loreto: Alto Amazonas, Nuevo San Juan, Gálvez River (AMNH 272761, 273155, 273156, 273157*, 273160, 273161*; MUSM 13262, 13263, 15278–15282); Estación Biológica Isla Muyuy (MUSM 21332); Estirón, Río Ampiyacu (MZUSP 14180, 14181); Mariscal Ramón Castilla, Río Yavarí Mirím, Quebrada Esperanza (FMNH 89117); Maynas, Orosa, Amazon River (AMNH 74022, 74024, 74025). San Martín: Moyobamba, Tingana (FMNH 203644). TRINIDAD AND TOBAGO— Trinidad: (FMNH 53054); Saint Andrew, Sangre Grande (AMNH 185342); Saint George, Heights of Aripo, Cerro Aripo (AMNH 29693); Saint George, Port of Spain (AMNH 183860). VENEZUELA—Amazonas: Río Negro, Muyrapinimo (AMNH 77556*, 77558).
Thyroptera wynneae (N = 3). BRAZIL—Minas Gerais: Campolina, about 25 km E Marliéria, Parque Estadual do Rio Doce (CMUFLA 694*, 1160*). PERU—Loreto: Requena, Jenaro Herrera, Centro de Investigaciones Jenaro Herrera (CEBIOMAS 237*).
APPENDIX 2
Gazetteer of Collecting Localities
Below we list all localities from which we personally examined specimens of Thyroptera. Names of the largest administrative unit (department, state, etc.) within each country are italicized, and geographic coordinates are provided.
BRAZIL
1. Amapá, Rio Amapari (2deg;00′N, 52°30′W).
2. Bahia, probably near Salvador (12°58′S, 38°30′W). See Gregorin et al. (2006) for a discussion of this problematic locality.
3. Minas Gerais: Parque Estadual do Rio Doce, Campolina (19°42′S, 42°30′W), about 25 km E Marliéria, Parque Estadual do Rio Doce.
4. Pará, Abaetetuba (1°43′S, 48°52′W).
5. Pará, Caxiricatuba (2°45′S, 55°02′W).
6. Pará, Rio Tocantins, Ihla do Taiuna (2°15′S, 49°29′W).
7. Pará, Santarém, Tapajós River, Tauari (3°05′S, 55°06′W).
8. Pará, Santarém, Tapajós River, Piquiatuba (2°34′S, 54°42′W).
9. Piaui, Estaçao Ecológica Uruçui-Una, Bom Jesus (8°52′S, 44°57′W).
10. Rio de Janeiro, Angra dos Reis (23°00′S, 44°19′W).
11. São Paulo, Iguape (24°42′S, 47°33′W).
12. Tocantins, Estaçao Ecológica Serra Geral do Tocantins (Jalap˜o) (10°33′S, 46°45′W).
COLOMBIA
COSTA RICA
ECUADOR
FRENCH GUIANA
PERU
22. Loreto, Alto Amazonas, Nuevo San Juan, Gálvez River (5°15′S, 73°10′W).
23. Loreto, Estación Biológica Isla Muyuy (3°58′S, 73°047prime;W).
24. Loreto, Estirón, Río Ampiyacu (3°19′;S, 71°51′W).
25. Loreto, Mariscal Ramón Castilla, Río Yavarí Mirím, Quebrada Esperanza (4°20′S, 71°55′W).
26. Loreto, Maynas, Orosa, Amazon River (3°32′S, 72°11′W).
27. Loreto, Requena, Jenaro Herrera, Centro de Investigaciones Jenaro Herrera (4°55′S, 73°40′W).
28. San Martín, Moyobamba, Tingana (5°54′S, 77°06′W).
29. Ucayali, Atalaya, Cumaría (9°51′S, 74°01′W).
TRINIDAD AND TOBAGO
VENEZUELA
Notes
[1] 3 Certain other foliage-roosting bats have adhesive pads rather than well-developed disks (Thewissen and Etnier, 1995).
[2] 4 The collector was probably referring to the swamp palm Mauritia flexuosa, locally known (in Guyana) as ité (Henderson et al., 1995).
[3] 5 Interestingly, although Thyroptera tricolor usually occupies upright-tubular or -conical living leaves, roosts of this species are occasionally found elsewhere. For example, colonies of T. tricolor have sometimes been reported roosting in pendant dead leaves of Heliconia (Goodwin and Greenhall, 1961) and Phenakospermum (Simmons and Voss, 1998), and individual bats are sometimes found roosting on the underside of unfurled leaves (Vonhof and Fenton, 2004). Such behavioral variability within species, if heritable, might have facilitated the evolution of taxonomie differences in roosting habits.
[4] 6 See Wilson (2008: 395) for relevant comments on publication dates.