HOLOTYPE: An adult female (fig. 2), preserved as skin and cleaned skull, deposited at the Coleção Adriano Lúcio Peracchi (ALP 9667), in the Laboratório de Mastozoologia of the Universidade Federal Rural do Rio de Janeiro, Rio de Janeiro, Brazil. It was collected on June, 3, 2011, by Adriano L. Peracchi, Marcelo R. Nogueira, Isaac P. Lima, and Paulo Cesar at the Estrada Oiticica (19°09′17″ S and 40°00′18″ W), Reserva Natural Vale, in the municipality of Linhares, state of Espírito Santo, southeastern Brazil (field number AP 2847). Liver tissue is preserved in ethanol at the Coleção Adriano Lúcio Peracchi. Skull and mandible in good condition, but the left upper canine and the left first premolar are both missing, the anterolabial portion of the left first upper molar is broken, and the distal portion of both pterygoids is broken.

PARATYPES: Two adult males, both also collected in the municipality of Linhares. The first specimen was collected on December 15, 1977, at the Floresta Nacional de Goytacazes (fig. 1) (19°25′51″ S and 40°04′20″ W) by Adriano L. Peracchi, Sila T. Albuquerque, Sansão D.L. Raimundo, and Onézimo F. Fraga (field number AP 1782). It is a dried skin with skull, initially deposited at the Coleção Adriano Lúcio Peracchi, under the number ALP 3268, but now incorporated into the mammal collection of the Museu Nacional, Universidade Federal do Rio de Janeiro, municipality of Rio de Janeiro, Brazil, under the accession number MN 78305. The skin is in good condition (except for some patches lacking fur on the ventral region), but the skull is in poor condition: the basicranium, the occipital region, and the left basal portion of the braincase are broken. The mandible is represented only by the left dentary, in which the ascending ramus is missing. The second specimen is preserved in fluid, with skull removed and cleaned, and was deposited at the Coleção Adriano Lúcio Peracchi under the number ALP 9599. It was collected on November 9, 2010, at 22:30 hours, by Isaac P. Lima (field number IPL 1285). The skull and mandible are in perfect condition, and the tongue is preserved separately in fluid. A photograph of this specimen appeared in Peracchi et al. (2011b: 166) under the name Choeroniscus minor.

TYPE LOCALITY: Brazil, state of Espírito Santo, municipality of Linhares, Reserva Natural Vale, 19°08′29″ S and 40°04′08″ W, approximately 60 m elevation.

ETYMOLOGY: In the first phytogeographic system proposed for Brazilian vegetation, the 19th-century German naturalist Karl Friedrich Philipp von Martius ascribed names of nymphs from Greek mythology to each of the major provinces he described (Martius, 1824). To the province we now recognize as Atlantic Forest, he gave the name “Dryades” (from the Greek Dryas, a wood nymph) for the immortal nymphs of the forest. The generic epithet proposed here means, therefore, “bat from Dryades,” hence “bat from the Atlantic Forest,” in reference to the endemic character of this taxon. The species name capixaba, a noun in apposition, is from the Tupi language and designates natives from the state of Espírito Santo.

DESCRIPTION: Body size small (forearm 29–32 mm; tables 1–2) for a Glossophaginae and for bats in general; muzzle elongated, but shorter than braincase; mandibular prognathism evident; spear of noseleaf nearly triangular (3 mm wide × 3.38 mm tall in the fluid paratype) with no central rib; internarial region with narrow, unbeaded ridge along midsagittal line; horseshoe of noseleaf continuous with upper lip with no ridges or other structures separating it from lip; lateral edges of horseshoe confluent with face inferiorly, but with superior edge free; dorsal snout with pair of well-developed padlike papillae separated from each other and located just behind noseleaf; two columns of vibrissae present at each side of snout, lateral to noseleaf, the more anterior with its two vibrissae standing entirely on the padlike papilla, and the more posterior with three of its four vibrissae standing on this pad (the lowest vibrissa stands on its own smaller pad); anterior half of the upper lip, at each side of mouth, also ornamented with well-developed vibrissae, five of which are nearly evenly distributed in an irregular line parallel to lip, beginning close to posterior border of the large padlike papilla described above, and one located just above the more posterior of these vibrissae; one genal vibrissa present halfway between the corner of the mouth and the base of the ear notch; chin with deep central cleft and with a pair of narrow, scalloped dermal pads, one on each side of the cleft; two interramal vibrissae located posterior to the chin cleft; chin skin adjacent to the cleft, along each side of the mandible, ornamented with five well-developed vibrissae arranged, from anterior to posterior region, in a 1-2-1-1 scheme (additional, smaller vibrissae also present); ears small, with rounded distal tip, strongly convex inner margin, and outer margin nearly straight; tragus about one third the length of the pinna, spatulate in form, noticeably thicker medially, thinner and translucent laterally, bearing a basal expansion with a lobule on its upper margin that is folded posteriorly; forearm with basal two-thirds covered with fur on both the dorsal and ventral surfaces; metacarpal and proximal phalange of thumb nearly equal in length; metacarpal formula V < IV < III; plagiopatagium dark brown and inserted at the base of toes (fig. 3); uropatagium relatively long (15 mm at midline), dark brown, bearing sparse, short hairs on dorsal surface and on upper one-third of ventral surface, and naked along its posterior border (no fringe of hairs); calcar length nearly equal to that of foot (fig. 3); tail short (6.50 mm); dorsal fur about 7 mm long on upper back, tricolored, with pale brown base, wide beige middle band, and narrow dark brown band distally (fig. 3); dorsal fur with additional pale frosting at tips; ventral fur on upper chest similar to dorsal fur in both length and banding pattern, but distal band similar in color to basal band, leading to a slightly paler general ventral appearance to the bat (brown venter versus dark brown in the dorsum).

Skull (figs. 4–5) with rostrum shorter than braincase; rostrum gently tapering anteriorly to a marked constriction at level of P4, then expanded in an unusual, autapomorphic inflation of anterior portion of maxillary bones (less developed in males); postorbital constriction present in the female (holotype), but poorly developed or absent in males (paratypes); premaxillae relatively well developed and with two circular and small midline foramina, the more posterior of which is small and located between antero-internal margin of incisive foramina; orbital portion of frontal marked by five diagonal lines marking the internal insertion points of nasal turbinals; braincase oval in dorsal view and low in lateral view; sagittal crest absent, but lambdoidal crests developed; anterior border of foramen magnum V-shaped; basisphenoid pits shallow and separated by low and wide septum; palate almost flat from premaxillae to the level of P3, then concave until the end of anterior portion of palate (just behind the dentition); posterolateral border of palate, behind molars, concave; posterior (interorbital) palate long and with U-shaped posterior border; pterygoid wings short, not touching the tympanic bulla (visible only in the paratypes; wings broken in holotype); presphenoid with a central ridge; basisphenoid smooth except for a small central ridge at junction with the wide interbasioccipital septum; zygomatic arches incomplete, small anterior projection of the zygoma turned upward.

Dentary (fig. 4) long and delicate; anterior extremity of dentary projected well beyond canines, but in the same plane with the body of the dentary (not downturned ventrally); mandibular condyle and coronoid process extend dorsally to approximately the same level above the dentary; angular region with discrete angular process at posterior extremity and no ventral projection (as a consequence, the ventral margin of the ascending ramus forms an almost straight line with the body (tooth-bearing portion) of the mandible.

Dentition (figs. 5–6) with dental formula I 2/0, C 1/1, P 2/3, M 3/3 × 2 = 30; upper incisors small, inner pair mesiodistally wider than outer and with flat crowns; outer incisors varying from pointed to peglike (variation probably related to wear); both upper incisor teeth completely visible in lateral view; upper canines delicate, laterally compressed, with flat anterior face, and strongly recurved posterior edge; lingual surface of upper canines slightly concave, whereas anterior surface lacks any longitudinal ridge or groove; posterior accessory cusp of upper canines poorly developed or absent; upper canines laterally bowed (more conspicuous in holotype than paratypes); two upper premolars, the second (P4) larger than first (P3) in anteroposterior length (0.79 versus 0.65 mm in the holotype) and also slightly higher; P3 with anterior and posterior cuspules usually absent or, when present, poorly developed; P4 similar to P3 with respect to presence and development of cuspules, but posterior crest of P4 protocone longer and less steeply inclined than anterior crest; both P3 and P4 laterally compressed; M1 and M2 greater in mesiodistal length than M3; all upper molars relatively similar in structure, with parastyle as main cusp, paracrista, paracone, and postparacrista (= precentrocrista of Philips, 1971) either reduced to a single low ridge (holotype) or with poorly defined, anteriorly displaced paracone (paratypes); precentrocrista runs toward lingual side of tooth almost parallel to the labial margin of the tooth, so that the anterior half of the W-shaped ectoloph essentially is not discernible; anterolabial portion of right M1 in the holotype (the left is broken) with an anterior basal projection or shelf that is absent in both M1s of paratypes; anterolabial portion of M3 retaining a nearly triangular shape, with paracone always present and not anteriorly displaced as in M1 and M2 of paratypes; posterolabial portion of all upper molars less laterally compressed than the anterior one and also retaining a nearly triangular shape typical of the ectoloph dentition (poorly defined in MN 78305, but clear in the holotype); metacone conspicuous on all upper molars, but mesostyle and metastyle variable in development (both conspicuous in the holotype, but less developed or reduced to a low ridge in paratypes); protocone relatively low, but present on all upper molars and connected to the base of metacone via postprotocrista; hypocone and hypoconal shelf absent.

Lower incisors absent; lower canines simple and small, straight and not recurved in lateral view, straight at outer margin and slightly recurved internally in anterior view, with poorly developed cingular shelf on the posterior side of the tooth; first lower premolar (p2) characterized by a low central cusp, only slightly higher than anterior and posterior stylar cusps, which are also poorly developed; p2 with lingual outline concave in occlusal view (more pronounced in the holotype); second lower premolar (p3) with central cusp conspicuously higher than anterior and posterior stylar cusps, both of which are better developed than those on first premolar; third lower premolar (p4) with the main cusp similar in size to the main cusp of p3, and with posterior crest dorsoventrally compressed and ending in a conspicuous posterior cuspule; degree of slope of anterior crest of the primary cusp of p4 varying from steep (anterior style visible; ALP 9599) to more gradual in most specimens (no anterior style visible); the three lower molars are laterally compressed, but retain all typical cusps; first two lower molars (m1 and m2) similar in mesiodistal length and slightly longer than m3; metaconid is the most prominent cusp in the lower molars; m1 with paralophid (connecting paraconid and protoconid) higher than on other molars; metacristid (connecting paraconid and metaconid) low on all molars; hypoconid and entoconid low but distinct; hypoconulid absent.

Multiple diastemata present in both upper and lower dentitions; upper incisors clearly separated by a wide median gap; diastema between inner and outer upper incisors slightly smaller than that between outer incisor and canine, which is about the same size as the mesiodistal length of inner incisor; both P3 and P4 widely separated from adjacent teeth; distance between canine and P3 large, similar to anteroposterior length of canine; diastema between P3 and P4 similar to anteroposterior length of P3; diastema between P4 and M1 similar to that between P3 and P4; three upper molars separated by conspicuous diastemata, the second (between M2 and M3) smaller than the first; first lower premolar (p2) separated from canine by a small diastema (usually smaller than those separating the lower premolars from each other); m1 separated from m2 by a conspicuous gap, and m2 separated from m3 by a small gap or in contact with this latter tooth (holotype).

COMPARISONS AND DIAGNOSTIC TRAITS: Dryadonycteris capixaba can be easily differentiated from all other described Neotropical nectar-feeding bats on the basis of external and craniodental morphology (table 3). The absence of lower incisors is shared only with members of the tribe Choeronycterini (sensu Baker et al., 2003), to the exclusion of all Lonchophyllini (Lionycteris, Lonchophylla, Platalina, and Xeronycteris) and Glossophagini (Glossophaga, Monophyllus, and Leptonycteris). Within the tribe Choeronycterini, two subtribes were recognized by Baker et al. (2003): Anourina and Choeronycterina. The former includes only the genus Anoura, from which Dryadonycteris is promptly differentiated by its long uropatagium and tail. Dryadonycteris is also distinguished from Anoura by morphology of the upper part of the lateral edges of the horseshoe, which are free from the face in Dryadonycteris instead of being confluent with skin of the face as in Anoura. In this respect Dryadonycteris resembles members of the subtribe Choeronycterina (data not available for Scleronycteris) (Wetterer et al., 2000). Among the seven genera currently recognized as comprising Choeronycterini, Musonycteris and Choeronycteris are remarkably different and are easily distinguished from all other genera, including Dryadonycteris, by virtue of possessing greatly elongated rostra that are approximately equal to or longer than their respective braincases (fig. 7). This character results in a very long muzzle in Musonycteris and Choeronycteris, which contrasts with the moderately long muzzle of Dryadonycteris (rostrum shorter than braincase). Externally, Dryadonycteris is also differentiated from Musonycteris and Choeronycteris, as well as from Choeroniscus and Scleronycteris, by the color banding pattern of dorsal fur. Dryadonycteris shares exclusively with Lichonycteris and Hylonycteris a tricolored fur pattern, i.e., each hair has a pale middle band and darker basal and distal bands. The condition present in most glossophagines and lonchophyllines is a bicolored pattern in which each hair has a pale base and a dark tip (Wetterer et al., 2000; Carstens et al., 2002). A tricolored dorsal pelage is also found in other genera of Phyllostomidae, but only outside the lineages of specialized nectar-feeding bats (e.g., Carollia). It must be noted, however, that the basal band in the dorsal fur of D. capixaba is paler than the distal band, which leads to a less visible contrast between the two basal bands. This is similar to the condition reported for Hylonycteris (Reid, 1997), while the prevalent condition in Lichonycteris is a more clearly tricolored pattern (Reid, 1997). A particular caution is, therefore, necessary when analyzing this character, particularly under field conditions.

Dryadonycteris capixaba has calcar and foot subequal in length, which departs from the condition reported for all other glossophagines and lonchophyllines, in which the calcar is shorter than the foot (Wetterer et al., 2000; Carstens et al., 2002). Additionally, in most glossophagines the insertion of the posterior edge of the plagiopatagium occurs at the ankle, whereas in Dryadonycteris it extends distally to the base of the toes. This latter condition is also found in Lichonycteris (Reid, 1997; LaVal and Rodriguez-H., 2002), which is the genus most prone to be externally confused with Dryadonycteris. The similarity between Dryadonycteris and Lichonycteris is particularly relevant because these taxa are the same general size and potentially occur sympatrically in the Atlantic Forest of eastern Brazil. Hylonycteris is also small sized and has tricolored dorsal fur, but its wings attach to the ankle, its calcar is shorter than its foot, and its known distribution is restricted to Central America.

Though generally similar to Lichonycteris in external characters, Dryadonycteris is markedly different from this genus in several craniodental features. The most obvious are: rostral length (shorter in Lichonycteris) and shape (anteriorly inflated in Dryadonycteris, but not in Lichonycteris); development of the premaxillae (well developed in Dryadonycteris, but not in Lichonycteris); position of upper incisors (separated by a wide median gap and convergently positioned in Dryadonycteris, versus evenly spaced and forming an almost straight line in Lichonycteris); and number of upper molars (only two in Lichonycteris, but three in Dryadonycteris). Though some of these features will not be evident in the field, we expect that the relatively longer muzzle, well-developed premaxillae, and convergently positioned upper incisors (both outer and inner incisors are visible in lateral view) will prove useful in field identification.

The skull of Dryadonycteris capixaba can be promptly differentiated from those of all other glossophagines by its unique inflation of the anterior maxillary bones, anterior to a conspicuous constriction of the rostrum (fig. 7). Among Choeronycterini genera, it is also unique in having well-developed premaxillae (figs. 7, 8), with both inner and outer incisors visible in lateral view (fig. 9). Dryadonycteris also has two small foramina on the midline of the premaxillae (in addition to the two larger foramina located more laterally), while in most glossophagines a single foramen is present or, as in the case of Scleronycteris ega and Lichonycteris obscura, no foramina are present in the midline (fig. 8) (Carstens et al., 2002). The diagonal lines present in the orbital region of Dryadonycteris seem to be shared only with Anoura geoffroyi and with the rare Amazonian endemic Scleronycteris ega, which makes this another character useful for distinguishing the skull of Dryadonycteris (fig. 9). In Anoura caudifer and Glossophaga soricina some lines are present, but they are not developed as in the aforementioned species. The posterior palate in Dryadonycteris is proportionally longer than that in all other Choeronycterini, and the pterygoid wings are short and not inflated at the distal margin as is the case in Choeroniscus, Choeronycteris, and Musonycteris (fig. 8).

In Dryadonycteris the paracone is subject to individual variation. The holotype lacks a distinct paracone on M1 and M2 (fig. 6), but this cusp is more evident in the paratypes, where it appears as a low cusp adjacent to the parastyle. This type of within-taxon variation in cusp development has also been described for Lichonycteris (Phillips, 1971). In contrast, the paracone is uniformly absent in Scleronycteris, Choeroniscus, Hylonycteris, Choeronycteris, and Musonycteris (Phillips, 1971), and it is uniformly present in Anoura, which retains a conspicuous W-shaped ectoloph in the upper molars. The mesostyle is another character that is typically informative regarding the evolution of Choeronycterini, as it is present in Anoura and Scleronycteris, poorly developed in Lichonycteris, and absent in Hylonycteris, Choeroniscus, Choeronycteris, and Musonycteris (Phillips, 1971; Jones and Carter, 1976; Koopman, 1994). In this respect, the female of Dryadonycteris capixaba (holotype; fig. 6) described here is more similar to Scleronycteris, whereas the two males resemble Hylonycteris and the other taxa listed above. In the lower dentition, the first premolar of Dryadonycteris is separated from adjacent canine by a small but conspicuous diastema, whereas in Lichonycteris, Scleronycteris, and Hylonycteris this gap is barely visible or absent (fig. 9). In Anoura, Choeroniscus, Choeronycteris, and Musonycteris, the development of this diastema seems to be variable. Dryadonycteris capixaba is similar to Lichonycteris, Scleronycteris, and Hylonycteris in having small diastemata separating the lower premolars, whereas in Anoura, Choeroniscus, Choeronycteris, and Musonycteris the diastema between the first two molars is generally larger than that separating the second and third premolars. Still in relation to lower dentition, Dryadonycteris is similar to most Choeronycterini in having a low main cusp on the first lower premolar (fig. 9). In Anoura and in most Hylonycteris we were able to examine, however, the main cusp is conspicuously higher than the adjacent stylar cusps. In describing the first lower premolar of glossophagines, Phillips (1971) reported that the lingual surface of this tooth is concave in Hylonycteris, a condition that is shared only with Dryadonycteris.

The mandible of Dryadonycteris exhibits a unique feature: the mandibular condyle is positioned approximately at the same height as the upper border of the coronoid process, whereas in all other Choeronycterini the condyle is lower (more ventral) than the coronoid process (fig. 9). The projection of the anterior part of the mandible is directed relatively ventrally in most Choeronicterini (Scleronycteris, Lichonycteris, Choeroniscus, Hylonycteris), but projects straight forward in Dryadoncyteris. In most Choeronycterini, the angular process is also poorly developed (exception in Anoura and Musonycteris), and there is a secondary process on the ventral surface of the angular region. Dryadonycteris capixaba is similar to Lichonycteris in having a poorly developed angular process with no secondary process on the ventral aspect of this region.

Size is another important character in the discrimination of Dryadonycteris and other Choeronycterini. The range for forearm length of D. capixaba (29–32 mm) overlaps only with those for Lichonycteris (31–35.5 mm; Griffiths and Gardner, 2008), Hylonycteris (31–37 mm; Koopman, 1994), and Choeroniscus (26.5–42 mm; Genoways et al., 1973; Griffiths and Gardner, 2008). Some members of Anoura (33–48 mm; Griffiths and Gardner, 2008) also are near Dryadoncyteris in forearm length, but the lowest value reported for the smaller species living in eastern Brazil, Anoura caudifer, is 35 mm (Taddei, 1975; Dias et al., 2002; Dias and Peracchi, 2008), which is significantly larger than D. capixaba. The poorly known Scleronycteris ega seems to be similarly larger (34.5–35.5 mm; Griffiths and Gardner, 2008; E. Bernard, personal commun.), and the other Choeronycterini are much larger than Dryadonycteris, e.g., Choeronycteris (42–47 mm) and Musonycteris (40–43 mm; Koopman, 1994). Lichonycteris degener, which according to Griffiths and Gardner (2008) is the only representative of this genus in eastern Brazil, can be confused with Dryadonycteris capixaba in forearm size, but these two species exhibit no overlap in cranial measurements (tables 1–2). Dryadonycteris has a longer but thinner skull, which is clearly reflected in linear measurements related to these aspects (e.g., condylocanine length, maxillary toothrow length, and braincase breadth). Hylonycteris is from Central America and will not be considered in detail here, but Choeroniscus deserves particular attention. The relevant species in this case is C. minor, which occurs in eastern Brazil (Aguiar et al., 1995; Faria et al., 2006) and is potentially syntopic with Dryadonycteris capixaba. Specimens of Dryadonycteris described here are consistently smaller than specimens of Choeroniscus collected by Simmons and Voss (1998) in French Guiana, but agree in all respects with those found at an Atlantic Forest site in the state of Minas Gerais, 200 km westward from the type locality of Dryadonycteris capixaba (Aguiar et al., 1995) (tables 1–2). It may be the case that Choeroniscus minor is geographically variable in size—smaller in the southern part of its range—but a taxonomic reevaluation of the sample from Minas Gerais is advisable.