Character 1

Pelage differentiated into over hair and under hair (0); or pelage uniform, over hairs apparently absent (1). In most phyllostomids, the pelage is of uniform length and density (Benedict, 1957) and the over hairs are apparently absent. In contrast, the pelage is clearly differentiated into fine, wavy under hairs and thicker, straighter, longer over hairs in Desmodus, Diaemus, Phyllostomus, Phylloderma, Brachyphylla, phyllonycterines, Lionycteris, Ardops, and Sturnira (Benedict 1957). The pelage is also differentiated into under hairs and over hairs in mormoopids and Noctilio (Benedict, 1957), suggesting that this is the primitive condition for phyllostomids.

Benedict (1957) described differentiation of the pelage and Straney (1980: character F2) first used this character in a component analysis. Our character states are equivalent to Straney's (1980) characters; however, we disagree with his characterization of desmodontines and brachyphyllines (s.l.). In his data matrix, Straney (1980) scored desmodontines “+”, indicating that all desmodontines share the derived condition of differentiated pelage; however, Diphylla does not have differentiated pelage. Our scoring reflects Benedict's (1957) original descriptions. In addition, Straney (1980) scored brachyphyllines “++”, indicating that not all members of the taxon have the derived state (pelage differentiated into under hairs and over hairs). However, according to Benedict (1957), Brachyphylla, Erophylla, and Phyllonycteris all have pelage that is differentiated into under and over hairs.

We based our conclusion that the over hairs are absent in bats with undifferentiated pelage on a comparison of cuticular scale types. In most bats with distinct over and under hair, the alternate annular arrangement of hair scales (see character 3) does not occur on the over hair; thus, it may be that in phyllostomids in which over and under hair cannot be distinguished, the over hair may be absent.

Character 2

Basal bulb on hair shaft absent (0); or present (1). In most phyllostomids, the proximal third of the hair filament is of roughly uniform diameter and there is no basal swelling, or bulb (Benedict, 1957). In contrast, a swollen basal bulb is present on the proximal third of the hairs in Desmodus and Diaemus, some phyllostomines (e.g., Lonchorhina, Tonatia), and some stenodermatines (e.g., Ametrida, Vampyrodes; Benedict, 1957; fig. 16, detail). In mormoopids and Noctilio, the basal bulb is absent (Benedict, 1957), suggesting that this is the primitive condition for phyllostomids.

Straney (1980: character F1) first used this character in a component analysis. Our character states are identical to his, although our scoring of desmodontines is different. Benedict (1957) described Diphylla as lacking the basal bulb that is present in the other two desmodontine genera. Although Straney (1980: 35) accurately described the taxonomic distribution of this character in his text, he scored desmodontines “+” in his matrix, indicating that all desmodontines are derived in having a basal bulb. Our scoring reflects Benedict's (1957) description.

Character 3

Consecutive cuticular scales appressed to hair shaft (0); or entire margin equally divergent (1); or consecutive scales with most divergent margins on alternate sides (“alternate annular” condition) (2). In most phyllostomids, each cuticular scale encircles the entire hair shaft and the bodies of successive scales diverge most from the opposite sides of the filament (Benedict, 1957). This type of arrangement, known as “alternate annular,” makes every other scale appear “hastate” (with a slight V-shaped emargination) on any given side of the filament, while the remaining scales appear to lack the V-shaped emargination (Benedict, 1957; see character 4 below; fig. 16). In contrast, the scales are completely appressed to the filament in the desmodontines (fig. 16); there is no difference in degree of divergence between the sides of each scale (Benedict, 1957). In Mormoops, consecutive cuticular scales have margins that are entirely divergent from the shaft; there is no difference in degree of divergence between the sides of each scale (Benedict, 1957; fig. 16). In Pteronotus and Noctilio, consecutive cuticular scales are appressed to the shaft (Benedict, 1957). This distribution of character states suggests that the appressed scale pattern seen in Pteronotus and Noctilio may be the primitive condition for phyllostomids.

Straney (1980: characters F3–F5) first used the divergence of the margin of the cuticular scales as a character in a component analysis, based on Benedict's (1957) descriptions. Straney (1980) defined three binary characters whose derived conditions are equivalent to our character states. However, Straney (1980) scored taxa for the alternate annular character (F5) as well as one of the other two characters (F3 or F4, scales either divergent or appressed). Benedict (1957) characterized the degree of divergence of alternate annular scales from the shaft along what appears to be a continuum (appressed, slightly divergent, divergent, divaricate). Because of the continuous nature of this description, and the fact that the alternate annular condition describes a particular kind of divergence of the cuticular scales from the shaft of the hair, we did not include the additional information Straney (1980) used in his characters F3 and F4 in our version of this character.

Straney (1980) was not clear about which type of hair or which part of it he used to make his comparisons. Benedict (1957) noted that cuticular scale type changes over the length of the filament, and may be different on over hair and under hair, when these can be identified. Although Straney (1980) stated that all phyllostomids have entire, irregular emarginate, or denticulate scales alternating with hastate cuticular scales, this “alternate annular” condition occurs only on the middle part of the under hairs in Phylloderma (Benedict, 1957). On the over hairs and the tip of the under hairs, the cuticular scales are sinuate to erose coronal. Benedict (1957: 287) noted that only those scale types in the midregion of the hair shaft are deemed to be “mature” cuticular scale types. For this reason we used her descriptions of the midregion of all hairs, or the under hairs (when these could be identified; see character 1 above), to score this character and character 4.

Character 4

Majority of scale margins on each hair entire (0); or irregular (1); or toothed (2); or entire and irregular (3); or entire and hastate (4). In most phyllostomids, cuticular scale margins are entire (scales with a smooth straight margin; Benedict, 1957). However, in Desmodus, Diaemus, Phylloderma, Phyllostomus, and Ardops, the scale margins are irregular (Benedict, 1957; fig. 16). In Lonchorhina, Macrophyllum, Anoura, Leptonycteris, Lionycteris, and many stenodermatines (e.g., Ariteus, Platyrrhinus), the scales have toothed margins (Benedict, 1957; fig. 16). Chrotopterus seems to be unique among phyllostomids in consistently possessing scales of two different morphologies on a single filament: both entire and irregular shaped scales appear on individual hairs (Benedict, 1957). We scored this taxon with state 3 in the matrix. Phyllonycteris aphylla has irregular scale margins while Phyllonycteris poeyi has entire scale margins. We scored the genus Phyllonycteris with states 0 and 1 in the matrix. In Mormoops and Pteronotus quadridens, the scale margins are toothed (Benedict, 1957), whereas Pteronotus parnellii has both entire and hastate scale margins on a single filament (Benedict, 1957; note that this is not the alternate annular condition described in character 3 above). In Pteronotus gymnotus and Noctilio the scale margins are entire (Benedict, 1957). We scored Pteronotus with states 0, 2, and 4 in the matrix. The primitive condition for phyllostomids cannot be assessed a priori given the distribution of these states in the outgroup taxa.

Benedict (1957) first described the morphology of the scale margins on bat hairs; this character has not appeared in previous phylogenetic analyses. Benedict (1957) described considerably more variation than we have recognized. For example, Benedict (1957) described our “irregular” scale margins (state 1) as being “repand” (slightly irregular at some points along the margin), “sinuate” (irregular along the entire scale margin with deeper notches), or “erose” (very irregular, with some sides of the scale being longer proximodistally than others). These scale types seem to illustrate states on a continuum. Thus, they may not represent discrete character states. For this reason, we decided to pool such states under the condition “irregular.” Similarly, we included both “dentate” and “denticulate” scale types within the character state “toothed” (state 2). We also scored taxa in which some scale margins occasionally differed from the most common margin type as having a single state because it seemed likely that occasional deviations may have resulted from damage to the scale margin (particularly because these deviations were almost always “irregular” in nature).

Character 5

Dorsal fur unicolored (0); or distinctly bicolored, hairs with pale bases and dark tips (1); or tricolored, hairs with distinct dark bases, a pale median band, and dark tips (2). In most phyllostomids, the dorsal fur is distinctly bicolored, with dark tips and paler bases. Especially in phyllostomines with this feature, the pale bases of the hairs tend to be most evident over the shoulders and upper back and appear less distinct caudally. The white base may occupy more than one-half of the hair shaft (e.g., Macrotus) or be confined to a narrow stripe at the base of the fur (e.g., Lonchorhina). In contrast, Micronycteris nicefori, M. sylvestris, Hylonycteris, Lichonycteris, Carollia, and many stenodermatines (e.g., Ametrida, Mesophylla) have dorsal fur that is tricolored. A dark basal band is succeeded by a pale stripe and a dark tip. Micronycteris brachyotis, Phyllostomus hastatus, Monophyllus, and Lionycteris are the only phyllostomids that have unicolored fur (dark from tip to base). Because other species of Phyllostomus have bicolored fur, we scored this genus 0 and 1 in the matrix. The dorsal fur in Mormoops has a pale base and darker tip, but in Pteronotus and Noctilio the dorsal fur is unicolored. This distribution of character states leads us to conclude that unicolored fur is primitive for phyllostomids. Although we have not scored Ectophylla for most of the characters related to pelage color, we did score this taxon for this character. In Ectophylla, the posterior half of the dorsum is a grayish-brown color, and the individual hairs in this region have pale bases and a darker tip.

Owen (1987: character 3) used an ordered character with states identical to ours, later modifying it slightly when he restricted his analysis to a subset of original taxa surveyed (Owen, 1991: character 7). Our observations agree with those of Owen (1987) with a few exceptions. Owen (1987) described the dorsal pelage of Platyrrhinus dorsalis, which we observed to have a pale base and dark tip like other Platyrrhinus species, as tricolored. Owen (1987) also scored Ectophylla as having unicolored fur. Ectophylla is unique among phyllostomids in having a white head and shoulders and a brownish lower body. The hairs on the anterior of the dorsum are white from base to tip, while the brownish fur on the lower body has a brown tip and pale base. We suggest that the presence of “unicolored” white fur on the anterior of the dorsum of Ectophylla alba is an autapomorphy of this species, and that the hair on the lower body that has a pale base and dark tip reflects a condition that Ectophylla may potentially share with other phyllostomids. Finally, Owen (1987) noted that Chiroderma improvisum has dorsal fur with a pale base and dark tip (in contrast to other Chiroderma species which have tricolored fur). We were unable to examine this species; however, Jones and Baker (1980: 1) described Chiroderma improvisum as having individual hairs that are “pale brownish gray throughout most of their length, slightly darker basally, and tipped with a dark, rich brown.” A juvenile individual examined by the same authors had distinctly tricolored fur (Jones and Baker, 1980). We therefore score Chiroderma as possessing tricolored fur. Owen's (1991) observations are identical to those of Owen (1987). Both Simmons (1996: character 1) and Marques-Aguiar (1994: character 4) employed a similar character in their analyses of Micronycteris and large-bodied Artibeus, respectively; our observations agree with theirs.

Character 6

Facial fur of uniform color (0); or facial stripes present (1). In most phyllostomids, facial fur is of uniform color and no facial stripes are present. In contrast, many stenodermatines (e.g., Artibeus, Uroderma) have facial stripes. There are two stripes on each side of the face in most taxa: a more intensely colored dorsomedial stripe, and a lighter ventrolateral stripe. The dorsomedial stripe begins at the posterolateral edge of the noseleaf and runs superior to the eye, generally terminating at some point along the medial external surface of the pinna. The ventrolateral stripe begins along the corner of the mouth and generally runs toward the tragus, ending at its base. Stripes vary in length, width, and color intensity within and between genera. Although the presence of dorsomedial stripes is invariant in genera, ventrolateral stripes are not always present in all species of a genus. For example, Chiroderma trinitatum has two bright pairs of facial stripes, whereas Chiroderma villosum lacks the ventrolateral pair and has lightly colored bands in the region of the dorsomedial stripes. Nevertheless, all species of Chiroderma have at least dorsomedial stripes. Other taxa that exhibit similar variation in the presence of ventrolateral facial stripes are Artibeus, Dermanura, Platyrrhinus, Uroderma. Mormoopids and Noctilio lack facial stripes, suggesting that this is the primitive condition for phyllostomids. We scored Ectophylla “−” for this character (see below).

Owen (1987: character 4) used facial stripes as part of a complex multistate character that he termed “pelage pattern.” This character incorporated facial striping, dorsal striping (see character 7 below), shoulder patches (see character 8 below), and striping on the wing (an autapomorphy of Centurio). Owen (1987) ordered the character to preserve potential homologies. In general (see below) we agree with Owen's (1987, 1991) scoring of facial striping; however, Owen's (1987) description of character state 2 suggests that all taxa with facial stripes have paired stripes, rather than the situation we described above. One of our disagreements with Owen's (1987) scoring of facial stripes concerns Artibeus; Owen (1987) reported that A. fuliginosus (= A. obscurus) lacks facial stripes whereas we observed faint dorsomedial facial stripes in all A. obscurus we examined. We also scored Ectophylla as “−”, whereas Owen (1987) regarded the pelage as uniformly colored. Owen (1991: character 8) considered facial stripes as a separate character for an analysis of a subset of stenodermatine taxa; however, scoring did not differ from his earlier work.

Lim (1993: character 1) used facial striping in his analysis of stenodermatine relationships. Although our descriptions of character states are identical to Lim's (1993), we do not agree with all his assessments concerning the taxonomic distribution of facial stripes. Lim (1993) noted that live specimens of Koopmania and Mesophylla have faint facial stripes similar to those in Artibeus obscurus. We could distinguish no trace of facial stripes in Mesophylla, either in color slides of live bats (AMNH 267281), or in well-preserved alcohol and skin preparations (e.g., AMNH 267281, AMNH 267563). Although our examinations of some color slides of live Koopmania suggested medial facial stripes were present in this genus, we believe that these “stripes” are the pale bases of the fur in this region, or frosted patches, because our examinations of skins of the bats appearing in these photos (e.g., AMNH 267193) did not reveal any facial striping. Frosted patches of fur are located on the forehead, face, neck, and shoulders of Mesophylla, and some specimens of Koopmania (e.g., AMNH 266267) also have frosted patches between the eyes, posterior to the noseleaf. These frosted patches may have been interpreted by Lim (1993) as facial stripes, but we do not believe they are homologous. Thus, we agree with Owen's (1987, 1991) assessment that Mesophylla and Koopmania lack facial stripes.

Lim (1993) scored Ectophylla as having facial and dorsal stripes and lacking a shoulder spot, whereas Owen (1987) scored the taxon as having uniformly colored pelage. We disagree with both scoring schemes. More than three quarters of the body of Ectophylla is grayish white; only the lower back is brown. Lim (1993) may have scored Ectophylla as possessing facial stripes because of a brown patch that extends from the lateral part of the noseleaf over the eye, but does not reach the ear. This brown fur visually separates the continuous white of the head into two parts that correspond to the areas in which the dorsomedial and ventrolateral stripes appear in other taxa. However, this color pattern is not strictly comparable to that seen in other phyllostomids with facial stripes, and we found it impossible to determine if stripes are unambiguously present. Accordingly, we scored Ectophylla “−” for this character.

Character 7

Dorsal stripe absent (0); or always present (1); or sometimes present; polymorphic within species (2). A pale dorsal stripe is present in the fur of one phyllostomine (Mimon crenulatum) and several stenodermatines (e.g., Platyrrhinus, Uroderma). In Micronycteris nicefori, a very pale dorsal stripe is present in most individuals (e.g., AMNH 266019) whereas it appears to be absent in the fur of others (e.g., AMNH 266015). We scored this species with state 2 in the matrix. Dorsal stripes usually begin at a point between the shoulder and the anterior margin of the medial pinnae, extend down the center of the back, and always reach the base of the tail. Length and color intensity of the dorsal stripe may vary among and within genera. For example, within Vampyressa, V. pusilla has no dorsal stripe, V. brocki has a faint stripe present on the lower back to the base of the tail, and V. bidens has a distinct stripe present from a point between the pinnae to the base of the tail. Such differences also exist among species of Chiroderma and Platyrrhinus, although no species in the latter genus completely lacks a dorsal stripe. Because some species of Chiroderma lack a dorsal stripe, whereas in others this feature is present, this genus is scored 0 and 1 in the matrix. As noted above (see character 5), the fur of the lower back is brown in Ectophylla and would therefore be expected show evidence of dorsal striping (if present) since dorsal stripes always reach the base of the tail. There is no evidence of a dorsal stripe in Ectophylla. In mormoopids, dorsal stripes are absent; however, some individuals of both Noctilio species have a pale to strongly colored dorsal stripe. Noctilio is scored with state 2 in the matrix. The primitive state for phyllostomids cannot be assessed a priori.

Owen (1987: character 4) originally used this feature in his external character “pelage pattern,” although he included several additional states to accommodate variation in stripe length. Because stripe length varies among species within genera, we choose to omit this information from our study. Our scoring of this character agrees with Owen's (1987) observations, with the exception of Ectophylla, which Owen (1987) scored as possessing unicolored fur.

Character 8

Fur on shoulder of uniform color, no white shoulder spot (0); or distinct white shoulder spot present (1). In most phyllostomids, the shoulder region is generally uniform in color and there is no distinct white spot. Although some specimens may appear to have pale patches on the shoulders, these patches do not have well-defined edges and are either extensions of the lighter ventral fur into the shoulder region (e.g., Chrotopterus AMNH 267131), or the pale bases of bicolored fur that are visible when the fur is slightly parted (e.g., Tonatia silvicola AMNH 267108). In Sturnira, many adults of both sexes have shoulder glands which produce a waxy secretion that may stain the fur in the shoulder region, producing what are termed “epaulettes” (Davis et al., 1964), but these patches are usually bright orange or brown, not white. In contrast, some stenodermatines (e.g., Ametrida, Sphaeronycteris) have a small, clearly defined patch of white fur on the shoulder. No white spot is present in mormoopids and Noctilio, suggesting that this is the primitive condition for phyllostomids. We scored Ectophylla “−” for this character because the shoulders in this species are completely white (see discussion under character 5 above).

Owen (1987: character 4; 1991: character 9) included shoulder spots in his phylogenetic analyses. Our character states and scoring are identical to those of Owen (1991) and Lim (1993: character 2).

Character 9

Fur on neck of uniform color, no white neck spot (0); or distinct white spot sometimes or always present on neck (1). In most phyllostomids, the area of the neck ventral and caudal to the lateral part of the pinna is roughly the same color as the fur on the neck and shoulders, and no white spot is present. In contrast, a white spot is present in Ametrida, Centurio, Phyllops, Pygoderma, Stenoderma, and Sphaeronycteris. In Pygoderma, the spot appears to be absent in some specimens (e.g., AMNH 234296). In Centurio, the spot is present on the face mask (see character 38 below). In the outgroups, mormoopids and Noctilio, no neck spot is present, suggesting that this is the primitive condition for phyllostomids. We scored Ectophylla, which has a completely white neck, “−” (see character 5 above).

This character has not been used in previous studies. We have chosen to treat polymorphism in this character conservatively (i.e., the derived condition is sometimes or always present) because Peterson (1965: 8), who was able to examine many more representatives of Ametrida than we did, found that this spot is not always present in this species. Our observations indicate that the spot varies in intensity, leading us to suspect that the spot may be polymorphic in more taxa than we report here.

Character 10

Uropatagium without fringe of hair along trailing edge (0); or with distinct fringe of hair along trailing edge (1). The phyllostomid uropatagium usually does not have a distinct fringe of hair that extends beyond the trailing edge of membrane. In contrast, a distinct fringe is present along most of the caudal edge of the uropatagium in desmodontines, Macrophyllum, Mimon crenulatum, Anoura, Leptonycteris, and many stenodermatines (e.g., Ardops, Platyrrhinus). In Rhinophylla fischerae, Artibeus hirsutus, Dermanura azteca, and D. tolteca, distinct fringes are present, whereas other species in these genera lack a fringe. We scored these genera with states 0 and 1 in the matrix. The uropatagium lacks a fringe in mormoopids and Noctilio, suggesting that this is the primitive condition for phyllostomids.

Lim (1993: character 13) used the presence of a fringe on the uropatagium as a character, erecting two states: “Edge of uropatagium naked or sparsely and irregularly haired (0); edge of uropatagium densely and evenly haired (1).” Only Platyrrhinus and Vampyrodes were scored with the derived condition. In contrast, we found dense and evenly haired fringes in many additional stenodermatine taxa.

Character 11

Superciliary vibrissae absent (0); or always present (1). Most phyllostomids lack superciliary (= supraorbital) vibrissae (fig. 17D–E), a group of one or more vibrissae that occurs above each eye in most mammalian groups (Pocock, 1914; Brown, 1971; fig. 17A). However, a single superciliary vibrissa is present above each eye in Desmodus (fig. 17B), Diaemus, some phyllostomines (e.g., Micronycteris megalotis, Vampyrum; fig. 17C), and one stenodermatine (Centurio). Taxonomic polymorphism exists in two genera: Lonchorhina (present in L. marinkellei, absent in other species), and Tonatia (present in T. silvicola and T. schulzi, absent in other species). We scored these genera with states 0 and 1 in the matrix. In one specimen of Chrotopterus (AMNH 267443), the superciliary vibrissae are absent, but these vibrissae are present in all other specimens of Chrotopterus we examined. We consider this an anomaly and score Chrotopterus with state 1 in the matrix. In some taxa (e.g., Micronycteris megalotis, Tonatia silvicola) the superciliary vibrissae are anterior to their position in other species. Rather than directly superior to the eye, they are anterior and superior to it. However, they occur in the same position relative to the other groups of facial vibrissae (fig. 17C). We therefore consider these vibrissae to be homologous to the superciliary vibrissae of other taxa. Superciliary vibrissae are absent in mormoopids, but present in Noctilio (fig. 17F); accordingly the primitive condition for phyllostomids cannot be reconstructed a priori.

Although this character has not been used in previous phylogenetic analyses, there is much evidence to suggest that homologies may be drawn among vibrissal groups, and potentially among individual vibrissae. The locations of various groups of vibrissae on the face is highly conservative among mammals, and presumably homologous groups or individual vibrissae may be identified on the basis of their relative position (Pocock, 1914; Brown, 1971; Wineski, 1985). Many and perhaps all facial vibrissae have individual representations in both the spinal trigeminal nucleus and the somatosensory cortex (Zucker and Welker, 1969; Waite, 1973a, 1973b, Woolsey, 1978), further supporting the idea that homologies may be drawn among individual vibrissae and vibrissal clusters in different taxa, as we do here.

Character 12

Genal vibrissae absent (0); or one vibrissa present in each cluster (1); or two genal vibrissae present in each cluster (2). Genal vibrissae occur on the sides of the face ventral and/or posterior to the eye (Pocock, 1914; Brown, 1971; fig. 17A). Genal vibrissae are present in most phyllostomids (fig. 17B–C, E). However, these vibrissae are absent in some phyllostomines (e.g., Macrophyllum, Mimon), Platalina, and several glossophagines (e.g., Glossophaga, Monophyllus). In most phyllostomids with genal vibrissae, two occur in each cluster. However, Brachyphylla, Erophylla, most glossophagines, some lonchophyllines, and Pygoderma have a single genal vibrissa on each cheek. Taxonomic polymorphism occurs in Lonchorhina (absent in L. fernandezi, two in L. aurita), Lonchophylla (fig. 17D; absent in L. mordax and L. thomasi, two in L. robusta), Anoura (one in A. caudifer and A. cultrata, two in A. geoffroyi), Choeroniscus (absent in C. minor [= C. intermedius], one in C. periosus), Carollia (absent in C. perspicillata, one in C. subrufa), and Sturnira (absent in S. bidens, one in S. bogotensis, two in other species we examined). We scored these taxa with the appropriate combinations of character states in the matrix (e.g., Lonchorhina is scored 0 and 2). In one specimen of Vampyrodes caraccioli (AMNH 239254), we found three genal vibrissae present in each cluster, whereas all other individuals we examined had two. We view this as an anomaly as all other phyllostomids we examined had no more than two of these vibrissae; consequently, we score Vampyrodes with state 2 in the matrix. The genal vibrissae are absent in Mormoops, whereas Pteronotus has a single vibrissa on each cheek and Noctilio has two genal vibrissae in each cluster (fig. 17F). The primitive state for phyllostomids cannot be reconstructed a priori.

This character has not appeared in previous phylogenetic analyses.

Character 13

Interramal vibrissae always absent (0); or none or one interramal vibrissa present, polymorphic within species (1); or one interramal vibrissa always present (2); or one or two interramal vibrissae present, polymorphic within species (3); or two interramal vibrissae always present (4); or none or two interramal vibrissae present, polymorphic within species (5); or three interramal vibrissae always present (6). Interramal vibrissae occur between the rami of the lower jaws well posterior to the mandibular symphysis in most mammals (Pocock, 1914; Brown, 1971; fig. 17A). Two interramal vibrissae are present in most phyllostomids (fig. 17B–C). However, interramal vibrissae are apparently uniformly absent in some phyllostomines (e.g., Chrotopterus, Vampyrum) and many stenodermatines (e.g., Pygoderma, Uroderma; fig. 17E). The three lonchophylline genera appear to be unique among phyllostomids in possessing three interramal vibrissae (fig. 17D). Choeronycteris and Musonycteris have a single interramal vibrissa. Among other genera, there is a large amount of both taxonomic and intraspecific polymorphism. Several species exhibit intraspecific variation: Lonchorhina aurita (absent or two), Macrophyllum macrophyllum (absent or one), Phyllostomus hastatus (absent or one), Trachops cirrhosus (absent or one interramal vibrissae), Choeroniscus intermedius (absent or one), Chiroderma villosum (absent or one), and Koopmania concolor (one or two), Platyrrhinus aurarius (absent or one). Taxonomic polymorphism occurs frequently; interramal vibrissae appear to be uniformly absent in Tonatia saurophila, Sturnirabidens, S. nana, and S. tildae whereas the congeners of these species have two interramal vibrissae. We scored these two genera with states 0 and 4 in the matrix. In Phyllostomus discolor, Chiroderma trinitatum, Platyrrhinus dorsalis and P. helleri, the interramal vibrissae are absent whereas other species in these genera are polymorphic (absent or one vibrissa). We scored these three genera with states 0 and 1 in the matrix. In Lonchorhina, L. fernandezi has two interramal vibrissae, whereas L. aurita is polymorphic (absent or two). We scored this genus with states 4 and 5 in the matrix. In Choeroniscus minor these vibrissae are either absent or a single vibrissa is present, whereas in C. periosus a single vibrissa is present. We scored the genus Cheroniscus with states 1 and 2 in the matrix. In Dermanura the interramal vibrissae are absent in D. cinerea, whereas D. anderseni has one and D. tolteca has two. We scored this genus with states 0, 2, and 4 in the matrix. Two interramal vibrissae are present in Mormoops and Noctilio (fig. 17F), whereas Pteronotus has one. This distribution of character states suggests that two interramal vibrissae are primitive for phyllostomids.

This character has not been used in previous phylogenetic analyses.

Character 14

Vibrissae lateral to nose/noseleaf arranged in two columns; medial column with three or more vibrissae, lateral column with two vibrissae (0); or single column with three or more vibrissae present, lateral column absent (1). In desmodontines, many phyllostomines (e.g., Chrotopterus, Phylloderma), Brachyphylla, phyllonycterines, glossophagines, and lonchophyllines, the vibrissae lateral to the noseleaf are arranged in two columns: a medial column (nearest the noseleaf), consisting of three or more vibrissae (see character 15), and a lateral column, composed of only two vibrissae (fig. 17B–D). The two vibrissae in the lateral column occur on the dorsum of the snout more or less posterior to the first and second vibrissae in the medial column (numbering from dorsal to ventral). The lateral column of vibrissae is absent in several phyllostomines (e.g., Macrophyllum, Phyllostomus), carolliines, and stenodermatines (figs. 17E, 18A–C). The lateral column is present in mormoopids and Noctilio (figs. 17F, 18D), suggesting that this is the primitive condition for phyllostomids.

This character has not appeared in previous phylogenetic analyses.

Character 15

Vibrissae in column adjacent to noseleaf/nose number two (0); or three (1); or more than 3 (2). In most phyllostomids there are more than three vibrissae in the column immediately adjacent to the noseleaf. These vibrissae are extremely long and well developed, with swollen bases (or pads as described in character 16 below; fig. 18A–B). In contrast, in Ametrida, Centurio, and Sphaeronycteris there are only three of these well-developed vibrissae surrounding the noseleaf (fig. 18C). In mormoopids and Noctilio, there are two vibrissae in this column surrounding the nose (due to the morphology of the face in mormoopids these vibrissae are posterolateral to the nose; fig. 18D). The presence of two vibrissae in the column lateral to the noseleaf appears to be primitive for phyllostomids.

This is the first use of this character in a phylogenetic analysis.

Character 16

Vibrissae surrounding the nose/noseleaf arise from small, separate vibrissal papillae (0); or vibrissal papillae connected, form padlike structure (1); or vibrissal papillae connected, form free skin flap (2); or vibrissal papillae separate, form elongate, fleshy, cylindrical projections (3). In most phyllostomids, the papillae of the individual vibrissae located immediately adjacent to the noseleaf are connected with each other, forming two padlike structures, one on each side of the noseleaf (fig. 18A). These pads incorporate the vibrissal papillae in the medial and, when present, the lateral columns described in character 14. In contrast, the vibrissal papillae form fleshy skin flaps in a several taxa: Rhinophylla, Ametrida, Ardops, Ariteus, Phyllops, and Stenoderma (fig. 18B). In Centurio and Sphaeronycteris, the vibrissal papillae are not connected with each other, and the papillae instead form elongate, fleshy cylinders that protrude from the face (fig. 18C). Chrotopterus, Tonatia, Trachops, and Vampyrum also have separate vibrissal papillae, but although these are often swollen, they are never elongate and cylindrical, nor do they form pads. In mormoopids and Noctilio, the vibrissal papillae are small and do not form padlike, flaplike, or cylindrical structures (fig. 18D); this distribution of character states suggests that the presence of small, separate vibrissal papillae is the primitive condition for phyllostomids.

This is the first use of this character in a phylogenetic analysis.

Character 17

Padlike or flaplike vibrissal papillae not in contact across dorsum of snout (0); or pads touch, or are confluent across dorsum of snout (1). Most taxa with padlike or flaplike vibrissal papillae have two discrete, clearly separated, pads or flaps on either side of the noseleaf. These structures are not in contact, nor are they continuous, across the dorsum of the snout. In contrast, in Brachyphylla, Erophylla, and Phyllonycteris the padlike structures meet, or are continuous, posterior to the noseleaf. In these three genera, the vibrissae that are generally more laterally placed in other phyllostomids due to the wider separation of these pads, are located somewhat more posteriorly. Mormoopids and Noctilio do not have padlike or flaplike like vibrissal papillae; therefore, we scored these taxa “−” for this character. The primitive condition for phyllostomids cannot be reconstructed a priori. Taxa scored 0 and 3 in character 16 are also scored “−” for this character.

This is the first use of this character in a phylogenetic analysis.

Character 18

Noseleaf pale cream to dark brown, not strongly bicolored (0); or noseleaf yellow (1); or noseleaf bicolored, well-defined cream patches present on proximal parts of the lateral lobes and edges of horseshoe, center of horseshoe and spear brown (2). In most phyllostomids the skin of the noseleaf is a shade of pale to dark brown, generally the same color as the lips and face. Although the noseleaf appears lighter in color in some taxa (e.g., Chrotopterus), the range of colors seen in different species and within populations of a single species varies to such a degree that they cannot be coded as discrete character states. Mimon crenulatum presents an excellent example of the range of variation seen within species. The noseleaf in AMNH 267112 is a pale, creamy pink color; however, other individuals from the same locality have noseleaves that range from light brown around the edges of the spear with pale creamy pink in the center and on the horseshoe (AMNH 267883) to entirely brown (AMNH 267888). We thus recognize “pale cream to dark brown” as a single character state. Ectophylla and Mesophylla are unique among phyllostomids in having a noseleaf that is yellow in life (this generally fades to a much paler yellow color in preserved specimens). Uroderma, Vampyressa bidens, V. nymphaea, V. pusilla, and Vampyrodes exhibit yet another pattern, a consistently bicolored noseleaf. In these taxa the central rib of the noseleaf is always brown, but the lateral lobes and horseshoe edges have distinct patches of cream or yellow. In some species of Chiroderma (e.g., C. trinitatum), Dermanura (e.g., D. cinereus), and Platyrrhinus (e.g., P. helleri) the noseleaf is also distinctly bicolored, but congeners of these species have a brown or cream noseleaf. Thus, these three genera are scored 0 and 2 in the matrix. We scored mormoopids and Noctilio, which clearly lack a noseleaf, “−” for this character; thus, the primitive state for phyllostomids cannot be assessed a priori. We also scored Centurio “−” for this character (see below).

This is the first use of this character in a phylogenetic analysis. Although one might imagine that state 2 might be correlated with presence of strong facial striping (see character 6), our observations suggest that this is not the case. In Artibeus lituratus, for example, strong facial stripes are present but the noseleaf is uniformly brown. However, we did not use Lim's (1993: character 9) character for yellow coloration of the thumb because it has an identical distribution with the yellow noseleaf (found only in Ectophylla and Mesophylla); this suggests that these characters may not be independent.

Although presence of a noseleaf is generally regarded as a diagnostic feature of Phyllostomidae, a noseleaf with a well-defined horseshoe and spear may not be present in all phyllostomids. Although some authors have indicated that a noseleaf is present in Centurio (e.g., Hill and Smith, 1984; Lim, 1993), others have suggested that this taxon lacks a true noseleaf (Miller, 1907; Paradiso, 1967; Snow et al., 1980). We found that a spearlike structure is present between the nostrils in Centurio, but it is not confluent internarially with a horseshoe. Instead, this structure forms the medial part of the upper lip. The horseshoe portion of the noseleaf (if present) surrounds only the lateral portions of the nostrils. One way to settle the question of the homology of these structures might be to examine ontogeny of the narial region, but this has not yet been attempted. Pending such a study, we chose to score Centurio “−” for all characters related to the noseleaf.

Character 19

Noseleaf spear long, greater than twice the height of the horseshoe (0); or spear truncated, equal to or less than the height of the horseshoe (1). The phyllostomid noseleaf is divided into two parts, the horseshoe and the spear. The horseshoe surrounds the lateral and inferior aspects of the nostrils. The spear is a fleshy protuberance that rises off the surface of the snout from the superior aspect of the nostrils. The spear, which is continuous with the horseshoe both internarially and laterally, is generally delineated in some way from the alae of the horseshoe. The spear of the noseleaf is long and tapers to a point in most phyllostomids. In terms of relative proportions, the spear (as measured from the inferior margin of the nostril to the tip of the spear) is greater than twice the height of the horseshoe (as measured from inferior to superior horseshoe border; e.g., fig. 19B). In contrast, the spear is much shorter (truncated) in desmodontines, Brachyphylla, and phyllonycterines (figs. 19A, 20A). In these taxa, relative height of the spear ranges from equal to less than half the height of the horseshoe. We scored mormoopids and Noctilio, which clearly lack a noseleaf, “−” for this character; thus, the primitive condition for phyllostomids cannot be determined a priori. We also scored Centurio “−” for this character.

Owen (1987: character 2, 1991: characters 3–5) and Marques-Aguiar (1994: character 9) both described the length of the noseleaf in relation to its width, within stenodermatines and Artibeus (Artibeus), respectively. We have redefined this character to account for the variation we observed among all phyllostomids.

Character 20

Spear of noseleaf with pointed or rounded distal tip (0); or with U-shaped notch in distal tip (1). The spear of the noseleaf has a pointed or rounded distal tip in most phyllostomids (figs. 19B–C, 20A–C, 21A–C). In contrast, a U-shaped notch is present in the distal tip of the noseleaf in desmodontines and Brachyphylla (fig. 19A). Presence of this notch may be dependent upon truncation of the spear (see character 19 above), but these features are not fully correlated. Phyllonycterines, which have a truncated spear, lack a notch in the distal tip (fig. 20A). We scored mormoopids and Noctilio, which clearly lack a noseleaf, “−” for this character; thus, the primitive condition for phyllostomids cannot be determined a priori. We also scored Centurio “−” for this character.

This character has not appeared in previous phylogenetic analyses.

Character 21

Central rib absent (0); or rib restricted to proximal part of spear (1); or rib extends to distal tip of spear (2). A thick, fleshy central rib is a prominent feature of the noseleaf of many phyllostomids. However, the central rib is absent and the noseleaf is flat in desmodontines, Macrotus, Brachyphylla, phyllonycterines, Platalina, and all glossophagines (figs., 19A, 20A, 20C). When the rib is present, it originates between the nostrils and extends distally into the spear. The rib extends into the proximal part of the spear but does not reach the tip in species of Micronycteris, Lionycteris, Lonchophylla, carolliines, Sphaeronycteris, and Sturnira (figs., 20B, 21A). In these taxa, the rib is well defined laterally but not distally, where it loses thickness and grades into the proximal surface of the spear. In contrast, the rib extends to the tip of the spear in most stenodermatines and phyllostomines (figs. 19B–C, 21B–C). We scored mormoopids and Noctilio, which clearly lack a noseleaf, “−” for this character; thus, the primitive condition for phyllostomids cannot be determined a priori. We also scored Centurio “−” for this character.

This is the first use of this character in a phylogenetic analysis.

Character 22

Internarial region smooth, no midsagittal ridge or papillae (0); or narrow fleshy ridge or line of papillae always present along midsagittal line (1); or internarial ridge or papillae variably present, polymorphic within species (2). A variety of structures occur between the nostrils in phyllostomids. The central rib of the spear continues into the internarial region in all taxa that have a rib (see character 21 above; figs., 19B–C, 20B, 21A–C). In most, the rib is the only structure present between the nostrils and it presents a smooth, slightly convex surface in this region. The internarial region is generally smooth (although flatter) in taxa that lack a rib on the spear (desmodontines, Macrotus, brachyphyllines, phyllonycterines, Platalina). In contrast, a well-delineated, narrow, fleshy ridge or line of papillae is consistently present along the midsagittal line of the internarial region in some phyllostomines (e.g., Lonchorhina, Vampyrum), Lionycteris, Lonchophylla, glossophagines, and Carollia (figs. 19B, 20B–C, 21A). In one phyllostomine species that we examined (Micronycteris hirsuta), two out of 10 specimens exhibited an internarial ridge; consequently, we score this taxon with state 2 in the matrix. We did not detect any other within-species polymorphism in any other taxon. We scored mormoopids and Noctilio, which clearly lack a noseleaf, “−” for this character; thus, the primitive condition for phyllostomids cannot be determined a priori. We also scored Centurio “−” for this character.

Presence/absence of an internarial ridge or line of papillae is not correlated with presence/absence of the central rib; both are present in some taxa (e.g., Vampyrum) and absent in others (e.g., Platalina), whereas only a rib is present in some forms (e.g., Phyllostomus) and only an internarial ridge or line of papillae is present in others (e.g., Glossophaga). When both rib and ridge are present, the former is distinguished by its greater width and continuity with the rib of the spear; the internarial ridge is located on the rib between the nostrils in these taxa (e.g., Vampyrum).

The form of the internarial structures varies among taxa in which they are present (taxa scored 1 and 2 for this character). For example, Chrotopterus and Vampyrum always have a narrow, undivided internarial ridge, whereas a line of distinct papillae is present in the same location in other taxa (e.g., Macrotus, Trachops). Although we initially considered these to represent distinct conditions, further examination of glossophagines and lonchophyllines led us to discover a series of intermediates between the “ridge” and “papillae” conditions. Many glossophagines have a ridge that is subdivided by small grooves (e.g., Glossophaga), whereas in others the internarial structure more closely resembles a series of partly fused papillae (e.g., Lionycteris). The latter condition is also seen in some phyllostomines (e.g., Micronycteris hirsuta). Morphology of internarial structures also appears to vary within species. For example, within Lonchophylla robusta, some individuals (e.g., AMNH 235760) have a solid internarial ridge, whereas others (e.g., AMNH 235761) have an internarial ridge that consists of segments (three in this case) that appear to be elongate papillae. This type of variation leads us to conclude that the ridge and line of papillae represent endpoints of a continuum. As a result, we have not scored “ridge” and “papillae” as separate character states.

This character has not been used by other authors.

Character 23

Sella absent (0); or present (1). The sella (sensu Hernandez and Cadena, 1978) is a fleshy globular or trilobed structure that occurs on the horseshoe just inferior (or anterior) to the proximal end of the internarial ridge or line of papillae, with which it is continuous. Most phyllostomids lack a sella; it is present only in Chrotopterus, Lonchorhina, and Vampyrum. In Chrotopterus and Vampyrum (fig. 22A), the sella is a simple spherical structure. In contrast, in Lonchorhina the sella is a complex, trilobed structure with one anterior and two lateral lobes. It is located just inferior to the nostrils on the edge of the horseshoe (fig. 22B). Because the horseshoe is completely bounded inferiorly by an extensive flap of skin in Chrotopterus and Vampyrum (see characters 25 and 26), the sella appears to occur in the middle of the horseshoe; however, it maintains the same position relative to the nostrils as seen in Lonchorhina. We scored mormoopids and Noctilio, which clearly lack a noseleaf, “−” for this character; thus, the primitive condition for phyllostomids cannot be determined a priori. We also scored Centurio and those phyllostomids that are scored with state 0 of character 22 (no internarial ridge or line of papillae) with “−” for this character.

Straney (1980: character G15) described the morphology of the noseleaf in Lonchorhina and Macrophyllum as having “lateral and medial projections on basal plate.” However, our observations of this region in Macrophyllum indicate that most individuals do not have a trilobed structure (e.g., AMNH 222040). Therefore we do not consider differences in sella morphology as a separate character because we observed the trilobed sella in only one taxon.

Character 24

Lateral edges of horseshoe thin and free (0); or superior portion of swollen edge of horseshoe forms free, flaplike edge (1); or swollen lateral edges of horseshoe ridgelike, fused to face along entire length with no free edge (2). The lateral edges of the horseshoe consist of thin, free flaps of skin in most phyllostomids (figs. 19A–C, 20A, 21 A–C). In contrast, in some glossophagines (e.g., Choeronycteris, Hylonycteris) only the superior part of the lateral edge of the horseshoe forms a free, thickened flap, while the inferior part of the horseshoe is fused to the face (fig. 20C). In lonchophyllines and some glossophagines (e.g., Anoura, Glossophaga), both the superior and inferior parts of the horseshoe are fused to the face and there is no free edge (fig. 20B). In some species of Tonatia (e.g., T. silvicola), both the superior and inferior portions of the horseshoe are fused to the face, whereas in other species (e.g., T. saurophila) the entire lateral edge of the horseshoe is a thin, free flap. Consequently, we score Tonatia with states 0 and 2 in the matrix. We scored mormoopids and Noctilio, which clearly lack a noseleaf, “−” for this character; thus, the primitive condition for phyllostomids cannot be determined a priori. We also scored Centurio “−” for this character.

This is the first use of this character in a phylogenetic analysis.

Character 25

Inferior border of horseshoe is thin, free flap of skin (0); or inferior horseshoe is thickened ridge with no free edge (1); or inferior horseshoe grades smoothly into upper lip, no distinct boundary between lip and horseshoe (2). The inferior border of the horseshoe is defined by a thin, free flap of skin in many phyllostomines (e.g., Phyllostomus, Vampyrum) and most stenodermatines (e.g., Ametrida, Uroderma; figs. 19B-C, 21B). In other phyllostomines (e.g., Micronycteris megalotis, Mimon benettii) and stenodermatines (e.g., Ardops, Pygoderma), the inferior border of horseshoe is formed by thickened ridge with no free edge (fig. 21C). In contrast, the inferior border of the horseshoe grades smoothly into the upper lip in desmodontines, some phyllostomines (e.g., Phylloderma, Tonatia, Trachops), Brachyphylla, phyllonycterines, glossophagines, lonchophyllines, carolliines, Enchisthenes, Sphaeronycteris, and Sturnira (figs. 19A, 20A–C, 21A). There is no clear boundary between the horseshoe and the upper lip in these taxa. We scored mormoopids and Noctilio, which clearly lack a noseleaf, “−” for this character; thus, the primitive condition for phyllostomids cannot be determined a priori. We also scored Centurio “−” for this character.

Straney (1980: characters G16–17) described two alternative conditions for the labial part of the horseshoe: fused to the upper lip (G16) or free (G17). We disagree with Straney's (1980) assessments of the labial horseshoe in Mimon bennettii, Phylloderma, glossophagines, and carolliines. Straney reported that in all of these taxa the ventral margin of the horseshoe is a free flap, contra our observations. Simmons (1996: character 7) also used a character similar to this one, although she did not consider our state 0 because it did not occur among members of the genus Micronycteris. Our scoring of this character in Micronycteris species is identical to hers. Marques-Aguiar (1994: character 10) also used this character, describing three conditions of the labial horseshoe (free, bound down and rimmed, and bound down and not rimmed), which she ordered 0 ↔ 1 ↔ 2. Although we agree with her observations concerning most species, we found that the conditions she described for both Artibeus fimbriatus (bound down and rimmed) and Artibeus amplus (bound down and not rimmed) appeared very similar to the conditions we observed in all other species of Artibeus (s.s.; flaplike: character state 0); thus we do not score Artibeus as polymorphic for this character.

Character 26

Thin, free edge of inferior horseshoe lies flat over upper lip (0); or forms upright cup around nostrils (1). Among those phyllostomids that have a horseshoe with a thin, free inferior edge (most phyllostomines and stenodermatines; see character 25), this flap lies flat over the upper lip (figs. 19C, 21B). In contrast, the flap forms an upright cup around the nostrils in Chrotopterus and Vampyrum (fig. 19B). We scored mormoopids and Noctilio, which lack a noseleaf, “−” for this character. The primitive condition for phyllostomids thus cannot be determined a priori. We also scored Centurio and those phyllostomids that lack a thin, free flap of skin along the inferior border of the horseshoe (states 1 and 2 of character 25) “−” for this character.

This character has not been used in previous phylogenetic analyses.

Character 27

Inferior border of thickened horseshoe smoothly curved or straight, no V-shaped notch or projection (0); V-shaped notch sometimes present (1); or V-shaped notch always present (2); or long, broad, V-shaped projection present (3). The form of the inferior edge of the horseshoe varies among those taxa with a thickened labial border (see character 25 above). The inferior edge of the horseshoe generally forms a smoothly rounded U-shape or a straight line under the nostrils. However, a V-shaped notch is sometimes present in the center of the inferior border of the horseshoe in Micronycteris megalotis (e.g., AMNH 267411) and Micronycteris minuta (e.g., AMNH 233221) and is always present in Micronycteris hirsuta. An alternative condition is seen in Ardops and Ariteus, which have a long, broad, V-shaped projection that extends toward the lip from the center of the labial border of the horseshoe and gradually grades into the upper lip (fig. 21C). We scored mormoopids and Noctilio, which lack a noseleaf, “−” for this character. The primitive condition for phyllostomids thus cannot be determined a priori. We scored Centurio and those phyllostomids that do not have a thickened labial border (states 0 and 2 for character 25) “−” for this character.

This character has not been used in previous phylogenetic analyses.

Character 28

Ridge of skin absent on dorsum of snout (0); or present (1). No special structures occur behind the spear of the noseleaf in most phyllostomids (although see characters 17 and 29). However, in desmodontines a simple ridge of skin runs across the dorsum of the snout just posterior to the spear. The ridge extends between the superior margins of the lines of vibrissae described in character 14, but is not associated with any vibrissae. In Desmodus and Diaemus, this ridge of skin is connected to the posterior part of the noseleaf by two thin ridges, one from each of the lateral edges of the U-shaped notch of the spear. However, this connection is usually absent in Diphylla (one individual, AMNH 91283, had a single central connection running between the noseleaf and the ridge). In Diaemus and Diphylla, the ridge is generally relatively thin, and in Desmodus it is typically thicker and somewhat higher. However, there is significant within-species variation in ridge morphology, and the conditions seen in former taxa overlap that seen Desmodus. In Mormoops and Noctilio, there is no ridge of skin on the dorsum of the snout; however, in Pteronotus, a ridge is present on the dorsum of the snout. Based on this distribution of character states, it seems that absence of the ridge is the primitive condition for phyllostomids.

This is the first use of this character in a phylogenetic analysis.

Character 29

No outgrowth from posterior spear (0); or small scalloped outgrowth present, connected to base of spear by small ridge (1); or large sexually dimorphic outgrowth from base of spear present (2). Although most phyllostomids do not have structures directly posterior to the spear (but see characters 17 and 28), an outgrowth of the posterior base of the rib is present in Mesophylla and Vampyressa pusilla. There is within-species variation in morphology of this flap, which always has a scalloped appearance, but may appear bifid or trifid in different individuals. This outgrowth is connected to the posterior base of the rib by a small central connective ridge. The only other stenodermatine which has an outgrowth from the posterior spear is Sphaeronycteris. In this taxon, the outgrowth is large and sexually dimorphic (better developed in males). In Sphaeronycteris, there are also two parts to the outgrowth. The first is directly posterior to the spear, and attaches to almost the entire posterior face of spear, leaving a free border around the spear's edge. This primary attachment is a small, rounded ridge that connects the spear with the larger outgrowth, commonly termed the “visor” due to its appearance in male Sphaeronycteris. The “visor” connects with the entire posterior face of the first ridge in both sexes, but has a different morphology in each. In female Sphaeronycteris, the “visor” is small and has its origin over the center of each eye. In males, the origin is from the lateral corner of the eye and the structure is about four times the size of that present in females. We scored mormoopids and Noctilio, which lack a noseleaf, “−” for this character. Therefore, the primitive state for phyllostomids cannot be assessed a priori. We also scored Centurio “−” for this character.

This is the first use of this character in a phylogenetic analysis.

Character 30

Chin with pair of dermal pads, one present on each side of midline (0); or chin with multiple, well-developed dermal papillae (1); or chin smooth or with a few poorly developed papillae (2); or chin partly or completely covered with skin flaps (3). In desmodontines, many phyllostomines (e.g., Macrotus, Vampyrum), phyllonycterines, glossophagines, lonchophyllines, and Rhinophylla, a pair of naked dermal pads are present on the chin (figs. 23A, 23C–E). These pads are typically oblong and are arranged in a U or V pattern just ventral to the lower lip. The space between the pads is always hairless and may lack other dermal structures, although small or large dermal papillae may be present in this region (e.g., Rhinophylla, see character 33 below).

The chin morphology of phyllonycterines and glossophagines is superficially different from that of other phyllostomids with chin pads (fig. 23D). The chin in these genera is slightly (e.g., Erophylla) to deeply cleft (e.g., Choeroniscus) at the midline and the surfaces of the cleft are lined with smooth, hairless skin. The anterior borders of the cleft are fleshy and scalloped; these small projections resemble papillae when the cleft is closed and the bat is viewed anteriorly. Muscular action apparently keeps the cleft closed much of the time in living bats (personal obs.), but the cleft is often open in preserved specimens. In such cases, the fleshy walls of the cleft can be seen to resemble the chin pads typical of other bats with these structures, leading us to conclude that they are homologous. This hypothesis is further supported by the morphology of the chin in lonchophyllines, which is not cleft. In some individuals, only the superior margins of the pads appear to be scalloped as in glossophagines and phyllonycterines (e.g., Lionycteris AMNH 145504); however, substantial variation in edge ornamentation exists and the entire edge is scalloped in many individuals (e.g., Lionycteris AMNH 202295). The chin pads in lonchophyllines are arranged in the same U- or V-shaped pattern seen in desmodontines, phyllostomines, and Rhinophylla. Based on these observations, we score phyllonycterines, glossophagines, and lonchophyllines as having a pair of chin pads; differences in chin morphology among taxa with chin pads are treated separately in characters 31 and 32.

Many phyllostomines, Brachyphylla, Carollia, and most stenodermatines exhibit a markedly different pattern in which a U- or V-shaped row of small dermal papillae occurs in place of paired chin pads (figs. 23B, F). These papillae differ from the padlike structures described above because they arise from the flat anterior surface of the chin rather than the scalloping of the free edge of a chin pad. The space between the two arms of the U or V may be bare, or may contain some small papillae whose position and development varies from individual to individual. However, one large central papilla is consistently present in some taxa (e.g., most stenodermatines; see character 33). Centurio lacks chin papillae and pads altogether, and instead has a smooth chin. Most specimens of Sphaeronycteris also have a completely smooth chin, but one individual (USNM 494538) that we examined has an incomplete row of very small, poorly developed papillae on the chin. However, we consider the condition in both Centurio and Sphaeronycteris to be potentially homologous and score both with state 2 in the matrix.

Mormoops, Pteronotus, and Noctilio have chins that are partly or completely covered with complex folds of skin (fig. 23G). These flaps and folds do not appear to be comparable with the chin pads or papillae seen in phyllostomids, but instead seem to be outgrowths of the lower lip. Pteronotus has a papillated lip pad that covers the area along the lip and is continuous with the pinnae. The papillation is confined to the broader central area of the lip pad. The labial border of the pad is arched, and the mental border has a U-shaped notch. At the lateral corner of the mouth there is a low ridge of skin that is continuous with the pinna and with the lip pad. On the ventral surface of the chin, caudal to the lip pad is a caudal chin flap—a simple skin flap that encompasses the area between the canines. In Mormoops, the lip pad and caudal chin flap are similar to those in Pteronotus, although they are slightly smaller, but there are more complicated foliations connecting these elements. Two sets of lip flaps are present. The superior lip flaps are broad, ruffled foliations that course almost from the corner of the mouth to the posterior face of the lip pad, which they join. The caudal chin flap runs between the posterior surfaces of the superior lip flaps, encompassing the area just below the lip pad. Narrower inferior lip flaps, which are connected to the pinnae, course to just lateral of the point of attachment of the caudal flap to the superior lip flaps, where they too join the posterior surface of the superior lip flaps. Noctilio lacks many of the foliations seen in mormoopids. A caudal chin flap runs between the flaplike lips. There are no additional foliations on surface of the chin. Because the unique conditions in the outgroup taxa do not occur among the ingroup, the primitive condition for phyllostomids cannot be assessed a priori.

Straney (1980: characters G18–19) used this character, describing two conditions: presence of simple chin pads (derived state of G18), roughly equivalent to our state 0, or complex chin pads (derived state of G19), roughly equivalent to our state 1. Our character is not identical to Straney's (1980), as he considered the chin flaps of mormoopids and noctilionids to be similar to chin papillae. Also, Straney (1980) scored Lonchorhina, Brachyphylla, and Rhinophylla as possessing multiple papillae on the chin, whereas we consider these taxa to have chin pads. This difference may be due to the presence of a large central papilla on the chins of these genera and others. We consider the presence of this papilla as a separate character (see character 33).

Character 31

Chin pads simple, not scalloped (0); or chin pads with scalloped lateral edges (1). As described above, the chin pads, when present, are simple (not scalloped) in desmodontines, phyllostomines, and Rhinophylla (figs., 23A, E). The lateral edges of the pads are smooth in these taxa. In contrast, the lateral edges of the chin pads are partly or completely scalloped in phyllonycterines, glossophagines, and lonchophyllines giving the pad edge a papillated appearance (see character 30 above; figs. 23C–D). We scored mormoopids and Noctilio, which lack paired chin pads, “−” for this character. The primitive condition for phyllostomids thus cannot be determined a priori. We scored phyllostomids that have either multiple chin papillae or smooth chins (states 1 and 2 of character 30 above) “−” for this character.

This character has not been used in previous phylogenetic analyses.

Character 32

Chin without central cleft (0); or with slight to deep central cleft (1). As described above (see character 30), the chin is relatively flat (not cleft) in most phyllostomids (fig., 23A–C, E–F). In contrast, the chin is slightly to deeply cleft at the midline in phyllonycterines and glossophagines (fig., 23D). Depth of the cleft appears to vary continuously however, and is therefore not considered as a separate character. Mormoopids and Noctilio have a chin that lacks a central cleft, suggesting that this is the primitive condition for phyllostomids.

This is the first use of this character in a phylogenetic analysis.

Character 33

Central papilla absent from chin (0); or central dermal papilla present on chin just ventral to midline of lower lip (1). A central papilla is absent from the chin in desmodontines, most phyllostomines, Brachyphylla, and lonchophyllines (fig., 23A–C). In contrast, this papilla is present just ventral to the midline of the lower lip in Macrophyllum, carolliines, and stenodermatines (fig., 23E–F). The central papilla is surrounded laterally and inferiorly by either chin pads (e.g., Rhinophylla) or rows of dermal papillae (e.g., Carollia). We scored mormoopids and Noctilio, which have a unique chin morphology involving chin flaps (see character 30), “−” for this character. Thus, the primitive condition for phyllostomids cannot be determined a priori. We scored taxa with smooth chins (state 2 in character 30) and those with a chin cleft (state 1 of character 32) “−” for this character.

Although Straney (1980) discussed the distribution of the central papilla in phyllostomids, he did not use this as a character in his analysis. Our observations are largely in agreement with Straney (1980), with the exception of Phylloderma, Phyllostomus, and Brachyphylla. Straney (1980) described the central pad as doubled in Phylloderma and Phyllostomus. We did not observe a doubled central pad in these taxa, rather we found that these two genera, and many others, showed evidence of having two lateral central pads (see fig. 23B, F). However, due to marked within- and between-species differences, we were unable to use this as a character at this hierarchical level. In Brachyphylla, we did not observe a central papilla, while Straney (1980) indicated that this papilla was present.

Character 34

Internal labial papillae absent (0); or internal labial papillae limited to lip line (1); or internal labial papillae more widely distributed, cover most of inside of cheeks (2). In most phyllostomids, the inner surface of the lips and cheek are smooth, but in stenodermatines these areas are studded with numerous small projections which Silva-Taboada and Pine (1969) called “internal labial papillae.” These fleshy papillae range in shape from somewhat flat, triangular projections to more conical structures. The extent to which internal labial papillae cover the inside of the mouth varies among taxa in which these structures are present. All stenodermatines have internal labial papillae along the upper lip, but the pattern of distribution is variable within species and no clear distinctions can be made among taxa. This is not true of the lower lip and cheeks, where the internal labial papillae are distributed in one of two patterns. Most stenodermatines have papillae located only along the lip line, and lack internal labial papillae on the inside of the cheeks. In contrast, Ametrida, Ardops, Ariteus, Centurio, Phyllops, Pygoderma, Sphaeronycteris, and Stenoderma have internal labial papillae that more or less cover the inside of the cheeks. Mormoopids and Noctilio lack internal labial papillae, suggesting that absence of these structures is the primitive condition for phyllostomids.

This is the first use of this character in a phylogenetic analysis.

Character 35

Pinna margin not smoothly rounded, concavity located on lateral border (0); or pinna with smoothly rounded margin, no concavity on lateral border (1). In most phyllostomids, a concavity is present at some point along the lateral margin of the pinna. However, in Chrotopterus, Macrotus, Micronycteris hirsuta, M. megalotis, M. minuta, Tonatia, Trachops, and Vampyrum, there is no concavity on the lateral margin of the pinna and the margin of the pinna is smoothly rounded. In mormoopids and Noctilio, a concavity is present on the lateral margin of the pinna, suggesting that this is the primitive condition for phyllostomids.

Owen (1987: character 1; 1991: characters 1–2) described the emargination of both the medial and lateral edges of the pinna. Although we agree with Owen (1987, 1991) that the pinna presents variation that may be useful in a phylogenetic context, our survey of this character suggests that Owen (1987) subdivided a continuous trait. Our character has been designed to avoid dealing with the continuous nature of pinnal emargination while still describing relevant variation. Although we concur with Owen's (1987, 1991) observation of an emarginate lateral border in all stenodermatines and carollines, we disagree on the morphology of the pinna in Macrotus. Owen (1987: 58) described the pinna in Macrotus as “Completely emarginated; pronounced indentation as in state 1.” We did not use Owen's (1991: character 1) character “Anterioproximal lobe on pinna,” because this feature is an autapomorphy of Centurio. Simmons (1996) used presence or absence of a concavity in the lateral edge of the pinna in an analysis of relationships of Micronycteris, and our scoring agrees with hers.

Character 36

Interauricular band absent (0); or present (1). In most phyllostomids, the ears are not connected by an interauricular band. In contrast, a band of skin extends between the anterior external surfaces of the medial pinnae in Macrotus, Micronycteris hirsuta, M. megalotis, and M. minuta. All of these species have ears that, for phyllostomids, are large in proportion to the size of the skull: when folded forward they reach approximately to the end of the snout. The skin band in Macrotus and these species of Micronycteris is slightly less than one quarter the height of the ears. This band is flexible and is typically folded back against the braincase in preserved specimens. The band in each taxon originates from the external surface of each pinna just posterior to the medial edge, and runs transversely across the skull just anterior to the slope of the braincase. Although Koopman (1994) reported that Tonatia silvicola has these skin flaps, our examination of this species revealed that these structures do not appear similar to those present in Macrotus and Micronycteris. Instead, we interpret the “flaps” in Tonatia as slightly larger versions of the ridges of skin which cover the auricular muscles in all species (these are particularly visible if the pinna is pulled forward). We therefore score Tonatia as lacking an interauricular band. Although there are extensive foliations between the ears in Mormoops (see below), Pteronotus and Noctilio lack an interauricular band. The distribution of character states in the outgroup taxa suggests that absence of the interauricular band is primitive for phyllostomids.

In Mormoops the extensive foliations between the ears differ significantly from the morphologies described above. A flap of skin originates from each ear, but these flaps are much more extensive and complex than those described above. Each flap arises from the medial external surface of the pinna and courses anteromedially toward the midsagittal line. Near the midpoint, each flap divides, giving rise to a poorly defined branch that connects with its counterpart across the top of the cranium and a much larger flap that courses anteriorly until it terminates on the dorsum of the nose halfway between the tip of the snout and the slope of the braincase. These large right and left flaps are separated throughout their course along the rostrum. At the point of termination there is a roughly triangular lateral expansion in each of these flaps. Because of the differences between the simple skin band seen in some phyllostomids and the complex branched flaps found in Mormoops, we suspect that these structures are not homologous. Pteronotus and Noctilio lack skin flaps between the ears, and presence of these structures is generally interpreted as autapomorphy of the genus Mormoops (Smith, 1972). Accordingly, we score Mormoops “−” for this character.

Straney (1980: character G20) used presence of an interauricular band in his component analysis. However, he observed a band between the ears in Lonchorhina and several species of Tonatia. As we noted above, we interpret these flaps as being slightly larger versions of the ridges of the auricular muscles that connect the pinna with the fascia covering the skull. Simmons (1996: character 5) previously used presence of an interauricular band as a character in an analysis of relationships of Micronycteris species; our assessments agree with hers.

Character 37

Notch in interauricular band shallow (0); or deep (1). The interauricular band is characterized by a shallow, V-shaped midsagittal notch in Micronycterishirsuta and M. megalotis. In contrast, the midsagittal notch is much deeper in Macrotus and Micronycteris minuta. In these taxa, the notch effectively divides the interauricular band into two triangular flaps, one associated with each ear. We scored mormoopids and Noctilio, which lack an interauricular band, “−” for this character. The primitive condition of this feature thus cannot be determined a priori. We scored other phyllostomids that lack an interauricular band (state 0 of character 36) “−” for this character.

Simmons (1996: character 6) previously used depth of the interauricular band as a character in an analysis of relationships of Micronycteris species. Due to the inclusion of several species of Micronycteris that do not appear in our analysis, Simmons (1996) subdivided the character more finely than we have done here, recognizing three states (shallow, moderate, and deep). However, our observations agree with hers.

Character 38

Facial hood absent in both sexes (0); or present in males (1). The skin of the neck is relatively taut and smooth in most phyllostomids. In contrast, folds of skin are present around the neck in Centurio and Sphaeronycteris. These skin bands are smaller in females than in males, and they reach their greatest development in male Centurio. In males of both Centurio and Sphaeronycteris, the throat folds apparently operate as a hood, or mask, which drops over the face when the bat is roosting (Hill and Smith, 1984; Paradiso, 1967; Goodwin and Greenhall, 1961; Nowak, 1991: photograph, p. 314). The facial hood is absent in mormoopids and Noctilio, suggesting that this is the primitive condition for phyllostomids.

Lim (1993: character 3) first used this character in a cladistic analysis. Our character states and scoring are identical to his.

Character 39

Vomeronasal tube well developed with neuroepithelium present (0); or rudimentary, neuroepithelium absent (1); or absent (2). The vomeronasal epithelial tube (VET) is “an elongate, cigar-shaped structure lined with epithelium” that is associated with nerve fibers, vomeronasal glands, blood vessels, and venous sinuses (Wible and Bhatnagar, 1996: 294). In desmodontines, Macrotus, glossophagines, Carollia, and stenodermatines the VET is well-developed and has a neuroepithelial lining medially (Wible and Bhatnagar, 1996). In contrast, in Brachyphylla the VET occurs in a rudimentary form and lacks the neuroepithelium (Wible and Bhatnagar, 1996). In mormoopids, the VET is rudimentary in Mormoops, well developed in Pteronotus parnellii, and completely absent in Pteronotus personatus (Wible and Bhatnagar, 1996). We scored Pteronotus with states 0, and 2 in the matrix. The VET is absent in Noctilio (Wible and Bhatnagar, 1996). Thus, the primitive state for phyllostomids cannot be assessed a priori.

Based primarily on the work of Bhatnager and his colleagues (e.g., Bhatnager and Kallen, 1974; Cooper and Bhatnager, 1976; Bhatnager 1980, 1985; Frahm and Bhatnager, 1980; Frahm 1981; Bhatnager et al., 1982), Wible and Bhatnagar (1996) defined this character (and characters 40, 41, and 135), mapping it onto preexisting hypotheses of chiropteran relationships. Our character states and scoring are identical to theirs.

Character 40

Vomeronasal cartilage bar-shaped (0); or curved, J-, C-, U-, or O-shaped in cross section (1). The vomeronasal cartilage, which supports the VET, is present and J-, C-, U-, or O-shaped in cross section in most phyllostomids (Wible and Bhatnagar, 1996). Because the curved vomeronasal cartilage changes shape over its length, more than one of these shapes may occur in a single individual. In contrast, Brachyphylla possesses a bar-shaped vomeronasal cartilage (Wible and Bhatnagar, 1996). Among the outgroup taxa, a curved vomeronasal cartilage is present in Mormoops, whereas a bar-shaped vomeronasal cartilage is present in Pteronotus and Noctilio (Wible and Bhatnagar, 1996). Given this distribution of character states, the presence of a bar-shaped vomeronasal cartilage appears to be primitive for phyllostomids.

Wible and Bhatnagar (1996) defined and mapped this character onto preexisting phylogenies of bats. Our scoring is identical to theirs. In the context of the phylogenies Wible and Bhatnager (1996) examined, they described the primitive state as our state 1. Their state 2 (vomeronasal cartilage absent) does not occur in any of our taxa and we therefore excluded it from consideration.

Character 41

Nasopalatine duct present (0); or absent (1). The nasopalatine duct, which connects the nasal and oral cavities through the incisive foramen in the palate, is present in most phyllostomids (Wible and Bhatnagar, 1996). In contrast, the nasopalatine duct is absent in Macrotus (Wible and Bhatnagar, 1996). The nasopalatine duct is present in Mormoops and Pteronotus personatus, but is absent in Pteronotus parnellii and Noctilio (Wible and Bhatnagar, 1996). We scored Pteronotus with states 0 and 1 in the matrix. Due to this distribution of character states among the outgroup taxa, the primitive state for phyllostomids cannot be reconstructed a priori.

Wible and Bhatnagar (1996) first used this character. Our character states and scoring are identical to theirs.

Character 42

Zygomatic arch always complete (0); or always incomplete (1); or polymorphic within species (2). Although most phyllostomids have a complete zygomatic arch, this structure is incomplete in Phyllonycteris, many glossophagines (e.g., Choeroniscus, Lichonycteris), lonchophyllines, and carolliines. In all taxa that lack the zygomatic arch, a distinct zygomatic process projects anteriorly from the lateral margin of the mandibular fossa and posteriorly from dorsal or slightly posterior to the last molar. Both processes taper and have rounded tips. Among brachyphyllines, phyllonycterines, glossophagines, and lonchophyllines, a thin but complete arch is present only in Brachyphylla, Erophylla, Glossophaga, Leptonycteris, and Monophyllus. Anoura is the only phyllostomid that exhibits polymorphism in zygomatic development. Some specimens of Anoura eoffroyi have two complete zygomatic arches whereas others lack both, and one specimen (AMNH 78288) has a complete zygomatic arch on the right side and an incomplete arch (that does not appear to have been broken) on the left side. However, in Anoura caudifera it appears that both zygomatic arches may be complete in all individuals (Miller, 1907; personal obs.). Those specimens without a complete arch show clear evidence of breakage (e.g., AMNH 176347). Based on the states defined above, we scored Anoura with states 0 and 2. Mormoopids and Noctilio have a complete zygomatic arch, suggesting that this condition is primitive for phyllostomids.

Lim (1993: character 19) first used this character in a cladistic analysis. We have followed his character state definitions but have added state 2 (polymorphic). Although Koopman (1994: 81, 86) expressed some reservation about the status of the zygomatic arch in Lichonycteris (“more or less complete”) and in a subgenus of Sturnira (Corvira: “weak or incomplete”), our examinations indicate that the zygomatic arch is never complete in Lichonycteris, but is always complete in Sturnira (Corvira).

Character 43

Mastoid breadth less than zygomatic breadth (0); or greater than zygomatic breadth (1). In most phyllostomids, breadth of the skull measured across the mastoid region is less than breadth measured across the zygoma. In contrast, mastoid breadth is greater than zygomatic breadth in Lonchorhina, Micronycteris minuta, and some species of Tonatia (e.g., T. silvicola, T. evotis). We scored Tonatia with states 0 and 1 in the matrix. We scored this character in some taxa with an incomplete zygomatic arch (see character 42) when the posterior process of the zygoma clearly extended laterally beyond the level of the mastoids (e.g., Carollia). However, in species where the posterior process of the zygoma does not extend past the level of the mastoid, we did not attempt to score this character; these taxa are scored “−” in the matrix. Mastoid breadth is less than zygomatic breadth in mormoopids and Noctilio, suggesting that this is the primitive condition for phyllostomids.

Simmons (1996: character 14) first used mastoid versus zygomatic breadth as a character in a phylogenetic analysis of relationships of Micronycteris species; our assessments agree with hers.

Character 44

Hard palate long, extends into interpterygoid space (0); or no palatal extension or emargination (1); or palate with shallow posterior emargination that extends maximally to middle of upper second molar (2); or with deep emargination that extends minimally to posterior border of first molar (3). The posterior part of the hard palate extends into the interpterygoid space in most phyllostomids (fig. 24A). In some taxa (e.g., Pygoderma) there is no extension of the hard palate into the interpterygoid space, nor is the palate emarginate (fig., 24B). However, in some stenodermatines the hard palate lacks a posterior extension and is emarginate. In Centurio and Sphaeronycteris, the palatal emargination is shallow, extending to approximately the middle of M2 measured at the labial margin of the tooth (one specimen [AMNH 175651] of Centurio in which the emargination appears deeper has clearly been damaged; fig. 24C). The emargination is deeper in Ardops, Ariteus, Phyllops, and Stenoderma ending somewhere along the labial border of M1 (fig. 24D). In Ametrida, the pterygoids are very small (compressed in the anterior-posterior dimension) and the internal nasal aperture is reduced by bony laminae extending medially from the pterygoids in both the coronal and transverse planes (fig. 24E). This is a unique condition among phyllostomids (Miller, 1907; personal obs.), and does not appear to be homologous with other conditions described above; we therefore scored Ametrida “−” for this character. In Mormoops blanvillii, Pteronotus, and Noctilio, the hard palate extends into the interpterygoid space; however, there is no palatal emargination or extension in Mormoops megalophylla. This distribution of character states suggests that possession of a hard palate that extends into the interpterygoid space may be primitive for phyllostomids.

Owen (1987: character 9) measured the length of the posterior margin of the hard palate from the anterior border of the orbit, and defined three states: “Extends farther caudad than least width of palate between orbits” (0), “Extends caudad to anterior border of orbits, but no farther than least width of palate” (1), “Does not extend caudad to anterior border of orbit” (2). This character was ordered (0 ↔ 1 ↔ 2). Subsequently, Owen (1991: characters 21–22) split his (1987) states 1 and 2 into separate characters. Although Owen's (1991) scoring remained unchanged, he transformed the multistate character he had previously described (Owen, 1987: appendix 2) into several characters using additive binary coding. We have described this character in different terms than those used by Owen (1987, 1991) to accommodate some of the variation his character did not account for. Thus, our character is not directly comparable to his.

Lim (1993) recognized that Owen's (1987) state 2 encompassed more variation than Owen (1987) had described. Accordingly, Lim (1993) erected several states to account for this variation. However, Lim (1993: character 4) incorporated shape and size in a single ordered (0 → 1 → 2 → 3 → 4 → 5) character: “Hard palate extending into interpterygoidal space (0); no palatal extension or emargination (1); palatal emargination shallow and V-shaped (2); palatal emargination moderate and converging anteriorly (3); palatal emargination deep and narrow (4); palatal emargination very deep and wide (5).” We have chosen to score shape separately from length (see character 45). In addition, we do not recognize moderate emargination (Lim's state 3) as a distinct character state. Lim (1993) stated that this condition appears in Phyllops, but our observations suggest that the “moderate” condition, which would be an autapomorphy of Phyllops and therefore uninformative in a phylogenetic analysis, more closely resembles the conditions seen in Ardops, Ariteus, and Stenoderma. We also do not score the palate of Ametrida for this character; Lim (1993) considered the palate in this genus to lack an extension or emargination.

Character 45

Posterior border of hard palate always U-shaped (0); or always V-shaped (1). The shape of the caudal border of the palate does not appear to be a useful character among phyllostomids with long palates (state 0 in character 44) because there is a continuous range of variation among the U-, V-, and W-shaped conditions both within genera and individual species. For example, Lonchophylla thomasi has a V-shaped emargination, whereas other Lonchophylla species, such as L. robusta, have a U shape (Koopman, 1994; personal obs.). In Carollia subrufa, the posterior palate is V shaped in some specimens (e.g., AMNH 164023), but is U shaped in others (e.g., AMNH 186375). In those taxa that do not have a palatal extension or emargination (state 1 in character 44), the posterior edge of the palate is straight to barely curved (fig. 24B). The shape of the medial posterior border of the hard palate appears to be potentially informative in stenodermatines with emarginate hard palates (states 2 and 3 in character 44). In Ardops, Ariteus, and Stenoderma, the palatal emargination is U-shaped (fig. 24D). The emargination is V-shaped in Centurio, Phyllops (we examined both P. falcatus and P. haitiensis; see below), and Sphaeronycteris (fig. 24C). The primitive condition for phyllostomids cannot be assessed a priori because mormoopids and noctilionids do not have emarginate palates. We scored taxa that do not have an emarginate palate (states 0 and 1 of character 44 above) “−” for this character.

Owen (1987: character 10) first described this character using four states, including the “U-” and “V-shaped” conditions we recognize. The other two states, “square” (his state 0) and “median projection” (his state 3) do not apply to taxa with an emarginate palate. Owen (1987) scored his palatal character for all taxa, including those with long palates, and ordered it in a complex fashion. Subsequently, Owen (1991: character 23) again used a similar character but only recognized two states: “square” and “V-shaped.” Our observations often do not agree with those of Owen (1987, 1991). Owen (1987) considered Ametrida to have a V-shaped posterior border of the hard palate and described this border as U-shaped in Pygoderma and Phyllops falcatus. Owen (1991) grouped taxa into two categories. “Square” included Ametrida, Centurio, Phyllops haitiensis, and Sphaeronycteris. “V-shaped” included the remaining taxa with emarginate palates and Pygoderma.

Lim (1993: character 4) considered palatal shape and length as a single ordered character, but scored the shape of the caudal border only in taxa with palatal emargination. Marques-Aguiar (1994: character 21), who also considered whether the “mesopterygoid fossa” had U- or V-shaped borders, scored this character for taxa with long palates. We have followed Lim (1993), because posterior border shape is variable within species with long palates, and because the posterior border of the palate may not be homologous in taxa with emarginate palates and those with longer palates. We have described the states of this character in slightly different terms, recognizing only two general shape categories, “V-” and “U-shaped,” rather than the four states Lim (1993) recognized (V-shaped; moderate and converging anteriorly; deep and narrow; very deep and wide). Two of the four conditions Lim (1993) identified are autapomorphies. Phyllops was the only genus Lim (1993) scored with the “moderate and converging anteriorly” condition, while only Ardops was scored as “palatal emargination very deep and wide.” Because these are autapomorphies, they are not informative in a phylogenetic analysis at this level. In addition, we observed that these two taxa share more general conditions (either “V-” or “U-shaped”) with other taxa. Consequently, we have recognized only these two conditions. We did not score genera with a palate that has no extension or emargination (state 1 in character 44) because a straight to slightly curved posterior border of the hard palate and state 1 in character 44 have identical distributions.

Character 46

Pterygoid lamina sheetlike (0); or pterygoid lamina greatly inflated posterior to anterior margin of mandibular fossa (1). The pterygoid lamina is a relatively thin sheet of bone that projects ventrally from the basicranium. In most phyllostomids, the pterygoid may curve slightly at its posterior and distal end, but over most of its course it diverges from the midline of the basicranium. In most phyllostomids, the pterygoid lamina is not inflated or ballonlike (fig. 25A). In contrast, in Choeroniscus, Choeronycteris, and Musonycteris, the pterygoid lamina is greatly inflated, ballooning laterally posterior to the anterior margin of the mandibular, or glenoid, fossa (fig. 25B). Anterior to this point, the pterygoid diverges from the midline of the basicranium. In mormoopids and Noctilio the pterygoid does not balloon laterally, suggesting that this is the primitive condition for phyllostomids.

This character has not been used in previous cladistic analyses, although it has been described by other authors (e.g., Miller, 1907; Phillips, 1971; Koopman, 1993).

Character 47

Rostrum sloping; facial processes of premaxilla and maxilla above teeth lie perpendicular to palatal processes (0); or facial processes of premaxilla and maxilla oriented horizontally above teeth (1). The rostrum slopes gently from the forehead to the external nares in most phyllostomids. Just above the teeth, the facial processes of the premaxilla and maxilla lie roughly perpendicular to the palatal processes of these bones in these taxa (fig. 26A). In contrast, the facial bones that compose the normally sloping rostrum in most phyllostomids (nasal and maxilla) are upturned and largely oriented vertically in Ametrida and Sphaeronycteris. The nasal aperture in these two taxa is located at the base of the cranium, giving the skull a distinctly “apelike” appearance. Just above the teeth, the facial processes of the premaxilla and maxilla lie in roughly the same plane as the palatal processes (rather than being roughly perpendicular to the palatal processes), an arrangement that makes this area appear dorsoventrally compressed (fig. 26B). In mormoopids and Noctilio, the rostrum is gently sloping and the facial surfaces of the maxilla and premaxilla just above the teeth are roughly perpendicular to the palatal processes, suggesting that this arrangement represents the primitive condition for phyllostomids.

Lim (1993: character 7) originally scored this character with two states (“gradually sloping rostral profile” and “dorso-ventral compression of the rostrum”), which correspond to our character states; our observations agree with his.

Character 48

I2 present (0); or absent (1). No bat has more than two upper incisors (Slaughter, 1970). Although there has been some debate regarding the homologies of these teeth to those in other mammals, most workers agree that these teeth probably represent I1 and I2 (Thomas, 1908; Anderson, 1912; Slaughter, 1970). Most phyllostomids retain both I1 and I2. In contrast, Desmodus and Diaemus lack I2. Mormoopids and Noctilio retain I2, suggesting that this condition is primitive for phyllostomids.

Owen (1987: character 21) used complete dental formula as a character in his examination of stenodermatine relationships. Under Owen's (1987) coding scheme, different dental formulae were scored as different character states and the transformation series was ordered in a complex fashion. The result of this coding was that homologies of individual teeth and tooth regions were not always preserved. For example, Owen's (1987) coding precluded the possibility that presence of a single lower incisor might be a shared, derived feature of two taxa with different overall dental formulae. Owen (1991) recoded these data, using characters considering specific tooth regions separately. Lim (1993) did not use dental formula in his analysis of stenodermatine relationships. We consider dental formula to be potentially informative, but prefer to score presence/absence of teeth at individual tooth loci to better preserve initial assessments of putative homologies.

Character 49

I1 distinctly larger than I2 (0); or I1 and I2 approximately equal in size (1). In most phyllostomids that retain both I1 and I2, the inner tooth (I1) is noticeably larger than the outer tooth (I2; fig. 27B, D–G). In contrast, I1 and I2 are approximately equal in size in glossophagines (fig. 27C). In mormoopids and Noctilio, I1 is noticeably larger than I2, suggesting that this is the primitive condition for phyllostomids. We scored taxa that lack I2 (state 1 in character 48) “−” for this character.

Lim (1993: character 10) described a single ordered character for size and shape of I1. The criterion Lim (1993) used to judge the size of I1 appears to have been arbitrary. Many of the taxa that Lim (1993) scored as possessing “reduced” teeth of various morphologies (his states 0, 2, and 3 of character 10) have an I1 roughly two times as large as I2 (e.g., Ametrida; fig. 27E). We prefer to define two characters that describe size and shape separately (this character and character 50) because this method allows clear definition of potential homologies. Phillips (1971: 26), who commented at length on the dentition of glossophagine (s.l.) bats, considered the I1 of Lonchophylla to be “about as large as” I2. The representatives of Lonchophylla we have examined (see appendix 1 for specimens) all possessed an I1 that was noticeably larger (often 2X as large) than I2.

Character 50

I1 occlusal margin generally straight or slightly rounded (0); or occlusal margin concave, C-shaped (1); or occlusal margin evenly bifid (2); or main cusp on occlusal margin pointed (3); or mesially and distally projecting lobes present (4). I1 in most phyllostomids has a roughly straight or somewhat rounded occlusal margin (fig. 27B–C). In desmodontines, I1 has a strongly concave (C-shaped) occlusal margin (fig. 27A). Yet another pattern appears in some stenodermatines (e.g., Artibeus, Uroderma), which have an I1 with a well-developed notch in the occlusal margin near the center of the tooth (fig. 27D). Presence of this notch gives the tooth a bifid or bilobed appearance. Although the notch may be slightly offset from the center of the tooth, both lobes are approximately the same size. Finally, other stenodermatines (e.g., Ametrida, Stenoderma) have a pointed main cusp on I1 (fig. 27E–G). Rhinophylla pumilio appears to have a unique I1 morphology. In this species, the occlusal margin of the tooth has two projecting lobes, one which projects mesially and one distally, with a slight depression between them. However, both R. alethina and R. fischerae have a straight occlusal margin on I1; thus we have scored Rhinophylla with states 0 and 4 for this character. I1 is bifid in mormoopids, but has a straight or somewhat rounded occlusal margin in Noctilio. The primitive state for phyllostomids thus cannot be reconstructed a priori.

Lim (1993: character 10) described a multistate ordered character based on the size and shape of I1. We do not agree with Lim's (1993) coding of this character. Mormoops and Pteronotus, which both have a bifid I1, were coded solely on the basis of tooth size by Lim (1993), who apparently ignored the similarity between the condition in these taxa and those seen in other taxa with bifid incisors (e.g., Artibeus). Lim (1993) also described two conditions for I1 of the “short-faced” stenodermatines (see discussion in character 51). In this character, we have chosen to emphasize the similarity in the shape of the main cusps of I1 in all “short-faced” stenodermatines, while describing certain differences among them in the characters below (51 and 52). Dividing the characters in this way best reflects our hypotheses of homology.

Marques-Aguiar (1994: character 24), in her analysis of relationships among species of the subgenus Artibeus, described a character based on morphology of I1. The two character states were “(0) bilobed (bifid) and not pointed,” and “(1) simple (not bifid) and pointed.” Although we agree with her characterization of all members of the subgenus Artibeus, Dermanura gnoma, and Koopmania as possessing a bilobed I1, we do not agree that I1 in Enchisthenes is pointed.

Character 51

I1 with pointed main cusp offset mesially, lateral part of tooth forms poorly to well-developed cusp (0); or, pointed cusp roughly in center of tooth (1). Among those phyllostomids with a pointed main cusp on I1, there appear to be two distinct morphologies. In Ardops, Ariteus, Centurio, Pygoderma, Phyllops, and Stenoderma, the most well-developed part of I1 is the mesial part of the tooth. This part of the tooth forms a distinct, pointed cusp in all these taxa (figs. 27F–G). Thus, the main cusp appears offset from the center (longitudinal axis) of the tooth. The lateral part of the tooth may form a well-developed (e.g., Ardops) or poorly developed (e.g., Pygoderma) cusp. In contrast, the main cusp comes to a point in roughly the center of the tooth in Ametrida and Sphaeronycteris (fig. 27E). In one specimen (AMNH 187225) of Ametrida out of 10 we examined for this character, we found an exceptionally well-developed accessory cusp near the tip of the main cusp. We interpret this as an anomaly, and score Ametrida with state 1. Mormoopids and Noctilio do not have pointed inner upper incisors and are therefore scored “−” for this character, as are all other taxa lacking pointed incisors (states 0, 1, 2, and 4 in character 50). The primitive state for phyllostomids thus cannot be reconstructed a priori.

Lim (1993: character 10) originally scored morphology of the inner upper incisors in his analysis of stenodermatine relationships. He scored Ametrida, Centurio, Pygoderma and Sphaeronycteris as having an I1 that is triangular in shape. He described the remaining “short-faced” taxa as having an I1 with a small secondary outer cusp (we call this cusp the “lateral cusp” to avoid confusion with the “accessory cusp”; see below). We observed the lateral cusp in most “short-faced” taxa, including Pygoderma. In addition, we found it difficult to characterize the teeth of Centurio and Pygoderma as similar to those of Ametrida and Sphaeronycteris, when both Centurio and Pygoderma clearly had mesially offset main cusps.

Although Miller (1907) reported that the main cusp in Centurio is in the center of the tooth, our observations indicate that it is placed towards the mesial edge of the tooth. Finally, Miller (1907) observed, as we did, an often poorly developed accessory cusp toward the tip of the main cusp in Pygoderma. This condition appears to be an autapomorphy of this genus. In the other taxa with a lateral cusp, as in Pygoderma, the lateral cusp arises from the base of the tooth, it does not arise from the main cusp.

Character 52

I1 with pointed main cusp only slightly longer than lateral cusp (0); or main cusp 2 × the length of the lateral cusp (1). In those stenodermatines with a lateral cusp, there are two patterns of main cusp development. In Stenoderma and Pygoderma, the main cusp is very long, roughly twice the size of the smaller lateral cusp (fig. 27G). In the remaining taxa with a mesially displaced main cusp, this cusp is only slightly longer than the lateral cusp (fig. 27F). We scored mormoopids and Noctilio, which do not have a pointed I1, “−” for this character. The primitive state for phyllostomids thus cannot be reconstructed a priori. We scored all other taxa lacking pointed incisors (states 0, 1, 2, and 4 in character 50) and well-developed mesial cusps (state 1 in character 51) “−” for this character.

Although this character has not appeared in previous cladistic analyses, this feature of the teeth was noted by previous authors (e.g., Miller, 1907).

Character 53

i1 present (0); or absent (1). Most phyllostomids have two lower incisors, which we identify as i1 and i2 following Slaughter (1970; although see Thomas, 1908 for a different view). However, i1 is absent in many glossophagines (e.g., Anoura, Choeronycteris, Hylonycteris). Mormoopids and Noctilio retain i1, suggesting that presence of i1 is the primitive condition for phyllostomids.

This character has not been used in previous phylogenetic analyses, although numerous authors have described dental formulae of glossophagines (e.g., Miller, 1907; Phillips, 1971; Koopman 1994).

Character 54

i2 present (0); or absent (1). Although most phyllostomids retain i2, this tooth is absent in some phyllostomines (e.g., Chrotopterus, Tonatia), many glossophagines (e.g., Anoura, Choeronycteris), and some stenodermatines (Sturnira bidens, S. nana, and Vampyressa bidens). We scored Sturnira with states 0 and 1 in the matrix. Mormoopids retain both lower incisors, but i2 is absent in Noctilio. The primitive condition for phyllostomids cannot be assessed a priori.

Straney (1980: character K3) based a binary character on presence or absence of i2. Our character states are equivalent to his, as is our scoring with the exception of our observation that not all stenodermatines have i2. Owen (1991: character 44) defined a character for loss of a lower incisor (from two incisors to one) in some stenodermatines, but our observations are not consistent with his coding of this character. Owen's (1987) character for dental formula correctly scored the two Sturnira species and Vampyressa bidens as possessing a single lower incisor. However, Owen (1991) scored Ariteus and some species of Dermanura as having only one lower incisor; all specimens of these taxa that we examined retained both i1 and i2.

Character 55

I1 and i1 in contact or separated by small gap when cheek teeth occlude (0); or I1 and i1 separated by marked gap (1); or I1 occludes with cingular shelf on lower canine well posterior to i1 (2); or I1 occludes posterior to i1 in mandibular fossae (3). When the jaw is closed and the canines and postcanine teeth are in occlusion, the I1 and i1 contact each other in most phyllostomids, though in some taxa these teeth may be separated by a very small gap. In contrast, I1 and i1 are separated by an extremely large gap when the canines and postcanine teeth are in occlusion in most stenodermatines. The only stenodermatines with a “normal” occlusal pattern (state 0) are two of the three species of Chiroderma we examined (C. salvini and C. trinitatum), Ectophylla, Mesophylla, Sturnira, and Vampyressa species. We scored Chiroderma with states 0 and 3 in the matrix. Another pattern occurs in Chrotopterus and Vampyrum, where I1 occludes with a broad cingular shelf that encircles the posterolingual base of the lower canine. Thus, I1 lies well posterior to i1 when the canines and postcanine teeth are in occlusion. Similarly, in desmodontines, the I1 lies posterior to the i1 when the canines and postcanine teeth occlude. However, in these taxa I1 rests in a fossa located in the mandible posterior to i1. In mormoopids and Noctilio, the I1 and I1 occlude or are separated by a very small gap when the canines and postcanine teeth occlude, suggesting that this is the primitive condition for phyllostomids. We scored taxa lacking both lower incisors (state 1 of characters 53 and 54) “−” for this character.

To our knowledge, patterns of incisor occlusion have not been previously used as characters in phylogenetic analyses of phyllostomid taxa, although the occlusal pattern and fossae of desmodontines have been described before (e.g., Miller, 1907).

Character 56

P3 present (0); or absent (1). No bat is known to have more than three upper premolars (Slaughter, 1970). Although there has been continuing debate regarding the homologies of these teeth to those in other mammals (e.g., Miller, 1907; Thomas, 1908; see review in Slaughter, 1970), most recent workers have settled on numbering the premolars “P2-P3-P4” in bats (e.g., Slaughter, 1970), and we follow this usage mainly for convenience⁴. Of the three upper premolars found primitively in bats, most phyllostomids retain two, which are apparently homologous to P3 and P4 of other bats (Slaughter, 1970). Only one noctilionoid, Anoura, has P2 in addition to P3 and P4; this appears to be an autapomorphy. In contrast, desmodontines have only one upper premolar, having lost P3 (Slaughter, 1970). Mormoopids retain both P3 and P4, whereas Noctilio has apparently lost P3. The primitive state for phyllostomids thus cannot be reconstructed a priori.

Straney (1980: character K6) described a binary character based on presence of P3; our character states and scoring are identical to his.

Character 57

Second distal accessory cusp on P4 absent (0); or second distal accessory cusp sometimes present (1); or second distal cusp always present (2). P4 lacks a second distal accessory cusp in most phyllostomids. In contrast, two distal accessory cusps are often present in the stenodermatines Ardops, Ariteus, Artibeus, Dermanura, Phyllops, Platyrrhinus, and Pygoderma. The second distal accessory cusp is always present only in Uroderma and Vampyrodes. Mormoopids and Noctilio do not have a second distal accessory cusp, suggesting that this is the primitive state for phyllostomids.

Lim (1993: character 6) first described this character; however, he scored only Platyrrhinus and Uroderma as having two distal accessory cusps, and did not note the polymorphism we have discovered in almost every species in which this feature is present. Despite our consideration of the “sometimes present” and “always present” conditions as separate character states, we suspect that the second distal accessory cusp is not uniformly present in any species (this suspicion is strengthened because Lim [1993] did not report the presence of this cusp in Vampyrodes).

Character 58

p3 absent (0); or p3 greatly reduced, tooth peglike, cusps poorly or not developed (1); or moderately reduced compared to p2 and p4, cusps present and well developed (2); or p3 crown height roughly equivalent to p2 and p4 (3). As with the upper premolar dentition (see above), the three lower premolars of bats are typically numbered “p2-p3-p4” (Slaughter, 1970), and we follow this usage. Only two premolars (presumed to be p2 and p4) are present in desmodontines, some phyllostomines (e.g., Mimon, Phyllostomus), Brachyphylla, phyllonycterines, carolliines, and stenodermatines. The hypothesis that p3 is absent in these taxa is supported by the observation that p3 is reduced in size relative to p2 and p4 in many phyllostomids. In these taxa (e.g., Lonchorhina, Macrophyllum), this tooth is extremely small (less than one quarter the overall size of p2 and p4) and peglike with cusps that are poorly developed or absent. In two specimens of Lonchorhina (AMNH 149233 and AMNH 183850) we suspect that p3 may be absent; however, dried tissue is present in this region, making it impossible for us to unambiguously determine whether p3 is present in these two individuals. We therefore score p3 as present in Lonchorhina. This tooth is absent from both sides in one specimen of Macrophyllum (AMNH 209320), but we consider this an anomaly, and score Macrophyllum with state 1. In some phyllostomines (e.g., Phylloderma, Tonatia), the crown of p3 is reduced in size, but the height of p3 is less than or equal to one third the height of p2 and p4. However, the tooth is still large and has well-developed cusps. In all remaining phyllostomids in which this tooth is present, the lower premolars have crowns that are subequal in height. In mormoopids, p3 is moderately reduced in Mormoops and is greatly reduced and peglike in Pteronotus. Noctilio has only two premolars. Although it is not clear which tooth has been lost, we score Noctilio as having state 0 to leave open the possibility that this is a derived trait shared with some phyllostomids. The primitive condition for phyllostomids thus cannot be reconstructed a priori.

Straney (1980: characters K8, 10–12) first used premolar loss and reduction in a phylogenetic analysis. His character K8 dealt with the presence or absence of this tooth. Our observations for this character are identical with his, although Straney (1980: 68) scored Noctilio as possessing p3. However, he noted that this tooth was absent in this taxon in his text. To evaluate degree of reduction of p3, Straney (1980) used the height of the cingula on p2 and p4; thus, our results are somewhat different from his. If we consider our character states 3, 2, and 1, equivalent to Straney's (1980) K10 (subequal), K11 (p3 lower than p2, 4 but above cingulum), and K12 (p3 lower than cingulum on p2,4), respectively, the only taxa that we disagree on are Phylloderma, Tonatia, and Trachops. Straney (1980) considered both Phylloderma and Trachops to have the derived state of K12 (p3 lower than cingula on p2, 4), and found that different species of Tonatia had different states for this character (derived for K10, K11, and K12). Our disagreement with Straney (1980) may be due to the presence of a better developed lingual cingulum in these three genera. As noted above, we consider p3 to be moderately reduced in these three taxa.

Owen (1991: character 43) used a character whose derived state was defined as the loss of a lower premolar (reduction in number from three to two premolars). Among his outgroups, Owen (1987) correctly scored only Macrotus as possessing three lower premolars. However, in his 1991 paper, Owen incorrectly scored the following taxa as having three lower premolars (state 0 for character 43): Carollia perspicillata, Ametrida, Ardops, Enchisthenes, Koopmania, Phyllops, Sphaeronycteris, Stenoderma, and Sturnira ludovici.

Simmons (1996: character 18) described an ordered character similar to this in her analysis of relationships among Micronycteris species. Our character state 0, p3 absent, was not used by Simmons (1996) because all Micronycteris retain p3. We have further modified this character by comparing the crown heights of the premolar teeth, rather than the overall size of tooth. Consequently, our character is not directly comparable to hers.

Character 59

Postcanine teeth including p3 aligned in row roughly parallel to long axis of mandible (0); or p3 sometimes lingually displaced from toothrow (1); or p3 always lingually displaced from toothrow (2). In most phyllostomids, all of the postcanine teeth, including p3, are aligned in a row that runs roughly parallel to the long axis of the jaw. However, p3 is always lingually displaced from the toothrow in Chrotopterus. In Lonchorhina, Macrophyllum, and Trachops, some specimens retain p3 in the toothrow (e.g., Lonchorhina: AMNH 184701; Macrophyllum: AMNH 94551; Trachops: AMNH 266081), whereas in others p3 is lingually displaced (e.g., Lonchorhina: AMNH 230122; Macrophyllum: AMNH 262424; Trachops: AMNH 267442). In those specimens with lingually displaced p3s, p2 and p4 are in contact to the labial side of this tooth. This tooth is not reduced in Mormoops. Pteronotus retains p3 in the toothrow, consequently p2 and p4 are not in contact in this genus. This tooth is absent in Noctilio. Thus the primitive state for phyllostomids cannot be reconstructed a priori. It seems likely that lingual displacement can only occur in taxa in which p3 is greatly reduced in size (e.g., taxa scored 1 for character 58). We are unaware of any examples from other microchiropteran families where a tooth of normal size is displaced, but there are examples of lingual displacement of greatly reduced teeth (e.g., Myotis seabrai and M. lesueuri; Koopman, 1994). Therefore, we score this character only in those taxa that have a greatly reduced p3; we scored genera in which p3 is either absent, slightly reduced, not reduced, or (states 0, 2, and 3 of character 58) “−” for this character.

Straney (1980: character K13) previously used this feature as a binary character and scored it in all phyllostomines in his analysis. Although we have found that this tooth is sometimes lingually displaced in Lonchorhina, our other observations agree with Straney's (1980) scoring of this character. However, we found that this feature is often variable, and have taken this into account in constructing our character states. We disagree with Smith (1972: 56) who stated that p3 is “markedly reduced to a small peg-like unicuspid tooth and is almost always excluded (lingually), or nearly so, from the toothrow” in the genus Pteronotus.

Character 60

W-shaped ectoloph present on M1–M2 (0); or absent (1). A W-shaped ectoloph is prominent on the upper molar teeth of most microchiropterans (Slaughter, 1970; Koopman and MacIntyre, 1980; Koopman, 1994), and many phyllostomids retain this characteristic (e.g., fig 28B, E). The W-shaped ectoloph is best developed on M1 and M2, but is generally present, although often altered, on M3 when this tooth is present (see character 64). However, desmodontines, brachyphyllines, phyllonycterines, several glossophagines (e.g., Lichonycteris, Musonycteris), Platalina, carollines, and stenodermatines do not have a W-shaped ectoloph on M1–M2 (fig. 28A, C–D, F). A W-shaped ectoloph is present on M1–M2 of mormoopids and Noctilio, suggesting that this is the primitive condition for phyllostomids.

Although the presence of a W-shaped ectoloph has not been formally defined as a character in a cladistic analysis, the phylogenetic implications of the presence/absence of the W-shaped ectoloph in phyllostomids have been discussed by Slaughter (1970) and Phillips (1971). Slaughter (1970) viewed the primitive condition as retention of the W-shaped ectoloph and noted various conditions which evolved from a dilambdodont condition. Phillips (1971) divided glossophagine (s.l.) bats into three groups based on molar morphology, primarily on the presence or absence of the W-shaped ectoloph, and noted that presence of a W-shaped ectoloph was more primitive. We agree with most of Phillips' (1971) observations, but have found that Hylonycteris has an ectoloph, whereas Lichonycteris does not, contra Phillips (1971). Finally, in considering the homologies of the cusps of the molar teeth of Brachyphylla, Griffiths (1985: 546) commented that “In every stenodermatine genus but two, the W-shaped ectoloph pattern was clearly evident … displaced laterally up against the stylar cusps of the tooth ….” We consider a strongly W-shaped ectoloph pattern to be absent in all stenodermatine genera. Although there are crests that are displaced laterally against the stylar cusps, these do not form a W shape (compare fig. 28B and F).

Although we are aware that we may be lumping several conditions together under our state 1 (i.e., the W-shaped ectoloph may have been lost in different taxa due to the loss or movement of different crests or cusps), we defined the character in this manner due to the problems associated with drawing homologies among the cusps of the upper molar teeth in taxa with unique molar morphologies. Often, homologies among cusps on the upper molars are drawn in consideration of the taxonomic affinities of the genus in question (e.g., Griffiths' [1985] interpretation of molar morphology in Brachyphylla), an approach that is not suited for cladistic analysis.

Character 61

Hypocone basin distinct on M1, hypocone indistinct to well developed (0); or basin and cusp both indistinct or absent (1); or hypocone wing present (2). The hypocone is conspicuously absent on the upper first molar of many phyllostomids, including desmodontines, brachyphyllines, phyllonycterines, glossophagines, lonchophyllines, carollines, and some stenodermatines (e.g., Mesophylla, Platyrrhinus; fig 28A, C–D). In most of these taxa, the hypocone basin is poorly developed and the tooth triangular, rather than square, in occlusal outline (some stenodermatines like Platyrrhinus and Sturnira are exceptions, but in these genera there is no cusp in the hypocone region). In all other genera, a variably developed hypocone is present in the well-developed hypocone basin. In many phyllostomines this cusp is incorporated into a crest that runs into the hypocone region and may encircle it; thus the cusp itself is often indistinct (fig. 28B). In stenodermatines, the cusp is usually not incorporated into a crest and is more distinctly conical than in the phyllostomines (fig. 28F). Monophyllus appears to be unique among phyllostomids in having what Phillips (1971) termed a hypoconal wing, a buttresslike structure that slopes towards the palate rather than connecting with the metacone (fig. 28E). In mormoopids and Noctilio, a distinct hypocone is present, suggesting that this is the primitive condition for phyllostomids.

Straney (1980: characters K19–23) scored hypocone development in what we interpret as two series of characters. The first series (character K19 and K20) dealt with whether this cusp was indistinct (the derived condition of character K19) or distinct (the derived condition of character K20). Straney (1980) scored most taxa, but not all, for these two characters. The second series (characters K21 to K23) dealt with the height of the hypocone: very low (derived condition of character K21), low (derived condition of character K22), or relatively high (derived condition of character K23). Straney scored all taxa for this second series of characters.

Owen (1987: character 14) described hypocone development on the first upper molar in a five-state ordered character. Marques-Aguiar (1994: character 26) also described the development of the hypocone on M1, defining three states. Although the fine grade distinction between a moderately developed hypocone and a well-developed hypocone might be appropriate for a species level analysis in some taxa (e.g., Marques-Aguiar, 1994), the character states defined by Owen (1987) appear to subdivide a continuous series of hypocone development. Our observations of the presence or absence of this cusp conflict with several of Owen's (1987) observations. Owen (1987) found that some species of Chiroderma, Platyrrhinus, Sturnira, and Vampyrodes have at least a moderately developed hypocone; our observations suggest that this cusp is absent in these taxa. Owen (1991: character 30, 31, 32) transformed his original multistate character into three separate binary characters. One of Owen's (1991) characters (30) is equivalent to ours. Our observations of the presence/absence of this cusp agree with those of Owen (1991).

Lim (1993: character 5) redefined this character as a discrete state ordered (0 ← 1 → 2) character by scoring the presence of either three or four cusps on the first upper molar (the fourth cusp being the hypocone). Our states and scoring are identical to his with one exception. Lim (1993) described Sturnira as having an autapomorphic state in which the protocone and hypocone are obscured by a longitudinal groove. Although we observed a longitudinal trough in Sturnira between the buccal and lingual cusps, this trough does not obscure the protocone and hypocone regions. Sturnira appears to lack a hypocone; accordingly, we score this taxon 1.

Phillips (1971) considered a hypocone basin to be present on the upper molars of Anoura, Glossophaga, Leptonycteris, and Lonchophylla. Although this region of the tooth may be present in these taxa, it is more poorly developed in these genera than this region is in phyllostomines and stenodermatines. The outline of the first upper molar in these genera remains much more triangular in outline than those of the stenodermatines and phyllostomines we examined. We also disagree with Slaughter (1970), who observed that all stenodermatines have a hypocone.

Considerable disagreement exists concerning the presence and location of many cusps (especially the paracone and metacone) on the upper first molars of some taxa (e.g., Brachyphylla, Erophylla, and Phyllonycteris). However most authors (e.g., Miller, 1907; Slaughter, 1970; Koopman and MacIntyre, 1980; Griffiths, 1985) have considered the hypocone to be lacking in these genera. Our observations support this interpretation, consequently we have scored these taxa 1 for this character.

Character 62

M1–M2 lacking extensive elongation in metastylar region (0); or with exceptionally long metastyle present (1). In most phyllostomids, the metastylar region of the upper first two molars is poorly to moderately developed and is not exceptionally elongate (fig 28A, C–F). In contrast, in Chrotopterus, Trachops, and Vampyrum, an exceptionally long metastyle is present at the distolabial border of these teeth (fig. 28B). Although the metastylar region of M1–M2 is elongate in some glossophagines (e.g., Glossophaga), the loph present on this region of the tooth (the postmetacrista) is approximately equal in length to the other lophs present on the tooth. This situation is unlike the condition seen in Chrotopterus, Trachops, and Vampyrum, where the postmetacrista is much longer than any other loph on the tooth. Thus we do not consider these two conditions homologous and score taxa like Glossophaga with state 0. The labial margin of M1–M2 in mormoopids and Noctilio does not have an exceptionally elongate metastylar region, suggesting that this is the primitive state for phyllostomids. We scored this character on the morphology of M1 alone in taxa lacking an M2 (e.g., Desmodus).

Straney (1980: character K26) used this character previously in a component analysis, and our scoring is identical to his. Slaughter (1970) also described this feature of the teeth in phyllostomids.

Character 63

M3 present (0); or sometimes or always absent (1). Most bats have three upper molars (“M1-M2-M3”) as do most other mammals. The third upper molar (M3) is present in most phyllostomids (fig. 28B–E). In contrast, M3 is absent in desmodontines, Leptonycteris, Lichonycteris, and many stenodermatines (e.g., Centurio, Ectophylla; fig 28A, F). In both Artibeus and Chiroderma, M3 is almost always absent in some species (e.g., A. lituratus, C. villosum), whereas it is consistently present in others (e.g., A. obscurus, C. trinitatum). Consequently, we scored these two taxa with both 0 and 1 in the matrix. In addition, we found that M3 was present on both sides in one specimen each of Enchisthenes (AMNH 233791; we examined 6 for this character), and Mesophylla (AMNH 262540; we examined 10). We scored these taxa with state 1. In Sphaeronycteris, we found that M3 was frequently absent from one side of the dentition (e.g., AMNH 24379), although all specimens we examined had M3 present on at least one side. Consequently, we scored Sphaeronycteris with state 0 for this character. Mormoopids and Noctilio have three upper molars, suggesting that presence of M3 is the primitive condition for phyllostomids.

Owen (1987: character 15) first used a character to describe the development of M3. In this character, Owen (1987) considered not only the presence or absence of this tooth but its size (well developed with cusps present 0; or poorly developed, small, and peglike 1). Our observations of the size and cusp development of M3 suggest that there is no simple way to characterize tooth size and development in phyllostomids; these features appear to vary continuously. Subsequently Owen (1991: character 34) used a character identical to ours. Our observations of the presence/absence of this tooth agree with those of Owen (1987, 1991), with the exception of our observation that this tooth is sometimes present in Artibeus jamaicensis, Enchisthenes, and Mesophylla, and may be absent in Sphaeronycteris. Although Owen (1991: 19) noted that “the condition of both M3 and m3 was miscoded in the table [1987: table 1], even though properly coded in the analyses,” our observations do not differ greatly from Owen's (1987) table 1.

Although we have usually erected a unique character state for polymorphic taxa, we have not done so here because we suspect that polymorphism is more widespread than we have reported, particularly with the very reduced teeth present in some stenodermatines. Thus, we have been conservative, pending further evaluation of polymorphism in these taxa.

Character 64

Ectoloph on M3 W-shaped (0); or V-shaped (1). Among phyllostomids that have a W-shaped ectoloph on M1 and M2 and a third upper molar (see characters 60 and 63), there are two distinct morphologies of the M3 ectoloph. In glossophagines and lonchophyllines with a W-shaped ectoloph on M3, a virtually complete W-shape is present (fig. 28E). In contrast, in phyllostomines the ectoloph appears V-shaped on M3 (fig. 28B). In mormoopids and Noctilio, a complete W-shaped ectoloph is present on M3, suggesting that this is the primitive condition for phyllostomids. We scored taxa lacking both a W-shaped ectoloph and an M3 (state 1 of character 60 and state 1 of character 63) “−” for this character.

Straney (1980: character K24) first used this character in a component analysis. Although our scoring and Straney's are identical, we have described this character differently. Straney (1980) described his character as the presence or absence of only the postmetacrista on M3. Although we suspect that the premetacrista, metacone, and postmetacrista have been lost, a view proposed by Slaughter (1970), we have avoided erecting any character (except see character 61) that would force us to draw homologies among the cusps on the upper molars. There is too much ongoing debate concerning the homology of the cusps on the upper molars in many genera (e.g., Brachyphylla, Phyllonycteris, and Erophylla).

Phillips (1971) indicated that the W-shaped ectoloph is absent from M3 in Anoura, Glossophaga, Lonchophylla, and Monophyllus, usually because one of the styles (e.g., mesostyle, metastyle) or cristae (e.g., metacrista = postmetacrista) are lacking. However, our observations of M3 in these taxa indicate that the W-shaped lophs are still distinctly visible on M3 in these taxa. Thus, our observations agree with Slaughter (1970: 68), who noted that “glossophagines have a well developed premetacrista and a short metacrista.”

Character 65

m2 present (0); or absent (1). As in the upper dentition, most phyllostomids have three lower molars (designated “m1-m2-m3”). Most phyllostomids retain m2; however, this tooth is absent in Desmodus and Diaemus. Mormoopids and Noctilio have three lower molars, suggesting that presence of m2 is the primitive condition for phyllostomids.

This character has not been used in previous phylogenetic analyses, although numerous authors have described dental formulae in phyllostomids (e.g., Miller, 1907; Phillips, 1971; Owen, 1987; Koopman, 1994).

Character 66

m3 present (0); or sometimes or always absent (1). Most phyllostomids retain m3. In contrast, m3 is absent in desmodontines, Leptonycteris, Lichonycteris, and many stenodermatines (e.g., Centurio, Ectophylla). In Chiroderma, Dermanura, and Sturnira, m3 is present in some species (e.g., C. doriae, D. glauca, S. lilium) and is absent in others (e.g., C. villosum, D. cinerea, S. thomasi). Consequently, we score these taxa with both 0 and 1 for this character. In Pygoderma, a single m3 is present on the left side in two individuals (AMNH 261761, AMNH 246408), one specimen has both the left and right m3 present (AMNH 234295), and the remaining individuals lack m3 entirely (a total of seven). We score Pygoderma with state 1. In Sphaeronycteris, we found that one m3 was frequently absent (e.g., AMNH 76251), although all specimens we examined had an m3 on at least one side. Consequently we score Sphaeronycteris with state 0 for this character. Mormoopids and Noctilio have three lower molars, suggesting that presence of m3 is the primitive condition for phyllostomids.

Owen (1987: character 22) first used a character to describe the development of m3. In this character Owen (1987) considered not only the presence or absence of this tooth but its size (well developed with cusps present 0; or poorly developed, small, and peglike 1). Our observations of the size and cusp development of m3 suggest that there is no simple way to characterize tooth size and development in phyllostomids; these features appear vary in a continuous fashion. Owen (1991: character 46) and Marques-Aguiar (1994: character 29) both used the presence of the third lower molar as a character. With the exception of our observations of the variability of this feature in Pygoderma and Sphaeronycteris, we agree with Owen (1987, 1991) and Marques-Aguiar (1994).

Although absence of M3 (see character 63 above) and absence of m3 are clearly correlated in some taxa, this is not true of all taxa. For example, some species of Artibeus, Ariteus, some species of Dermanura, Mesophylla, Vampyressa bidens, and Vampyrodes lack M3 but retain m3. Accordingly, we score presence/absence of upper and lower third molars as separate characters. We also score polymorphism in this character conservatively (see discussion in character 61).

Character 67

m1 does not form high shearing ridge, lingual aspect of tooth well developed (0); or m1 laterally compressed, lingual aspect of tooth poorly developed, tooth incorporated into continuous shearing ridge with anterior cheek teeth (1). Like most bats, most phyllostomids have lower molars that do not form a straight, continuous, shearing ridge with the anterior cheek teeth. In contrast, desmodontines have laterally compressed mandibular cheek teeth that are incorporated into a continuous, straight (i.e., parallelling the long axis of the jaw) shearing ridge that includes the canines, premolars, and any additional molar teeth (present in Diphylla). Mormoopids and Noctilio have molars that do not form a straight, continuous, shearing ridge, suggesting that this is the primitive condition for phyllostomids.

This character has not been used in previous cladistic analyses, although many authors have described the morphology of the teeth in desmodontines (e.g., Miller, 1907; Slaughter 1970).

Character 68

Paraconid always present on m1 (0); or sometimes absent (1); or always absent (2). A paraconid is present on m1 in most phyllostomids (fig. 29B, E). In contrast, the paraconid is always absent from m1 in Brachyphylla, phyllonycterines, and most stenodermatines, including some species of Sturnira (e.g., S. erythromos; other Sturnira species, like S. oporophilum, have a distinct paraconid; fig. 29C–D, G–I). In some individuals of Platyrrhinus helleri (e.g., AMNH 263616; appears in six of 10 individuals examined for this character) a small paraconid is occasionally present but is absent, or appears less frequently, in other species (e.g., P. dorsalis, where it was absent in four of four specimens). This situation also occurs in Vampyrodes (e.g., AMNH 186381; appears in one of 10 individuals examined), and some species of Dermanura (e.g., D. cinerea AMNH 97075; appears in one of 10 individuals examined for this character). We scored Dermanura and Platyrrhinus with states 1 and 2 in the matrix. Due to difficulties in drawing homologies among the cusps of the lower molars in Rhinophylla and the desmodontines, we scored these taxa “?” in the matrix for this character and characters 69–70 (fig. 29A, F). The paraconid is present on m1 in mormoopids and Noctilio, suggesting that this is the primitive condition for phyllostomids.

Lim (1993; character 9) first introduced this character. However, Lim's (1993) original description did not draw any homologies between cusps, describing the character states as four prominent cusps (0), or three prominent cusps (1). We find substantial disagreement between our observations and those reported by Lim (1993), who found the derived condition of this character (three prominent cusps) in only four taxa: Chiroderma, Ectophylla, Mesophylla, and Vampyressa. Our observations indicate that there are two prominent cusps in these taxa, an enormous protoconid and a smaller hypoconid. In the remaining taxa Lim (1993) scored with the primitive condition (four prominent cusps), we observed from three to five distinct cusps. We prefer to divide observed variation in m1 cusp patterns into three separate characters (68–70), in order to account for this variation and putative cusp homologies.

Although Slaughter (1970) hypothesized that it is the lingual cusps that are lost in Rhinophylla and desmodontines, much of this argument rests on the relationships of these taxa to other phyllostomids. Therefore, we have chosen to score these taxa “−” until more detailed information is available concerning the homologies of the cusps in these taxa.

Character 69

Metaconid present on m1 (0); or absent (1). A metaconid is present on the lower m1 of most phyllostomids (fig. 29B–C, E, I). In contrast, the metaconid is absent from m1 in phyllonycterines and many stenodermatines (e.g., Ardops, Uroderma; fig. 29D, G–H). In some species of Sturnira (e.g., S. lilium) a metaconid is present, whereas in others this cusp is absent (e.g., S. bidens). A metaconid is present on m1 in mormoopids and Noctilio, suggesting that this is the primitive condition for phyllostomids. We scored desmodontines and Rhinophylla “?” for this character.

Character 70

Entoconid present on m1 (0); or absent (1). An entoconid is present on the lower m1 of most phyllostomids (fig. 29B–C, H–I). In contrast, the entoconid is absent from m1 in phyllonycterines and many stenodermatines (e.g., Chiroderma, Mesophylla; fig. 29D, G). In Sturnira, an entoconid is present in some species (e.g., S. tildae) and absent in others (e.g., S. erythryomos). An entoconid is present on m1 in mormoopids and Noctilio, suggesting that this is the primitive condition for phyllostomids. We scored desmodontines and Rhinophylla “?” for this character.

Character 71

P3 and P4 in contact, no diastema present (0); or diastema sometimes or always present between P3 and P4 (1). In most phyllostomids, the upper third and fourth premolars are in contact and there is no gap (diastema) present between these teeth. In contrast, in some phyllostomines (e.g., Micronycterishirsuta, Vampyrum), phyllonycterines, glossophagines, lonchophyllines, Rhinophylla, and many stenodermatines (e.g., Chiroderma, Mesophylla), a diastema is sometimes or always present between P3 and P4. In Mormoops a diastema separates these teeth, but in Pteronotus and Noctilio the upper third and fourth premolars are in contact, suggesting that this is the primitive condition for phyllostomids. We scored taxa in which P3 is absent (state 1 in character 56) “−” in the matrix.

Lim (1993: character 11) defined a single ordered (0 → 1 → 2) character based on diastemata in the lower jaw: “No gaps between mandibular cheek teeth (0); space between mandibular premolars (1); additional space between mandibular premolar and molar (2).” We apply this character to the maxillary, rather than mandibular, tooth row to avoid problems stemming from the loss of p3 in some phyllostomids (see character 58). We have chosen to split Lim's (1993) original character into two separate characters (and one new character we developed after including glossophagines in our analysis) to preserve our ideas about the homology of the various gaps. With few exceptions, our scoring of this character is identical to Lim's (1993). The only difference is the appearance of a diastema between the third and fourth upper premolars in Platyrrhinus and Uroderma, which does not occur in the lower tooth row.

Character 72

P4 and M1 in contact, no diastema present (0); or diastema sometimes or always present between P4 and M1 (1). In most phyllostomids, the upper fourth premolar and first molar are not separated by a diastema. In contrast, in phyllonycterines, lonchophyllines, most glossophagines (e.g., Choeronycteris, Hylonycteris), Rhinophylla, and some stenodermatines (e.g., Chiroderma and Mesophylla), a diastema is sometimes or always present between P4 and M1. In mormoopids and Noctilio, the upper fourth premolar and first molar are in contact, suggesting that this is the primitive condition for phyllostomids.

As noted above, we broke Lim's (1993: character 11) original character, which had been based on diastemata in the lower jaw, into two separate characters. Our observations differ little from Lim's (1993). The only difference is the appearance of a P4–M1 diastema in all Vampyressa species included in our analysis. These gaps do not occur consistently in the lower tooth row. We also disagree with Phillips (1971) concerning this character in Lonchophylla. Phillips (1971) noted that a cingular style of P4 contacts the parastyle of M1 in Lonchophylla. However, we have found that there is no contact between these two teeth in at least some individuals of Lonchophylla.

Character 73

M1 and M2 in contact, no diastema present (0); or diastema sometimes or always present between M1 and M2 (1). In most phyllostomids, the upper first and second molars are in contact and there is no gap (diastema) present between these two teeth. However, many glossophagines (e.g., Choeronycteris, Hylonycteris), Platalina, Rhinophylla, Ectophylla, Mesophylla, and all Vampyressa species, a diastema is sometimes or always present between M1 and M2. In mormoopids and Noctilio, the upper first and second molars are in contact, suggesting that this is the primitive condition for phyllostomids. Desmodus lacks an M2 , consequently we scored this character “−” in this taxon.

This character has not been used in previous phylogenetic analyses of phyllostomids, but was described by other authors (e.g., Phillips, 1971).

Character 74

Distal tip of clavicle attached by ligaments to coracoid process of scapula (0); or attached to both coracoid and acromion processes (1). The distal tip of the clavicle is attached to the coracoid process of the scapula by ligaments in Desmodus, Diphylla, Macrotus, Phyllostomus, Leptonycteris, Choeronycteris, and Carollia (Vaughan, 1959; Strickler, 1978). There is no ligamentous connection to the acromion process in these taxa. In contrast, the distal tip of the clavicle is attached by ligaments to both the coracoid and acromion processes of the scapula in Glossophaga and Centurio (Strickler, 1978). The distal tip of the clavicle is attached to the coracoid only in mormoopids, but is attached to both the coracoid and acromion in Noctilio (Strickler, 1978). The primitive condition for phyllostomids cannot be assessed a priori.

This feature has not been used previously in a phylogenetic analysis.

Character 75

M. occipitopollicalus with single cranial muscle belly (0); or with a cranial and a distal muscle belly (1). M. occipitopollicalus is a complex muscle system that originates on the occiput of the skull and inserts into the metacarpal or first phalanx of the pollex (Strickler, 1978). In some species, a second tendon may insert onto the ventral metacarpal of digit one. This muscle complex includes a variable number of muscle bellies, segments of tendon, and segments of elastic tissue (Strickler, 1978). M. occipitopollicalus has a single cranial muscle belly in phyllostomines, Brachyphylla, Choeronycteris, Glossophaga, Leptonycteris, Carollia, Centurio, Sturnira, and Uroderma (Vaughan, 1959; Strickler, 1978; Straney, 1980; fig. 30A–B). This cranial muscle belly is also present in Desmodus, Diphylla, Artibeus, and Chiroderma; however, in these taxa an additional muscle belly is present distal to the first (Strickler, 1978; Straney, 1980; fig. 30C). In these taxa, the two muscle bellies are separated by a small proximal tendon and a distal stretch of elastic tissue (Straney, 1980; Strickler, 1978). Only the cranial belly is present in mormoopids; however, both cranial and distal muscle bellies are present in Noctilio (Strickler, 1978). Thus, the primitive state for phyllostomids cannot be assessed a priori.

Straney (1980: characters C1–2) developed two binary characters whose derived states are equivalent to our states 1 and 0, respectively. Our scoring is identical to Straney's (1980).

Character 76

M. occipitopollicalus has one connection to ventral flight musculature (0), or two attachments to ventral flight musculature (1). In most phyllostomids, a single tendinous connection exists between the proximal tendon of m. occipitopollicalus and the ventral flight musculature (Vaughan, 1959; Strickler, 1978; Straney, 1980; fig. 30A, C). The single exception is Phyllostomus, which has two tendinous attachments to the ventral flight musculature (Strickler, 1978; Straney, 1980; fig. 30B). In mormoopids, there is a double tendinous attachment to the ventral flight musculature (Strickler, 1978; Straney, 1980). Noctilio has a single connection to the ventral flight musculature (Strickler, 1978; Straney, 1980). Due to this distribution of character states, the primitive state for phyllostomids cannot be assessed a priori.

Straney (1980: characters C3–4) developed two binary characters whose derived conditions are equivalent to our states 1 and 0, respectively. Our scoring is identical to Straney's (1980).

Character 77

M. spinodeltoideus originates from vertebral border of scapula only (0); or from vertebral border of scapula and transverse scapular ligament (1). In bats, m. deltoideus comprises three separate muscles: m. clavodeltoideus, m. acromiodeltoideus, and m. spinodeltoideus (Strickler, 1978). M. spinodeltoideus arises from the vertebral border of the scapula in Choeronycteris, Glossophaga, and Leptonycteris (Strickler, 1978). However, m. spinodeltoideus has an enlarged origin in Desmodus, Diphylla, Macrotus, Phyllostomus, Carollia, and Centurio, where it arises from the vertebral border of the scapula and part of the transverse scapular ligament (Vaughan, 1959; Strickler, 1978). In mormoopids, m. spinodeltoideus arises from the vertebral border of the scapula and the transverse scapular ligament (Strickler, 1978). In Noctilio, m. spinodeltoideus arises from the vertebral border of the scapula only (Strickler, 1978). Thus, the primitive condition for phyllostomids cannot be reconstructed a priori.

This feature has not been previously used in a phylogenetic analysis.

Character 78

Caput mediale of m. triceps brachii inserts on both elbow sesamoid and caput laterale tendon (0); or on caput laterale tendon only (1); or on elbow sesamoid only (2). In bats, as in most mammals, the m. triceps brachii is composed of three heads: caput longum, caput laterale, and caput mediale (Strickler, 1978). In Diphylla, Phyllostomus, and Carollia the caput mediale of m. triceps brachii inserts on the elbow sesamoid (Strickler, 1978). In contrast, the caput mediale of m. triceps brachii inserts on the ventral edge of the tendon of the caput laterale in Desmodus, Choeronycteris, Glossophaga, Leptonycteris, and Centurio (Strickler, 1978). In Mormoops, the caput mediale of m. triceps brachii is absent (Vaughan and Bateman, 1970, Strickler, 1978), and we scored Mormoops “−” for this character. In Pteronotus and Noctilio, the caput mediale inserts on both the elbow sesamoid and the ventral edge of the caput laterale tendon (Vaughan and Bateman, 1970; Strickler, 1978). This distribution of character states suggests that the primitive condition for phyllostomids is the insertion of the caput mediale of m. triceps brachii on the elbow sesamoid and the caput laterale tendon.

This character has not been used previously in a phylogenetic analysis. Vaughan (1959), who described the postcranial musculature of Macrotus, noted only that the caput mediale of the m. triceps brachii inserted on the olecranon process of the ulna. This description is not detailed enough for us to score Macrotus for this character. Vaughan and Bateman (1970) described the origin and insertion of this muscle after examining all mormoopid species and 18 phyllostomid species, but did not describe the different insertions noted by Strickler (1978).

Character 79

M. palmaris longus inserts on digit II (0); or does not insert on digit II (1). M. palmaris longus originates on the distal tip of the spinous process of the humerus, divides into several tendons at the wrist, and inserts on various digits and tendons in the manus (Vaughan and Bateman, 1970). This muscle inserts on digit II in Desmodus, Erophylla, Glossophaga, and Artibeus (Altenbach, 1979; Vaughan and Bateman, 1970). In contrast, m. palmaris longus has no insertion on digit II in Macrotus, Phyllostomus, Carollia, and Sturnira (Vaughan, 1959; Vaughan and Bateman, 1970). In Mormoops and Pteronotus parnellii, m. palmaris longus inserts on digit II. In Pteronotus quadridens, P. davyi, and P. personatus the distal tendons of this muscle are small and difficult to trace (Vaughan and Bateman, 1970); therefore, we have scored Pteronotus with the state Vaughan and Bateman (1970) described for Pteronotus parnellii. Based on the distribution of character states among outgroup taxa, presence of an insertion of m. palmaris longus on digit II appears to be the primitive condition for phyllostomids.

Straney (1980: characters C5–7) based three characters on his descriptions of the insertion of m. palmaris longus: “Palmaris longus with insertion on digits one and two;” “Palmaris longus with insertion on digits one and three;” and “Palmaris longus with insertion on digits one, three, four, and five.” However, Straney (1980) appears to have been unaware of the earlier study by Vaughan and Bateman (1970), and did not discuss how his results compared to theirs. Although Straney (1980) described the distribution of this character in many more phyllostomids than those examined by Vaughan and Bateman (1970), his descriptions frequently conflict with theirs, particularly the description of the insertion of m. palmaris longus onto digit II. Straney (1980) found that m. palmaris longus inserts on digit II in stenodermatines, Phyllostomus, and Macrotus, in contrast to Vaughan's (1959) and Vaughan and Bateman's (1970) reports that these taxa (or representatives of them, e.g., Sturnira) do not have an insertion on digit II. Further, Straney (1980) found that Glossophaga and mormoopids lack an insertion on digit II, whereas Altenbach (1979) and Vaughan and Bateman (1970) reported that an insertion on digit II was present in these taxa. Because Straney's (1980) descriptions differ so greatly from the reports by Vaughan (1959), Vaughan and Bateman (1970), and Altenbach (1979), we did not use Straney's (1980) descriptions to score this character.

Character 80

M. palmaris longus does not insert on digit III (0); or inserts on digit III (1). M. palmarislongus inserts on digit III in Desmodus, most phyllostomines (e.g., Macrotus, Trachops), Erophylla, Anoura, Glossophaga, Lonchophylla, and Carollia (Vaughan, 1959; Vaughan and Bateman, 1970; Altenbach, 1979; Straney, 1980). In contrast, there is no insertion on digit III in Phyllostomus and stenodermatines (Vaughan and Bateman, 1970; Straney, 1980). M. palmaris longus does not insert on digit III in Mormoops, but does insert on this digit in Pteronotus parnellii (Vaughan and Bateman, 1970). In Noctilio there is no insertion on digit III (Straney, 1980). This distribution of character states among the outgroup taxa suggests that the absence of an insertion on digit III is the primitive condition for phyllostomids.

Unlike Straney's (1980) descriptions of the insertion of m. palmaris longus onto digit II (see character 79), his descriptions of the insertion of m. palmaris longus onto digit III agree, for the most part, with those of previous authors. The only differences are found in descriptions of this insertion in Macrotus and mormoopids. In Macrotus, Straney (1980) found no insertion on digit III, whereas Vaughan (1959) reported that an insertion was present on this digit. Straney (1980) reported that m. palmaris longus inserted on digit III in all mormoopids, whereas Vaughan and Bateman (1970) found that it did not insert on this digit in Mormoops. Although these disagreements do exist, we decided to utilize Straney's (1980: 53) text descriptions to score this character in taxa that were not described by previous authors.

Character 81

M. palmaris longus does not insert on digit IV (0); or inserts on digit IV (1). There is no insertion of m. palmaris longus on digit IV in Desmodus, Macrotus, Phyllostomus, Carollia, Artibeus, and Sturnira (Vaughan, 1959; Vaughan and Bateman, 1970; Altenbach, 1979). In contrast, m. palmaris longus inserts on digit IV in Erophylla and Glossophaga (Vaughan and Bateman, 1970). Both Mormoops and Pteronotus parnellii lack an insertion of m. palmaris longus on digit IV (Vaughan and Bateman, 1970), suggesting that this is the primitive condition for phyllostomids.

This character has not been used previously in a phylogenetic analysis. Straney (1980) did not describe the insertion of m. palmaris longus on digit IV.

Character 82

M. palmaris longus does not insert on digit V (1); or inserts on digit V (1). M. palmaris longus does not insert on digit V in Macrotus, Phyllostomus, Artibeus, and Sturnira (Vaughan, 1959; Vaughan and Bateman, 1970). However, in Desmodus, Erophylla, Glossophaga, and Carollia, m. palmaris longus inserts on digit V (Vaughan and Bateman, 1970; Altenbach, 1979). M. palmaris longus does not insert on digit V in Mormoops and Pteronotus parnellii (Vaughan and Bateman, 1970), suggesting that this is the primitive condition for phyllostomids.

This character has not been used previously in a phylogenetic analysis. Straney (1980) did not describe the insertion of m. palmaris longus on digit V.

Character 83

M. flexor digitorum profundus inserts on digit IV (0); or does not insert on digit IV (1). M. flexor digitorum profundus, which originates on the distal tip of the spinous process of the humerus, divides into multiple tendons at the carpus (Vaughan and Bateman, 1970). These tendons insert on digits I and III in all phyllostomids (Vaughan, 1959; Vaughan and Bateman, 1970; Altenbach, 1979). M. flexor digitorum profundus has an additional insertion on digit IV in Phyllostomus (Vaughan and Bateman, 1970). In other phyllostomids, there is no insertion on digit IV (Vaughan, 1959; Vaughan and Bateman, 1970; Altenbach, 1979). Mormoops and Pteronotus have an insertion on digit IV, in addition to the insertions on digits I, III, and V (Vaughan and Bateman, 1970). Presence of an insertion on digit IV appears to be the primitive condition for phyllostomids.

This is the first use of this character in a phylogenetic analysis, although Straney (1980: characters C8–9) based two characters on the insertion of this muscle into digits I, II, and III. Straney (1980), unlike other previous authors, did not find that all phyllostomids had insertions on digits I and III. Instead, Straney (1980: 53) reported that four taxa, Lonchorhina, Macrophyllum, Macrotus, and the outgroup taxon Noctilio, had insertions on digits I and II. Because Straney's (1980) descriptions differ so greatly from the reports by Vaughan (1959), Vaughan and Bateman (1970), and Altenbach (1979), we did not use Straney's (1980) descriptions to score this character.

Character 84

Third metacarpal longer than fourth or fifth (0); or third and fourth metacarpals subequal in length, both longer than fifth (1); or fourth metacarpal longest (2); or fourth and fifth metacarpals subequal in length, both longer than third (3); or fifth metacarpal longest (4); or third and fifth metacarpals subequal in length, both longer than fourth (5); or third, fourth, and fifth metacarpals all subequal in length (6). In Lonchorhina, Micronycteris nicefori, Phyllostomus, glossophagines, lonchophyllines, Ametrida, and Centurio, the third metacarpal is the longest of metacarpals three, four, and five. In Macrophyllum, Mimon crenulatum, Brachyphylla, phyllonycterines, Carollia, and many stenodermatines (e.g., Ariteus, Uroderma) the third and fifth metacarpals are subequal in length and both are always longer than the fourth metacarpal. In contrast, the third and fourth metacarpals are subequal in length and both are always longer than the fifth metacarpal in desmodontines and Micronycteris brachyotis. Finally, in most phyllostomines, Rhinophylla, and several stenodermatines (e.g., Sturnira, Phyllops) the fifth metacarpal is the longest. In only one taxon, Enchisthenes, did we observe the third, fourth, and fifth to all be subequal in length. In mormoopids, the third metacarpal is longer than the fourth and fifth, while in Noctilio, the fourth metacarpal is the longest. Thus, the primitive condition for phyllostomids cannot be determined a priori.

Simmons (1996: character 8) scored the metacarpal formula of species of Micronycteris. We had originally attempted to code this character in a similar fashion to that developed by Simmons (1996; the complete metacarpal formula, e.g., 3 > 4 > 5); however, due to the number of unique metacarpal formulas we discovered (more than nine), we found it more useful to express this diversity by creating a character for the longest metacarpal(s). We cannot create an additional character for the shortest metacarpal because these characters would not be independent in taxa like Brachyphylla, which has two metacarpals that are equal in length and longer than a third.

Character 85

First phalanx of digit III of wing shorter than second phalanx (0); or first and second subequal (1). In most phyllostomids, the first (proximal) phalanx of digit III is noticeably shorter than the second phalanx. In contrast, the first and second phalanges are subequal in Micronycteris hirsuta, M. megalotis, M. minuta, Tonatia, and phyllonycterines. In mormoopids and Noctilio, the first phalanx is shorter than the second, suggesting that this is the primitive condition for phyllostomids.

Simmons (1996: character 9) used this character in her study of relationships among species of Micronycteris; our character states and scoring are identical to hers.

Character 86

First phalanx of digit IV of wing shorter than second phalanx (0); or subequal to second phalanx (1); or longer than second phalanx (2). In most phyllostomids, the first (proximal) phalanx of digit IV is noticeably shorter than the second phalanx. In contrast, the first and second phalanges are subequal in many phyllostomines (e.g., Chrotopterus, Mimon bennettii), carolliines, Ectophylla, Mesophylla, and Vampyressa pusilla. The first phalanx is distinctly longer than the second in some phyllostomines (e.g., Micronycteris hirsuta, Macrotus), Erophylla, Phyllops, and Centurio. Taxonomic polymorphism occurs in three genera. In Tonatia, some species (e.g., T. schulzi) have the first and second phalanges subequal, whereas others (e.g., T. saurophila) have a first phalanx that is longer than the second. We scored Tonatia with states 1 and 2 in the matrix. In Chiroderma and Platyrrhinus, some species have a first phalanx shorter than the second (e.g., C. salvini, P. helleri), whereas others have the first and second phalanges subequal (e.g., C. villosum, P. infuscus). We scored Chiroderma and Platyrrhinus with states 0 and 1 in the matrix. In Mormoops, the first and second phalanges are subequal in length. In Pteronotus and Noctilio, the first phalanx is shorter than the second. This distribution of character states suggests that a first phalanx that is shorter than the second is the primitive condition for phyllostomids.

Simmons (1996: character 10) used this character in her study of relationships among species of Micronycteris. Although our character states and scoring are identical to hers, we have not ordered the character as she did (0 ↔ 1 ↔ 2).

Character 87

Calcar present, equal to or longer than foot (0); or shorter than foot (1); or vestigial or absent (2). The length of the calcar is equal to or greater than the length of the foot (including claws) in most phyllostomines. In contrast, the calcar is noticeably shorter than the foot in Diphylla, several phyllostomines (e.g., Micronycteris nicefori, Phylloderma), Erophylla, glossophagines, lonchophyllines, carolliines, and most stenodermatines (e.g., Artibeus, Uroderma). The calcar is either vestigial or entirely absent in Desmodus, Diaemus, Brachyphylla, Phyllonycteris, and Sturnira. In mormoopids and Noctilio, the calcar is equal to or longer than the foot, suggesting that this is the primitive state for phyllostomids.

Straney (1980: G6–8) developed three binary characters based on the length of the calcar: “Calcar longer than foot,” “Calcar subequal to foot,” and “Calcar shorter than foot.” We found that Straney's (1980) second character (G7) included what we consider to be two distinct conditions. Straney (1980) described the calcar as subequal to the foot in Chrotopterus, Macrotus, some species of Micronycteris, Phylloderma, some species of Phyllostomus, some species of Tonatia, Trachops, Vampyrum, Choeronycteris, Platalina, and unidentified stenodermatines. Although it may be reasonable to describe the calcar as subequal to the foot in Choeronycteris and Platalina, in all glossophagines that we examined, the calcar was consistently shorter than the foot (the same is true of the stenodermatines we examined). The calcar in many phyllostomines is apparently subequal to the foot; however, in several taxa the calcar is clearly longer than the foot (e.g., Macrophyllum), and in at least some species the calcar varies from equal to slightly longer than the foot (e.g., Micronycteris brachyotis; Simmons, 1996). As a result, we recognize “calcar equal to or longer than foot” as a single character state. Due to our revision of character states, our scoring of this character differs from Straney's (1980) for those taxa that he considered to have a calcar subequal to the foot. Our observations agree with those of Simmons (1996: 11) who used this character in an analysis of relationships among species of Micronycteris. We have added a new state “calcar vestigial or absent” that was not used previously by Straney (1980) or Simmons (1996). We confirmed these observations with measurements on pickled specimens where necessary.

Character 88

Posterior edge of plagiopatagium inserts onto lower leg or ankle region (0); or onto calcar (1); or onto lateral surface of first metatarsal (2); or onto dorsal surface of first metatarsal (3). The posterior edge of the plagiopatagium inserts at, or proximal to, the ankle region in desmodontines, many phyllostomines (e.g., Macrophyllum, Macrotus), Brachyphylla, phyllonycterines, glossophagines, lonchophyllines, Carollia, and Sturnira. A unique condition among phyllostomids occurs in Lonchorhina, where the plagiopatagium is attached to the proximal part of the calcar creating a small “pocket” between the plagiopatagium and uropatagium. The posterior edge of the plagiopatagium inserts onto the lateral metatarsal in Micronycteris hirsuta, Phylloderma, Phyllostomus, Rhinophylla, and stenodermatines (except Sturnira). The plagiopatagium is attached to the dorsal surface of the first metatarsal in Chrotopterus, Tonatia, and Vampyrum. In mormoopids, the plagiopatagium is attached to the proximal part of the calcar creating a small “pocket” between the plagiopatagium and the uropatagium; however, the plagiopatagium is attached proximal to the ankle on the lower leg in Noctilio. Thus, the primitive condition for phyllostomids cannot be assessed a priori.

Straney (1980: characters G9–13) scored attachment site of the plagiopatagium in a series of binary characters. We have subsumed Straney's (1980) characters G9 (“wing attaches to lateral side of ankle”) and G13 (“wing attaches on lower leg”) into a single state (our state 0) because we were unable to divide the continuous range of variation of ankle/lower leg attachments that we observed into these discrete states. In addition to changing the character coding, we did not observe the attachment of the plagiopatagium to the calcar (scored as a vento-medial ankle attachment of the plagiopatagium in Straney's [1980] matrix) in Macrotus and Mimon bennettii, and found that Tonatia shares the same condition seen in Chrotopterus and Vampyrum, an attachment to the dorsal surface of the first metatarsal.

Owen (1987: character 6) also used this character, recognizing four states: attachment to the tibial region 0, attachment to the tarsal region 1, attachment to the metatarsal region 2, and attachment to the metatarsal-phalangeal joint 3 (ordered 0 ↔ 1 ↔ 2 ↔ 3). We have combined his first two states into our state 0, and the latter two conditions into our state 2, because of our finding that these conditions were variable within genera, or our inability to separate the variation we observed into these discrete states. However, our observations largely agree with his, with the exception of Sturnira nana. Owen (1987) observed that the plagiopatagium in S. nana is attached to the metatarsal region. However, we found that it inserts at the ankle, as in other Sturnira species. Owen (1987) also reported that in Chiroderma doriae, the attachment of the plagiopatagium is to the ankle region rather than the metatarsals as is the case in other Chiroderma. Unfortunately, we were unable to examine specimens of C. doriae and therefore score Chiroderma with state 2 in the matrix. Owen (1991: characters 14, 15, 16) used additive binary coding, but both his characters and scoring are equivalent to his earlier work.

Marques-Aguiar (1994: character 7) also used this character in her analysis of relationships among large-bodied Artibeus species. Marques-Aguiar (1994) recognized three states: plagiopatagium attaches at side of foot, attaches at ankle, or attaches at base of toes. We combined her state 0 (side of foot) with 1 (base of toes), because both states are often present within genera. Our observations agree with those of Marques-Aguiar (1994) with one exception. Marques-Aguiar (1994) reported that the plagiopatagium attaches to ankle in Enchisthenes hartii, whereas we observed (as did Owen, 1987) that in this species the attachment is to the metatarsals.

Character 89

Tail of medium length, extends into uropatagium but is shorter than hind legs (0); or tail long, approximately equal to length of hind legs (1); or tail effectively absent, caudal vertebrae do not extend into uropatagium (2). In most phyllostomines, phyllonycterines, most glossophagines, lonchophyllines, and Carollia, the tail is of moderate length and is shorter than the hind legs. A very long tail, approximately equal in length to the hind legs, is present in Lonchorhina, Macrophyllum, and Macrotus. The tail is effectively absent in desmodontines, Vampyrum, Brachyphylla, Leptonycteris, Rhinophylla, and stenodermatines. Taxonomic polymorphism occurs in Anoura where some species have a medium length tail (e.g., A. caudifera), whereas others lack a tail (e.g., A. geoffroyi). We scored Anoura with states 0 and 2 in the matrix. Examination of caudal vertebrae reveals that one or two small vertebrae may extend beyond the ischia in some individuals of these taxa. For example, some specimens of Anoura geoffroyi (e.g., USNM 49356) lack postischial vertebrae, whereas others have two vestigial postischial vertebrae (e.g., USNM 385799). As originally noted by Miller (1907), all specimens of Brachyphylla that we examined had three long, cylindrical postischial vertebrae similar to those found in all taxa with external tails; however, these vertebrae do not extend into the uropatagium. Because the condition in Brachyphylla externally appears identical to the condition in other taxa lacking tails, we scored this genus with state 2 in the matrix. In mormoopids and Noctilio, the tail is of moderate length, suggesting that this is the primitive condition for phyllostomids.

Straney (1980: G1–3) developed three binary characters devoted to tail morphology: “Tail present,” “Tail short,” and “Tail to edge of uropatagium,” whose derived conditions correspond to our character states. Our scoring agrees with that of Straney (1980) with the exception of Rhinophylla. Straney (1980) scored all carolliines as possessing a medium-length tail; however, an externally visible tail is absent in Rhinophylla. We have chosen to define tail length in relation to the hind legs rather than the uropatagium (as Straney [1980] did) because the uropatagium is more variable in length.

Character 90

M. mylohyoideus undivided (0); or partly divided into anterior and posterior parts by a fleshy aponeurosis (1); or with a pronounced break, clearly divided into distinct anterior and posterior parts (2). M. mylohyoideus consists of a single, undivided sheet of muscle in Desmodus, glossophagines, lonchophyllines, and Carollia (Wille, 1954; Griffiths, 1982; fig. 31A). Sturnira also has a single muscle belly (Sprague, 1943). In Brachyphylla and phyllonycterines, m. mylohyoideus is partly divided into anterior and posterior parts by a fleshy aponeurosis (Griffiths, 1982). In contrast, m. mylohyoideus exhibits a pronounced break and is clearly divided into distinct anterior and posterior parts in phyllostomines and most stenodermatines (Wille, 1954; Griffiths, 1982; fig. 31C). M. mylohyoideus is undivided in Pteronotus and Noctilio (Sprague, 1943), suggesting that this is the primitive condition for phyllostomids.

Straney (1980: characters E9–11) used three binary characters to describe variation in the morphology of m. mylohyoideus. Our character states 0 and 2 correspond to the derived conditions of Straney's (1980) characters E9 and E11, respectively. We did not include the information contained in Straney's (1980) character E10, which was based on Sprague's (1943) description of the entire anterior portion of the m. mylohyoideus as aponeurotic in Phyllonycteris and Glossophaga (a condition that is not equivalent to having two muscle masses separated by a fleshy aponeurosis). Griffiths (1982) observed that the mylohyoid was sometimes thin enough to reveal the geniohyoid deep to it, but that this condition was variable within glossophagine species. Morphology of m. mylohyoideus was not included in Griffiths' (1982) character analysis, or in that of Gimenez et al. (1996).

Character 91

Medial fibers of m. sternohyoideus originate from medial manubrium (0); or from mesosternum (1); or from xiphoid process of sternum (2). The medial fibers of m. sternohyoideus originate from the medial manubrium in phyllostomines, Carollia, and stenodermatines (including Sturnira; Sprague, 1943; Griffiths, 1982). The origin is shifted posteriorly to the mesosternum in Desmodus, Brachyphylla, and phyllonycterines (Griffiths, 1982). In glossophagines and lonchophyllines, the medial fibers of m. sternohyoideus originate entirely from the xiphoid process (Griffiths, 1982). Among the outgroups, the manubrium is the origin of the medial fibers of m. sternohyoideus in Pteronotus (Sprague, 1943). In Noctilio, m. sternohyoideus is not clearly differentiated into groups of medial and lateral muscle fibers. Instead, the single origin of all fibers in this taxon is the manubrium (Sprague, 1943). Nevertheless, the homologs of the medial and lateral fibers found in other phyllostomids are presumably present in Noctilio, thus we have scored Noctilio for this character and character 92. The distribution of character states in the outgroups suggests that a manubrial origin is the primitive condition for phyllostomids.

Griffiths (1982: characters 1–2) scored variation in the origin of m. sternohyoideus as two binary characters: “posterior shift of sternohyoid origin,” and “xiphoid origin of sternohyoid.” Gimenez et al. (1996) recently used this character and apparently followed Smith and Hood (1984) in regarding Griffiths' (1982) characters 1 and 2 as states of the same character. Gimenez et al.'s (1996: character 17) single multistate character is identical to ours in construction and coding. Griffiths' (1982) characters (which are subsumed under states 1 and 2 above) did not account for differences in the origins of medial versus lateral fibers of m. sternohyoideus. Accordingly, we chose to score variation in the origin of m. sternohyoideus as two multistate characters (our characters 91 and 92) which separately describe the origin of the two sets of fibers.

Character 92

Lateral fibers of m. sternohyoideus originate from manubrium (0); or originate from manubrium and clavicle (1); or originate from clavicle and first rib (2); or originate from xiphoid process (3). The lateral fibers of m. sternohyoideus originate from the manubrium in Sturnira (Sprague, 1943). These fibers originate from the lateral manubrium and clavicle in phyllostomines, Erophylla, Carollia, and stenodermatines (except Sturnira; Griffiths, 1982). The lateral fibers of m. sternohyoideus originate from the clavicle and the proximal head of the first rib in Desmodus, Brachyphylla, and Phyllonycteris (Griffiths, 1982). In glossophagines and lonchophyllines, the lateral fibers of m. sternohyoideus originate from the xiphoid process (Griffiths, 1982). Sprague (1943) noted that the lateral fibers of m. sternohyoideus originate from the manubrium in Pteronotus and Noctilio; thus, a manubrial origin appears to be primitive for phyllostomids.

Character 93

M. sternohyoideus inserts via tendon on basihyal (0); or via raphe into the fibers of m. hyoglossus and m. genioglossus (1). M. sternohyoideus inserts via tendon to the basihyal in Desmodus, phyllostomines, Brachyphylla, phyllonycterines, Carollia, and stenodermatines (Griffiths, 1982). In contrast, this muscle inserts via a raphe into the fibers of m. hyoglossus and m. genioglossus in glossophagines and lonchophyllines (Griffiths, 1982). Typically, there is no muscular connection to the hyoid apparatus in these taxa. A remnant of a basihyal tendon was found by Griffiths (1982) in one individual each of Lionycteris spurelli and Lonchophylla robusta, but the principal insertion was identical to that of other lonchophyllines. Accordingly, we scored these taxa with state 1 in the matrix. Although Sprague (1943) reported the insertion of m. sternohyoideus in Pteronotus, Noctilio, and Sturnira, these three taxa are scored “?” in the matrix (see discussion below). Thus, the primitive condition for phyllostomids cannot be reconstructed a priori.

Straney (1980), Griffiths (1982), and Gimenez et al. (1996) previously used this character. Our character states are equivalent to those recognized by these authors, although Griffiths (1982) and Gimenez et al. (1996) described what they believed to be the derived condition (our state 1) differently. Griffiths (1982: character 3) described the derived condition as “loss of sternohyoid's connection to hyoid bone,” whereas Gimenez et al. (1996: character 18) described the derived condition of their character (“connection of sternohyoideus muscle to hyoid bone”) as “dissociated.” Straney (1980: characters E19–20) erected two binary characters whose derived conditions are equivalent to our character states: “Sternohyoid insertion on basihyal,” and “Sternohyoid insertion on raphe ventral [to] basihyal.”

Although our scoring of this character agrees with the work of Griffiths (1982) and Giminez et al. (1996), Straney (1980) scored Desmodus and glossophagines as possessing an insertion on the basihyal element, while he scored all other taxa for which data was available as possessing an insertion on the basihyal raphe. In the text, however, Straney (1980: 32) noted that Desmodus was unique among noctilionoid bats in possessing the insertion on the basihyal. Straney's (1980) description agrees with Sprague's (1943) discussion of this character. Sprague (1943) described an insertion on the basihyal raphe for Phyllostomus, Phyllonycteris, Glossophaga, Carollia, Artibeus, Sturnira, Pteronotus, and Noctilio. Desmodus possessed an insertion on the basihyal by means of a tendon (Sprague, 1943). Although Griffiths' (1982) examination of the hyoid apparatus demonstrated that Sprague's (1943) description of an insertion in the basihyal raphe is inaccurate for most taxa, no additional data are available for Pteronotus, Noctilio, or Sturnira. Because it seems unlikely that Sprague's (1943) description of this character is correct, we have chosen to score these taxa “?” in the matrix.

Wille (1954) reported that this muscle originates on the xiphoid process of the sternum and inserts into the tongue in glossophagines and lonchophyllines. As Griffiths (1978) noted, Wille (1954) seemed unaware that m. hyoglossus is greatly elongated in glossophagines and lonchophyllines. The muscle segment that Wille (1954) considered part of the sternohyoideus is actually part of m. hyoglossus (Griffiths, 1978).

Character 94

Part of m. ceratohyoideus inserts on ceratohyal (0); or m. ceratohyoideus does not insert on ceratohyal (1). Part of m. ceratohyoideus inserts on the ceratohyal element in Micronycteris nicefori, Brachyphylla, phyllonycterines, glossophagines, lonchophyllines, Carollia, and stenodermatines (including Sturnira; Sprague, 1943; Griffiths, 1982). In contrast, this muscle does not insert on the ceratohyal in Desmodus, Macrotus, and Phyllostomus (Griffiths, 1982). Part of m. ceratohyoideus inserts on the ceratohyal element in Pteronotus (Sprague, 1943; Griffiths, 1983b) and Noctilio (Sprague, 1943), suggesting that this condition is primitive for phyllostomids.

Straney (1980), Griffiths (1982), and Gimenez et al. (1996) did not include any characters based on m. ceratohyoideus in their analyses, although Griffiths (1982) clearly described the variation that we have scored in our characters 94 and 95. Although Sprague (1943) described the attachment of m. ceratohyoideus to the ceratohyal in both Pteronotus and Noctilio, he referred to this connection as an origin, and noted that the insertion of this muscle in these taxa was on the thyrohyal.

Character 95

M. ceratohyoideus does not insert on stylohyal (0); or part of m. ceratohyoideus inserts on stylohyal (1). M. ceratohyoideus does not insert on the stylohyal element in Brachyphylla, phyllonycterines, many glossophagines (e.g., Choeroniscus, Choeronycteris), lonchophyllines, Carollia, and some stenodermatines (e.g., Artibeus, Uroderma; Griffiths, 1982; Sturnira; Sprague, 1943). In contrast, part of m. ceratohyoideus inserts on the stylohyal in Desmodus, phyllostomines, some glossophagines (e.g., Glossophaga, Leptonycteris), Platyrrhinus, and Vampyressa pusilla (Griffiths, 1982). M. ceratohyoideus does not insert on the stylohyal element in Pteronotus (Griffiths, 1983b) and Noctilio (Sprague, 1943), suggesting that this is the primitive condition for phyllostomids.

Character 96

M. hyoglossus originates via tendon from basihyal bone (0); or from raphe which forms insertion of m. sternohyoideus (1). M. hyoglossus originates via tendon from the basihyal bone in Desmodus, phyllostomines, Brachyphylla, phyllonycterines, Carollia, and stenodermatines (Griffiths, 1982). In contrast, m. hyoglossus originates from the raphe, which forms the insertion of m. sternohyoideus (the former basihyal raphe, now disconnected from the basihyal bone) in glossophagines and lonchophyllines (Griffiths, 1982). There is no direct connection of m. hyoglossus to the bones of the hyoid apparatus in these taxa. Although Sprague (1943) reported the origin of m. hyoglossus in Pteronotus, Noctilio, and Sturnira, these three taxa are scored “?” in the matrix (see discussion below). The primitive condition for phyllostomids cannot be reconstructed a priori.

Straney (1980: characters E21–22) used origin of m. hyoglossus as two binary characters, whose derived conditions are identical to our character states. However, Straney (1980) scored the raphe origin (our character state 1) as occurring in phyllostomines, brachyphyllines, carolliines, stenodermatines, Pteronotus, and Noctilio. Straney (1980) reported an origin from the basihyal (our state 0) in glossophagines and desmodontines. Based on Sprague's (1943) descriptions, only Straney's coding of Noctilio, which Sprague described as having an origin from the basihyal and thyrohyal, was inaccurate. Although Griffiths' (1982) examination of the hyoid apparatus demonstrated that Sprague's (1943) description of the origin of m. hyoglossus from the basihyal raphe was inaccurate for most taxa, no additional data are available for Pteronotus, Noctilio, or Sturnira. Because it seems unlikely that Sprague's (1943) description of this character is correct, we have chosen to score these taxa “?” in the matrix. Our scoring of this character was identical to that of Griffiths (1982: character 4), although he described the derived condition (our state 1) as “hyoglossus elongated and loses connection to hyoid bone.” Our character states are also equivalent to those used by Gimenez et al. (1996: character 19), who described the derived condition in terms similar to those used by Griffiths (1982). Wille (1954: 317) reported that this muscle extends “from basihyal to basihyal raphe in geniohyoideus” in glossophagines and lonchophyllines. Given the observations of Griffiths (1982), this description appears to be incorrect.

Character 97

M. geniohyoideus has single insertion via tendon to basihyal or basihyal raphe (0); or muscle splits near insertion, deep fibers insert directly on anterior surface of basihyal, superficial fibers insert in association with m. hyoglossus and m. sternohyoideus (1). M. geniohyoideus has a single insertion via tendon to the basihyal bone in Desmodus, phyllostomines, Brachyphylla, phyllonycterines, Carollia, and stenodermatines (Griffiths, 1982). In contrast, m. geniohyoideus splits as it passes posteriorly in lonchophyllines and glossophagines (Griffiths, 1982). The deep fibers of m. geniohyoideus insert directly on the anterior surface of the basihyal bone, while the superficial fibers form an insertion in association with m. hyoglossus and m. sternohyoideus (Griffiths, 1982). Although Sprague (1943) reported the origin of m. geniohyoideus in Pteronotus, Noctilio, and Sturnira, these three taxa are scored “?” in the matrix (see discussion below). The primitive condition for phyllostomids cannot be reconstructed a priori.

Straney (1980: characters E17–18) defined two binary characters to describe the insertion of m. geniohyoideus: “Geniohyoid insertion on basihyal” and “Geniohyoid insertion on raphe ventral to basihyal.” The derived conditions of these characters are essentially equivalent to our character states, although Straney (1980), and Sprague (1943), who was his source, were apparently unaware that deep fibers of m. geniohyoideus retain their attachment to the basihyal in forms with the “raphe” insertion of the superficial fibers. Additionally, Straney (1980) scored all noctilionoids, except Phyllostomus and Noctilio, as having the insertion on the basihyal, which conflicts with Sprague's (1943) description. Sprague (1943) found that in Phyllostomus, Phyllonycteris, Carollia, Artibeus, Sturnira, Pteronotus, and Noctilio, the insertion of m. geniohyoideus was on the basihyal raphe, whereas in Desmodus and Glossophaga the insertion was on the entoglossal process of the basihyal. Sprague (1943) found no evidence of split insertions in these taxa. Although Griffiths' (1982) examination of the hyoid apparatus demonstrated that Sprague's (1943) description of the insertion of m. geniohyoideus was inaccurate for most taxa, no additional data are available for Pteronotus, Noctilio, or Sturnira. Because it seems unlikely that Sprague's (1943) description of this character is correct, we have chosen to score these taxa “?” in the matrix.

Our scoring of this feature was identical to that of Griffiths (1982: character 5) and Giminez et al. (1996: character 20), although both described the derived condition (our state 1) as “double insertion of geniohyoid.” Wille (1954) also discussed the morphology of m. geniohyoideus in glossophagine and lonchophylline bats, describing an anterior and posterior part to this muscle. For the anterior part of the muscle, Wille (1954) only noted an insertion into the basihyal raphe; he was apparently unaware that there was a deep insertion of this muscle to the basihyal bone.

Character 98

Superficial fibers of m. geniohyoideus pass ventral to basihyal and insert into fibers of m. hyoglossus and m. sternohyoideus via raphe (0); or superficial fibers insert in well-developed loop around ventral and dorsal surfaces of the intersection of m. hyoglossus and m. sternohyoideus (1). For taxa with a split insertion of m. geniohyoideus, two conditions are possible for the insertion of the superficial fibers of this muscle. In some glossophagines (e.g., Glossophaga, Lichonycteris) and lonchophyllines, the superficial fibers pass ventral to the basihyal to form a weak insertion into fibers of m. hyoglossus and m. sternohyoideus via the former basihyal raphe (Griffiths, 1982). In the remaining glossophagines (e.g., Choeronycteris, Hylonycteris), the superficial fibers of m. geniohyoideus insert in a well-developed loop around ventral and dorsal surfaces of the intersection of m. hyoglossus and m. sternohyoideus. This “tunnel insertion” is unique among mammals (Griffiths, 1982). The primitive state for phyllostomids cannot be determined a priori because we scored Pteronotus and Noctilio “?” for character 97. We scored all taxa lacking a split insertion of m. geniohyoideus (state 0 of character 97), “−” for this character.

Griffiths (1982: character 6) scored this feature as “tunnel insertion of geniohyoid” (present'absent). Griffiths (1982) scored taxa lacking a split insertion and those with a split insertion but lacking a “tunnel” insertion as “absent.” Gimenez et al. (1996: character 21) scored this character similarly; their character states correspond to ours. Sprague (1943) did not describe the tunnel insertion, consequently this character was not used by Straney (1980).

Wille (1954) described the posterior part of m. geniohyoideus as forming a muscular tube that surrounds the sternohyoideus (= m. hyoglossus; Griffiths, 1978). Wille (1954) reported that this tube was present in all taxa he examined except Glossophaga and Monophyllus, where the posterior part of m. geniohyoideus formed a median belly between the two bellies of the sternohyoideus (= m. hyoglossus; Griffiths, 1982). Wille (1954) therefore does not agree with Griffiths (1982) on the taxonomic distribution of this character, because Griffiths (1982) reported that the tunnel insertion does not occur in Lonchophylla, whereas Wille (1954) reported the presence of the tunnel insertion in this taxon. We have chosen to follow Griffiths (1982) in this instance.

Character 99

Right and left m. geniohyoideus muscles partly or completely fused across midline (0); or muscles not fused (1). M. geniohyoideus is partly or completely fused with its counterpart across the midline in Desmodus and Glossophaga (Sprague, 1943). In contrast, the right and left m. geniohyoideus are not fused in Phyllostomus, Phyllonycteris, Carollia, Sturnira, and Artibeus (Sprague, 1943). These muscles are fused in Pteronotus and Noctilio (Sprague, 1943), suggesting that this condition is primitive for phyllostomids.

Straney (1980: characters E12–14) described fusion of m. geniohyoideus with three binary characters. We combined two of his characters, “Geniohyoids fused into single muscle” and “Geniohyoids unfused anteriorly,” into a single character state because Sprague (1943) found that these muscles are never fused near their origins. Sprague's (1943) taxonomic descriptions disagree with the distributional account of this character appearing in his discussion section. We have used the description given in the discussion section (Sprague, 1943: 463) to score this character. Griffiths (1982) did not mention fusion of m. geniohyoideus with its counterpart and consequently this character was not used by Gimenez et al. (1996).

Character 100

M. styloglossus inserts on lateral surface of tongue along much of its length (0); or inserts on posterolateral “corner” of tongue (1). M. styloglossus inserts on the lateral surface of the tongue along much of its length in Desmodus, phyllostomines, Brachyphylla, phyllonycterines, lonchophyllines, Carollia, and stenodermatines (Griffiths, 1982; including Sturnira; Sprague, 1943; fig. 31C–D). In contrast, m. styloglossus inserts on the posterolateral “corner” of the tongue in glossophagines (Griffiths, 1982; fig. 31A–B). M. styloglossus inserts on the lateral surface of the tongue along much of its length in Pteronotus and Noctilio (Sprague, 1943), suggesting that this is the primitive condition for phyllostomids.

Our scoring of this character is identical to that of Griffiths (1982: character 7) and Gimenez et al. (1996: character 22), although Griffiths (1982) described the derived condition (our state 1) as “posterior shift of styloglossus insertion.”

Character 101

M. genioglossus inserts into ventral surface of tongue along more than half of its length (0); or inserts into posterior half to third of ventral surface of tongue (1); or inserts into posterior quarter of ventral surface of tongue (2). M. genioglossus inserts into the ventral surface of the tongue along much of its length in Desmodus, phyllostomines, Brachyphylla, phyllonycterines, Carollia, and stenodermatines (Griffiths, 1982). In contrast, this muscle inserts into the posterior half to third of the ventral surface of tongue in lonchophyllines (Griffiths, 1982). In glossophagines, this muscle inserts into the posterior quarter of the ventral surface of tongue (Griffiths, 1982). Although Sprague (1943) provided a description of the insertion of m. genioglossus in the bats he surveyed, the description is not detailed enough to permit us to score Pteronotus, Noctilio, and Sturnira. Because we cannot score our outgroups, the primitive state for phyllostomids cannot be reconstructed a priori.

In Griffiths' (1982: character 8) matrix the derived condition of this feature was “posterior shift of genioglossus insertion” (present'absent). Griffiths' (1982) descriptions, however, suggest that multiple conditions exist. Therefore, we followed Gimenez et al. (1996: character 23) in recognizing the three character states defined above.

Character 102

M. stylohyoideus absent (0); or present (1); or sometimes present; polymorphic within species (2). M. stylohyoideus is consistently absent in Desmodus, phyllostomines, Brachyphylla, most glossophagines (e.g., Anoura, Choeronycteris), lonchophyllines, and stenodermatines (including Sturnira; Sprague, 1943; Wille, 1954; Griffiths, 1982). In contrast, m. stylohyoideus is present in phyllonycterines, some glossophagines (e.g., Glossophaga, Lichonycteris), and Carollia (Griffiths, 1982). This muscle is variably present within species of Leptonycteris (Griffiths, 1982). M. stylohyoideus is present, but reduced and fused to m. stylopharyngeus, in Pteronotus parnellii (Griffiths, 1983b). This muscle is absent in Noctilio (Sprague, 1943). Thus, the primitive condition for phyllostomids cannot be established a priori.

Although our character states 0 and 1 are identical to those used by Griffiths (1982: character 9) and Gimenez et al. (1996: character 24), we have erected a separate character state to deal with the variation found within the species of Leptonycteris that Griffiths (1982) examined. Both Griffiths (1982) and Gimenez et al. (1996) scored Leptonycteris as 0'1, which is not an appropriate coding for situations of within-species variation (see Simmons, 1993).

Character 103

Anterolateral slip of m. sphincter colli profundus present (0); or absent (1). In bats, m. sphincter colli profundus originates from the fascia of the mylohyoid region, forms a variable number of slips, and inserts on the inner surface of the skin of the cervical region (Sprague, 1943; Griffiths, 1982). An anterolateral slip is present in Desmodus, phyllostomines, Brachyphylla, lonchophyllines, Carollia, and stenodermatines (Griffiths, 1982; including Sturnira; Sprague, 1943). In contrast, this slip is absent in phyllonycterines and glossophagines (Griffiths, 1982). Among the outgroups, an anterolateral slip is absent in Pteronotus, but present in Noctilio (Sprague, 1943), so the primitive condition for phyllostomids cannot be determined a priori.

Griffiths (1982: character 10) treated the condition of m. sphincter colli profundus as one character, “reduction of m. sphincter colli profundus” (present'absent). However, this arrangement did not accommodate the variation Griffiths (1982) described in this muscle (see discussion in Smith and Hood, 1984). Gimenez et al. (1996: character 25) attempted to score the variation described by Griffiths (1982) using four character states: “(0) well-developed, with two (or three) slips; (1) reduced to a few fibers (originated from the fascia of posterior milohyoid [sic] region); (2) almost complete reduction (eventually vestigial); (3) completely absent.” We have chosen to divide the character somewhat differently and treat presence or absence of anterolateral and lateral slips in m. sphincter colli profundus as two characters (characters 103 and 104).

Character 104

Lateral slip of m. sphincter colli profundus present (0); or absent (1). The lateral slip of m. sphincter colli profundus is present in phyllostomines, Brachyphylla, many glossophagines (e.g., Glossophaga, Leptonycteris), lonchophyllines Carollia, and stenodermatines (Griffiths, 1982), except Sturnira where this slip is absent (Sprague, 1943). This slip is also absent in Desmodus, phyllonycterines, Choeroniscus, and Choeronycteris (Griffiths, 1982). These slips are absent in Pteronotus and Noctilio (Sprague, 1943), suggesting that this condition is primitive for phyllostomids.

Character 105

Lateral slip of m. sphincter colli profundus passes laterally (0); or passes anterolaterally to insert on skin of cervical region (1). The lateral slip of m. sphincter colli profundus passes laterally from its origin in the fascia of the mylohyoid region to insert on the inner surface of the skin of the cervical region in phyllostomines, Brachyphylla, glossophagines that have this slip, lonchophyllines, and Carollia (Griffiths, 1982). In contrast, the lateral slip of m. sphincter colli profundus passes anteriorly and laterally in stenodermatines that have this slip (Griffiths, 1982). The primitive condition of this character for phyllostomids could not be established a priori because both Pteronotus and Noctilio lack the lateral slip of this muscle (Sprague, 1943). We scored taxa lacking the lateral slip of m. sphincter colli profundus (state 1 of character 104) “−” for this character.

This feature was not treated in Griffiths' (1982) or Gimenez et al.'s (1996) character analyses because stenodermatines, the only taxa with state 1, were not the focus of their studies.

Character 106

M. cricopharyngeus consists of a single large slip (0), or two slips (1), or three slips (2), or more than three slips (3). M. cricopharyngeus consists of a single large slip in Desmodus. In Brachyphylla, phyllonycterines, and lonchophyllines, this muscle is composed of two distinct slips or bellies (Griffiths, 1982). M. cricopharyngeus consists of three slips in phyllostomines, Carollia, and stenodermatines (including Sturnira; Sprague, 1943; Griffiths, 1982). In glossophagines, the muscle consists of “several slips” (although Griffiths [1982:18] is not clear, we interpret this as more than three slips). A single undivided slip is present in Pteronotus (Griffiths, 1983b) and Noctilio (Sprague, 1943), suggesting that this condition is primitive for phyllostomids.

Griffiths (1982: character 11) scored the morphology of m. cricopharyngeus as “cricopharyngeus reduced to two bellies” (present'absent), apparently assuming three slips was the primitive condition. Gimenez et al. (1996: character 26) recognized three states of this character that are equivalent to our states 1, 2, and 3. Although Griffiths (1982: 18) noted that this muscle consists of “several slips” in glossophagines (s.s.), he did not give the number of slips present. The single-slip condition seen in Desmodus and the outgroups was not included in Griffiths' (1982) or Gimenez et al.'s (1996) character analysis.

Character 107

Medial circumvallate papillae present (0); or absent (1). In most phyllostomids, two medial circumvallate papillae (MVPs) are present proximally on the dorsum of the tongue, one on either side of the midline (figs. 33A, 34A–C). In contrast, the MVPs are absent in desmodontines and some glossophagines (e.g., Anoura, Choeroniscus; fig. 32C). Mormoopids and Noctilio have two MVPs on the dorsum of the tongue (fig. 32A–B), suggesting that this is the primitive condition for phyllostomids.

Griffiths (1982: character 15) first used this character in a cladistic analysis, erecting states equivalent to ours. However, Griffiths (1982) did not score lonchophyllines with any state for this character (“see text” appeared in his matrix instead of “+” or 0) to indicate that lonchophyllines lack both medial and lateral circumvallate papillae (see below). Gimenez (1993: characters 3.3.2.1, 3.3.5.1) and Gimenez et al. (1996: character 4) also used this character. Gimenez (1993) described the derived condition (absence of MVPs, her state 1) as MVPs vestigial or absent, whereas Gimenez et al. (1996) erected states identical to ours.

Griffiths (1982) reported the absence of MVPs not only in the taxa noted above, but also in Lionycteris, Lonchophylla, and Platalina. Griffiths and Criley (1989) also noted that Lonchophylla lacks all vallate papillae. In contrast, we observed MVPs in every specimen of Lionycteris, Lonchophylla, and Platalina that we examined (see appendix 1 for species and specimen numbers). Our results agree with the findings of Smith and Hood (1984), who reported the presence of these structures in Lionycteris spurelli and Lonchophylla robusta, and with those of Gimenez (1993) and Gimenez et al. (1996), who observed MVPs in Lionycteris purelli, Lonchophylla bokermanni, L. dekeyseri, and L. mordax, although Gimenez (1993) had scored these papillae as “vestigial or absent” in L. bokermanni.

Griffiths (1982) reported that Lichonycteris has MVPs, whereas Gimenez (1993) and Gimenez et al. (1996) did not find MVPs in the specimens that they examined. Giminez (1993) coded Lichonycteris 0'1 to account for her observations and those of Griffiths (1982). We agree with Gimenez (1993) and Gimenez et al. (1996) that Lichonycteris lacks MVPs. Although Smith and Hood (1984) noted that Centurio lacks MVPs, the specimens that we examined all possess very small MVPs that are located at the proximal extreme of the tongue. In addition, the papillary bodies of the MVPs are partly fused to the valla (see character 108), which makes observation very difficult and may account for the report of Smith and Hood (1984).

Gimenez (1993) and Gimenez et al. (1996) reported that MVPs were variably present in the specimens of Lonchophylla thomasi they examined (present in only three of the six specimens; Gimenez et al., 1996). We examined six specimens of L. thomasi and found MVPs present in all. Griffiths (1982: table 1) scored Leptonycteris as “variable” for this character, though in the text he noted that only one specimen of five L. sanborni (= L. curasoae) lacked one MVP. Our observations of Leptonycteris curasoae and L. nivalis agree with those of Park and Hall (1951), Greenbaum and Phillips (1974), Smith and Hood (1984), Gimenez (1993) and Gimenez et al. (1996), who reported that the MVPs were always present in Leptonycteris. Based on our observations, we have chosen to score both Leptonycteris and Lonchophylla as having MVPs (state 0).

Character 108

Papillary bodies of medial circumvallate papilla separate from valla, fossae complete (0); or papillary bodies fused anterolaterally and posteromedially to valla, fossae incomplete (1). In most phyllostomids, the fossa of each MVP is complete and the papillary body is not fused to the vallum (fig. 35A). In contrast, in Centurio and Sphaeronycteris each MVP papillary body is fused with the vallum anterolaterally and posteromedially (fig. 35B). Consequently, the fossae are not circular furrows, but appear as paired slits on each side of the papillary body. In mormoopids and Noctilio, the MVP bodies are not fused to the valla and the fossae are complete, suggesting that this condition is primitive for phyllostomids. We scored the eight taxa lacking medial vallate papillae (state 1 of character 107) “−” for this character.

Smith and Hood (1984) examined Centurio, but incorrectly reported that the MVPs were absent in this genus (see character 107). No previous study of tongue morphology has included Sphaeronycteris. Consequently, this character has not been used in previous cladistic analyses.

Character 109

Lateral circumvallate papillae present (0); or absent (1). Most phyllostomids have a pair of circumvallate papillae on the proximal part of the lateral surfaces of the tongue (figs. 33A–C, 34A–C). Desmodontines, Brachyphylla, and Centurio lack these lateral circumvallate papillae (LVPs; fig. 32C). LVPs are present in mormoopids, but absent in Noctilio (fig. 32A–B). Thus, the primitive condition for phyllostomids cannot be reconstructed a priori.

Griffiths (1982) did not use presence of LVPs as a character, but did note the loss of all circumvallate papillae in lonchophyllines in his matrix (see discussion under character 108). Gimenez (1993: characters 3.3.2.2, 3.3.5.3) and Gimenez et al. (1996: character 6) used presence of LVPs as a character and defined states identical to ours.

Griffiths (1982) indicated that lonchophyllines, Brachyphylla, and desmodontines lacked LVPs, and Griffiths and Criley (1989) noted that Lonchophylla lacks all vallate papillae. However, our observations agree with those of Smith and Hood (1984), Gimenez (1993), and Gimenez et al. (1996) who found that LVPs are present in all Lionycteris and Lonchophylla specimens they examined. Because specimens of Lonchophylla robusta and Platalina were not available for study, Giminez (1993) accepted Griffiths' (1982) conclusion that these papillae appeared to be absent in these taxa. Our observations of LVPs in all specimens of Lionycteris, Lonchophylla, and Platalina that we examined (see appendix 1 for species and specimen numbers) confirm that these papillae are present in lonchophyllines. Unlike Greenbaum and Phillips (1974), who reported the absence of one of the two LVPs in one specimen each of Leptonycteris curasoae and L. nivalis, we did not find missing LVPs in any representative of Leptonycteris we examined. Park and Hall (1951) reported the presence of only two vallate papillae in most taxa they examined (including Leptonycteris); however, as Greenbaum and Phillips (1974) noted, it appears these authors simply overlooked the lateral vallate papillae, reporting only on the presence of medial vallate papillae. Finally, Gimenez et al. (1996) indicated that all stenodermatines possess lateral vallate papillae. This study, however, did not include Centurio, which lacks these papillae (Smith and Hood, 1984; this study).

Character 110

Lateral circumvallate papillae located on lateral surface of tongue (0); or in dorsolateral position, at border between lateral and dorsal surfaces of tongue (1); or on dorsal surface of tongue (2). In most phyllostomids, LVPs are located on the lateral surface of the tongue (figs. 33A–C, 34A). However, in some stenodermatines (e.g., Uroderma, Vampyrodes), the LVPs are located at the steeply curved “border” between the dorsal and lateral surfaces of the tongue (fig. 34B). In other stenodermatines (e.g., Ametrida, Artibeus), the LVPs are located on the dorsal surface of the tongue (fig. 34C). In some of these taxa (e.g., Pygoderma), the LVPs are almost directly adjacent to the MVPs. Whereas mormoopids have LVPs that are located on the lateral surface of the tongue (fig. 32A), LVPs are absent in Noctilio; therefore, the primitive condition for phyllostomids appears to be a lateral position for the LVPs. We scored taxa lacking LVPs (state 1 of character 109) “−” for this character.

Gimenez (1993: character 3.3.4.2) first used the position of the LVPs as a character in cladistic analysis, and we follow her state definitions. Smith and Hood (1984: 450) characterized the LVPs as “dorsal” in all stenodermatines after examining Ametrida, Artibeus, Centurio (which lacks LVPs), and Sturnira. Smith and Hood's (1984) assessment of LVP position in Ametrida and Artibeus appears to be supported by our observations; however, we agree with Gimenez (1993) that Sturnira has dorsolateral LVPs.

Gimenez (1993) characterized the LVPs as being located laterally in Vampyressa bidens and V. pusilla. Our observations of LVPs in these species and V. nymphaea indicate that only V. pusilla has laterally located LVPs. In the other Vampyressa species that we examined, these papillae are located in a dorsolateral position. In Artibeus, Dermanura, and Koopmania, Gimenez (1993) described the LVPs as being dorsolateral. In contrast, we found that the LVPs were located on the dorsum of the tongue all representatives of these taxa that we examined (see appendix 1 for specimens examined).

Character 111

Pharyngeal region of tongue completely covered with papillae (0); or covered laterally with papillae and medially bare (1); or wholly unpapillated (2). The pharyngeal region of the tongue lies proximal to the circumvallate papillae (Sonntag, 1925). In taxa lacking circumvallate papillae (e.g., desmodontines), this region apparently lacks fungiform papillae (Sonntag, 1925), providing a means of identification. Most phyllostomids have fleshy mechanical papillae covering the lateral sections of the pharyngeal region of the tongue (figs. 32C, 33A–C, 34B). In these taxa, a smooth, unpapillated bare patch is present medially, often extending distally between the MVPs. In some phyllostomines (e.g., Chrotopterus, Phyllostomus) and some stenodermatines (e.g., Mesophylla, Vampyrodes) the tongue lacks this medial bare patch and has a carpet of mechanical papillae over the entire pharyngeal region (fig. 34A). In contrast, many stenodermatines (e.g., Ametrida) have an entirely bare pharyngeal region (fig. 34C). One individual of Pygoderma (AMNH 248334) has a single medial line of papillae bisecting the otherwise bare pharyngeal region, whereas all other specimens we examined have a completely bare pharyngeal region. One individual of Phyllops (AMNH 176190) has approximately 14 papillae present laterally on both sides in the pharyngeal region, whereas all other specimens that we examined do not have any papillae present in the pharyngeal region. Similarly, a single specimen (USNM 522707) of Stenoderma has several papillae present laterally on both sides of the pharyngeal region, whereas all other specimens do not. We interpret these conditions as anomalies and score these genera with state 2 in the matrix. Mormoopids have a pharyngeal region that is bare (fig. 32A), whereas Noctilio has fleshy mechanical papillae surrounding a central bare patch (fig. 32B). The primitive state for phyllostomids cannot be reconstructed a priori.

Gimenez (1993: characters 3.3.2.3; 3.3.3.3; 3.3.4.7–8) first used the location of papillae in the pharyngeal region in a cladistic analysis. One character (Gimenez, 1993: character 3.3.4.7) accounted for presence or absence of the pharyngeal papillae, and appeared only in the analysis of Stenodermatini genera (see Historical Background for details of Gimenez's [1993] analyses). The other character (Gimenez, 1993: character 3.3.2.3, 3.3.3.3, 3.3.4.8) described the two patterns of papillation (pharyngeal region completely covered or covered only laterally), and appeared in most analyses, except the analysis of Glossophagini genera. We have chosen to combine the two characters recognized by Gimenez (1993) to avoid the problems of missing data often associated with hierarchical characters.

In her analyses, Gimenez (1993) scored those Stenodermatini genera with bare pharyngeal regions (character 3.3.4.7: pharyngeal papillae absent) as having the laterally papillated state in the second character (character 3.3.4.8), rather than “−” as appropriate for a hierarchical character. In Gimenez's (1993) analysis of desmodontines, basal phyllostomines, and the tribes Phyllostomini, Glossophagini, and Stenodermatini, she scored Phyllostomini and Stenodermatini as having only the laterally papillated condition. In fact, these taxa are polymorphic, with both laterally papillated and totally papillated conditions occurring in Phyllostomini and Stenodermatini genera. Although mormoopids have a completely bare pharyngeal region, Gimenez's (1993) character did not include this condition as a state, and she scored mormoopids as having a medially bare pharyngeal region.

Gimenez (1993) described the pharyngeal region of Phylloderma as being fully papillated. We observed a smooth medial pharyngeal region laterally flanked by fleshy, filiform papillae in all specimens we examined (see appendix 1 for specimens examined). In Mesophylla and Vampyrodes, Gimenez (1993) described the pharyngeal region as being laterally papillated and medially bare. We observed a completely papillated pharyngeal region in both genera.

Character 112

Lingual sulci absent (0); or lateral lingual sulci present (1); or ventral lingual sulci present (2). Lingual sulci are present in Desmodus, Diaemus, and lonchophyllines, but are absent in all other phyllostomids. In Desmodus and Diaemus, the sulci are located on the ventral aspect of the tongue and run anteriorly from the frenulum (a median vertical fold present on the floor of the mouth) along the distal half of the tongue (fig. 36C). The sulci do not meet at the tongue tip. In lonchophyllines, sulci are found on the lateral surface of the tongue, beginning just distal to the LVPs and coursing to the tongue tip, where the right and left sulci meet (fig. 36D). The sulci in lonchophyllines appear to be relatively deeper than those of Desmodus and Diaemus and are bordered by a fringe of hairlike papillae (see character 113), which is absent in the desmodontine taxa. Mormoopids and Noctilio lack lingual sulci, suggesting that this is the primitive condition for phyllostomids.

Griffiths (1982: character 12) defined a character for presence or absence of a papilla-lined lingual sulcus. Our character states 0 and 1 are roughly equivalent to those of Griffiths (1982), although we have added an extra state to include the variation introduced by the desmodontines, which Griffiths (1982) described but did not consider in his analysis. Unlike Griffiths (1982), we consider the hairlike papillae lining the sulcus in lonchophyllines in the context of a different character (see character 113). Wille (1954) reported the presence of a groove along the lateral margins of half the length of the tongue in Leptonycteris nivalis. This groove does not appear to be present in Leptonycteris, as was previously noted by Park and Hall (1951), Winklemann (1971), and Greenbaum and Phillips (1974).

Griffiths and Criley (1989) found that the sulci of Desmodus and Diaemus are ventral to the lingual nerve and the transverse muscle mass (fig. 36A), but in lonchophyllines the sulci are dorsal to these structures (fig. 36B). On this basis, these authors suggested that the sulci in desmodontines and lonchophyllines are not homologous. Gimenez (1993: characters 3.3.2.8, 3.3.5.14) erected two different lingual sulcus characters, one for the presence of a ventral sulcus, the other for the presence of a lateral sulcus. Her character 3.3.2.8 is equivalent to our state 2 as a binary (present'absent) character. Gimenez's (1993) second character (3.3.5.14), “presence of a lateral lingual sulcus bordered by filiform papillae,” is equivalent to our character state 1. Gimenez et al. (1996: character 2) defined and scored a character for the presence of a lateral lingual sulcus, following Griffiths and Criley's (1989) and Gimenez's (1993) assessment that the lateral and ventral lingual sulci are not homologous. We prefer to make no such a priori assessments and included both types of sulci as states in a single character to allow all possible transformations.

Character 113

Brush of hairlike papillae around the distal margin of tongue absent (0); or present (1). Long, thin, filiform papillae, known as “hairlike” papillae (Park and Hall, 1951), are present laterally on the distal part of the tongue in phyllonycterines, glossophagines, and lonchophyllines (fig. 33A–C). The hairlike papillae form a brushlike tongue tip in these taxa. Hairlike papillae are absent in all other phyllostomids (figs. 32C, 34A–C). Hairlike papillae are also absent in mormoopids and Noctilio (fig. 32A–B), suggesting that absence of these papillae is the primitive condition for phyllostomids.

Griffiths (1982: characters 12, 14) defined two binary (present'absent) characters based on hairlike papilla morphology and distribution: “groove in tongue lined with hairlike papillae,” and “brush tip formed by hairlike papillae.” He coded phyllonycterines and glossophagines “absent” for the former and “present” for the latter, whereas the reverse was true for lonchophyllines. Smith and Hood (1984: 448–449) objected to these characters, which appeared to score presence of “Glossophaga-type” and “lonchophylline-type” tongues, obscuring the potential homology of the hairlike papillae, which are not found on the tongues of other phyllostomids.

Gimenez (1993: character 3.3.5.11) used a single unordered multistate character with three states (absent, 0; present but short and few in number, 1; present and abundant, 2) to describe variation of hairlike papillae. Gimenez et al. (1996: character 14) retained states 0 and 1 (Gimenez, 1993), but added two new states: brushlike tip present, with abundant hairlike papillae 2, and brushlike tip extremely well developed, with abundant and very long hairlike papillae 3 (see character 115). Gimenez et al. (1996) ordered the transformations in this character (0 ↔ 1 ↔ 2 ↔ 3). In addition to this character, Gimenez et al. (1996: character 3) included a separate character describing the line of filiform papillae that runs along the lateral lingual sulcus in lonchophyllines.

We prefer to recognize the presence of hairlike papillae, derived relative to the outgroups, as a single state, with additional hierarchically dependent characters (characters 114 and 115) to accommodate variation in the distribution and shape of the papillae. This approach best reflects our views of potential homologies.

Character 114

Hairlike papillae confined to lateral margin of distal third of tongue, with a single line of papillae that extends roughly to LVPs (0); or hairlike papillae distributed around lateral margin and dorsum of distal third of tongue, not arranged in a single line (1). The distribution of hairlike papillae differs substantially among taxa in which they are present. In lonchophyllines, the papillae are restricted to the distal part of the lateral surface of the tongue, with a single line of hairlike papillae running just ventral to the lingual sulci, extending posteriorly roughly to the LVPs (fig. 37B). These single lines of papillae are not developed in phyllonycterines and glossophagines (fig. 37A, C). In these taxa, the papillae are found on the lateral and dorsal surfaces of the tongue, and a narrow channel (“median trough,” or “anterior groove” of Park and Hall, 1951) that lacks hairlike papillae runs down the midline of the dorsum of the tongue. Mormoopids and Noctilio lack hairlike papillae, so the primitive state for phyllostomids cannot be assessed a priori. We scored taxa that lack hairlike papillae (state 0 of character 113) “−” for this character.

As noted above, Griffiths (1982: characters 12, 14) was the first to define characters based on distribution of hairlike papillae. Gimenez's (1993: character 3.3.5.11) character states 1 (present but short and few in number) and 2 (present, long and abundant) are roughly equivalent to our states 0 and 1, respectively and our scoring is identical to hers. Gimenez et al.'s (1996: character 14) character incorporated aspects of both distribution pattern and shape of individual papilla, which we have chosen to treat as two separate characters (this character and character 115).

Character 115

Hairlike papillae fleshy and conical (0); or fleshy and conical with filamentous tips (1); or cylindrical with ellipse-shaped distal end (2). Phyllonycterines, glossophagines, and lonchophyllines are each characterized by a different shape of hairlike papilla. The individual hairlike papillae of lonchophyllines are short, fleshy, and conical (fig. 37B, inset). In glossophagines, the papillae are longer and thinner than in lonchophyllines, and have filamentous tips (see scanning electron micrographs in Vaughan, 1986; fig. 37A, inset). These papillae most closely resemble hairs. Phyllonycterines have cylindrical hairlike papillae (fig. 37C, inset). The distal ends of these papillae appear as though they have been sliced obliquely, producing ellipsoid ends. Mormoopids and Noctilio lack this papilla type; thus the primitive state for phyllostomids cannot be established a priori. We scored taxa without hairlike papillae (state 0 of character 113) “−” for this character.

Although previous workers have not used shape of the hairlike papillae as a separate character, Gimenez et al. (1996: character 14) incorporated papilla presence, morphology, and distribution into a single character. Gimenez et al. (1996) recognized three conditions, roughly corresponding to our character states (see character 113).

Character 116

All or most medial-posterior mechanical papillae inclined toward the pharyngeal region of the tongue (0); or all oriented toward tongue tip (1); or all oriented toward midline of tongue (2); or papillae not inclined (3). Gimenez (1993) defined the medial-posterior mechanical papillae as those mechanical papillae that lie between the MVPs and the anterior mechanical papillae. In taxa that lack MVPs (e.g., desmodontines), we define this region as roughly the middle third of the tongue. All or most of the medial-posterior mechanical papillae are inclined toward the pharyngeal region of the tongue in most phyllostomids (figs. 32C, 33A–C, 34A). In contrast, all of the medial-posterior mechanical papillae are inclined toward the tip of the tongue in all stenodermatines with the exception of Sturnira (fig. 34B–C). In Mormoops these papillae are directed pharyngeally. In Pteronotus, the medial-posterior mechanical papillae are unlike the fleshy, filiform papillae in other noctilionoids (fig. 32A). Instead, these papillae are basin-shaped and directed dorsally (i.e., they are not inclined). In Noctilio, the posteriormost medial-posterior mechanical papillae are on ridges that curve anteriorly towards the lateral margins of the tongue (fig. 32B). The ridges disappear as the papillae on them become more sharply pointed (roughly one third of the area of the medial-posterior mechanical papillae is ridged). The individual medial-posterior mechanical papillae are directed toward the midline of the tongue. Thus, the primitive condition for phyllostomids cannot be reconstructed a priori.

Park and Hall (1951) noted that the medial-posterior mechanical papillae incline anteriorly in Artibeus among the phyllostomids they studied, but Gimenez (1993: character 3.3.4.11) was the first to use this information in a cladistic analysis. Our character states and scoring are equivalent to hers.

Character 117

Small patch of anteriorly directed medial-posterior mechanical papillae always absent, all papillae oriented toward pharyngeal region (0); or medial patch present in some individuals; polymorphic within species (1); or medial patch always present (2). Among those phyllostomids in which the medial-posterior mechanical papillae are principally directed toward the pharyngeal region (state 0 of character 116), a small patch of papillae near the midline of tongue may be oriented toward the tip of the tongue. This patch is always present in Micronycteris sylvestris, Lionycteris, and Platalina, and is sometimes present in Lonchophylla (e.g., Lonchophylla thomasi AMNH 266107; fig. 33C). In Mormoops, the small patch of anteriorly directed papillae is not present. The medial-posterior mechanical papillae are not directed toward the pharyngeal region in Noctilio and Pteronotus, and we scored these genera “−” for this character. Thus the primtive state for phyllostomids appears to be the absence of the anteriorly directed patch of papillae. We scored taxa in which the medial-posterior mechanical papillae are not principally directed toward the pharyngeal region (states “1,” “2,” or “3” for character 116) “−” for this character.

Gimenez (1993: characters 3.3.5.9) first used this feature in a cladistic analysis, and our character states are equivalent to hers. Gimenez (1993) left Lonchophylla unscored because the medial-posterior mechanical papillae were too small and rudimentary to permit confident observation. Gimenez et al. (1996: 49) indicated that a patch of distally directed papillae may be seen not only in Lionycteris, but in “some of the best preserved tongues of Lonchophylla species.” This description corroborates our observation of such a patch in Lonchophylla thomasi.

Character 118

Long-tipped, bifid anterior mechanical papillae absent (0); or present at border between anterior mechanical papillae and medial-posterior mechanical papillae (1). In most phyllostomids, the small, keratinized anterior mechanical papillae and the larger, fleshy medial-posterior mechanical papillae meet near the middle of the tongue (figs. 32C, 33A–C, 34A, C). No other papilla type is present at this juncture. In contrast, this juncture is covered by a narrow, transverse band of keratinized bifid papillae in many stenodermatines (e.g., Artibeus, Uroderma; fig. 34B). The individual papillae are erect and directed toward the pharyngeal region. Each papilla tip has a long, thin, cornified filament. These papillae seem to be uniformly present in the species that possess them, although they may be absent in one poorly preserved specimen of Koopmania (AMNH 80340) of the five we examined. We interpret this possible condition as an aberration, and therefore score Koopmania as having long-tipped, bifid anterior mechanical papillae. Long-tipped, bifid anterior mechanical papillae are absent in mormoopids and Noctilio (fig. 32A–B), suggesting that this is the primitive condition for phyllostomids.

Gimenez (1993: character 3.3.4.10) first used presence of long-tipped, bifid anterior mechanical papillae as a character. Gimenez (1993) noted that in most taxa, these papillae are not well differentiated, and the junction between these papillae and the fleshy medial-posterior mechanical papillae is more or less transverse. In contrast, Gimenez (1993) indicated that Mesophylla macconelli, Vampyressa bidens, and V. pusilla have well differentiated long-tipped, bifid anterior mechanical papillae, and that the junction between the anterior and medial-posterior mechanical papillae is V-shaped with the medial-posterior papillae penetrating into the anterior region medially. We found that the shape of the juncture between the anterior and medial-posterior mechanical papillae shows continuous variation from transverse or slightly V-shaped to strongly V-shaped. Accordingly, we have scored only the presence of these papillae, not the shape of the zone in which they occur. We also found well-differentiated, long-tipped, bifid anterior mechanical papillae in several additional taxa (see above).

Character 119

Basketlike medial-posterior mechanical papillae absent (0); or present (1). In most phyllostomids, the medial-posterior mechanical papillae are thick, fleshy, and have one or more rounded points. However, some of the medial-posterior mechanical papillae are what Park and Hall (1951) termed “basketlike.” Basketlike papillae occur in phyllostomines, Brachyphylla, some glossophagines (e.g., Hylonycteris, Lichonycteris), Lionycteris, Lonchophylla, carolliines, and Sturnira (figs. 33C, 34A). These papillae have central concavities and fleshy, cylindrical bases. The rim of each basketlike papilla is usually uneven, with the posterior part of the rim higher than the anterior. The entire rim, or just the posterior part, may be adorned with from one to more than 10 pointed projections, which appear to form a fringe along the edge of the rim. Basketlike papillae are present in Pteronotus and Mormoops (fig. 32A), but are absent in Noctilio (fig. 32B), making the state primitive for phyllostomids impossible to reconstruct a priori.

Gimenez (1993: characters 3.3.2.5, 3.3.3.5, 3.3.4.9, 3.3.5.7) erected four separate characters to describe morphology of the medial-posterior mechanical papillae in her analyses of various subsets of taxa. Gimenez et al. (1996: character 10) scored presence'absence of basketlike papillae, which they called “posteromedian conical papillae,” in a separate binary character. The multiple characters used by Gimenez (1993) included eight states describing the basketlike papillae and other papilla types, many of which are apparently autapomorphic. For example, Gimenez (1993: character 3.3.2.5) erected a state for the “craterlike” papillae of Pteronotus, which she considered to be autapomorphic, describing them as “circular and differentiated in the form of a crater … with a depression and fringed border.” However, Gimenez (1993: character 3.3.3.5) reported similar papillae, which she described as “circular, in the form of a crater … with a reduced fringe,” in Lonchorhina, Mimon renulatum, and unspecified species of Tonatia. Gimenez (1993) scored the “craterlike” papillae of Lonchorhina and Tonatia with a separate state from those of Mimon crenulatum, in which she called the “craterlike” papillae “huge and irregular.” We treat all “basketlike” papillae as potentially homologous, including Gimenez's (1993) “craterlike” papillae, because they are very similar in appearance (e.g., presence of fringe, and central concavity), and position on the tongue, despite some differences in relative size (e.g., as in Mimon crenulatum). Furthermore, because each tongue includes a variety of different basketlike types ranging from smooth to fringed, we chose to subsume all of the varieties in a single state rather than attempting to subdivide what appears to be a range of variation in the number of points in the fringe.

Despite the differences in scoring we noted above, we agree with virtually all of Gimenez's (1993) assessments, except her finding that basketlike papillae are present in Glossophaga and are present but reduced in size in Platalina (data apparently taken from Griffiths, 1982). We find that these papillae are absent in both genera. Gimenez et al. (1996) also scored basketlike papillae as present in Glossophaga, as well as Monophyllus, which was not mentioned in their text or list of specimens examined. Our observations indicate that Monophyllus lacks basketlike papillae. Giminez et al. (1996: 49) scored Lichonycteris “1'v,” indicating that this character is polymorphic in this taxon. The reason for this scoring given in their text was that Lichonycteris has other kinds of papillae in this region in addition to basketlike papillae, a factor that does not come into consideration in our version of this character. Gimenez et al. (1996) scored Lonchophylla as “0'–,” indicating that basketlike papillae are present in some species, but that the character is not applicable in others because these papillae are so reduced (e.g., L. dekeyseri). Unfortunately, we were unable to examine Lonchophylla dekeyseri, but our observations of L. robusta and L. thomasi agree with Gimenez et al.'s (1996) reports of tiny basketlike papillae in L. mordax and L. bokermanni; thus, we decided to score Lonchophylla with state 1. In addition, we found that Sturnira has proximally fringed basketlike papillae similar to those of phyllostomines, rather than the anteriorly fringed, posteriorly smooth papillae described for this taxon by Gimenez (1993). Park and Hall (1951) first described the presence of basketlike papillae in Carollia, but did not observe these papillae in any other phyllostomid they examined (including Macrotus, Leptonycteris, Glossophaga, Choeronycteris, Artibeus, and Desmodus), whereas we found these papillae to be present in many of these taxa.

Character 120

Cluster of horny papillae located near tip of tongue (0); or located significantly proximal to tongue tip (1). In most phyllostomids, the horny papillae (sensu Park and Hall, 1951) are located near the tip of the tongue; less than 20% of tongue length is distal to the horny papillae (figs. 32C, 33A, C, 34A–C). In contrast, the horny papillae are located in a more proximal position on the tongue in glossophagines (fig. 33B). In these taxa, more than 25% of the tongue length is distal to the horny papillae. In mormoopids and Noctilio the horny papillae are located near the tip of the tongue (fig. 32A–B), suggesting that this condition is primitive for phyllostomids.

Griffiths (1982) indicated that lonchophyllines lack horny papillae, but Smith and Hood (1984), Gimenez (1993), and Gimenez et al. (1996) noted the presence of distinct apical horny papillae in Lionycteris and Lonchophylla; our results confirm that they are indeed present in all lonchophyllines. Gimenez (1993: character 3.3.5.12) was the first to use this character, and our states are equivalent to hers. However, Gimenez (1993) considered the horny papillae of phyllonycterines to be located in a relatively proximal position from the lingual apex, and scored them with the same state as glossophagines. Gimenez et al. (1996: character 13) defined and scored the character the same way. In contrast, we found that the horny papillae are located near the tongue tip in phyllonycterines, whereas only glossophagines have the more proximally located horny papillae. Rough quantification supports this scoring, as the largest gap in the distribution of values for percentage of tongue length distal to the horny papillae falls between the phyllonycterines and the glossophagines (data available on request).

Character 121

Horny papillae arranged in lines or elliptical cluster (0); or large V-shaped cluster (1). In most phyllostomids, the horny papillae form an elliptical cluster (which may be reduced to a few rows of horny papillae) near the tip of the tongue (figs. 33A–C, 34A–C). In contrast, desmodontines have a large V-shaped arrangement of horny papillae which follows the general outline of the tongue tip, spreading laterally and proximally from the near apical position to cover the tongue's width near the tip (31C). In mormoopids and Noctilio, the patch of horny papillae forms an elliptical cluster (fig. 32A–B), suggesting that this is the primitive condition for phyllostomids.

Gimenez (1993: character 3.3.2.7) first used the V-shaped distribution of horny papillae as a binary character. Griffiths (1982) and Gimenez et al. (1996) did not use this character, because desmodontines were not the focus of their studies. Although Park and Hall (1951), who described an elliptical rather than V-shaped cluster of horny papillae in desmodontines, and Vierhaus (1983) included desmodontine papillae under the description “horny papillae,” Smith and Hood (1984) suggested that the keratinized papillae in desmodontines are not comparable to the horny papillae of other phyllostomids. Based on our observations, the only difference between these papillae in desmodontines and those in other phyllostomids is the pattern of distribution, not the actual structure of the individual papillae.

Character 122

All horny papillae present in elliptical cluster equal in size (0); or some papillae markedly larger than others (1). In most phyllostomids, there are one or two horny papillae that are larger than the others (fig. 38C–G). In contrast, Lonchorhina, Macrophyllum, Vampyrum, and Rhinophylla do not have one or two horny papillae that are larger than others; rather, all the horny papillae are roughly equal in size (fig. 38A–B). In mormoopids and Noctilio, the horny papillae are all equal in size, suggesting that this is the primitive condition for phyllostomids. We scored taxa that have a V-shaped cluster of papillae (state 1 of character 121) “−” for this character.

This character has no direct counterpart in previous phylogenetic analyses (see discussion under character 124).

Character 123

Horny papillae all large (0); or all small (1). Among taxa with horny papillae that are all roughly equal in size, Vampyrum and Rhinophylla have more than 15 very large papillae in the center of the cluster, which are surrounded by additional smaller horny papillae (fig. 38A). In contrast, Lonchorhina and Macrophyllum have an elliptical cluster composed of equally sized horny papillae, all of which are very small (fig. 38B). Mormoopids have very small horny papillae. In Noctilio, the papillae in the center of the cluster, more than 15 in number, are all equally large. Because of this distribution of character states, the primitive condition for phyllostomids cannot be assessed a priori. We scored taxa that have a V-shaped cluster of horny papillae (state 1 of character 121) or those with papillae of different sizes in the cluster (state 1 of character 122) “−” for this character.

This character has no direct counterpart in previous phylogenetic analyses (see discussion in character 124). To assess the relative sizes of these papillae, we drew representative tongues to the same scale for comparison.

Character 124

Single large horny papilla present in center of elliptical cluster (0); or two large horny papillae present in center of elliptical cluster (1). A single large papilla is present in the elliptical cluster of horny papillae in most phyllostomines (e.g., Phyllostomus, Tonatia), Brachyphylla, phyllonycterines, Anoura, lonchophyllines, Carollia, and several stenodermatines (e.g., Artibeus, Sturnira; fig. 38C–E). This large papilla is located approximately on the midsagittal line of the tongue. In Micronycteris minuta, Mimon crenulatum, most glossophagines (fig. 38F), and some stenodermatines (e.g., Ardops, Pygoderma; fig. 38G), there are two large papillae in the cluster, one on either side of the midline of the tongue. In mormoopids and Noctilio, all horny papillae are roughly equal in size. Consequently, these taxa are scored “−” for this character and the primitive state for phyllostomids cannot be determined a priori. We scored taxa that have a V-shaped cluster of horny papillae (state 1 of character 121) or those with papillae of uniform size in the cluster (state 0 of character 122) “−” for this character.

Griffiths (1982; character 16) was the first to use arrangement of the horny papillae in a cladistic analysis, although Smith and Hood (1984: 449) later dismissed this effort as a “Rorschach analysis” of horny papillae patterns. Our observations suggest that, as Griffiths (1982) initially noted, these patterns are potentially phylogenetically informative. It is evident that the number and distribution of horny papillae are consistent within species'genera and vary among them. Other authors have also found consistent patterns in phyllostomid horny papillae. For example, Greenbaum and Phillips (1974) noted that all specimens of Leptonycteris they examined had two large central horny papillae. Park and Hall (1951: 65) reported that there were “Generally … seven papillae in a group, two to four of which are always larger than the others and horny instead of fleshy.” Although, we found that there was no difference in keratinization for the large and small horny papillae, Park and Hall's (1951) comments demonstrate that they also observed some consistent pattern. Even Smith and Hood (1984) indicated that a consistent pattern appeared to be present in Lionycteris and Lonchophylla. These authors observed a large anterior papilla with two large papillae posterior to it.

Our definitions of characters and character states dealing with these papillae differ substantially from Griffiths' (1982) treatment. Griffiths (1982: character 16) identified a single pattern of interest, the “3'4 pattern” (one row of 3 large papillae and a second row of 4 large papillae), which he scored as a binary character. Thus, Griffiths (1982) grouped several unique patterns (see characters 122 to 127) into a single, plesiomorphic state defined by absence of a recognizable 3'4 pattern (Smith and Hood, 1984). We found that many of the taxa in which Griffiths (1982: 31) recognized a “less orderly” patch of horny papillae (e.g., Macrotus, many stenodermatines) are characterized by distinctive patterns. Griffiths (1982) observed two large papillae in Carollia, whereas we found one. Gimenez (1993: character 3.3.5.13) followed Griffiths (1982) in recognizing only the 3'4 pattern, again scoring this as a binary character. She scored only Choeroniscus, Choeronycteris, and Hylonycteris with the derived condition (the presence of the 3'4 pattern). In all other taxa, she reported that this pattern was absent—clearly at odds with our description of horny papillae patterns, and the previous report by Griffiths (1982).

Gimenez et al. (1996: character 12) subdivided the patterns observed among nectar-feeding phyllostomids and their allies into four states. They described state 1 as a triangular arrangement of three papillae, one small papilla anterior to two larger ones (Gimenez et al., 1996). They scored Glossophaga, Monophyllus, Leptonycteris, and lonchophyllines as having this state. Based on Griffiths' (1982) descriptions, Gimenez et al. (1996) also scored Erophylla and Phyllonycteris with this state, although these taxa have one large papilla anterior to two smaller ones. Gimenez et al. (1996) erected a separate state for Anoura (state 3), which they described as having an autapomorphic pattern of three to four papillae in a transverse row with smaller papillae anterior to this row. However, the Anoura that we examined have a single large central papilla surrounded by smaller papillae. Gimenez et al. (1996) scored Choeroniscus, Choeronycteris, Hylonycteris, and Lichonycteris as having the 3'4 pattern (state 2), three small papillae anterior to four larger papillae, the central two of which are largest. Finally, they described all other phyllostomids as having a poorly defined state (0): “8–15 main papillae arranged in a circular to nearly square cluster” (Gimenez et al., 1996: character 12).

We divided the horny papillae pattern into several characters (124 to 127) in such a way as to allow scoring of the consistent variation that we observed. The first of these (this character) describes the number of large papillae (what we term the “main papillae”), which provide a landmark for descriptions of the remaining horny papillae (anterior, posterior, or lateral to the main papillae). We believe, based on position and overall similarity, that each of the character states defined here describes a potentially homologous pattern.

Character 125

Three small papillae present anterior to main papilla(e) (0); or one papilla present (1); or no papillae present (2). Among phyllostomids with an elliptical cluster of horny papillae, many have one or more papillae located anterior to the main papilla(e). Three small papillae are present anterior to the main papilla(e) in phyllostomines (fig. 38C), Brachyphylla (except AMNH 213731 in which the anteriormost appears to be missing), some glossophagines (e.g., Choeroniscus, Lichonycteris), Carollia, and all stenodermatines (fig. 38G). The middle of the three papillae is usually slightly anterior to the other two and is centered roughly on the midline of the tongue. We considered Brachyphylla AMNH 213731 aberrant, and scored this taxon with state 0 in the matrix. In contrast, in some glossophagines (e.g, Glossophaga, Leptonycteris) there is only a single papilla anterior to the main papilla (fig. 38F). No papillae are present anterior to the main papilla(e) in phyllonycterines and lonchophyllines (fig 38D–E). Mormoopids and Noctilio have horny papillae of uniform size, so the primitive state for phyllostomids cannot be determined a priori. We scored taxa that have a V-shaped cluster of horny papillae (state 1 of character 121) or those with papillae of uniform size in the cluster (state 0 of character 122) “−” for this character.

The “3'4” pattern used by Griffiths (1982: character 16) and Gimenez (1993; character 3.3.5.13) included the presence of three small anterior papillae. Gimenez et al. (1996: table 2) scored Erophylla and Phyllonycteris as having one anterior papilla. However, the “anterior” papilla in phyllonycterines is much larger than other surrounding horny papillae, so we interpret it as the main papilla of the cluster (see character 124). Consequently, we score phyllonycterines as lacking small anterior papillae. Gimenez et al.'s (1996; character 12) state “8–15 main papillae arranged in a circular to nearly square cluster,” scored as present in phyllostomines, Brachyphylla, and stenodermatines, apparently included three papillae that we observed anterior to the central largest papilla(e) in these taxa. If this is the case, our observations are largely in agreement with those of Gimenez et al.(1996).

Character 126

Two or more small horny papillae present posterior to main papilla(e) (0); or absent (1). Two or more small horny papillae are present posterior to the main papilla(e) in most phyllostomids (fig. 38C–E). These are typically aligned on either side of the midline of the tongue. In contrast, there are no horny papillae posterior to the main papilla(e) in all glossophagines (fig. 38F), except Anoura, and many stenodermatines (e.g., Uroderma, Sphaeronycteris; fig. 38G). Mormoopids and Noctilio have horny papillae of uniform size, so the primitive state for phyllostomids cannot be determined a priori. We scored taxa that have a V-shaped cluster of horny papillae (state 1 of character 121) or those with papillae of uniform size in the cluster (state 0 of character 122) “−” for this character.

Character 127

Main horny papilla(e) flanked by a pair of smaller horny papillae, one on each side (0); or no papillae present lateral to main papilla(e) (1). Most phyllostomids have one small horny papillae located on each side of the main papilla(e) (fig. 38C–D, G). In contrast, these are absent in Glossophaga, Leptonycteris, Monophyllus, and lonchophyllines (fig. 38E–F). In some specimens of Phyllonycteris (e.g., AMNH 23762) these papillae are occasionally absent. Mormoopids and Noctilio have horny papillae of uniform size; thus, the primitive state for phyllostomids cannot be determined a priori. We scored taxa that have a V-shaped cluster of horny papillae (state 1 of character 121) or those with papillae of different sizes in the cluster (state 1 of character 122) “−” for this character.

Griffiths' (1982: character 16) 3'4 character included the presence of four papillae in the posterior row: two central large papillae and two smaller flanking papillae. Similarly, the state described by Gimenez et al. (1996: character 12) as “8–15 main papillae arranged in a circular to nearly square cluster,” apparently included a flanking pair. Our observations agree with those of previous authors.

Character 128

Paired lingual arteries present, lingual veins not enlarged (0); or single, midline lingual artery present, lingual veins enlarged (1). A pair of lingual arteries is present in the tongue of Desmodus, phyllostomines, Brachyphylla, lonchophyllines, Carollia, and stenodermatines (Griffiths, 1982). The lingual veins are not enlarged in these taxa (Griffiths, 1982). A single, midline lingual artery and a pair of enlarged lingual veins are present in phyllonycterines and glossophagines (Griffith, 1982). The condition of the lingual vascular system has not been reported for mormoopids or noctilionids, but most mammals have paired lingual arteries (Griffiths, 1982), suggesting that this condition may be primitive for phyllostomids.

Griffiths (1982: characters 17–18) treated the lingual vascular system in two characters: “single midline lingual artery” (present'absent) and “enlarged lingual veins” (present'absent). However, these features may be functionally correlated, and they have identical taxonomic distributions. Accordingly, we chose to treat the lingual arteries and veins as part of a single character.

Character 129

Brunner's glands present at gastroduodenal juncture (0); or absent (1). Brunner's glands, which secrete mucus, are present at the gastroduodenal juncture in phyllostomines, glossophagines, and Sturnira (Forman, 1973; Forman et al., 1979). Brunner's glands are absent in this region of the stomach in Brachyphylla and the remaining stenodermatines (Forman, 1973; Forman et al., 1979). In some species of Artibeus, Dermanura, and Platyrrhinus, Brunner's glands are present, whereas in other species in these genera these glands are absent. We score these genera with both states 0 and 1 in the matrix. Mormoopids and Noctilio have “unusually” abundant Brunner's glands at the gastroduodenal juncture (Forman, 1971), suggesting that presence of Brunner's glands in this region may be the primitive state for phyllostomids.

Straney (1980: character A1) defined a single binary character based on the number of Brunner's glands distributed in the proximal part of the duodenum, describing the derived state as “Brunner's glands present but not numerous.” Straney (1980) scored the taxon “Phyllostomatinae” with the primitive condition of this character (i.e., Brunner's glands present and numerous). Straney (1980) scored all glossophagines, stenodermatines, carolliines, and brachyphyllines as having Brunner's glands that were less numerous than those in the phyllostomines. However, it is not easy to determine how abundance of these glands should be treated. Forman (1971: 274) described Brunner's glands in mormoopids and noctilionids as “unusually abundant,” whereas glands in four phyllostomines were described by Forman et al. (1979: 225) as “relatively numerous.” Anoura and Glossophaga have “relatively more abundant” glands than Choeroniscus and Lichonycteris do (Forman, 1971: 277), whereas some Artibeus species have “sparse” Brunner's glands and some Dermanura species have “numerous” Brunner's glands at the gastroduodenal juncture (Forman et al., 1979: 225). Because it is difficult to determine exactly which states should be considered homologous, and it seems likely to us that such descriptions may actually describe states on a continuum, we have chosen to disregard such information until such data can be more readily quantified.

Character 130

Male accessory gland includes compact white body situated anteromedially (0); or compact white body absent (1). All phyllostomids have an accessory gland (seminal vesicles and prostate) composed of two parts: a pink, vesiculate, anterior portion, and a white, compact, posterior portion (Straney, 1980). However, in some phyllostomines (e.g., Phyllostomus, Trachops), Glossophaga, Carollia, and Sturnira, the accessory gland includes a third part: a compact white body situated anteromedially (this body may be the seminal vesicles; Straney, 1980). In mormoopids, the accessory gland is also tripartite with a pink, vesiculate, anterior part; a compact, white, posterior part; and a compact, white anteromedial part. The accessory bulb in Noctilio is not tripartite and is primarily pink and vesiculate. However, a compact white body is present and it is situated anteromedially (Straney, 1980). This distribution of character states suggests that presence of a compact white anteromedial body is the primitive condition for phyllostomids.

Straney (1980: characters B1–3, and 239) developed three binary characters based on the morphology of the male accessory gland: “(1) Prostate of two parts; caudal part dark, vesiculate; cranial part small, white, compact; (2) Prostate three parted; as in (1) but caudal part with a large, compact white region; (3) Prostate of two parts; caudal part white, compact; cranial part dark vesiculate.” We did not use information pertaining to other parts of the male accessory gland because they do not appear to be informative at this taxonomic level (i.e., within phyllostomids). We followed Straney's (1980: 23–28) text descriptions to score taxa rather than his data matrix, because he seems to have scored mormoopids incorrectly in the matrix (indicating they have only a bipartite, not a tripartite accessory gland). Although Straney (1980) scored Macrophyllum in his matrix for this character, he did not describe or list specimen numbers for this taxon; therefore, we did not score Macrophyllum for this character.

Character 131

Uterine horns approximately one half length of common uterine body (0); or one quarter length of common uterine body (1), or uterus fully simplex, uterine horns not distinct from common body (2). The length of the uterine horns is approximately one half the length of common uterine body in Desmodus (Hood and Smith, 1982, 1983; fig. 39A). External uterine fusion is somewhat greater in some phyllostomines (e.g., Macrotus, Trachops), which have uterine horns that are one quarter the length of the common uterine body (Hood and Smith, 1982, 1983; fig. 39B). In contrast, the uterus is simplex (the uterine horns are not distinct from common body of the uterus) in other phyllostomines (e.g., Phylloderma, Phyllostomus), Brachyphylla, Phyllonycteris, Lonchophylla, glossophagines, carolliines, and stenodermatines (Hood and Smith, 1982, 1983; fig. 39C–D). The length of the uterine horns is approximately one half the length of common uterine body in mormoopids and Noctilio (Hood and Smith, 1982, 1983; fig. 39A), suggesting that this condition is primitive for phyllostomids.

Our scoring of this character follows that proposed by Hood and Smith (1982: character 1); however, we have eliminated their character state 0 (“uterine horns 3'4 length of body”), because this state does not appear in any of the taxa in our study. Hood and Smith (1982) had ordered transformations in this character.

Character 132

Cornual lumina distinct, join within the common uterine body (0); or reduced to tubular intramural cornua (1); or completely fused to form a single common lumen, no remnant of cornual lumina (2). The cornual lumina are distinct and join the common uterine lumen within the common uterine body in Desmodus and some phyllostomines (e.g., Macrotus, Trachops; Hood and Smith, 1982, 1983; fig. 39A–B). In contrast, the cornual lumina are reduced to tubular intramural uterine cornua (IUC) in other phyllostomines (e.g., Phyllostomus, Phylloderma), Brachyphylla, Phyllonycteris, Lonchophylla, glossophagines, and carolliines (Hood and Smith, 1982, 1983; fig. 39C). There is a single common uterine lumen and no remnants of the cornual lumina in stenodermatines (Hood and Smith, 1982, 1983; fig. 39D). The cornual lumina are distinct and join the common uterine lumen within the common uterine body in mormoopids and Noctilio (Hood and Smith, 1982, 1983), suggesting that this condition is primitive for phyllostomids.

Our scoring of this character generally follows that proposed by Hood and Smith (1982: character 3), although we combined two character states defined by those authors (state 0: “short common uterine lumen, cornual lamina join within the common uterine body” and state 1: “large common uterine lumen, cornual lumina join immediately within the common uterine body”) because the former condition is found only in one outgroup taxon (Noctilio). Our state 0 includes these two conditions. Hood and Smith (1982) ordered transformations in this character.

Character 133

Oviducts enter lateral border of uterine horns or body (0); or enter fundic border of uterine body (1); or enter fundic border of uterine body near the midsagittal line (2). The oviducts enter the lateral border of the uterine horns or body in Desmodus and phyllostomines (Hood and Smith, 1982, 1983; fig. 39A–B). The location of this intersection is shifted anteromedially so that the oviducts enter the fundic border of the uterine body in Brachyphylla, Phyllonycteris, glossophagines, Lonchophylla, and carolliines (Hood and Smith, 1982, 1983; fig. 39C). In stenodermatines, the oviducts enter the uterus more medially, near the midsagittal line on the fundic border (Hood and Smith, 1982, 1983; fig. 39D). The oviducts enter the lateral border of the uterine horns or body in mormoopids and Noctilio (Hood and Smith, 1982, 1983; fig. 39A), suggesting that this is the primitive condition for phyllostomids.

This character was previously used in a phylogenetic analysis by Hood and Smith (1982: character 5); our character is identical to theirs, except we have not ordered transformations.

Character 134

Ovarian ligament extends from ovary to external entry of oviduct (0); or extends from ovary to lateral border of uterus (1). The ovarian ligament extends from the ovary to the external entry of the oviduct into the uterus in Desmodus, phyllostomines, Brachyphylla, Phyllonycteris, glossophagines, and Lonchophylla (Hood and Smith, 1982, 1983; fig. 40A). In contrast, the ovarian ligament extends from the ovary to the lateral border of uterus in carolliines and stenodermatines (Hood and Smith, 1982, 1983; fig. 40B). The ovarian ligament extends from the ovary to the external entry of the oviduct into the uterus in mormoopids and Noctilio (Hood and Smith, 1982, 1983; fig. 40A), suggesting that this is the primitive condition for phyllostomids.

This character is identical to that used by Hood and Smith (1982: character 6) in their phylogenetic analysis.

Character 135

Accessory olfactory bulb absent (0); or present (1). In most phyllostomids, the accessory olfactory bulb, the part of the forebrain receiving sensory information from the vomeronasal organ, is well developed (Wible and Bhatnagar, 1996). In contrast, the accessory olfactory bulb is absent in Brachyphylla (Wible and Bhatnagar, 1996), Choeroniscus, and Rhinophylla (Baron et al., 199630). The accessory olfactory bulb is also absent in Mormoops, Pteronotus personatus, and P. gymnotus, and Noctilio (Wible and Bhatnagar, 1996). The accessory olfactory bulb is present in Pteronotus parnellii (Wible and Bhatnagar, 1996). Therefore, the primitive state for phyllostomids appears to be absence of the accessory olfactory bulb.

Wible and Bhatnagar (1996) first defined and mapped this character onto preexisting phylogenies of bats; however, they expressed some reservations concerning its use because the absence of the accessory olfactory bulb is perfectly correlated with a reduced or absent vomeronasal epithelial tube (VET) in mammals (i.e., a VET that lacks neuroepithelium; see character 39). However, because reduced VET have structures that indicate a sensory function (e.g., vomeronasal nerves, paravomeronasal ganglia, and receptorlike cells), Wible and Bhatnagar (1996) considered this character separately from the VET character. We do the same here, adding that the accessory olfactory bulb has been described in many more taxa than has the vomeronasal organ. Because we are not certain of the state of the VET in these taxa (i.e., the VET could be well developed in one of these taxa while the accessory olfactory bulb is absent), we felt it unwise to combine these two characters. Our scoring and character construction are equivalent to those used by Wible and Bhatnagar (1996), although due to choice of outgroups, their primitive and derived states differ from ours.

Character 136

Cerebellar vermis does not cover medial longitudinal fissure or inferior colliculi (0); or cerebellar vermis completely covers longitudinal fissure between inferior colliculi, inferior colliculi exposed dorsally only along lateral edges of cerebellar vermis (1); or inferior colliculi completely covered by cerebellar vermis and cerebral hemispheres, colliculi not visible in dorsal view (2). The inferior colliculi are fully to mostly exposed dorsally in many phyllostomines (e.g., Mimon crenulatum, Trachops; fig. 41A; McDaniel, 1976; Baron et al., 1996a, 1996b). The medial longitudinal fissure is visible between the inferior colliculi in these taxa, and the right and left colliculi appear to be located directly adjacent to one another. In contrast, a rostral extension of the cerebellar vermis covers the medial longitudinal fissure between the inferior colliculi in Diphylla, some phyllostomines (e.g., species of Micronycteris), phyllonycterines, and many stenodermatines (e.g., Centurio, Mesophylla; McDaniel, 1976; Bhatnager, 1988; Baron et al., 1996a, 1996b; fig. 41B). The inferior colliculi are partly covered medially in these genera, and lateral portions of the inferior colliculi are visible along the edges of the cerebellar vermis. As a result, the right and left colliculi appear to be separated from one another when viewed dorsally. Finally, the inferior colliculi are completely covered by the cerebellar vermis and cerebral hemispheres in Desmodus, Diaemus, some phyllostomines (e.g., Phylloderma), Brachyphylla, Lonchophylla, glossophagines, carolliines, and many stenodermatines (e.g., Enchisthenes, Uroderma; McDaniel, 1976; Bhatnager, 1988; Baron et al., 1996a, 1996b; fig. 41C). No portion of the inferior colliculi is visible in these taxa. Taxonomic polymorphism occurs in some genera. In Phyllostomus, only P. elongatus has dorsally exposed inferior colliculi whereas all other species have complete coverage (McDaniel, 1976). Accordingly, we score Phyllostomus with states 0 and 2 in the matrix. In Artibeus, Dermanura, Platyrrhinus, and Sturnira some species in each genus have partly exposed inferior colliculi, whereas their congeners have completely covered inferior colliculi (McDaniel, 1976); we score these taxa with states 1 and 2 in the matrix. The inferior colliculi are only partly exposed in Mormoops, Pteronotus, and Noctilio albiventris, but are fully exposed dorsally in Noctilio leporinus (Baron et al., 1996a, 1996b). This distribution of character states suggests that partial exposure of the inferior colliculi is the primitive condition for phyllostomids.

Straney (1980: characters D1–2) treated dorsal exposure of the inferior colliculi in two binary characters: “Inferior colliculi contiguous at midline” and “Inferior colliculi separate at midline.” Although the latter is misleading (the inferior colliculi are not actually separated, just covered medially; fig. 42), the derived conditions of these characters correspond to our states 0 and 1. Straney (1980) did not score complete coverage of the inferior colliculi by cerebral and cerebellar tissues. However, Straney (1980) left blank spaces in his matrix to indicate that the condition of the inferior colliculi was unknown in glossophagines and carolliines because they were completely covered by the cerebellar vermis. In his matrix, Straney (1980) appears to have mistakenly scored desmodontines, brachyphyllines, and stenodermatines for the derived conditions of both of his characters even though he correctly notes in his text (Straney, 1980: 39) that the inferior colliculi, when they can be viewed in these taxa, are separated.

Our scoring of this character agrees with McDaniel's (1976) descriptions, with the exception of Artibeus jamaicensis, which we have scored as state 1, and McDaniel described as having state 2. We scored Artibeus jamaicensis, and several other taxa, based on figures (e.g., McDaniel, 1976; Bhatnager, 1988) when text descriptions were not clear about the exposure and contiguity of the inferior colliculi. Our scoring of Artibeus jamaicensis agrees with the description of this taxon by Schober and Brauer (1974).

Character 137

Sex-determining system XX'XY; X chromosomes never translocated to autosomes (0); or sex determining system XX'XY₁Y₂; X chromosomes translocated to autosomes (1); or X chromosomes sometimes translocated to autosomes; polymorphic within species (2). Most phyllostomids, like most other mammals, have an XX'XY sex-determining system (Baker, 1967, 1970, 1973, 1979; Hsu et al., 1968; Forman et al., 1968; Baker and Hsu, 1970; Baker and Lopez, 1970; Baker and Bleier, 1971; Baker et al. 1973; Nagorsen and Peterson, 1975; Cadena and Baker, 1976; Gardner, 1977a; Bass, 1978; Patton and Baker, 1978; Baker and Bass, 1979; Baker et al., 1979; Johnson, 1979; Honeycutt et al., 1980; Baker et al., 1981a, 1981b, 1982). In species with this system, males and females have an equivalent number of chromosomes. However, some species of Choeroniscus, some species of Carollia, and many stenodermatines (e.g., Artibeus, Ametrida) have an XX'XY₁Y₂ sex system (Baker, 1967, 1970, 1973, 1979; Hsu et al., 1968; Baker and Hsu, 1970; Baker and Lopez, 1970; Baker and Bleier, 1971; Greenbaum et al., 1975; Stock, 1975; Gardner, 1977a; Baker et al., 1979, 1981b, 1982; Johnson, 1979; Myers, 1981). In these taxa, the X chromosomes are translocated to autosomes. Thus males have an unpaired Y chromosome (because the X is translocated to an autosome), and the homolog of the autosome carrying the X is referred to as a second Y. Some individuals of Carollia castanea, and C. brevicauda, C. perspicillata, and C. subrufa have an XX'XY₁Y₂ system (Baker, 1967, 1979; Hsu et al., 1968; Baker and Bleier, 1971; Patton and Gardner, 1971; Stock, 1975; Baker et al., 1982). Choeroniscus godmani has an XX'XY₁Y₂ system, whereas C. minor (= C. ntermedius) does not (Baker, 1967, 1970, 1979; Patton and Gardner, 1971; Stock, 1975; Baker et al., 1982). Dermanura cinerea has an XX'XY₁Y₂ system, whereas other species in this genus do not. Thus, we score Carollia with states 1 and 2 in the matrix, and score Choeroniscus and Dermanura with states 0 and 1. Mormoopids (Baker, 1967; Baker and Hsu, 1970; Baker and Lopez, 1970; Patton and Baker, 1978; Sites et al., 1981) and Noctilio (Patton and Baker, 1978) have XX'XY sex chromosomes, suggesting that this is the primitive condition for phyllostomids.

Although several authors have drawn conclusions about phylogenetic relationships using the sex chromosomes (e.g., Baker and Lopez, 1970; Greenbaum et al., 1975), this is the first use of this feature as a character in a cladistic analysis.

Character 138

Restriction site 28 present (0); or absent (1). In most phyllostomids, an Xba I restriction site (Van Den Bussche's character 28) is present in the external transcribed spacer (ETS; Van Den Bussche, 1991, 1992). However, in Platyrrhinus, Uroderma, and Vampyrodes, this restriction site is absent (Van Den Bussche, 1991, 1992). This site is present in Mormoops and Noctilio (Van Den Bussche, 1991, 1992), suggesting that this is the primitive condition for phyllostomids.

Our scoring of this character and those that follow are identical to that of Van Den Bussche (1991, 1992: character 28); however, Van Den Bussche (1991, 1992) scored Hylonycteris “9” for this character and characters 139, 141, 142, 143, 146, and 148 below to indicate that there was missing data for this taxon, whereas we use “?” for this purpose.

Character 139

Restriction site 32 present (0); or absent (1). A Hinc II restriction site (character 32) is present in the ETS of most phyllostomids (Van Den Bussche, 1991, 1992). However, in Chrotopterus, Mimon crenulatum, Artibeus, Chiroderma, and Sturnira this restriction site is absent (Van Den Bussche, 1991, 1992). The Hinc II (character 32) restriction site is present in Mormoops and Noctilio (Van Den Bussche, 1991, 1992), suggesting that this is the primitive condition for phyllostomids.

Character 140

Restriction site 36 present (0); or absent (1). In most phyllostomids a Xho I restriction site (character 36) is present in the ETS (Van Den Bussche, 1991, 1992). However, in Choeroniscus, Choeronycteris, Hylonycteris, and Musonycteris, this restriction site is absent (Van Den Bussche, 1991, 1992). This site is present in Mormoops and Noctilio (Van Den Bussche, 1991, 1992), suggesting that this is the primitive condition for phyllostomids.

Character 141

Restriction site 38 present (0); or absent (1). An Sst I restriction site (character 38) is present within the 18S gene in most phyllostomids (Van Den Bussche, 1991, 1992). In contrast, this restriction site is absent in Ametrida, Ardops, Ariteus, Centurio, Pygoderma, and Stenoderma (Van Den Bussche, 1991, 1992). This site is present in Mormoops and Noctilio (Van Den Bussche, 1991, 1992), suggesting that this is the primitive condition for phyllostomids.

Character 142

Restriction site 43 present (0); or absent (1). In most phyllostomids, an Nco I restriction site (character 43) is present in the internal transcribed spacer (ITS; Van Den Bussche, 1991, 1992). However, in Chrotopterus and Trachops, this restriction site is absent (Van Den Bussche, 1991, 1992). This site is present in Mormoops and Noctilio (Van Den Bussche, 1991, 1992), suggesting that this is the primitive condition for phyllostomids.

Character 143

Restriction site 46 present (0); or absent (1). Most phyllostomids have an Nco I restriction site (character 46) in the ITS (Van Den Bussche, 1991, 1992). In contrast, this restriction site is absent in Ariteus, Pygoderma, and Stenoderma (Van Den Bussche, 1991, 1992). This site is present in Mormoops and Noctilio (Van Den Bussche, 1991, 1992), suggesting that this is the primitive condition for phyllostomids.

Character 144

Restriction site 47 present (0); or absent (1). In most phyllostomines, a Stu I restriction site (character 47) is present in the ITS (Van Den Bussche, 1991, 1992). However, this restriction site is absent in Macrophyllum, Brachyphylla, Phyllonycteris, lonchophyllines, glossophagines, carollines, and stenodermatines (Van Den Bussche, 1991, 1992). This site is present in Mormoops and Noctilio (Van Den Bussche, 1991, 1992), suggesting that this is the primitive condition for phyllostomids.

Character 145

Restriction site 49 present (0); or absent (1). In most phyllostomids, a Pvu II restriction site (character 49) is present in the ITS (Van Den Bussche, 1991, 1992). However, in Choeroniscus, Choeronycteris, Hylonycteris, and Musonycteris, this restriction site is absent (Van Den Bussche, 1991, 1992). This site is present in Mormoops and Noctilio (Van Den Bussche, 1991, 1992), suggesting that this is the primitive condition for phyllostomids.

Character 146

Restriction site 50 present (0); or absent (1). A BamH I restriction site (character 50) is present in the ITS in most phyllostomids (Van Den Bussche, 1991, 1992). In contrast, in many phyllostomines (e.g., Phyllostomus, Trachops), Leptonycteris, Monophyllus, and lonchophyllines this restriction site is absent (Van Den Bussche, 1991, 1992). This restriction site is present in Mormoops and Noctilio (Van Den Bussche, 1991, 1992), suggesting that this is the primitive condition for phyllostomids.

Character 147

Restriction site 52 present (0); or absent (1). Most phyllostomids have a Stu I restriction site (character 52) present in the ITS (Van Den Bussche, 1991, 1992). However, this restriction site is absent in Choeroniscus, Choeronycteris, Hylonycteris, and Musonycteris (Van Den Bussche, 1991, 1992). This site is present in Mormoops and Noctilio (Van Den Bussche, 1991, 1992), suggesting that this is the primitive condition for phyllostomids.

Character 148

Restriction site 53 present (0); or absent (1). In most phyllostomids, a Bgl II restriction site (character 53) is present in the ITS (Van Den Bussche, 1991, 1992). In contrast, this restriction site is absent in Lonchophylla and Lionycteris (Van Den Bussche, 1991, 1992). This restriction site is present in Mormoops and Noctilio (Van Den Bussche, 1991, 1992), suggesting that this is the primitive condition for phyllostomids.

Character 149

Restriction site 54 present (0); or absent (1). In glossophagines, lonchophyllines, and stenodermatines except Sturnira, a Xho I restriction site (character 54) is present in the ITS (Van Den Bussche, 1991, 1992). However, in desmodontines, phyllostomines, Brachyphylla, Phyllonycteris, and Sturnira, this restriction site is absent (Van Den Bussche, 1991, 1992). This restriction site is present in Mormoops and Noctilio (Van Den Bussche, 1991, 1992), suggesting that this is the primitive condition for phyllostomids.

Character 150

900 base pair EcoRI-defined nuclear satellite DNA repeat absent (0); or present (1). An EcoRI-defined nuclear satellite DNA repeat is present in Artibeus, Dermanura, and Koopmania (Van Den Bussche et al., 1993). However, this repeat is absent in all other phyllostomids (Van Den Bussche et al., 1993). The repeat is absent in Mormoops and Noctilio, suggesting that this is the primitive condition for phyllostomids.

Van Den Bussche et al. (1993) discussed the phylogenetic implications of the taxonomic distribution of this feature, but did not formally define and score it as a character.