Phylogeny of Phyllostomid Bats (Mammalia: Chiroptera): Data from Diverse Morphological Systems, Sex Chromosomes, and Restriction Sites

Home , Brachyphylla, Carolliinae, Erophylla, Glossophaginae, Lonchophylla, Lonchorhina

PHYLOGENY OF PHYLLOSTOMID BATS (MAMMALIA: CHIROPTERA): DATA FROM DIVERSE MORPHOLOGICAL SYSTEMS, SEX CHROMOSOMES, AND RESTRICTION SITES

ANDREA L. WETTERER Graduate Fellow, Division of Vertebrate Zoology, Department of Mammalogy, American Museum of Natural History; Graduate Student, Program in Ecology and Evolutionary Biology Columbia University

MATTHEW V. ROCKMAN Intern, Division of Vertebrate Zoology, Department of Mammalogy, American Museum of Natural History

NANCY B. SIMMONS Associate Curator, Division of Vertebrate Zoology, Department of Mammalogy, American Museum of Natural History

BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY Number 248, 200 pages, 66 ®gures, 8 tables, 4 appendices Issued March 1, 2000 Price: $17.40 a copy

Abstract ...... 4 Introduction ...... 5 Historical Background ...... 7 Phyllostomid Classi®cation from 1758 to Present ...... 7 Summary ...... 35 Materials and Methods ...... 36 Sampling and Character Coding ...... 36 Assessing Congruence Among Data Sets ...... 39 Phylogenetic Analyses ...... 41 Character Descriptions ...... 42 Pelage and Integument ...... 46 Skull and Dentition ...... 67 Postcranium ...... 87 Hyoid Apparatus ...... 93 Tongue ...... 101 Digestive Tract ...... 115 Reproductive Tract ...... 116 Brain ...... 119 Sex Chromosomes ...... 122 Restriction Sites ...... 122 EcoRI-De®ned DNA Repeat ...... 124 Results ...... 125 Taxonomic Congruence Analysis ...... 125 Pelage and Integument ...... 125 Skull and Dentition ...... 125 Postcranium ...... 125 Hyoid Apparatus ...... 125 Tongue ...... 129 Digestive Tract ...... 131 Reproductive Tract ...... 131 Brain ...... 131 Sex Chromosomes ...... 131 Restriction Sites ...... 131 EcoRI-De®ned DNA Repeat ...... 131 Summary ...... 132 Character Congruence Analysis ...... 132 Discussion ...... 135 Comparison of Results of Taxonomic and Character Congruence Analyses ...... 135 Classi®cation of Phyllostomid Bats ...... 136 Higher-level Classi®cation ...... 136 Generic Considerations ...... 140 Interpretation of Chromosomal Data ...... 142 Interpretation of Immunological Data ...... 145 Uterine Fusion: Progressive and Unidirectional? ...... 147 Evolution of Facial Features: A Phylogenetic Perspective ...... 148 Vibrissae ...... 148 The Noseleaf ...... 155 Tracing the Diversi®cation of Feeding Habits ...... 162 2 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 3

Summary and Conclusions ...... 170 Acknowledgments ...... 172 References ...... 173 Appendix 1: Specimens Examined ...... 184 Appendix 2: Data Matrix ...... 188 Appendix 3: Discrete-State Characters Not Included In This Study ...... 192 Appendix 4: Taxonomic Diagnoses ...... 192 4 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

ABSTRACT Phyllostomidae is a large (Ͼ 140 species), diverse clade of Neotropical bats. Different species in this family feed on blood, insects, vertebrates, nectar, pollen, and fruits. We investigated phylogenetic relationships among all genera of phyllostomid bats and tested monophyly of several genera (e.g., Micronycteris, Mimon, Artibeus, Vampyressa) using 150 morphological, karyological, and molecular characters. Results of parsimony analyses of these combined data indicate that all traditionally recognized phyllostomid subfamilies are monophyletic and that most taxa that share feeding specializations form clades. These results largely agree with studies that have used a taxonomic congruence approach to evaluate karyological, immunological, and limited sets of morphological characters, although our ®nding that Phyllostominae is monophyletic is novel. Our results indicate that several genera (Micronycteris, Artibeus, and Vampyressa) are not monophyletic. We propose a new classi®cation for Phyllostomidae that better re¯ects hypothesized evolutionary relationships. Important features of this new classi- ®cation include: (1) formal recognition of two clades that group nectarivorous and frugivorous subfamilies, respectively, (2) rede®nition of Glossophaginae and recognition of two tribal-level taxa within that subfamily, (3) recognition of several tribal-level taxa in Phyllostominae, (4) formal recognition of two clades that have been colloquially referred to as ``short-faced'' and ``long-faced'' stenodermatines, (5) elevation of the subgenera of Micronycteris to generic rank, (6) recognition of Mesophylla as a junior synonym of Ectophylla, (7) recognition of Enchis- thenes as a distinct genus, and (8) retention of Dermanura and Koopmania as subgenera of Artibeus. Although Vampyressa is not monophyletic in our tree, we recommend no nomenclatural change because we did not include all Vampyressa species in our study. Comparisons of character and taxonomic congruence approaches indicate that character congruence provides improved resolution of relationships among phyllostomids. Many data sets are informative only at limited hierarchical levels or in certain portions of the phyllostomid tree. Although both chromosomal and immunological data provide additional support for several clades that we identi®ed, these data sets are incongruent with many aspects of our phylogenetic results. These con¯icts may be due to methodological constraints associated with the use of karyological and immunological data (e.g., problems with assessing homologies and distinguishing primitive from derived traits). Among other observations, we ®nd that Macrotus waterhousii, which has been thought to have the primitive karyotype for the family, nests well within the phyllostomine clade. This suggests that results of previous analyses of chromosomal data may need to be reevaluated. Mapping characters and behaviors on our phylogenetic tree provides a context for evaluating hypotheses of evolution in Phyllostomidae. Although previous studies of uterine evolution in phyllostomids and other mammals have generally supported the unidirectional progressive fusion hypothesis, our results indicate that intermediate stages of external uterine fusion are often derived relative to the fully simplex condition, and that reversals also occur with respect to internal uterine fusion. Uterine fusion therefore appears to be neither completely unidirectional nor progressive in Phyllostomidae. Evolution of the vibrissae and noseleaf is similarly complex and homoplasy is common in these structures; however, many transformations in these systems diagnose clades of phyllostomids. Within Phyllostomidae, there is considerable derived reduction in numbers of vibrissae present in various vibrissal clusters. The phyllostomid noseleaf seems to have become a much more elaborate and complex structure over evolutionary time. Primitively within the family, the spear was short, the internarial region was ¯at, and the horseshoe was undifferentiated from the upper lip. Subsequently, within the various subfamilies, the spear became more elongate, the central rib and other internarial structures evolved, and the labial horseshoe became ¯aplike or cupped in some taxa. Dietary evolution in phyllostomids appears somewhat more complex than previously thought. We ®nd that most of the major dietary guilds (e.g., frugivory, sanguivory) are represented by a single large clade within Phyllostomidae, indicating that each feeding specialization evolved once. However, reversals do occur (e.g., loss of nectar- and pollen-feeding in many phyllostomines and stenodermatines), and some specializations may have evolved more than once (e.g., carnivory). 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 5

INTRODUCTION Of the 17 families of extant microchirop- is little agreement about how Phyllostominae teran bats, Phyllostomidae is the largest fam- should be subdivided. Considerable attention ily endemic to the New World, with 49 gen- has been focused on the monophyly and re- era and more than 140 species (Koopman, lationships of the nectar-feeding subfamilies 1993, Simmons, 1998). Feeding habits are (e.g., Baker and Lopez, 1970; Forman, 1971; unusually diverse in this family; dietary spe- Phillips, 1971; Smith, 1976; Wilder, 1976; cializations include sanguivory (blood-feed- Gardner, 1977a; Baker and Bass, 1979; Bak- ing), insectivory, carnivory, omnivory, nec- er et al., 1981a; Grif®ths, 1982; Haiduk and tarivory, palynivory (pollen-feeding), and Baker, 1982; Warner, 1983; Smith and Hood, frugivory. Interest in the evolutionary origins 1984; Honeycutt and Sarich, 1987a; and of these feeding habits has motivated nu- Baker et al., 1989). Monophyly of several merous studies of phyllostomid relationships. phyllostomid genera, including Micronycter- Data sets that have been applied to this prob- is, Mimon, Phyllostomus (sensu Baker et al., lem include allozymes, chromosomal mor- 1988a) Artibeus, and Vampyressa, has also phology, host-parasite associations, immu- been questioned (Anderson, 1906; Miller, nological distances, morphology, rDNA re- 1907; Straney et al., 1979; Honeycutt, 1981; striction sites, and mitochondrial DNA se- Straney 1980; Koop and Baker, 1983; Owen, quences (see table 1). Analyses of these data 1987, 1991; Baker et al., 1988a; Van Den sets have produced a large number of com- Bussche, 1992; Van Den Bussche and Baker, peting hypotheses of phyllostomid relation- 1993; Van Den Bussche et al., 1993, 1998; ships. Few attempts have been made to in- Simmons, 1996; Jassal and Simmons, 1996). vestigate congruence and explore con¯icts This study principally addresses phyllos- among data sets; consequently, there are tomid relationships at and above the generic many disagreements concerning phyllostom- level. We examine phyllostomid relation- id relationships at all taxonomic levels. ships using characters of the integument, pel- Historically, Phyllostomidae has been di- age, skull, dentition, postcranium, hyoid ap- vided into as few as two and as many as paratus, tongue, digestive tract, urogenital eight subfamilies (see table 2). Koopman's tract, brain, sex chromosomes, and rDNA re- (1993) classi®cation, which recognized the striction sites. Previous studies that examined largest number of subfamilial groupings pro- multiple data sets relied soley on taxonomic posed to date, included one subfamily of san- congruence, an approach that may have se- guivores (Desmodontinae), one of insecti- rious limitations when used as the only meth- vores, carnivores, and omnivores (Phyllosto- od of phylogenetic reconstruction (see Ma- minae), four of nectarivores and palynivores terials and Methods below). Here, we use (Brachyphyllinae, Phyllonycterinae, Glosso- taxonomic congruence to explore con¯icts phaginae, and Lonchophyllinae), and two of among data sets, but do not rely on this frugivores (Carolliinae and Stenodermati- method to reconstruct phylogeny; instead, we nae). Monophyly of some of these subfami- employ character congruence (``total evi- lies has been questioned. For example, many dence'') to resolve the branching pattern authors agree that Phyllostominae (sensu within the family and to test the monophyly Koopman, 1993) is not monophyletic (e.g., of previously recognized clades. Our goal is Walton and Walton, 1968; Slaughter, 1970; to develop a robust, well-resolved phylogeny Smith, 1972, 1976; Hood and Smith, 1982; of phyllostomid genera that will serve as a Honeycutt and Sarich, 1987a; Baker et al., framework for future studies of the biology 1989). However, this consensus view has not and evolution of this unusually diverse group been re¯ected in classi®cations because there of bats. 6 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 7

HISTORICAL BACKGROUND We present a summary of more than 200 cluding carnivorans, insectivorans, lago- years of higher-level phyllostomid classi®- morphs, primates, rodents, and xenarthrans. cation (and see table 2). Although we con- By the early 19th century, bats were gener- centrate on relationships among subfamilies ally recognized as a separate higher-level and genera, we occasionally note information group, although this group often included relevant to the monophyly of certain genera. dermopterans. When discussing historical classi®cations we LaceÂpeÁde (1799) recognized the ®rst phyl- use the original names with the spellings lostomid genus, Phyllostomus. Although used by the author(s) of each study, including only four species and one genus of phyllos- the incorrect Chilonycterinae, Lobostominae, tomids were named prior to 1800, the pace Phyllostomatinae, Hemiderminae, Stenoder- of discovery quickened remarkably after this minae, Desmodidae, and Anthorhina, Lon- date. More than 20 genera and 34 species choglossa, Vampyriscus, Vampyrops (see were described between the years of 1800 Miller, 1924; Cabrera, 1958; Handley, 1960; and 1850 (see ®g. 1). Most authors classi®ed Smith, 1972; Jones and Carter, 1976; Hand- these newly discovered phyllostomids in ley, 1980; Gardner and Ferrell, 1990; for in- families or tribes with taxa that are today rec- formation on other synonymys see Miller, ognized as members of other families (e.g., 1924 and Koopman, 1993). We have, how- Pteropodidae, Emballonuridae, Megaderma- ever, corrected obvious spelling errors. We tidae, Molossidae, Mormoopidae, Noctilion- use the subfamilial names proposed by idae, Rhinolophidae, Rhinopomatidae, and Koopman (1993, 1994) in some discussions; Vespertilionidae). our use of these names is for convenience Dumeril (1806), in his ``natural classi®- only and does not imply that these taxa are cation of animals,'' was one of the ®rst au- monophyletic. thors to recognize multiple genera within his single family of bats, ChiropteÁres. Dumeril PHYLLOSTOMID CLASSIFICATION (1806) recognized Phyllostomes as one of FROM 1758 TO PRESENT these six genera. Linnaeus (1758) recognized seven bat spe- Fischer von Waldheim (1813) recognized cies, all placed in the genus Vespertilio.Two bats as the order Dactyloptera, which he split of these species, Vespertilio spectrum (ϭ into two divisions (``naso simpli'' or ``naso Vampyrum spectrum) and V. perspicillata (ϭ cristato''). The single phyllostomid genus Carollia perspicillata), are today placed in Phyllostoma with Megaderma and Rhinolo- the family Phyllostomidae. Subsequent early phus formed the ``naso cristato'' group. classi®cations of bats usually recognized Oken (1816) recognized four ``Gattung'' only one or two genera of bats (Vespertilio (genera) of bats in his natural history text. In and Noctilio or Pteropus). These genera were one genus, Oken (1816) placed most phyl- placed in groups with other mammals, in- lostomids and Pteropus minimus. Oken 8 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 9

these groups, two, Phyllostomes and SteÂn- odermes, included phyllostomids. In his treatise on mammalogy, Desmarest (1820) recognized three phyllostomid genera, Phyllostome, Glossophage, and StenodeÂrme, among the 16 genera of bats in his famille CheiropteÁres, tribu Chauve-souris. While Glossophage and StenodeÂrme contained only glossophagines and stenodermatines respectively, the genus Phyllostome included a variety of species now recognized as desmodontines, phyllostomines, carolliines, and stenodermatines. Goldfuss (1820) was the ®rst taxonomist to group bats into families, recognizing four in his classi®cation. Goldfuss (1820) included most phyllostomids in the family Phyllo- stomata within the genus Phyllostoma. Other members of this family were Megaderma, Nycteris, Rhinolophus, and Rhinopoma. Goldfuss (1820) placed Stenodermata within the family Noctiliones with molossids, noctilionids, and vespertilionids. Gray (1821) was the ®rst to divide the class Cheiroptera into two groups, the orders Fructivorae and Insectivorae. Gray (1821) placed the three phyllostomid genera, Phyl- lostoma, Vampyre, and Stenodermes,inthe family Noctilionidae, a member of Insecti- vorae. Other genera in this family were Mol- losses, Noctilio, and Nyctimones. In his monograph on Brazilian bats and primates, Spix (1823) recognized two families of bats, Anistiophori, for genera without noseleaves, and Istiophori, for those with Fig. 1. The pace of discovery of currently noseleaves. Only the four phyllostomid gen- recognized phyllostomid taxa described from era (Diphylla, Phyllostoma, Vampyrus, and 1750 to 1993. A. Genera. B. Species. There has Glossophaga) were members of Istiophori. been a steady decrease in the rate of description Lesson (1827) followed Spix (1823) by of new genera since the early 1800s, but the pace splitting his tribe Chauve-souris into the di- of description of new species has not declined at visions Anistiophori and Istiophori. Lesson the same rate. We used dates of publication from (1827) recognized seven genera of Phyllo- Koopman (1993). stomes in his family Istiophori: Phyllostoma, Vampirus, Glossophaga, Monophyllus, Arti- beus, Madateus (ϭ Artibeus), and Rhinopo- (1816) classi®ed Stenoderma in another ge- ma. Lesson (1827) did not consider Steno- nus with emballonurids, molossids, and ves- derma a member of this group, but placed it pertilionids. in Noctilonina in the family Anistiophori. Although Cuvier (1817) followed Linnae- Noctilionina included molossids, a mor- us (1758) in recognizing only a single genus moopid, noctilionids, and a vespertilionid. of bats (Vespertilio) in his classi®cation of Gray (1826) recognized a single family of the animal kingdom, he nevertheless divided bats (Vespertilionidae), which he divided into this genus into more than 12 groups. Among two sections, Istiophori and Anistiophori. 10 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Gray (1826) divided ®ve subfamilies among still in use today (see table 2). Gervais these two groups. The subfamily Phyllosto- (1856) recognized four tribes of phyllostom- mina, placed in Istiophori, included Rhino- ids: Desmodina, Vampyrina, Glossophagina poma and all phyllostomid genera except (including Hemiderma [ϭ Carollia]), and ``perhaps'' Stenoderma. Gray (1826: 243) Stenodermina (including Brachyphylla). suggested that this genus might belong in the These tribes were equivalent to the kinds subfamily Noctilionina, a group in Anistio- (``genre'') he had previously proposed (Ger- phori. vais, 1854). In a classi®cation of vertebrates, Bona- Koch's (1862±63) classi®cation included parte (1831) recognized Phyllostomina as two suborders, Carpophagen and Entomo- one of ®ve ``subfamilies'' within his order phagen. Within Entomophagen, Koch (1862± Chiroptera and its single family Vespertili- 63) recognized two families, Gymnorrhina ones. Phyllostomina included Phyllostoma and Istiophora. Istiophora contained three (and as recognized subgroups of this genus groups of phyllostomids: Diphyllata included Desmodus, Phyllostoma, and Vampyrus), Choeronycteris, Glossophaga, Phyllostoma Glossophaga, Megaderma, Mormops, Nyc- (this genus included as `Ùntergattung'' Nyc- teris (not Nyctinomus as reported by Miller tiplanus [ϭ Sturnira] and Sturnira), Nycteris, [1907]), Nyctophilus, and Rhinopoma. Later, and Nyctophilus; Monophyllata included Bonaparte (1838) recognized three families Desmodus, Diphylla, Brachyphylla, and of Chiroptera. The family Vampyridae con- Rhinopoma; and Pseudophyllata was com- tained the sole subfamily Vampyrina, but posed of a single genus, Stenodermata. Bonaparte (1838) did not name the genera he Peters (1865) recognized seven families of included in this subfamily. bats in his classi®cation. Five subfamilies Gray's (1838) classi®cation remained sim- comprised the family Phyllostomata: Vam- ilar to his classi®cation of 1826. However, at pyri included both carolliine and phyllosto- lower levels, Gray's (1838) tribe Phyllosto- mine genera; Glossophagae included glos- mina was composed of Lavia, Megaderma, sophagines and Phyllonycteris; Stenodermata Mormoops, Rhinopoma, and all phyllostomid included Brachyphylla and stenodermatines. genera save Ariteus, which he placed in the The remaining two subfamilies Peters (1865) tribe Rhinolophina. recognized were Desmodi and Mormopes. Wagner's (1840) classi®cation recognized Gray's (1866a±d) last classi®cation of bats Chiroptera as a suborder with three families was presented as a series of papers in which (Frugivora, Istiophora, Gymnorhina). In the he recognized ®ve families. In the family family Istiophora, Wagner (1840) included Phyllostomidae he included Desmodina, only two tribes (``Sippe''): Desmodina and Phyllostomina, Vampyrina, Glossophagina, Phyllostomata. Phyllostomata included four and Stenodermina as tribes. Gray (1866d) phyllostomid genera (Phyllostoma, Brachy- also introduced ®ve monotypic tribes: Lon- phylla, Glossophaga, and Stenoderma), as chorhinina, Macrophyllina, Trachyopina, well as Megaderma, Nycteris, Nyctophilus, Brachyphyllina, and Centurionina (Rehn Rhinolophus, and Rhinopoma. [1901] later supported recognition of the lat- Lesson (1842) abandoned Spix's (1823) ter group as Centurioninae). Gray's (1866d) divisions Anistiophori and Istiophori; how- Phyllostomina included Alectops (ϭ Phyllos- ever, he still recognized the same ®ve groups tomus), Guandira (ϭ Phylloderma), Phyllos- (now families) in the tribe Chiroptera. The toma, Schistozoma (ϭ Micronycteris), Tylos- family Phyllostomineae included phyllostoma (ϭ Mimon crenulatum), Carollia, Rhin- tomid genera, Megaderma, Mormoops, Nyc- ophylla, and Rhinops (ϭ Carollia). Vampyr- tophyllus, Nycteris, and Rhinopoma. ina included Chrotopterus, Lophostoma (ϭ Gervais (1854) recognized four families in Tonatia), Macrotus, Micronycteris, Mimon, his order CheiropteÁres. In both his natural and Vampyrus. history of mammals and his work on South Gill's (1872) classi®cation divided the or- American bats, Gervais (1854, 1856) restrict- der Chiroptera into two suborders, Animali- ed PhyllostomideÂs to New World leaf-nosed vora and Frugivora. All phyllostomids ap- bats and laid the basis for the classi®cation peared in families in the suborder Animali- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 11 vora. Although Gill (1872) recognized Phyl- recognized ®ve families1. Winge (1941) fol- lostomidae as a distinct family, he placed lowed Dobson (1875, 1878) in recognizing a Vampyrinae, Glossophaginae, and Stenoder- division between Mormopini (including Noc- minae as subfamilies of another family, Me- tilio) and Phyllostomatini within Phyllosto- gadermidae. Desmodidae and Mormopidae matidae. Winge (1941) subdivided Phyl- appear as separate families in Gill's (1872) lostomatini into four groups: Desmodontes, arrangement. Phyllostomata, Glossophagae, and Stenoder- In an attempt to arrange genera and fam- mata. Within Phyllostomata, Winge (1941) ilies of bats according to their ``natural af®n- placed Lonchorhina, Macrophyllum, and ities,'' Dobson (1875, 1878; see table 2) split Macrotus in a basal group because of their Chiroptera into two currently recognized primitive external and dental morphology. suborders (Megachiroptera and Microchirop- Winge (1941) considered two other lineages tera) and six families. Dobson (1875, 1878) relatives of the Lonchorhina group: (1) Lo- phostoma ( Tonatia), Phylloderma, Schis- recognized two subfamilies, Lobostominae ϭ tozoma (ϭ Micronycteris), Trachyops, and (ϭ Mormoopidae) and Phyllostominae, in his Vampyrus, and (2) Mimon (ϭ Mimon ben- family Phyllostomidae. Following Peters nettii), Phyllostoma, and Tylostoma (ϭ Mi- (1865), Dobson (1875, 1878) divided Phyl- mon crenulatum). Although Winge (1941) lostominae into Desmodontes, Vampyri (ϭ viewed these two lineages as successively Phyllostominae and Carolliinae), Glossopha- more advanced than the Lonchorhina group, gae, and Stenodermata (including Brachy- he recognized Hemiderma (ϭ Carollia) and phylla). Dobson (1875: 353±354) concluded Rhinophylla as the most derived members of that ``Rhinophylla leads from the Vampyri to Phyllostomata. the Glossophagae; and the close connexion Winge (1941) viewed Phyllostomata as of the Vampyri with the Stenodermata is seen less specialized than Glossophagae and Sten- in the similarity of the warts of the lower odermata. Within Stenodermata, Winge lip.'' Dobson (1875) also noted the close (1941) recognized Vampyrops (ϭ Platyrrhin- morphological similarity of Desmodus and us) and Sturnira as the most dentally primi- Brachyphylla. tive genera. Another group was composed of In a natural history text, Gill (1884) divid- Artibeus, Stenoderma, Centurio, and Pygo- ed bats into two suborders, Animalivora and derma; Chiroderma represented a separate Frugivora, and placed 10 families in these lineage. Winge (1892) considered Brachy- two suborders. Gill (1884) recognized Des- phylla to be a member of Desmodontes; modontidae, Mormopidae, and Phyllostomi- however, he (1941) later reclassi®ed Brachy- dae as separate families, and divided phyl- phylla as a special, primitive offshoot of lostomids into three subfamilies, Phyllosto- Stenodermata. mines, Glossophagines, and Stenodermines. Allen (1898) recognized three `àlliances'' Flower and Lydekker (1891) recognized within Glossophaginae in his study of this groups similar to those named by Dobson subfamily. Phyllonycteris was the sole member of the phyllonycterine alliance. A second (1875, 1878) in their classi®cation (they re- group included Glossophaga, Leptonycteris, lied primarily on ordinal accounts of bats and (``probably'') Monophyllus. Anoura, written by Dobson). However, these authors Choeronycteris, and Lonchoglossa (ϭ An- used the subfamily name Chilonycterinae rather than Lobostominae. 1 Winge wrote in Danish, which made his publications Allen (1892a) named an additional phyl- largely inaccessible to non-Danish scientists. However, lostomid subfamily for Natalus, concluding an English translation of Winge's work on interrelation- that presence of a rudimentary noseleaf in ships of mammalian genera was published in 1941. This late embryonic stages indicated a closer af- volume (which we cite simply as ``Winge, 1941'') was ®nity to Phyllostomidae than to Vespertilion- based both on Winge's published works and his unpublished notes. The core of the section on bats was taken idae. However, no subsequent authors fol- from Winge (1892); however, the notes that followed lowed this suggestion. included Winge's comments on papers published be- Within Chiroptera, Winge (1892, 1941) tween 1892 and 1922. 12 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 oura) composed the choeronycterine alli- but suggested that Ectophylla and Mesophyl- ance. Allen (1898: 238) derived the latter la formed a ``distinct group.'' Miller (1907: two groups from Vampyri (ϭ Phyllostominae 170) also found that Ametrida, Centurio, and and Carolliinae), but noted that Phyllonyc- Sphaeronycteris formed a closely related teris `ìs so near Brachyphylla that it would group within ``short-snouted'' stenoderma- be easy to effect the transition and remove tines2, and that Centurio could be linked to the genus to the alliance expressed by the ``typical'' stenodermatines through Phyllops term brachyphylline. It is akin, therefore, if and Pygoderma. not annectant to the subfamily Stenodermi- In 1945, Simpson published his in¯uential nae.'' classi®cation of mammals. The section on In his classi®cation, Weber (1904) recog- Phyllostomidae closely followed Miller's nized two chiropteran suborders and ®ve (1907) classi®cation. Simpson (1945) rec- families, one of which was Phyllostomatidae. ognized seven phyllostomid subfamilies: This family included three subfamilies: Lo- Chilonycterinae, Phyllostomatinae, Phyllo- bostominae (including Noctilio), Desmodon- nycterinae, Glossophaginae, Carolliinae, tinae, and Phyllostominae. Within Phyllo- Sturnirinae, and Stenoderminae. Simpson stominae, Weber (1904) identi®ed three (1945) placed desmodontines in their own groupings that corresponded to Dobson's family. Simpson's (1945) main alteration of (1875, 1878) Vampyri (ϭ Phyllostominae Miller's (1907) work was to synonymize and Carolliinae), Glossophagae (which in- many genera. For example, Simpson (1945) cluded Phyllonycteris), and Stenodermata synonomized Hylonycteris with Choeronyc- (including Brachyphylla). teris, Mesophylla with Ectophylla, and Ari- Miller's (1907) revision of Chiroptera teus, Ardops, and Phyllops with Stenoderma. formed the basis of many subsequent clas- These represent only some of Simpson's si®cations. Miller (1907) recognized two (1945) generic revisions. suborders and 17 families within Chiroptera. Savage (1951) described a new genus and Miller (1907) relied on his descriptions of species of extinct phyllostomid, Notonycteris craniodental, facial, and postcranial mor- magdalenensis, from late Miocene beds in phology to de®ne seven phyllostomid sub- Colombia. Savage (1951) compared Noto- families: Chilonycterinae (ϭ Mormoopidae), nycteris with Phyllostomus, Chrotopterus, Phyllostominae, Phyllonycterinae, Glosso- and Vampyrum, concluding that Notonycteris phaginae, Hemiderminae (ϭ Carolliinae), was more similar to the last two genera than Sturnirinae, and Stenoderminae (including to Phyllostomus. Savage (1951: 362) even Brachyphylla; see table 2). Miller (1907) rec- suggested that, ``The fossil makes an accept- ognized Desmodontidae as a separate family. able structural predecessor for Vampyr- Differences between Miller's (1907) classi- um....'' ®cation and those published previously in- In their study of facial histology in bats, cluded recognition of Phyllonycterinae (pre- Dalquest and Werner (1954: 159) offered a viously included within Glossophaginae), higher-level classi®cation that they consid- Hemiderminae (previously included within ered ``more nearly the true phylogenetic or- Phyllostominae), and Sturnirinae (previously der than that adopted by Miller'' (1907). Al- included within stenodermatines) as distinct though the authors noted no appreciable dif- subfamilies. Miller (1907) suggested that ferences between mormoopids and phyllos- hemidermines (carolliines) might be most tomids, Dalquest and Werner (1954) closely allied to Desmodontidae and Phyl- accorded familial status to Chilonycterinae lonycterinae. (Chilonycteridae ϭ Mormoopidae). Later Miller (1907) also discussed some rela- studies of echolocation calls (Grif®n and tionships within subfamilies, suggesting, for example, that Diphylla was the least special- 2 The ``short-snouted'' or ``short-faced'' stenoderma- ized member of Desmodontinae. Within tines are Ametrida, Ardops, Ariteus, Centurio, Phyllops, Stenoderminae, Miller (1907: 159) viewed Pygoderma, Stenoderma, Sphaeronycteris. However, Mesophylla as an intermediate between Ec- some authors (e.g., Smith, 1976) included Artibeus spe- tophylla and Vampyrops (ϭ Platyrrhinus), cies in this group. 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 13

Novick, 1955, Novick, 1963) and host-para- matine lineage. Despite this placement, de la site associations (Machado-Allison, 1967; Torre (1961) still classi®ed Brachyphylla as Wenzel et al., 1966) supported recognition of a member of Stenoderminae. In the second mormoopids as a distinct family. stenodermatine line, Chiroderma was the sis- In their classi®cation, Hall and Kelson ter taxon of Vampyriscus (ϭ Vampyressa) (1959) recognized seven phyllostomid sub- and Mesophylla. Vampyressa was also in- families: Chilonycterinae, Phyllostominae, cluded in this group, although de la Torre's Phyllonycterinae, Glossophaginae, Carolli- (1961) tree lacked a branch for this taxon. In inae, Sturnirinae, and Stenoderminae. Hall the third group, two pairs of sister taxa, Vam- and Kelson (1959) placed Brachyphylla pyrops (ϭ Platyrrhinus) and Enchisthenes within Stenoderminae and recognized des- and Uroderma and Artibeus, formed a clade. modontines as a separate family (Hall's 1981 Sturnira and Vampyrodes appeared as suc- revised edition differed only in recognizing cessively more basal branches of this group. Brachyphyllinae for Brachyphylla and the All ``short-faced'' taxa formed a clade in de phyllonycterines). la Torre's (1961) tree. Phyllops and Ariteus In his work on the systematics of Sturnira, were successive sister taxa to Pygoderma de la Torre (1961) wrote extensively on re- and Ardops. Sphaeronycteris and Ametrida lationships within Phyllostomidae. Based on appeared as sister taxa, and formed a clade his analysis of unspeci®ed ``cranial and gen- with Centurio. Stenoderma occupied the bas- eral morphology,'' de la Torre (1961: 140) al branch within the Ametrida group. proposed that mormoopids deserved familial Subsequently, de la Torre made signi®cant status. Using dental morphology to evaluate changes in this tree (the revised version was relationships, de la Torre (1961) recognized published in Wenzel et al., 1966: ®g. 144; ®ve subfamilies: Phyllostominae, Phyllo- our ®g. 3), dramatically rearranging relation- nycterinae, Glossophaginae, Carolliinae, and ships within Phyllostominae. Vampyrum, Stenoderminae (see table 2). Phyllostomines which de la Torre (1961) had placed in a represented the most basal branch of the fam- clade with Chrotopterus, was now grouped ily (de la Torre, 1961). In addition, de la Tor- with Barticonycteris (ϭ Micronycteris; see re (1961) recognized a close relationship be- Koopman, 1978), Macrotus, and Micronyc- tween phyllonycterines and glossophagines, teris, formerly members of de la Torre's and suggested that carolliines evolved from (1961) Phyllostomus line. Chrotopterus be- a lineage ``close to'' glossophagines (de la came the sister taxon of Tonatia, and togeth- Torre, 1961: 42). However, in a tree of ``es- er with Trachops, was united with the Vam- timated evolutionary relationships'' (®g. 2), pyrum group. Also in the Wenzel et al. the carolliines appeared to be more closely (1966) tree, de la Torre recognized Macro- related to stenodermatines than glossopha- phyllum and Lonchorhina as sister taxa and gines. placed them as the ®rst branch of the Phyl- Within Phyllostominae, de la Torre (1961) lostomus group. Anthorina (ϭ Mimon cren- identi®ed three groups: (1) Chrotopterus and ulatum), Mimon (M. bennettii), Phylloderma, Vampyrum; (2) Trachops; and (3), Macrotus, and Phyllostomus formed the other branches Micronycteris, Mimon, Phyllostomus, and of the Phyllostomus group. In the Wenzel et Tonatia. Within Glossophaginae, de la Torre al. (1966) tree, de la Torre aligned Choeron- (1961) recognized two evolutionary lines. In iscus, Hylonycteris, Lichonycteris, Loncho- one clade, Glossophaga and Lionycteris were glossa (ϭ Anoura), and Musonycteris (taxa successive sister taxa to Platalina and Lon- which had not been included in his previous chophylla. Anoura, Choeronycteris Lepto- tree) with the Monophyllus lineage of glos- nycteris, and Monophyllus formed a second sophagines (s.l.). Finally, Brachyphylla, clade. Phyllonycterines appeared as the sister which de la Torre still classi®ed as a sten- group of the glossophagines. odermatine, clearly split from the carolliine As de®ned by de la Torre (1961), Sten- branch (Wenzel et al., 1966). oderminae consisted of four lineages. The Machado-Allison (1967) described host- ®rst lineage, Brachyphylla, branched off be- parasite associations and interpreted data on fore the carolliines split from the stenoder- echolocation calls (Novick, 1963) as sug- 14 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 2. Intergeneric relationships of phyllostomid bats proposed by de la Torre (1961; redrawn from ®g. 4). This tree is based on dental morphology. Vampyressa is not connected to the tree in the original ®gure. gesting that Desmodidae should be consid- af®rmed this placement of desmodontines us- ered a subfamily of Phyllostomidae. Forman ing host-parasite associations of mites of the et al. (1968) reached the same conclusion genus Eudusbabekia. based on immunological, karyological, and Baker (1967) used karyological data to morphological comparisons of phyllostomids identify seven groups of phyllostomids: (1) and desmodontines. Uchikawa (1987) later Pteronotus; (2) Carollia, Choeronycteris, 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 15

Fig. 3. Tree from Wenzel et al. (1966; redrawn from ®g. 144) by de la Torre. and Choeroniscus; (3) Leptonycteris, Glos- that are super®cially identical, Baker (1967) sophaga, Phyllostomus, Trachops, and Mac- suggested that Sturnira should be included in rotus; (4) Micronycteris; (5) Anoura; (6) Stenodermatinae. Although Pteronotus has a Sturnira, Artibeus, Vampyrops (ϭ Platyr- karyotype unique among phyllostomids, rhinus), Chiroderma, Enchisthenes, and Cen- Baker (1967: 421) noted only that ``Pterono- turio; and (7) Uroderma. Because Artibeus, tus and related genera are suf®ciently distinct Vampyrops, and Sturnira have autosomes to merit at least subfamily status.'' Baker 16 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

(1967, 1970) did not discuss the relationships ilies of the Phyllostomatidae.'' They con- among these seven groups, but noted that the cluded that mormoopids were most closely similarity of fundamental number, diploid associated with the ``Macrotus-type'' of number, certain autosomes, and the XX/ phyllostomine, a group that included Chro-

XY1Y2 sex system in Carollia and Choeron- topterus and Vampyrum, genera that Savage iscus indicated that Glossophaginae (s.l.) (1951) allied with an extinct Miocene phyl- might not be monophyletic (this point was lostomid, Notonycteris. Walton and Walton also made by Hsu et al., 1968). However, (1968) accepted Savage's (1951) conclusion problems relating the karyotype of Carollia that there is a division between the genus to Choeroniscus arose when it was discov- Phyllostomus and his Vampyrum-Chrotopte- ered that not all Choeroniscus species have rus group (their ``Macrotus-type''). a translocated X chromosome (Baker, 1970; Walton and Walton (1968) hypothesized see discussion below). Further study dem- that Phyllonycterinae and Carolliinae onstrated that the karyotype of Rhinophylla evolved from the ``Macrotus-type'' of phyl- (unknown in 1967) is more similar to the lostomine. From the ``Phyllostomus-type,'' karyotypes of other glossophagines, phyllos- Walton and Walton (1968) derived Sturniri- tomines, and stenodermatines than to the kar- nae; however, they suggested that this sub- yotypes of Carollia, Choeroniscus, and Cho- family might be best recognized as a member eronycteris (Baker and Bleier, 1971). Baker of Carolliinae. Also associated with the Phyl- and Bleier (1971) concluded that either a lostomus-type was a second lineage that split great deal of evolution occurred within Car- into Glossophaginae (s.l.) and Stenodermi- olliinae, or that this subfamily was not mono- nae. Walton and Walton (1968) further divid- phyletic. ed Stenoderminae into a primitive ``Vampyr- Koopman and Cockrum (1967) presented ops-type'' (Vampyrops ϭ Platyrrhinus) and a classi®cation of phyllostomids, recognizing a derived ``Artibeus-type.'' Although Walton Chilonycterinae, Phyllostomatinae, Phyllo- and Walton (1968: 31) treated desmodontines nycterinae, Glossophaginae, Carolliinae, as a separate family, they noted that, ``Myo- Sturnirinae, and Stenoderminae (including logically and osteologically they are very Brachyphylla). These authors placed des- close to the phyllostomatids.'' modontines into their own family. In addition to studies that indicated that Gerber (1968) and Gerber and Leone the taxonomic positions of desmodontines (1971) used immunological comparisons to and mormoopids should be evaluated, evi- investigate relationships between some glos- dence introduced in the late 1960s suggested sophagines and Carollia. Both studies re- that the position of Brachyphylla also needed ported that Choeronycteris has a greater af- review. Although Allen (1898) had noted ®nity for desmodontines and phyllostomines that Brachyphylla is morphologically similar than for glossophagines, and that Carollia to Phyllonycteris and suggested a name for has the greatest af®nity for Glossophaga spe- a subfamily including the two genera cies. Immunological results suggested that a (Brachyphyllinae), the most persuasive evi- group including Desmodus, Chrotopterus, dence for an af®liation between Brachyphyl- Phyllostomus, and Choeronycteris was more la and Phyllonycterinae was presented by closely related to the Glossophaga-Carollia Silva-Taboada and Pine (1969). Data from group than to the stenodermatines. Results of behavior, parasites, and craniodental, exter- these studies supported earlier conclusions nal, and postcranial morphology indicated to that mormoopids should be recognized as a Silva-Taboada and Pine (1969) that Brachy- distinct family of bats, that desmodontines phylla, Erophylla, and Phyllonycteris should should be considered a phyllostomid subfam- be placed in a single subfamily. Later kary- ily, and that Sturnira should be placed within ological and immunological studies also sup- Stenodermatinae. ported this conclusion (Baker and Lopez, In a study of postcranial osteology, Walton 1970; Baker and Bass, 1979; Baker et al., and Walton (1968: 29) suggested that mor- 1981a). moopids, then recognized as Chilonycteri- Baker (1967, 1973), Hsu et al. (1968), nae, were the ``most primitive of the subfam- Baker and Hsu (1970), Baker and Lopez 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 17

(1970), Greenbaum et al. (1975), and Baker Chilonycterinae, Desmodontinae, Phyllosto- et al. (1982) discussed the phylogenetic im- matinae, Phyllonycterinae, Glossophaginae, plications of the sex chromosomes of phyl- Carolliinae, and Stenoderminae. Within Sten- lostomids. Several genera of stenodermatine odermatinae, Koopman and Jones (1970) bats (Ametrida, Ardops, Ariteus, Artibeus, identi®ed three tribes, placing Brachyphylla most Dermanura species, Enchisthenes, in Brachyphyllini, Sturnira in Sturnirini, and Koopmania, Phyllops, Pygoderma, and Sten- the remaining stenodermatines in Stenoder- oderma) have an XX/XY1Y2 sex chromo- mini. some system (see character 137 for descrip- Slaughter (1970) identi®ed two major tion; Baker, 1967, 1979; Hsu et al., 1968; groups of phyllostomines in an analysis of Baker and Lopez, 1970; Greenbaum et al., chiropteran dental evolution (®g. 4). These 1975; Gardner, 1977a; Baker et al., 1979, two groups were different from those previ- 1982; Johnson, 1979; Myers, 1981). Baker ously proposed by Walton and Walton (1968) and Lopez (1970) suggested that the XX/ and Savage (1951). According to Slaughter

XY1Y2 sex system delineates a group con- (1970), the dentally similar forms Lonchorhi- sisting of Ametrida, Artibeus, Enchisthenes, na, Mimon, Phyllostomus, Trachops, and and Stenoderma (sex chromosomes of other Vampyrum composed one group, while Mac- genera mentioned above were not known at rotus represented another lineage. Slaughter this time). Baker and Lopez, (1970) consid- (1970) suggested that Lonchorhina, Mimon, ered Centurio a relative of this group, and and Phyllostomus share more dental similar- interpreted its XX/XY sex system as a re- ities with each other than with the distinct versal. Greenbaum et al. (1975) suggested Trachops and Vampyrum lines. Slaughter that two ``short-faced'' groups could be iden- (1970) also noted that the dentitions of phyl- ti®ed: (1) Centurio and Sphaeronycteris, lostomines, glossophagines (s.l.), and sten- both of which have an XX/XY system, and odermatines could each be derived indepen- (2) the remaining ``short-faced'' taxa, all of dently from the prototypic phyllostomid den- which have an XX/XY1Y2 system. tition. Slaughter (1970) echoed de la Torre's Carollia brevicauda, C. perspicillata, C. (1961) conclusion that Phyllonycterinae and subrufa and some individuals of C. castanea Carolliinae could have easily arisen from also have an XX/XY1Y2 system (Baker, within Glossophaginae (s.l.). Brachyphylla 1967; Hsu et al., 1968; Baker and Bleier, and Sturnira were identi®ed as the most den- 1971; Patton and Gardner, 1971; Stock, tally primitive stenodermatines and formed 1975; Baker, 1979; Baker et al., 1982). Cho- separate branches of Slaughter's (1970) sten- eroniscus godmani has a translocated X odermatine tree. Slaughter (1970) also iden- chromosome, while C. minor (ϭ C. inter- ti®ed a lineage consisting of Vampyrops (ϭ medius) does not (Baker, 1967, 1970; Patton Platyrrhinus), Chiroderma, and Ectophylla and Gardner, 1971; Stock, 1975; Baker, (from least to most derived), and a lineage 1979; Baker et al., 1982). Both Baker (1967, composed of Uroderma, Stenoderma, Arti- 1970) and Hsu et al. (1968) had previously beus, and Centurio (from least to most spe- suggested that the presence of a similar sex cialized). Slaughter (1970) proposed that chromosome system in Carollia and Cho- desmodontids arose from a form intermediate eroniscus might indicate that Glossophaginae between Carollia (primitive) and Rhinophyl- (s.l.) is not monophyletic. Stock (1975) used la (derived), and recognized two distinct des- C- and G-banding techniques to test these as- modontid dental lineages: Diphylla (primi- sertions, and found that banding patterns of tive) and Desmodus (derived). Slaughter Choeroniscus show no recognizable similar- (1970) identi®ed Chilonycterinae (ϭ Mor- ity to those in Carollia, suggesting that the moopidae) as the most dentally primitive two genera are not closely related. Baker et phyllostomid subfamily. al. (1981a) also found that immunological Phillips (1971) investigated craniodental and electrophoretic comparisons provided no characters in glossophagine bats and pro- evidence for glossophagine (s.l.) diphyly. duced a fully dichotomous tree which divid- Koopman and Jones' (1970; see table 2) ed glossophagines (s.l.) into two groups. One classi®cation included seven subfamilies: group was composed of Choeroniscus, Cho- 18 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 4. Slaughter (1970; redrawn after ®g. 5) proposed this tree of phyllostomids relationships based on dental character evolution. The original caption read ``Dental morphology tree suggesting types of dentition possessed by ancestral forms of chiropteran groups. Generic names are used merely to denote certain types and/or grades of dental forms.`` eronycteris, and Musonycteris. The second ment of glossophagine monophyly impossi- group consisted of all other glossophagines. ble given the data available. Phillips (1971) found that this division also Forman (1971) used stomach morphology agreed with immunological associations to identify two groups of glossophagines (Gerber, 1968) and karyological data (Baker, (s.l.): (1) Anoura, Glossophaga, and Lepto- 1967). Within the Glossophaga group, Phil- nycteris; and (2) Choeroniscus and Licho- lips (1971) identi®ed three clades: (1) Glos- nycteris. Forman (1971: 282) suggested that sophaga, Leptonycteris, and Monophyllus; these two groups formed an ``unnatural as- (2) Anoura, Lionycteris, and Lonchophylla; semblage,'' and noted that Anoura was in- and (3) Hylonycteris, Lichonycteris, Platali- termediate in certain respects between his na, and Scleronycteris. In the ®rst clade, two groups. Glossophaga and Monophyllus were sister Smith (1972) raised Chilonycterinae to fa- taxa; Lonchophylla and Lionycteris appeared milial status as Mormoopidae, the older fam- as sister taxa in the second clade. Within the ily name having priority over Chilonycteri- third group, Scleronycteris and Platalina dae. Smith (1972) cited numerous characters were successive sister taxa to Lichonycteris from many different systems which together and Hylonycteris. Phillips (1971) found that indicated that Mormoopidae is both mono- craniodental, karyological, and immunologi- phyletic and distinct from Phyllostomidae. cal data provided con¯icting information Smith (1972: 21) used the morphology of the about relationships between glossophagines medial process of the distal humerus to di- and other phyllostomids, making an assess- vide phyllostomids into a ``Micronycteris- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 19 line'' (Lonchorhina, Macrotus, Micronycter- found that all ``short-faced'' stenodermatines is, glossophagines [s.l.], and carolliines) and formed a clade. Karyological data did not a``Phyllostomus-line'' (Phyllonycterinae, support previous associations of Ectophylla Phylloderma, Phyllostomus, Trachops, Stur- with Mesophylla, which were based on mor- nirinae, Stenoderminae [ϭ Stenodermatinae], phologic similarity (Laurie, 1955; Goodwin and ``probably'' Desmodontinae). Smith and Greenhall, 1961; Baker, 1973). Instead, (1972) allied Chrotopterus, Mimon, Tonatia, Greenbaum et al. (1975) proposed that Chi- and Vampyrum with the ``Phyllostomus-line'' roderma may be more closely related to Me- despite lacking postcranial skeletons of these sophylla and Vampyressa than is Ectophylla, genera. or that Ectophylla diverged from the Meso- Baker (1973) used chromosomal morphol- phylla-Vampyressa line before a reduction in ogy to resolve relationships among stenod- diploid number occurred. Greenbaum et al.'s ermatine bats, and proposed that the primi- (1975) conclusion also contrasted with Star- tive karyotype for stenodermatines is similar ett and Casebeer's (1968) opinion that Ecto- to that of Artibeus, Sturnira, Vampyrodes, phylla is more derived than Mesophylla and and Vampyrops (ϭ Platyrrhinus). He divided Vampyressa. these ``basal'' stenodermatines into three Smith (1976) integrated previously pub- groups based on morphology of the Y chro- lished dental, host-parasite, immunological, mosome. Baker (1973) divided species of karyological, and postcranial data in a ``ten- Sturnira and Vampyrops between groups one tative'' cladogram of phyllostomid relation- and two, and species of Artibeus between ships (®g. 5; see table 2). Noting that ``much groups two and three. Species of Vampyro- of the evidence up to this time is contradic- des appeared only in group two. From this tory and confusing,'' Smith's (1976: 62) basal stock (i.e., groups one, two, and three), summary interpretation was proposed as a Baker (1973) derived four lineages. The ®rst ``point of departure for future investiga- lineage consisted of Chiroderma, Vampyres- tions.'' Among the traditionally recognized sa, and Mesophylla (from least to most spe- subfamilies, only Phyllostominae and Glos- cialized), and was derived from groups one sophaginae were not monophyletic in and two. Although no chromosomal data Smith's (1976) tree. were available for Ectophylla, this genus was Smith (1976) recognized two primary lin- associated with the end of this lineage be- eages of phyllostomids. The ``Macrotus-lin- cause it is morphologically similar to Meso- eage'' consisted of Desmodontinae, a clade phylla, as had been previously noted by Star- of phyllostomines (Lonchorhina, Macrophyl- ett and Casebeer (1968). A second lineage lum, Macrotus, and Micronycteris), Phyllo- consisted only of Uroderma species and was nycterinae (s.l.), a clade of glossophagines derived from groups two and three, whereas (except Choeroniscus, Choeronycteris, and the third consisted of Ametrida, Centurio, Musonycteris), and Carolliinae (s.l.). Brachy- Stenoderma,and Sphaeronycteris and pre- phylla appeared as the sister taxon of Phyl- sumably evolved from group three. Baker lonycteris and Erophylla, and Desmodonti- (1973) associated the karyologically un- nae, Phyllonycterinae, and Carolliinae known genera Ardops, Ariteus, Phyllops, and formed a clade. Smith's (1976) other group, Pygoderma with the end of this lineage be- the ``Phyllostomus-lineage,'' included a cause they are morphologically similar to the monophyletic group composed of the re- other four taxa. Enchisthenes appeared as a maining phyllostomines (Chrotopterus, Mi- separate offshoot of group three. mon, the extinct Notonycteris, Phylloderma, Greenbaum et al. (1975) described the kar- Phyllostomus, Tonatia, Trachops, and Vam- yotypes of Ardops, Ariteus, and Phyllops, pyrum), a glossophagine clade consisting of and concluded that Baker (1973) had cor- Choeroniscus, Choeronycteris, and Muso- rectly identi®ed their af®nities. Greenbaum nycteris, and Stenoderminae. Smith (1976) et al. (1975) suggested that Ardops, Ariteus, left relationships among the ``Phyllostomus- Phyllops, and Stenoderma were more closely lineage'' clades unresolved. related to each other than to Centurio and At lower taxonomic levels, Smith's (1976) Sphaeronycteris. However, these authors summary tree (®g. 5) indicated that Diaemus 20 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

and Desmodus were sister taxa, forming a clade with Diphylla. Among phyllostomines of the ``Macrotus-lineage,'' Smith (1976) suggested that Macrotus and Micronycteris were sister taxa, as were Lonchorhina and Macrophyllum. Among phyllostomines of the ``Phyllostomus-lineage,'' Smith (1976) recognized three pairs of sister taxa: Phyllo- derma and Phyllostomus, Mimon and Tona- tia, and Chrotopterus and Trachops. These three clades formed a polytomy; Vampyrum and the extinct Notonycteris appeared as successive sister taxa to this group. In Smith's (1976) tree, relationships among genera within the two glossophagine (s.l.) clades were identical to those proposed by Phillips (1971; see above). Among stenodermines, Smith (1976) recognized a monophyletic group of ``long- faced'' genera consisting of Chiroderma, Ec- tophylla, Mesophylla, Sturnira, Uroderma, Vampyressa, Vampyrodes, and Vampyrops (ϭ Platyrrhinus). Within this group, the sister taxa Ectophylla and Mesophylla formed a clade with Chiroderma. Uroderma and Stur- nira were successive sister taxa to a trichotomy of Vampyressa, Vampyrodes,and Vam- pyrops. The second stenodermatine group recognized by Smith (1976) was composed of ``short-faced'' genera. Within this lineage, Smith (1976) identi®ed a monophyletic group composed of Ardops, Ariteus, Phyl- lops, and Stenoderma. These genera and the sister taxa Artibeus and Enchisthenes formed a clade. Finally, Sphaeronycteris and Pygo- derma were successive sister taxa to Ametri- da and Centurio. McDaniel (1976) described aspects of brain anatomy in phyllostomids and presented two trees of potential phylogenetic relationships. McDaniel (1976) recognized Phyl- lostomatinae as the basal phyllostomid subfamily. Brachyphylla appeared as the sister taxon of Desmodontinae, and together with stenodermatines these taxa formed a clade Fig. 5. ``Tentative'' phylogeny proposed by derived from within Phyllostomatinae. The Smith (1976; redrawn from ®g. 2) after an eval- ``Macrotus-type'' phyllostomine (Macro- uation of previously published data including den- phyllum, Macrotus, Tonatia, and Trachops) tal, immunological, karyological, parasitological, was derived from more basal phyllostomines and postcranial data. Asterisk indicates an extinct (see below), and gave rise independently to taxon. phyllonycterines (s.s.), glossophagines, and carolliines. McDaniel (1976) noted that the brains of Anoura, Choeronycteris, and Mon- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 21 ophyllus are very different from those of other glossophagines included in his study, suggesting that Glossophaginae might be diphy- letic. McDaniel's (1976) second tree depicted relationships among phyllostomines. The sister taxa Lonchorhina and Mimon formed the basal branch within Phyllostomatinae. McDaniel (1976) depicted an uncertain relationship between Vampyrum and the sister taxa Phylloderma and Phyllostomus by using a dashed line; however, he provided no comments about this tentative relationship. The ``Macrotus-type'' phyllostomine appeared as the sister taxon of Micronycteris. Wilder (1976) examined histology of the parotid gland in Anoura, Glossophaga, Lep- tonycteris,and Monophyllus. Based on parotid characters, Anoura appeared to be the most divergent glossophagine included in this study. Wilder (1976) suggested that an Anoura-like ancestor led to both Anoura and a group including the other three genera. In this latter group, Glossophaga and Leptonyc- teris were sister taxa. Cadena and Baker (1976) found chromosomal similarities between desmodontines and some members of the subfamilies Phyl- lostominae, Phyllonycterinae, and Glosso- phaginae. In addition, these authors suggested that Diaemus and Diphylla were less derived than Desmodus because they had karyotypes more similar to the proposed primitive karyotype for phyllostomids. Gardner (1977a; ®g. 6) presented an ``arbitrarily derived'' tree of phyllostomid relationships mainly based on chromosomal data. Gardner (1977a) hypothesized that Des- modontinae, Phyllostominae, Carolliinae, and Stenoderminae were each monophyletic. Although Gardner (1977a) discussed the possible independent origin of several glossophagine (s.l.) groups, his tree depicted a single clade of nectar feeders. Relationships among these clades were unresolved. Gardner's (1977a) Phyllostominae included three clades united in a polytomy: (1) Lonchorhina and Macrophyllum, (2) Chro- Fig. 6. Gardner's (1977a; redrawn from ®g. 8) topterus and Tonatia, and (3) the remaining ``arbitrarily derived'' tree of phyllostomid rela- phyllostomine genera. Within the third clade, tionships based on chromosomal similarities. As- Macrotus and Micronycteris appeared as sis- terisks indicate taxa that were karyotypically unknown and whose placement was conjectural. ter taxa. Phylloderma and Phyllostomus were also sister taxa, and formed a clade with Mi- mon. These two clades, Trachops, and Vam- 22 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 pyrum formed a polytomy. Placement of Ma- mormoopids, Patton and Baker (1978) pro- crophyllum was conjectural because its kar- posed that the karyotype of Macrotus water- yotype was unknown at that time. housii (FN ϭ 60, 2n ϭ 46) was primitive for Gardner (1977a) recognized three mono- phyllostomids. Patton and Baker (1978) phyletic groups within his nectar-feeder identi®ed two phyllostomine clades which clade: (1) Phyllonycterinae (s.l.), Glossopha- differed from the more primitive Macrotus ga, Leptonycteris, and Monophyllus; (2) An- waterhousii karyotype: (1) Micronycteris, oura; and (3) all remaining glossophagine and (2) Mimon, Phyllostomus, and Tonatia. genera. Within the ®rst group, Glossophaga In the second clade, the former two genera and Monophyllus were sister taxa and formed appeared as sister taxa. a clade with Leptonycteris. Also within this Johnson (1979) carried out additional kar- group, Brachyphylla appeared as the sister yological work and suggested that phyllos- taxon of Erophylla and Phyllonycteris. Gard- tomines diverged before other subfamilies. ner (1977a) identi®ed three clades within his This group was followed by desmodontines third group: (1) Choeroniscus, Choeronyc- and stenodermatines, which split from a glos- teris, and Musonycteris; (2) Hylonycteris, Li- sophagine-brachyphylline group before ei- chonycteris, Platalina, and Scleronycteris; ther of these latter subfamilies was distinct. and (3) Lionycteris and Lonchophylla. Kar- Although G-banded chromosomes provided yotypes of Lionycteris, Musonycteris, Platal- no resolution within much of the stenoder- ina, and Scleronycteris were unknown at this matine clade, Johnson (1979) did note that time and their placement was conjectural. Mesophylla is karyologically very similar to Within Stenoderminae, Gardner (1977a) Vampyressa pusilla (G-banded chromosomes left the position of Uroderma unresolved. of Ectophylla were not available). Chiroderma, Mesophylla, and Vampyressa Straney et al. (1979: 169) used allozyme formed a clade, and the latter two genera data to produce a ``phylogenetic estimate'' of were sister taxa. Within the group formed by relationships among 13 phyllostomid genera. the remaining stenodermatine genera, Gard- Phyllostomus and Carollia appeared as sister ner (1977a) left the position of Sturnira un- taxa in their tree. This group formed a clade resolved. The sister taxa Artibeus and En- with stenodermatines, except Sturnira. With- chisthenes formed a clade with Ectophylla, in the stenodermatine clade, Chiroderma and Vampyrodes, and Vampyrops. The ®nal Uroderma were sister taxa and grouped with group consisted of ``short-faced'' stenoder- Artibeus (Dermanura) cinereus, Artibeus li- matines. The positions of Stenoderma and turatus, and Vampyrops (ϭ Platyrrhinus). Pygoderma (which was karyologically un- The last three taxa formed a clade; Artibeus known) were unresolved, but Ametrida, Ar- cinereus and Vampyrops appeared to be dops, Ariteus, and Phyllops formed a clade, more closely related to each other than to A. as did Centurio and Sphaeronycteris. lituratus. Artibeus jamaicensis and Ametrida Bass' (1978) examination of G-banded were successive sister taxa to this larger sten- karyotypes suggested that Desmodontinae, odermatine group. Phyllonycterinae, and Glossophaginae (s.l.) The other large clade identi®ed by Straney formed a clade, with Phyllostominae as the et al. (1979) included Desmodus, glosso- sister group. Bass (1978) found that the phyl- phagines, and Sturnira. Desmodus was nest- lonycterines and glossophagines included in ed within Glossophaginae, as the sister taxon her study have identical karyotypes and was of Glossophaga. This group formed a clade therefore unable to draw conclusions con- with Anoura and Sturnira as successive sister cerning their relationships. However, Bass taxa. This assemblage and the phyllostomine (1978) was able to resolve relationships group described above were sister taxa. among the desmodontine genera, and found Straney (1980) used allozymic, karyolog- that karyologically, Desmodus and Diphylla ical, immunological, and morphological data are more similar to each other than either is in separate component analyses designed to to Diaemus. elucidate relationships among phyllosto- Using G-banded chromosomes and out- mines. Based on morphological data, Straney group comparisons with noctilionids and (1980) identi®ed four groups of phyllosto- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 23 mids de®ned by two or more derived char- technique, each contained at least one group acter states (replicated components). These whose position was ambiguous. Straney groups were: (1) Lonchorhina, Macrophyl- (1980) felt unable to choose among these lum, and Macrotus; (2) Lonchorhina and trees using only the allozyme data alone. His Macrophyllum; (3) Chrotopterus, Trachops, component analysis was more successful. and Vampyrum; and (4) Chrotopterus and Straney (1980) was able to identify three ad- Vampyrum. In addition to identifying the ditional phyllostomid clades above the ge- replicated components, Straney (1980) used neric level: (1) Macrophyllum and Phyllos- subsets of his derived, ``reliable'' characters tomus, (2) Carollia and Rhinophylla, and (3) (i.e., adequately sampled characters that did Desmodus and Diphylla. not involve loss or reduction) that de®ned Corbet and Hill (1980, 1986; see table 2) compatible groups to construct ``compatible presented a classi®cation of the family, rec- cladograms.'' In all but one of the 18 com- ognizing six subfamilies: Phyllostomatinae, patible cladograms that Straney (1980) iden- Phyllonycterinae (which included Brachy- ti®ed, Desmodontinae was the sister taxon of phylla), Glossophaginae, Carolliinae, Sturni- the remaining phyllostomids. Another fre- rinae, and Stenoderminae. Desmodontines quent grouping included Lonchorhina, Ma- appeared in a separate family. crophyllum, Macrotus, and Mimon bennettii. Handley (1980) standardized nomencla- Straney (1980) suggested that, based on his ture by correcting the misuse of certain fa- data set, the two subgenera of Mimon should milial and subfamilial names which were the not be considered congeneric; in 13 of the 18 result of improperly applied Greek and Latin cladograms Straney (1980) examined, these word-formation rules. Handley (1980) noted two taxa were not each other's closest rela- that Phyllostomidae and Phyllostominae are tive. correct, while Phyllostomatidae and Phyllo- Straney (1980) evaluated the compatibility stomatinae are not. Stenoderminae is an in- of his morphologically based phylogenetic correct name for Stenodermatinae. Desmo- hypotheses with the data then available from didae (and hence Desmodinae) is an incor- chromosomes (e.g., Patton and Baker, 1978; rect form of Desmodontidae (or Desmodon- Baker, 1979) and his own work in immunol- tinae). ogy. Although he did not ®nd any congru- Baker and Lopez (1970) and Baker and ence between the phylogenies based on kar- Bass (1979) documented the similarity of the yology and his ®ndings based on morpholo- chromosomes in Brachyphylla, phyllonycter- gy, he did ®nd that one of the clades he had ines, and glossophagines, but these authors identi®ed using immunology appeared in his were not able to clarify relationships among morphology tree. The base of Straney's these taxa. Baker et al. (1981a) used electro- (1980) immunology tree was a trichotomy. phoretic and immunological data to address Macrotus was associated with Desmodus, this question. Their results indicated that whereas Glossophaga, Carollia, and Arti- Brachyphylla was the sister taxon of Ero- beus formed a clade with Carollia and Arti- phylla and Phyllonycteris. Anoura, Choero- beus as sister taxa. The third group included niscus, Hylonycteris, and Leptonycteris were two ``tentatively'' placed taxa, Chrotopterus more closely related to Glossophaga and and Trachops, which were united with Vam- Monophyllus than to brachyphyllines. Final- pyrum. Chrotopterus and Vampyrum ap- ly, Baker et al. (1981a: 671) found that Lio- peared as sister taxa. Based on all the avail- nycteris and Lonchophylla were ``the most able data, Straney (1980) concluded that the divergent of all the glossophagine and bra- relationships of these three taxa, Chrotopte- chyphylline genera examined.'' However, the rus, Trachops, and Vampyrum, were most relationship between these two genera was likely to re¯ect phylogenetic relationships. described as ``tenuous,'' and Baker et al. Finally, Straney (1980) conducted an al- (1981a) were not able to determine whether lozyme analysis, using two methods, out- Lionycteris and Lonchophylla were derived group comparisons and component analysis, or primitive relative to other glossophagines. to produce trees. Although Straney (1980) Grif®ths (1982) developed a hypothesis of did ®nd several trees using the outgroup relationships among nectar-feeding phyllos- 24 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 7. A. Grif®ths' (1982; redrawn from ®g. 33) hypothesis of relationships among nectar-feeding phyllostomids based on hyoid and lingual data and additional consideration of karyological and dental data. The open base of the cladogram signifys the possible relationship of the lonchophylline clade to other nonnectar-feeding phyllostomids. Asterisks indicate taxa whose placement was based soley on craniodental data. B. Haiduk and Baker's (1982; redrawn from ®g. 8) tree depicting relationships among nectar-feeding phyllostomids based on G-banded chromosome morphology. C. Haiduk and Baker's (1982; redrawn from ®g. 9) reanalysis of Grif®ths' (1982) original data, without consideration of char- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 25 tomids based on a cladistic analysis of gines. Within Glossophaginae, Grif®ths tongue and hyoid morphology, and some (1982) found that Hylonycteris, Anoura, and consideration of craniodental morphology Leptonycteris appeared to be successive sis- and karyotypes (®g. 7A; see table 2). As ter taxa to Choeroniscus and Choeronycteris. many had before him, Grif®ths (1982) pro- The most basal glossophagine clade consist- posed that glossophagines were not mono- ed of Glossophaga and Monophyllus. Based phyletic. However, Grif®ths (1982) ignited on characters of the teeth and basicranial re- controversy by recognizing a new subfamily gion only, Grif®ths (1982) placed Musonyc- (Lonchophyllinae) for Lionycteris, Loncho- teris with Choeronycteris and Scleronycteris phylla, and Platalina (see table 2), and by with Hylonycteris. Grif®ths (1982) placed Li- suggesting that Lonchophyllinae might be chonycteris as the sister taxon of the Glos- more closely related to nonnectar-feeding sophaga-Monophyllus clade based on kar- phyllostomids than to phyllonycterines, glos- yological and dental evidence. sophagines (s.s.), or brachyphyllines. Win- Grif®ths' (1982) conclusions drew criti- kelmann (1971) and Thomas (1903) had pre- cism from several researchers. Haiduk and viously suggested a close relationship among Baker (1982) performed a cladistic analysis these three genera. Winkelmann (1971: 90) of G-banded chromosome morphology in indicated that Lionycteris, Lonchophylla, and nectar-feeding phyllostomids, and their re- Platalina formed a ``natural evolutionary sults did not support lonchophylline mono- unit,'' and suggested that this group was not phyly. In the tree based on G-band data (®g. closely related to other nectar feeders. In- 7B), the positions of Brachyphylla, Erophyl- stead Winkelmann (1971) suggested that la, Phyllonycteris, Glossophaga, Leptonyc- each group of nectar feeders had arisen from teris, and Monophyllus were unresolved. Hy- phyllostomine-like ancestors. lonycteris, Choeroniscus, Lionycteris, Lon- In Grif®ths' (1982) tree (®g. 7A), Lio- chophylla, and Anoura appeared as succes- nycteris and Lonchophylla were sister taxa sive sister taxa to Choeronycteris and and formed a clade with Platalina. The lon- Musonycteris. chophylline clade was not united with the Although lonchophyllines did not appear rest of the cladogram to signify the possible to be monophyletic based on these results, close relationship of lonchophyllines to other Haiduk and Baker (1982) noted that standard nonnectar-feeding phyllostomids. Brachy- karyotypes of Lonchophylla robusta ap- phylla appeared as the sister taxon of the peared very similar to the standard karyotype phyllonycterine-glossophagine clade, al- of Lionycteris spurelli. Unfortunately, G- though Grif®ths (1982) considered this banded karyotypes of this species were un- placement tentative because he could identify available, and Lonchophylla thomasi, which only one potential synapomorphy uniting this has a highly derived karyotype, was used. If group. the G-banded karyotype of L. robusta is Subsequently, Grif®ths (1985) changed his more similar to that of Lionycteris, Haiduk assessment of the relationship of Brachy- and Baker (1982) suggested that it would in- phylla to Phyllonycterinae when he described dicate that these two genera had shared a several derived dental features of a brachy- common ancestor, with L. thomasi subse- phylline (s.l.) clade. Grif®ths (1985) further quently becoming more derived. Thus, lon- suggested that this group shared a common chophyllines would be monophyletic. ancestor with Glossophaginae. However, in Haiduk and Baker (1982) also presented Grif®ths' (1982) tree, Phyllonycterinae (s.s.) the results of an unweighted parsimony anal- appeared as the sister taxon of glossopha- ysis based on the original coding of Grif®ths'

← acter weights employed by Grif®ths (1982) or karyological and dental data. D. Grif®ths' (1983; redrawn from ®g. 1) response to Haiduk and Baker (1982) was a cladogram based only on hyoid and lingual morphology. E. Smith and Hood's (1984; redrawn from ®g. 3B) reanalysis of Grif®ths' (1982) original data using an all zero outgroup and no character weights. 26 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

(1982) data. In this tree (®g. 7C), Brachy- Smith and Hood (1984: 457) discussed ob- phylla appeared as the sister taxon of the re- servational inaccuracies and methodological maining nectar feeders. Erophylla, Phyllo- ¯aws in Grif®ths' (1982, 1983a) studies and nycteris, and glossophagines (s.l.) formed a presented several trees based on Grif®ths' polytomy. Within Glossophaginae (s.l.), the (1982) original coding of the data. Their positions of Glossophaga, Lichonycteris, and most ``diagnostically ef®cient'' tree (®g. 7E), Monophyllus were unresolved, but the re- which was constructed using a hypothetical maining taxa formed a clade. The latter (all zero) outgroup, depicted slightly differ- group was composed of a monophyletic Lon- ent relationships than those found by other chophyllinae, Anoura, Leptonycteris, and a authors (Smith and Hood, 1984: 456). Ero- clade consisting of Choeroniscus, Choero- phylla and Phyllonycteris appeared as sister nycteris, and Hylonycteris. taxa and formed a clade with all other nectar Grif®ths (1983a) responded to Haiduk and feeders. Brachyphylla was the sister taxon of Baker's (1982) critique with another clado- a clade composed of glossophagines and lon- gram (®g. 7D) based exclusively on hyoid chophyllines, which were each monophylet- and tongue data and constructed without the ic. The only additional difference between character weights used in his previous study this tree and that presented by Grif®ths (i.e., Grif®ths, 1982). This tree was less re- (1983a) is the recognition of a clade com- solved than that presented by Grif®ths posed of Glossophaga, Lichonycteris, and (1982), but was similar to the hyoid data tree Monophyllus. presented by Haiduk and Baker (1982; ®g. Hood and Smith (1982, 1983; ®g. 8) pre- 7C). However, there was one important ex- sented a cladistic analysis of phyllostomid ception: in Grif®ths' (1983a; ®g. 6D) new subfamilial relationships using characters of tree, Lonchophyllinae appeared as the sister the female reproductive tract. A monophy- taxon of Glossophaginae (s.s.), rather than letic Stenoderminae, Carollia, and Rhino- nesting within Glossophaginae as suggested phylla formed a clade. This group formed a by Haiduk and Baker (1982; ®g. 7C). Grif- polytomy with brachyphylline (s.l.) and glos- ®ths (1982) had previously rejected a sister sophagine (s.l.) genera. The genera of the group relationship between glossophagines ``Phyllostomus-group'' (Phylloderma, Phyl- and lonchophyllines because he felt that bra- lostomus) formed a polytomy with the nec- chyphyllines, phyllonycterines, and glosso- tar-feeder and fruit-feeder clade. Genera in phagines shared more derived dental traits, the ``Macrotus-group'' (Macrotus, Micro- and that some of the tongue and hyoid char- nycteris, and Trachops) formed a polytomy acters deserved more weight than others. with the nectar-feeder, fruit-feeder, and Phyl- Warner (1983) favored the original Grif- lostomus-group clade. Desmodus was the sis- ®ths (1982) hypothesis, viewing Haiduk and ter taxon to all other phyllostomids. Baker's (1982) G-banding study with skep- Koopman (1984) presented a classi®cation ticism due to problems with identifying ho- of Phyllostomidae in which he recognized mologous chromosome arms. Warner (1983) six subfamilies: Phyllostominae, Glosso- suggested that Grif®ths' (1982) tree was sup- phaginae, Carolliinae, Stenoderminae, Bra- ported by immunological data presented by chyphyllinae (including Brachyphylla and Baker et al. (1981a, see above). Although phyllonycterines), and Desmodontinae. Baker et al. (1981a: 671) had noted that Pierson (1986; ®g. 9) used transferrin im- Lionycteris and Lonchophylla were the most munology to investigate relationships among immunologically divergent among nectar phyllostomids. Pierson (1986) found that feeders, they also noted that ``the proposition Phyllostominae was not monophyletic be- that two or more glossophagine lineages cause Brachyphylla and the stenodermatines arose independently from a non-nectar-feed- Artibeus and Sturnira appeared to be derived ing stock is highly suspect.'' Baker et al. from within phyllostomines. Other results in- (1981a) also added that they could not de- cluded: (1) Mimon and Tonatia were related termine if Lionycteris and Lonchophylla had to Phyllostomus; (2) Chrotopterus, Trachops, chromosomes that were derived or primitive and Vampyrum formed a clade; (3) Brachy- relative to those of other glossophagines. phylla and Glossophaga formed a monophy- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 27

Fig. 9. Pierson's (1986; redrawn after ®g. 35) proposed phylogeny of phyllostomid relationships based on transferrin immunology. This tree does not include all the taxa used by Pierson (1986); see the text for discussion of additional relationships. Asterisks indicate taxa for which antisera were unavailable.

lution in desmodontines or equal yet independent divergence in albumins from those of other phyllostomids might have caused this unusual association. However, they found no conclusive evidence to support either possibility. Honeycutt (1981) and Honeycutt and Sar- ich (1987a) also recognized a Phyllostomus- Tonatia group and a Micronycteris-Vampyr- um group. The Micronycteris-Vampyrum Fig. 8. Hood and Smith's (1982; redrawn group appeared as the sister taxon of a clade from ®g. 5) hypothesis of higher-level relation- that included the Phyllostomus-Tonatia ships among phyllostomid bats based on mor- group, Phyllonycteris, Brachyphylla, Glos- phology of the female reproductive tract. sophaga, Monophyllus, Carollia. Using re- ciprocal comparisons between Artibeus anti- letic group; (4) Artibeus, Sturnira, and possibly the Brachyphylla-Glossophaga clade were allied with the Vampyrum lineage. In some of the analyses, Macrotus formed the most basal phyllostomid branch, while in other analyses desmodontines were basal. Although Pierson (1986) found a relationship between Macrotus and Carollia, she noted that this ®nding was probably due to the conservative nature of the transferrin of Carol- lia. Honeycutt (1981) and Honeycutt and Sar- ich (1987a) used albumin immunology to investigate phyllostomid relationships (®g. 10). These authors found that Macrotus, which Fig. 10. Honeycutt's (1981; redrawn after ®g. 2) and Honeycutt and Sarich's (1987a; redrawn was the most immunologically divergent after ®g. 1) tree based on bidirectional immuno- phyllostomine, associated with desmodonti- logical comparisons of albumins. The authors ®t nes to form a basal phyllostomid lineage. other taxa into this tree using unidirectional com- Honeycutt (1981) and Honeycutt and Sarich parisons; placement of these taxa is described in (1987a) postulated that a slower rate of evo- the text. 28 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 sera and groups of antisera representing major lineages, these authors found that Arti- beus associated with Carollia. Unidirectional comparisons were used to ®t other taxa into this tree. Chrotopterus associated with the Micronycteris-Vampyrum group. Lonchorhi- na, Macrophyllum, and Mimon associated with the Phyllostomus-Tonatia group, but Honeycutt (1981) and Honeycutt and Sarich (1987a) noted that the association between Lonchorhina and the Phyllostomus-Tonatia clade was no closer than that seen between Carollia and this clade. Honeycutt (1981) found that Phylloderma associated with Phyllostomus. Honeycutt's (1981) and Honeycutt and Sarich's (1987a) results indicated that the nectar-feeding taxa formed a clade. However, these authors found that Glossophaginae was Fig. 11. Consensus tree from Honeycutt and paraphyletic because Glossophaga and Mon- Sarich (1987a; redrawn from ®g. 3) based on immunological, chromosomal, and morphological ophyllus were successive sister taxa to a data. clade containing Brachyphylla and Phyllo- nycteris. Honeycutt (1981) carried out unidirectional immunological comparisons with Brachyphylla, Phyllonycteris, Glossophaga, Diaemus were sister taxa within Desmodon- and Monophyllus to place Erophylla, An- tinae. Honeycutt and Sarich (1987a) also as- oura, Hylonycteris, Leptonycteris, Lionycter- sessed congruence of their results with mor- is, and Lonchophylla in the tree. Erophylla phological data (from Grif®ths, 1982; Hood associated with Phyllonycteris. Leptonycteris and Smith, 1982; and Smith and Hood, 1984) appeared closer to the Glossophaga lineage, and karyological data (from Patton and Bak- whereas Anoura and Choeroniscus associat- er, 1978; and Baker et al., 1979). A consen- ed with the Monophyllus lineage. Hylonyc- sus tree based on these three data sets (®g. teris could be placed at the base of either the 11) suggested to Honeycutt and Sarich Glossophaga or Monophyllus lineages. Ho- (1987a) that a major revision of Phyllostom- neycutt (1981) could not assess the relation- idae was necessary. In the consensus tree, ships of Lionycteris and Lonchophylla be- one clade was composed of a monophyletic cause it appeared that these two genera rep- group of nectar-feeding taxa and the fruit- resented independent lineages that had feeding clade of Carollia and Artibeus. With- branched off prior to the brachyphylline- in the nectar-feeding clade, Brachyphylla, glossophagine common ancestor. Phyllonycteris, and Glossophaga formed a Honeycutt's (1981) and Honeycutt and polytomy; Monophyllus appeared as the sis- Sarich's (1987a) results indicated that Car- ter taxon of this group. The nectar- and fruit- ollia and Artibeus formed a clade that was feeder clade was the sister taxon of a group the sister group of the Phyllostomus-Tonatia of phyllostomines (Lonchorhina, Macrophyl- clade. Honeycutt's (1981) results also indi- lum, Mimon, Phyllostomus, and Tonatia; see cated that Carolliinae might not be mono- ®g. 11). This large clade and Macrotus, Mi- phyletic, since Rhinophylla associated with cronycteris, and another phyllostomine clade either Carollia or Micronycteris. However, (Chrotopterus, Trachops, and Vampyrum) Honeycutt (1981) used only unidirectional formed a polytomy. Desmodontines occupied comparisons and could not resolve this ques- the basal branch within the family. tion. Finally, Honeycutt (1981) and Honey- Owen (1987) conducted a parsimony anal- cutt et al. (1981) reported that Desmodus and ysis of relationships of stenodermatine bats 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 29 using discrete and continuous craniodental the Ardops-Phyllops clade, became the sister and external characters. Owen's (1987) ``best taxon of Centurio and Pygoderma. The Ar- phylogenetic hypothesis'' (®g. 12A) indicat- dops and Centurio clades were sister taxa in ed that Artibeus (s.l.) was not monophyletic. Owen's (1991) revised tree. Owen (1987) suggested that small-bodied Lim (1993) recoded and reanalyzed the Artibeus species should be placed in a sep- discrete-state craniodental and external char- arate genus, Dermanura. He also suggested acters from Owen's (1987, 1991) data sets, that, if Vampyressa is monophyletic, Meso- and added some new characters. The ``work- phylla should be considered congeneric with ing hypothesis'' (®g. 12C) proposed by Lim it. The basal node of Owen's (1987) tree was (1993: 158) included two stenodermatine a polytomy that included two clades of Stur- clades. The ®rst clade included all ``short- nira and two larger clades (®g. 12A). The faced'' genera. Pygoderma appeared as the ®rst of these latter groups consisted of a sister taxon of all other ``short-faced'' sten- clade of ``short-faced'' stenodermatines, odermatines. Ametrida, Centurio, and Sphae- Dermanura concolor (ϭ Koopmania; Owen, ronycteris formed a clade, with the latter two 1991), a clade of all remaining Dermanura genera as sister taxa. Ardops and Phyllops species, and Enchisthenes. Within the ``short- were successive sister groups to a clade in- faced'' stenodermatine group, the positions cluding Ariteus and Stenoderma. In the sec- of Centurio, Pygoderma, and Stenoderma ond stenodermatine group, two pairs of sister were unresolved, but Ametrida and Sphae- taxa formed a clade: (1) Ectophylla and Me- ronycteris formed a clade, as did Ardops, Ar- sophylla, and (2) Chiroderma and Vampyr- iteus, and Phyllops. Ardops and Ariteus were essa. Successive sister taxa to this large clade sister taxa. Within the second large clade, were the sister genera Platyrrhinus and Vam- Owen (1987) recognized ®ve branches form- pyrodes, a clade consisting of Artibeus (s.l) ing a polytomy: (1) Vampyressa melissa; (2) and Uroderma, and Sturnira. V. bidens, V. brocki, and V. pusilla; (3) V. In Lim's (1993) analyses all members of nymphaea and Mesophylla; (4) Chiroderma, the genus Artibeus were scored identically, Vampyrodes, and Vampyrops (ϭ Platyrrhin- and although he found no unique synapo- us); and (5) Ectophylla, a clade of all large morphy uniting this group, trees in which Artibeus, and Uroderma. Dermanura was separated from Artibeus Subsequently, Owen (1991) used selected were two steps longer. Synonymy of Meso- discrete-state characters from his previous phylla with Vampyressa was also shown to study in a parsimony analysis that included be unwarranted, as these genera did not ap- all species of Dermanura and all ``short- pear to be sister taxa in Lim's (1993) tree. faced'' stenodermatines. Owen's (1991) Lim (1993) also conducted a strict parsi- ``proposed phylogeny'' (®g. 12B) was more mony analysis of Owen's (1991) data set that resolved than his previous hypothesis (Owen, used Owen's (1991) original character cod- 1987; ®g. 12A). Dermanura concolor, Der- ing. The resulting tree (®g. 12D) was very manura hartii, and a clade composed of the different from Owen's (1991) analysis of the remaining Dermanura species were succes- same data (®g. 12B). In Lim's (1993) tree sive sister taxa to the ``short-faced'' steno- (®g. 12D), Artibeus (s.s.) and Dermanura dermatines. Based on this outcome, Owen were sister taxa, further weakening Owen's (1991) recommended recognizing Enchis- (1987, 1991) contention that Dermanura thenes as a genus separate from Artibeus, and should be recognized as a genus distinct from proposed that Dermanura concolor should (and not particularly closely related to) Ar- be placed in a new genus, Koopmania. Owen tibeus. Although Enchisthenes and Koop- (1991) did not make nomenclatural recom- mania did not form a clade with Artibeus and mendations for the paraphyletic Phyllops, Dermanura in Lim's (1993) reanalysis, Lim which included Ardops in his tree. Steno- (1993: 152) suggested that Owen's (1991) re- derma was the sister taxon of Ametrida and sults should be viewed with caution because Sphaeronycteris in this tree, and the positions of Lim's ``concerns about the discrete-state of Centurio and Pygoderma were resolved character set,'' rampant homoplasy, and vi- (®g. 12B). Ariteus, previously a member of olation of the assumption of ingroup mono- 30 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 12. A. Hypothesis of stenodermatine generic relationships proposed by Owen (1987) based on his study of continuous and discrete craniodental and external characters (redrawn from Owen, 1987: ®g. 17). B. Result of Owen's (1991; redrawn from ®g. 1) study designed to clarify relationships among Dermanura, Enchisthenes, and Koopmania, which was based on a reanalysis of selected characters from Owen (1987). C. Lim's (1993; redrawn from ®g. 6) ``working hypothesis'' of stenodermatine relationships using discrete craniodental and external characters. Note that Artibeus includes Dermanura, En- chisthenes, and Koopmania. D. Lim's (1993; redrawn after ®g. 3) reanalysis of Owen's (1991) data set. 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 31 phyly (i.e., Artibeus, an outgroup according subsequently mapped his restriction site to Owen [1991], grouped with Dermanura). characters on the cladogram proposed by Baker et al. (1989) investigated higher-lev- Baker et al. (1989; see ®g. 13A) to resolve el phyllostomid relationships and produced a relationships among genera. Results of this consensus tree (®g. 13A) that incorporated analysis indicated that Micronycteris was al- previous work in immunology (Baker et al., lied with Vampyrinae. Within Phyllostomini, 1981a; Honeycutt et al, 1981; Arnold et al., Macrophyllum appeared as the sister taxon of 1982; Honeycutt and Sarich, 1987a; Honey- the remaining genera. Within Stenodermati- cutt and Sarich, 1987b), karyology (Bass, ni, Artibeus, Chiroderma, and Sturnira 1978; Patton and Baker, 1978; Baker, 1979; formed a clade that was the sister taxon of a Baker and Bass, 1979; Baker et al., 1979, monophyletic group formed by the remain- 1981a, 1987, 1988a; Johnson, 1979; Haiduk ing stenodermatine genera, Carollia, and and Baker, 1982; Baker, unpublished data cit- Rhinophylla (®g. 13B). Although the posi- ed in Baker et al., 1989), and morphology tions of many stenodermatines within the lat- (Hood and Smith, 1982). The tree presented ter clade were unresolved, the restriction site by Baker et al. (1989; ®g. 13A) was similar data did diagnose two larger groups: (1) to the consensus tree constructed by Honey- Uroderma, Vampyrodes, and Vampyrops (ϭ cutt and Sarich (1987a; ®g. 11) although it Platyrrhinus); and (2) Ametrida, Ardops, Ar- was less resolved. Unlike Honeycutt and Sar- iteus, Centurio, Pygoderma, and Stenoder- ich (1987a), Baker et al. (1989) presented a ma. Van Den Bussche (1992) was unable to revised classi®cation based on their tree (see resolve relationships among three clades of table 2). Desmodontinae was the only tradi- nectar-feeding bats: (1) Brachyphyllinae tionally recognized subfamily that Baker et (s.l.), (2) Anoura, Glossophaga, Leptonycter- al. (1989) found to be monophyletic. Des- is, Lionycteris, Lonchophylla, and Mono- modontinae, Macrotus, Micronycteris, Vam- phyllus; and (3) Choeroniscus, Choeronyc- pyrinae (consisting of Chrotopterus, Trach- teris, Hylonycteris, and Musonycteris. Based ops, and Vampyrum), and a rede®ned ``Phyl- on the restriction site data, and consideration lostominae'' formed a basal polytomy in of immunological, karyological, and mor- their tree (®g. 13A). Phyllostominae included phological evidence, Van Den Bussche three tribes: Phyllostomini (the remaining (1992) proposed the recognition of Macroti- phyllostomine genera), Glossophagini (bra- nae and Micronycterinae as monogeneric chyphyllines, phyllonycterines, glossopha- subfamilies. gines, and lonchophyllines), and Stenoder- Van Den Bussche et al. (1993) examined matini (carolliines and stenodermatines). Re- 402 base pairs of the cytochrome b gene and lationships among phyllostomine tribes were an EcoRI-de®ned nuclear satellite DNA re- left unresolved. Baker et al. (1989) classi®ed peat. The satellite DNA repeat was found all genera in these higher-level groups as sed- only in Artibeus, Dermanura, and Koopman- is mutabilis (taxonomic position unclear). ia, suggesting that these three taxa formed a Van Den Bussche (1991) analyzed restric- clade. Cytochrome b data also supported this tion site data from the ribosomal DNA gene hypothesis. Enchisthenes was found to lack complex, but found that the few informative the satellite DNA repeats seen in the other characters and the large amount of homopla- three taxa, and in a bootstrap tree of the cy- sy prevented the construction of a robust hy- tochrome b data, Enchisthenes appeared as pothesis on the basis of these data alone. To the sister taxon of Centurio rather than Ar- circumvent these problems, Van Den Bus- tibeus (s.l.). sche (1991) tested previously proposed phy- Van Den Bussche et al. (1998) examined logenies (from Smith [1976] and Baker et al. the entire cytochrome b gene (1,140 bases) [1989]) for congruence with the restriction in 26 species of Artibeus (s.l.). This work site data. The topology of the Baker et al. supported the monophyly of a clade includ- (1989) tree explained the restriction site data ing Artibeus, Dermanura, and Koopmania. better than the Smith (1976) topology, re- Artibeus and Dermanura appeared as sister quiring seven fewer steps (Van Den Bussche, taxa. Given these relationships, Van Den 1991). Van Den Bussche (1992; ®g. 13B) Bussche et al. (1998) retained Koopmania in 32 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 13. A. Baker et al.'s (1989; redrawn after ®g. 2) consensus tree based on morphology, immunology, and karyology. B. Van Den Bussche's (1992; redrawn from ®g. 2) tree derived from mapping restriction sites onto the Baker et al. (1989) topology in an attempt to improve resolution of relationships within the clades identi®ed by Baker et al. (1989). 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 33

Fig. 14. Gimenez (1993) used the topology of Baker et al. (1989) to partition her analyses of lingual morphology. A. Cladogram of relationships from Gimenez's (1993; redrawn from ®g. 22) analysis of desmodontines, vampyrines, Macrotus, Micronycteris, and Phyllostominae. B. Gimenez's (1993; redrawn from ®g. 24) proposed relationships among Stenodermatini genera. C. Gimenez's (1993; redrawn from ®g. 25) proposed relationships among Glossophagini genera.

Artibeus. Van Den Bussche et al. (1998) also sition of Micronycteris was unresolved, but supported the recognition of Enchisthenes as Gimenez (1993) found that Macrotus a genus distinct from Artibeus (s.l.) based on grouped with Vampyrinae. Macrotus, Trach- the results of the cytochrome b analysis and ops, and the sister taxa Chrotopterus and the earlier study of the nuclear satellite DNA Vampyrum formed a polytomy. Desmodon- repeat (Van Den Bussche et al., 1993). tines, Micronycteris, the vampyrine group, Gimenez (1993) examined lingual mor- and Phyllostominae (sensu Baker et al., phology in phyllostomids and conducted sev- 1989) formed a basal polytomy. Although in- eral parsimony analyses designed to address cluded in her matrix, Gimenez (1993) did not intergeneric relationships within the higher- depict relationships among Phyllostomatini, level taxa identi®ed by Baker et al. (1989). Glossophagini, and Stenodermatini in her Gimenez's (1993) ®rst analysis (®g. 14A) in- tree. cluded desmodontine genera, Macrotus, Mi- Gimenez's (1993) second analysis exam- cronycteris, vampyrine genera, Phyllostomi- ined relationships within Phyllostomini (sen- ni, Glossophagini, Stenodermatini, and the su Baker et al., 1989). This analysis included outgroups Noctilionidae and Mormoopidae. Phyllostomini genera, Stenodermatini (a sin- The last ®ve taxa were scored as single gle OTU), and Glossophagini (a single OTUs in this analysis. Within desmodon- OTU). Desmodontinae (a single OTU) and tines, Desmodus and Diaemus formed a clade Vampyrinae (a single OTU) were used as with Diphylla as their sister group. The po- outgroups. Gimenez (1993) found that lin- 34 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 gual characters were largely uninformative Glossophaginae, Lonchophyllinae, and Sten- within Phyllostomatini. The three clades odermatinae. Koopman (1994) also recog- identi®ed consisted of Phylloderma and nized two taxa within Stenodermatinae: Stur- Phyllostomus, Mimon bennettii and Mimon nirini included Sturnira, while Stenoderma- crenulatum, and Glossophagini and Steno- tini encompassed the remaining stenoder- dermatini. matine genera. Within Stenodermatini, lingual characters Based on a study of cranial characteristics were more informative (®g. 14B). Stenoder- and cranial allometry, Solmsen (1994) rec- matini genera and the outgroups Micronyc- ognized four subfamilies of nectar-feeding teris, Phyllostomini (a single OTU) and phyllostomids: Phyllonycterinae, Brachy- Glossophagini (a single OTU) were used in phyllinae, Glossophaginae, and Lonchophyl- this analysis. Carolliinae and Stenodermati- linae. Within Glossophaginae, Solmsen nae were each monophyletic and appeared as (1994) named the tribe Choeronycterini for sister taxa. Sturnira occupied the basal sten- the genera Choeroniscus, Choeronycteris (in- odermatine branch. Gimenez (1993) identi- cluding Musonycteris), and Hylonycteris.In ®ed three clades among the remaining sten- addition, Solmsen (1994) concluded that odermatine genera: (1) the ``short-faced'' Lonchophyllinae was most closely related to taxa Ametrida and Pygoderma, (2) Dermanu- Carolliinae, while other glossophagine gen- ra, Koopmania, Platyrrhinus, and the sister era seemed perhaps to share a closer evolu- taxa Artibeus and Uroderma, and (3) Chi- tionary history (e.g., ®g. 84 in Solmsen, roderma, Vampyressa macconnelli (ϭ Me- 1994). Like Gimenez (1993), Solmsen sophylla), and Vampyressa. (1994) found that Platalina appeared to nest The ®nal analysis conducted by Gimenez within the genus Lonchophylla. (1993) focused on relationships among nec- Gimenez et al. (1996; see ®g. 15A) further tar-feeding taxa (®g. 14C). Micronycteris, evaluated relationships among the nectar- Phyllostomini (a single OTU), and Stenod- feeding phyllostomids. The tree derived from ermatini (a single OTU) were used as out- lingual morphology (Gimenez et al., 1996; groups in this analysis. Gimenez (1993) cor- ®g. 15A) was similar to that presented by rected some observational inaccuracies in Gimenez (1993), but two previously partly Grif®ths' (1982) descriptions, but achieved resolved clades, including Lonchophyllinae, results similar to his. Gimenez (1993) found were completely unresolved. When Gimenez that Brachyphylla appeared as the sister tax- et al. (1996) combined their lingual data with on of a clade consisting of all other nectar- Grif®ths' (1982) hyoid data, resolution with- feeders. Phyllonycterinae, Glossophaginae, in the Choeroniscus group improved (®g. and Lonchophyllinae were each monophy- 15B). Hylonycteris, Anoura, and Lichonyc- letic, and phyllonycterines and glossophagi- teris appeared as successive sister taxa to nes formed a clade with lonchophyllines as Choeroniscus and Choeronycteris. Relation- their sister group. Within Glossophaginae ships among other taxa were identical to (s.s.), the positions of Glossophaga, Licho- those in the tree based on lingual data only. nycteris, and Monophyllus were unresolved, Gimenez et al. (1996) also described a sub- but the remaining taxa formed a clade. With- optimal tree (one step longer) in which Glos- in the latter group, Anoura and Leptonycteris sophaginae and Lonchophyllinae were sister were successive sister taxa to the unresolved taxa. When Desmodontinae was included in grouping of Choeroniscus, Choeronycteris, these analyses, desmodontines and Brachy- and Hylonycteris. Gimenez (1993) found that phylla appeared as sister taxa. within Lonchophyllinae, Lionycteris was the In a classi®cation of mammalian species, sister taxon of a clade of Lonchophylla spe- McKenna and Bell (1997), dramatically re- cies. Platalina nested within the Loncho- vised phyllostomid classi®cation. These au- phylla species (®g. 14C). thors expanded the de®nition of Glossophag- Koopman (1993, 1994) presented a clas- inae, which they recognized as including four si®cation of phyllostomids that included sev- tribes: Brachyphyllini, Phyllonycterini, Glos- en subfamilies: Desmodontinae, Phyllostom- sophagini, and Lonchophyllini. Similarly, inae, Brachyphyllinae, Phyllonycterinae, Stenodermatinae was expanded to include 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 35

that gradually added stability to classi®cations as their improvements were consistently used by other taxonomists. Goldfuss (1820) ®rst divided Chiroptera into families, including a family Phyllostomata. However, it was Gervais (1854, 1856) who restricted the use of Phyllostomidae to New World leaf-nosed bats and applied names to the sub- divisions of the family which have been used fairly consistently since his classi®cation. Af- ter Miller's (1907) in¯uential classi®cation, which recognized hemidermines (ϭ carollines) and phyllonycterines for the ®rst time, phyllostomid taxonomy remained virtually unchanged until the 1960s. During this next period of change, mormoopids were recognized as a distinct family (Gerber, 1968; Ger- ber and Leone, 1971; Smith, 1972), desmodontines were recognized as phyllostomids (Machado-Allison, 1967; Forman et al., 1968; Gerber, 1968; Gerber and Leone, 1971), Brachyphylla was removed from Stenodermatinae and allied with nectar-feeders (Silva-Taboada and Pine, 1969; Baker and Lopez, 1970; Baker and Bass, 1979; Baker et al., 1981a), Sturnira was placed within Stenodermatinae (Baker, 1967; Ger- ber, 1968; Gerber and Leone, 1971), and a new subfamily, Lonchophyllinae, was recognized (Grif®ths, 1982). These slight changes in taxonomy failed to re¯ect the controversies surrounding the evolutionary Fig. 15. A. Gimenez et al.'s (1996; redrawn relationships of these bats. With the advent from ®g. 3) cladogram depicting relationships of cladistic methods, and research into many among nectar feeders based on lingual characters. aspects of phyllostomid biology, both mor- B. Gimenez et al.'s (1996; redrawn from ®g. 4) phological and molecular, systematists have cladogram depicting relationships among nectar begun to evaluate the data supporting tradi- feeders based on lingual and hyoid characters. tionally recognized groups. Studies based on diverse types of data have supported monophyly of a number of the tribes Carolliini and Stenodermatini. higher-level phyllostomid taxa (see table 3) Sturnira was placed in its own subtribe, Stur- including Desmodontinae, Vampyrinae, nirina, while the remaining Stenodermatini Phyllostominae (sensu Baker et al. 1989), genera were placed in the subtribe Steno- Phyllostomini, Stenodermatinae/Stenoder- dermatina. Desmodontinae and Phyllostomi- matini, Glossophagini, and Phyllonycterinae. nae were recognized in their traditional form However, several aspects of phyllostomid by these authors. phylogeny remain controversial. Studies based on immunological and morphological SUMMARY data have indicated that Phyllostominae (sen- As was the case with taxonomy in general, su Koopman, 1993, 1994) is not monophy- phyllostomid classi®cation was in a state of letic; however, there is no agreement on in- ¯ux during the 18th and 19th centuries. Sev- tergeneric relationships of phyllostomines eral authors made important contributions because results of different studies did not 36 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 support congruent groups. Relationships incomplete (lacking many taxa; e.g., Honey- among nectar-feeding genera have also prov- cutt and Sarich, 1987a) or not resolved at the en dif®cult to resolve. Results of immuno- generic level (e.g., Baker et al., 1989). These logical and karyological studies have sug- problems have apparently originated from gested that Lonchophyllinae might not be the use of taxonomic congruence (see dis- monophyletic, although analyses based on cussion of this method under Materials and hyoid and lingual morphology have support- Methods below). Because few taxa have ed this grouping. Data from studies of allo- been sampled for all relevant data sets, use zymes, behavior, immunology, morphology, of taxonomic congruence made full resolu- and parasitology support the placement of tion of intergeneric relationships impossible. Brachyphylla as the sister taxon of Phyllon- The studies presented by Honeycutt and Sar- ycterinae; however, hyoid and lingual data ich (1987a) and Baker et al.(1989) excluded are not congruent with this interpretation. craniodental, external, hyoid, lingual, and Monophyly of Carolliinae has been support- morphology from formal consideration dur- ed by some data sets (allozymes, lingual ing tree-building. No previous study has used morphology), but immunological studies a character congruence approach applied to con¯ict with these results. Finally, interge- all available discrete data in a broad taxo- neric relationships within most higher-level nomic sample. This approach, described in groups of phyllostomids (e.g., stenodermati- detail below, may help to resolve the branch- nes, glossophagines) remain problematic. ing pattern of genera within Phyllostomidae Previous attempts to reach a phylogenetic and permit evaluation of the monophyly of consensus based on consideration of multiple traditionally recognized subfamilies and oth- data sets have produced trees that were either er higher-level groups.

MATERIALS AND METHODS SAMPLING AND CHARACTER 1970, 1973, 1979; Gardner, 1977a; Patton CODING and Baker, 1978), but we used none of these characters in our analysis. Identi®cation of We scored representatives of all currently chromosomal synapomorphies has relied on recognized phyllostomid genera (see Koop- assumptions of ingroup monophyly and di- man, 1993) for which specimens were avail- rection of character change because most able for most external, craniodental, lingual, chromosomal characters can be understood and postcranial osteological characters listed only in the context of a reference taxon des- under Character Descriptions below. Alto- ignated as primitive. For example, it is only gether, we examined more than 2,300 speci- by comparing karyotypes of Micronycteris mens (see appendix 1 for specimens exam- nicefori to Macrotus waterhousi that we can ined). When available, we examined 10 in- say that M. nicefori has ``Rearrangements of dividuals of each representative species (®ve biarmed waterhousi chromsomes 1/2, 23/24, males and ®ve females) for each type of [and] 26/25 . . .'' (Patton and Baker, 1978: preparation used (e.g., skins). We used ®ve 453). This approach may fail if the reference or fewer specimens to score the tongue char- taxon has a derived morphology (see Dis- acters because of the destructive aspect of cussion: Interpretation of Chromosomal such dissections. We scored characters of the Data, below). Because of this problem, and brain, chromosomes, digestive and reproduc- our inability to resolve it satisfactorily, we tive tracts, hair microstructure, hyoid appa- did not consider autosomal characters when ratus, postcranial musculature, and restriction constructing trees. However, once a tree is sites based soley on published descriptions constructed using other data, chromosomal (see Character Descriptions below). evolution may be evaluated in the context of Numerous studies of phyllostomid rela- the new tree. We adopted this a posteriori tionships were based on morphology of the approach to investigate autosomal evolution. autosomal chromosomes (e.g., Baker, 1967, We included three genera from other chi- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 37

ropteran families as outgroups: Pteronotus (Simmons, 1998; Simmons and Geisler, and Mormoops (Mormoopidae), and Noctilio 1998). We assumed monophyly of each of (Noctilionidae). Previous analyses of chirop- these families based on previous work teran interfamilial relationships have indicat- (Smith, 1972; Simmons, 1998) rather than ed that these families and Phyllostomidae directly investigating relationships among form a well-supported clade, Noctilionoidea outgroup genera and phyllostomids, and con- 38 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 strained our analyses to re¯ect the most re- from which the skull had been previously recent hypotheses of relationships among these moved. We examined all tongues using a Ni- taxa: ((Noctilionidae, Mormoopidae), Phyl- kon microscope (ϫ0.66±ϫ4.0 plus doubling lostomidae). Accordingly, our data set in- lens). To facilitate viewing of papillae, we cludes only characters relevant to relation- dried tongues using canned air (e.g., Beseler ships among the ingroup taxa. Dust Gun) or paper towels. We illustrated We assumed generic monophyly for most tongue characters of selected specimens us- ingroup taxa. However, we assessed intra- ing a camera lucida. generic variation for some characters by ex- We evaluated craniodental morphology us- amining multiple species within a genus ing cleaned skeletons and skulls, and exam- when we found previous reports of polymor- ined external characters using both ¯uid-pre- phism, or when our own suspicions were served specimens and skins. We used a sim- raised. For some genera suspected of being ilar Nikon microscope in these examinations, nonmonophyletic, we sampled at the subge- and illustrated characters of selected speci- neric and/or species level. For Mimon,we mens using a camera lucida. We occasionally sampled the two monotypic subgenera: M. took measurements for some characters (e.g., (Anthorhina) crenulatum and M.(Mimon) character 87: length of calcar) using digital bennettii. We sampled representatives of four calipers; discussion of methods used in these of the ®ve subgenera of Micronycteris, in- cases is provided in the character descrip- cluding M.(Lampronycteris) brachyotis, and tions below. M.(Trinycteris) nicefori, which represent We identi®ed characters of potential phy- monotypic subgenera, and M.(Glyphonycter- logenetic signi®cance based on comparisons is) sylvestris, M.(Micronycteris) hirsuta, M. among ingroup taxa. We utilized what Wil- (Micronycteris) minuta, and M.(Micronyc- kinson (1995) has termed ``reductive,'' as teris) megalotis. We included representatives opposed to ``composite,'' character coding. of each of the three subgenera of Vampyres- Reductive character coding breaks apparent- sa: V.(Vampyriscus) bidens, which is mono- ly independent attributes (e.g., shape and col- typic; V.(Metavampyressa) nymphaeae; and or pattern of a structure) into separate char- V.(Vampyressa) pusilla. We separated Arti- acters. In contrast, composite character cod- beus (s.l.) into the four currently recognized ing combines such apparently independent subgenera: Enchisthenes and Koopmania, which are monotypic, and Artibeus and Der- features into single multistate characters. Re- manura. In our character discussions, we use ductive coding seems to have some bene®ts the species names for all of these taxa, except over composite coding. For example, Wilkin- Artibeus (s.l.) where we consistently use the son (1995) noted that composite characters subgeneric names. We do this for conve- may conceal homoplasy and constrain recon- nience because both Artibeus and Derma- struction of ancestral states. Unless properly nura are speciose and we often sampled ordered, composite characters may not be an- broadly within them. alytically equivalent to reductive characters To examine tongue morphology, we dis- and may result in different trees (Wilkinson, sected at least one individual of each taxon. 1995). The skin of the lower jaw was re¯ected over We also coded partly dependent characters the mandible and muscles with origins or in- in a hierarchical fashion where one character sertions on the mandible (e.g., m. mylohy- concerns the presence or absence of a trait, oideus, and m. genioglossus) were cut, al- and subsequent dependent characters de- lowing the tongue to be separated from the scribe features of this trait (see Simmons, mandible and re¯ected ventrally. This tech- 1993). For example, degree of development nique allowed us to view structures on the of a hypocone on M3 is dependent on the proximal tongue, which are dif®cult to see in presence of M3. Using hierarchical coding, most intact specimens. Unless necessary, ad- taxa are ®rst scored for the presence or ab- ditional specimens were not dissected; in- sence of M3. Taxa possessing an M3 are then stead, we used individuals that were pre- scored for the size of the hypocone in a sec- served with their mouths open, or specimens ond character. Taxa lacking an M3 are scored 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 39

``Ϫ'' for the second character, indicating that ASSESSING CONGRUENCE AMONG this character does not apply to those taxa. DATA SETS Unfortunately, computer algorithms do not distinguish between an inapplicable state Systematists have used three approaches to (``Ϫ'') and missing data (``?''). Thus, using assess congruence among data sets: taxo- hierarchical characters may result in impos- nomic congruence, character congruence, sible state reconstructions at internal nodes and conditional combination. Taxonomic and for terminal taxa (Platnick et al., 1991), congruence involves separate analysis of data in the rejection of trees that would be more sets based on different character systems (or equally) parsimonious (Maddison, 1993), (e.g., molecular and morphological), and or in arbitrary resolutions (Wilkinson, 1995). compares the results using consensus indices It is possible to construct characters to handle or consensus trees. Character congruence, or hierarchical attributes in several alternative ``total evidence'' as it is often called, com- ways; however, some of these methods re- bines all available data sets in a single anal- quire that the absence of the character be ysis. Conditional combination uses statistical scored more than once (Maddison, 1993; tests for heterogeneity to determine if data Pleijel, 1995). If absence of the complex trait sets should be combined in a character con- is derived, this gives more weight to a single gruence analysis. evolutionary event by counting it multiple Hillis (1987) suggested that taxonomic times, which can bias searches for the most and character congruence may be used to parsimonious trees. Another option is to use achieve different goals. Taxonomic congru- composite character coding (Maddison, ence, which results in more conservative and 1993), which has other undesirable qualities stable topologies, might best be used to con- (see above). Both methods affect character struct classi®cations, while character congru- state reconstructions. Although imperfect, we ence, which results in more fully resolved prefer hierarchical coding because it best re- but perhaps less stable hypotheses, might be ¯ects our views on the logical independence used for character interpretation where the of attributes (see above) and allows us to use of poorly resolved trees is problematic. construct characters equivalent to our hy- Although stability is not necessarily a goal potheses of homology, without requiring that of systematics (Kluge, 1989; Kluge and characters be weighted or ordered. Wolf, 1993), classi®catory recommendations We constructed unique character states to made on the basis of weakly supported to- accommodate intraspeci®c variation (poly- pologies are clearly more likely to be over- morphism within a species). For example, if turned with the addition of new data. How- species A includes some individuals with ever, taxonomic congruence is not the only yellow pelage (state 0) and others with way to address the issue of stability. Boot- brown (state 1) pelage, we erected state 2, strapping and decay indices may be used to which we described as ``yellow or brown investigate support for clades produced in a pelage; polymorphic within species.'' We character congruence analysis, allowing clas- used combinations of two or more character si®catory recommendations to be based on states to indicate taxonomic polymorphism in only the most strongly supported clades. our composite terminal taxa. For example, if Lanyon (1993) and Miyamoto and Fitch two species within the same genus exhibit (1995) have argued that data sets should be different states (species A with state 0 and analyzed separately because each data set species B with state 1), we scored the genus provides an independent test of the others. 0/1 in the matrix. In addition to characters As Miyamoto and Fitch (1995: 69) noted: scored as inapplicable (``Ϫ''), we used ``?'' to indicate relevant data that had not been Independence is more likely for characters from dif- collected for a particular taxon. We recorded ferent data sets than for characters from the same set character states using MacClade version 3.0 (Swofford, 1991; Lanyon, 1993). As a result, characters from different data sets are less likely to sup- (Maddison and Maddison, 1992), which we port the same, and perhaps wrong, species phylogeny also used to examine character transforma- than are those from the same data sets (de Quieroz, tions and draft ®gures. 1993). 40 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

These authors have advocated use of consen- some parts of the tree and not others, com- sus indices or trees to compare results of in- bining data may provide more resolution. dependent analyses. Combining data may allow a weak phylo- Barrett et al. (1991) defended the use of genetic signal to overwhelm noise that ob- taxonomic congruence in the following cas- scured this signal when the data are analyzed es: to determine how much two data sets dis- separately (Barrett et al., 1991; de Quieroz, agree with each other, how results might dif- 1993). fer when applying different methods of anal- Recently, some authors (e.g., Bull et al., ysis (e.g., phenetic vs. cladistic), and how re- 1993; de Quieroz, 1993; Huelsenbeck et al. sults for different data sets compare if the 1996) have suggested that it may be inappro- data cannot be combined (e.g., discrete char- priate to simply combine entire data sets that acters and distance data; also noted by Lan- strongly support con¯icting tree topologies yon, 1993; Eernisse and Kluge, 1993). because these con¯icts may indicate that In contrast, Kluge (1989) has advocated there are problems with either the data sets character congruence, and has advanced sev- or method of analysis. This approach has eral criticisms of taxonomic congruence. been termed ``conditional combination.'' The Kluge (1989) and Kluge and Wolf (1993) chief concern of these authors (Bull et al., noted that choice of consensus technique is 1993; de Quieroz, 1993; Rodrigo et al., 1993; arbitrary, and that different consensus tech- de Quieroz et al., 1995; Huelsenbeck et al., niques and indices may give different results. 1996) seems to be that a data set may be Kluge (1989), Eernisse and Kluge (1993), ``positively misleading'' in that it always re- Kluge and Wolf (1993), and Simmons (1993) covers an incorrect phylogeny when a suf®- also criticized the notion that data sets can ciently large number of characters have been be partitioned into subsets that re¯ect ``real'' sampled (see Felsenstein, 1978). By combin- divisions. ing a positively misleading data set with oth- Although data sets are typically equally er data sets not af¯icted with this problem, weighted in taxonomic congruence, the char- the entire analysis is compromised. As Bull acters in each data set are unequally weight- et al. (1993: 385) noted ``it is better to obtain ed when the sizes of the data sets are differ- one right answer and one wrong answer from ent (Miyamoto, 1985). Miyamoto (1985) separate analyses than to get a single wrong also criticized taxonomic congruence be- answer in a combined analysis.'' cause it compares the topologies without ref- Other problems that may lead to disagree- erence to the strength of character data sup- ment of two or more data sets include inves- porting them, resulting in trees that are less tigator error (e.g., in establishing character parsimonious than those taking all evidence homology), computational error (e.g., failure into account. Lanyon (1993) suggested that to ®nd the shortest tree; Cracraft and Min- only the most strongly supported clades be dell, 1989; Rodrigo et al., 1993), sampling used to build taxonomic congruence consen- error (Bull et al., 1993; Rodrigo et al., 1993; sus trees. However, these trees will still be de Quieroz et al., 1995), violation of the as- less parsimonious than those taking all evi- sumptions of tree-building methods (e.g., dence into account. rate of character change, or character inde- Combining data seems to afford several pendence; Hillis, 1987; Bull et al., 1993; bene®ts. Kluge (1989) and Barrett et al. Rodrigo et al., 1993; de Quieroz et al., 1995), (1991) have argued that any hypothesis must and data sets that do not share the same his- account for all evidence, and as Miyamoto tory (e.g., horizontal transmission; Bull et al., (1985) noted, trees produced in a character 1993; Rodrigo et al., 1993; de Quieroz et al., congruence analysis more parsimoniously 1995). explain all the data. Furthermore, character Several new statistical tests are being used con¯icts are resolved in favor of the stron- to examine heterogeneity among data sets gest evidence when data sets are combined (e.g., Rodrigo et al., 1993; Farris et al., 1995) (Hillis, 1987; Kluge and Wolf, 1993). Hillis and provide information useful for deciding (1987) and de Quieroz (1993) noted that be- when data sets should be combined. In these cause some data sets are informative in only tests, data may be combined if the disagree- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 41 ment among the trees proves to be what one linked and gene products from one paritition would expect from sampling error (Bull et do not regulate gene expression in other par- al., 1993; de Quieroz et al., 1995; Huelsen- titions); however, our limited understanding beck et al., 1996). But what happens when of these types of genetic interactions make the null hypothesis (homogeneity) is reject- such conditions almost impossible to satisfy ed? at the present time. Furthermore, we may Although in practice it is not yet possible fail to correctly identify ``real'' partitions, to identify the cause of disagreements among and tests for heterogeneity will indicate that data sets, de Quieroz et al. (1995) advocated data are homogeneous when they are not this approach. These authors suggested that (Huelsenbeck et al., 1996). This appears to when different stochastic processes (e.g., dif- be a particular problem for morphological ferent rates of change) are involved, com- data sets, where the parameters governing bined data sets may produce an incorrect so- the evolution of particular characters are vir- lution (i.e., the data sets are positively mis- tually unknown. leading; Felsenstein, 1978; de Quieroz, We feel that all taxonomic congruence 1993). In a maximum parsimony analysis, rests on an arti®cial division of evidence the solution to this problem may be differ- and has other associated problems (see ential character weighting (also proposed by above). Therefore, we use taxonomic con- Barrett et al. [1991], Bull et al. [1993], and gruence as a tool for identifying clades in Chippendale and Wiens [1994] for some or which these character sets seem to be most all situations when heterogenous data are informative and providing comparisons with combined). When the true tree topology is previous work. We use character congruence different for each data set (i.e., different his- for phylogenetic reconstruction and classi- tories), de Quieroz et al. (1995) and Bull et ®cation recommendations. al. (1993) argued that data sets should not be combined and de Quieroz et al. (1995) ad- PHYLOGENETIC ANALYSES vocated removing offending taxa or data sets from a combined analysis. Huelsenbeck et al. We conducted phylogenetic analyses us- (1996) suggested that possible remedies ing PAUP* (Swofford, 1998: version 4.0b1) might include using taxonomic congruence on a Macintosh PowerBook 1400. To ex- (previously proposed as a solution when two amine patterns of taxonomic congruence, data sets strongly disagree by de Quieroz, we ran separate analyses of all data parti- 1993), reanalyzing the data using a different tions. We chose three of our 11 partitions phylogenetic model, or simply leaving the re- (pelage and integument; skull and dentition; lationships among the group in question un- postcranium) based on traditional anatomi- resolved. cal divisions. The remaining eight partitions Clearly, examining individual data sets correspond to character sets examined in for disagreement seems to be warranted be- previously published papers that focused on fore combining data in a character congru- speci®c anatomical systems (hyoid appara- ence analysis given the concerns of the au- tus; tongue; digestive tract; brain; reproduc- thors advocating conditional combination. tive tract), types of chromosomes (sex chro- However, we are left with perhaps the most mosomes), mitochondrial DNA (restriction basic problem: how do we identify ``real'' sites) and nuclear DNA (EcoRI-de®ned data partitions? Although Bull et al. (1993) DNA repeat). Dividing the characters into have championed ``process partitions'' as partitions in this way facilitates comparison ``real'' classes of data because partitions of our results with the results of previous evolve according to different rules (e.g., studies. In other analyses, we combined all ®rst, second, and third codon positions), the 150 characters in a character congruence best examples of characters that might rep- analysis after testing the paritions for het- resent process partitions are molecular. Mi- erogenity. yamoto and Fitch (1995: 66) suggested ®ve We could not use exact search methods other requirements that process partitions due to the size of the data matrix; instead, might meet (e.g., genes in a partition are not we used 1000 iterations of the heuristic 42 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 search algorithm with a random taxon addi- places ambiguous character state changes tion sequence, although we modi®ed this higher in the topology thereby favoring par- procedure when we found more than 30,000 allel evolution of the feature under investi- most parsimonious trees (our MaxTrees set- gation. In the bootstrap analyses, we per- ting), a common occurrence when we ana- formed 100 replicates. For each bootstrap lyzed the separate partitions. In these cases, replicate, we conducted 5 iterations with a we ®rst attempted to discover the shortest random taxon addition sequence, saving only tree. We saved 30,000 most parsimonious 10,000 trees. In the decay analyses, we used trees from a single replicate that used a ran- the heuristic search option with a random ad- dom taxon addition sequence. Then we ran dition sequence and 1000 repetitions to two additional searches using 1000 iterations search for suboptimal trees. We performed of the heuristic search algorithm with a ran- these searches in one-step increments (i.e., if dom taxon addition sequence, but saved no the shortest trees were of length X, our ®rst more than 1000 trees that were longer than decay analysis saved trees of length X and or equal to the length we discovered in our X ϩ 1; the second analysis saved trees of ®rst search. In this manner, we attempted to length X, X ϩ 1, and X ϩ 2, etc.). We con- sample different islands of most parsimoni- structed strict consensus trees based on each ous trees. We compared the results of the analysis, and compared these to determine three searches using strict consensus trees. the number of additional steps required to When we ran searches for the data partitions, collapse each clade that was present in the we eliminated those taxa for which no data shortest trees (those of length X). For groups were available (see table 4). with a decay value of more than 2, we used We used TBR branch swapping in most monophyly constraint trees and searched for analyses, collapsed branches with a mini- the shortest trees incompatible with the con- mum length of zero, and interpreted multi- straint tree. We performed only 10 iterations state taxa as polymorphism (see Swofford of these searches since the goal was not to and Begle, 1993 for discussion of these ®nd all the shortest trees, but simply to ®nd methods). We gave all characters equal one short tree that was incompatible with the weight in all analyses. We did not order char- constraint tree. acters in any analyses. We use the following institutional abbre- We used PAUP to construct consensus viations in this paper: AMNH, American trees, calculate summary tree statistics, and Museum of Natural History, New York; perform bootstrap and decay analyses. We FMNH, Field Museum of Natural History, used both accelerated transformation Chicago, Illinois; MVZ, Museum of Verte- (ACCTRAN) and delayed transformation brate Zoology, Berkley, California; UMMZ, (DELTRAN) to optimize characters on trees. University of Michigan Museum of Zoology, ACCTRAN forces state changes to a lower Ann Arbor, Michigan; USNM, National Mu- level (i.e., closer to the root) in the topology, seum of Natural History, Smithsonian Insti- favoring reversals. In contrast, DELTRAN tution, Washington, D.C.

CHARACTER DESCRIPTIONS We use the classi®cation of Koopman Our study would never have been possible (1993) throughout the following sections to without these prior contributions. Neverthe- facilitate discussion of the taxonomic distri- less, we often encountered dif®culty in com- bution of various features. In this context, paring observations from the literature with use of traditional subfamilial names (e.g., our own work. To simplify such compari- ``phyllostomines'') is for convenience only sons for future researchers, we include de- and does not imply that these taxa are nec- tailed discussion of the differences in ob- essarily monophyletic. servation, interpretation, and character con- For each character, we provide a sum- struction that exist between previous studies mary of previous use in systematic studies. and our own. 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 43 44 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 45 46 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

PELAGE AND INTEGUMENT length and density (Benedict, 1957) and the This section summarizes characters of the over hairs are apparently absent. In contrast, pelage, vibrissae, integument, and features of the pelage is clearly differentiated into ®ne, related structures including chin papillae, wavy under hairs and thicker, straighter, lon- noseleaf, and pinnae. Many of these charac- ger over hairs in Desmodus, Diaemus, Phyl- ters are described for the ®rst time, but some lostomus, Phylloderma, Brachyphylla, phyl- are based on features originally described by lonycterines, Lionycteris, Ardops, and Stur- Straney (1980: character numbers from ap- nira (Benedict 1957). The pelage is also dif- pendix 1), Owen (1987: character numbers ferentiated into under hairs and over hairs in from appendix 23; 1991: character numbers mormoopids and Noctilio (Benedict, 1957), from the appendix), Lim (1993: character suggesting that this is the primitive condition numbers from appendix 2), Marques-Aguiar for phyllostomids. (1994: characters from appendix 4), and Sim- Benedict (1957) described differentiation mons (1996). We have modi®ed their char- of the pelage and Straney (1980: character acter descriptions in many cases, and have F2) ®rst used this character in a component scored these characters for all taxa included analysis. Our character states are equivalent in our analysis when possible (¯uid speci- to Straney's (1980) characters; however, we mens of Musonycteris and Scleronycteris disagree with his characterization of desmo- were unavailable; see table 4 for taxa includ- dontines and brachyphyllines (s.l.). In his ed in previous analyses). Characters related data matrix, Straney (1980) scored desmo- to hair structure (characters 1 to 4 below) are dontines ``ϩ'', indicating that all desmodon- based on Benedict's (1957) descriptions (see tines share the derived condition of differ- table 4); we did not collect any new obser- entiated pelage; however, Diphylla does not vational data for these four characters. have differentiated pelage. Our scoring re- Character 1: Pelage differentiated into ¯ects Benedict's (1957) original descriptions. over hair and under hair (0); or pelage uni- In addition, Straney (1980) scored brachy- form, over hairs apparently absent (1). In phyllines ``ϩϩ'', indicating that not all mem- most phyllostomids, the pelage is of uniform bers of the taxon have the derived state (pelage differentiated into under hairs and over 3 Character state numbers in Owen's (1987) appendix hairs). However, according to Benedict 2 do not correspond with those in the discrete character (1957), Brachyphylla, Erophylla, and Phyl- matrix he presented on pages 8±9. In appendix 2, the lonycteris all have pelage that is differenti- ®rst assigned character state (presumably the primitive ated into under and over hairs. condition in most cases) is 1, whereas in the matrix the We based our conclusion that the over ®rst character state assigned is 0, except for the ®rst hairs are absent in bats with undifferentiated three characters where the character states in the matrix pelage on a comparison of cuticular scale and those from appendix 2 agree. To avoid confusion, types. In most bats with distinct over and un- we choose to use the numbers in his matrix (i.e. begin- der hair, the alternate annular arrangement of ning with state 0) in our discussions because they correspond with our method of scoring. Although Owen hair scales (see character 3) does not occur (1987) presented multistate characters and ®gures de- on the over hair; thus, it may be that in phyl- scribing the manner in which these characters were or- lostomids in which over and under hair can- dered, he nevertheless appears to have recoded these not be distinguished, the over hair may be characters using additive binary coding prior to running absent. his analyses. These recoded characters were not pre- Character 2: Basal bulb on hair shaft ab- sented in his 1987 text, and we therefore direct our com- sent (0); or present (1). In most phyllosto- ments to the multistate characters presented in his ap- mids, the proximal third of the hair ®lament pendix 2. However if Owen's (1987) additive binary is of roughly uniform diameter and there is characters were similar to those he presented in 1991, no basal swelling, or bulb (Benedict, 1957). some of our criticisms of his characters might not be valid. For example, while Owen (1987) presented dental In contrast, a swollen basal bulb is present formulae in a multistate characterÐsee our comments on the proximal third of the hairs in Des- under character 48ÐOwen (1991) broke this character modus and Diaemus, some phyllostomines into several that described loss of individual teeth at spe- (e.g., Lonchorhina, Tonatia), and some sten- ci®c loci (i.e., loss of a lower premolar). odermatines (e.g., Ametrida, Vampyrodes; 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 47

Fig. 16. Magni®ed views (ϫ100) of plastic impressions of hair shafts in the outgroup taxon Mor- moops and two phyllostomids. The detail of the basal bulb is at 430ϫ magni®cation and represents a ``typical basal bulb'' (Benedict, 1957: plate 24l) of the kind found in Desmodus (the drawing is of Rhinolophus). Note the ``alternate annular'' arrangement of the scales in Lonchorhina (see character 3; drawn after Benedict, 1957: plates 24l, 29v, 30a, v).

Benedict, 1957; ®g. 16, detail). In mormoop- gination) on any given side of the ®lament, ids and Noctilio, the basal bulb is absent while the remaining scales appear to lack the (Benedict, 1957), suggesting that this is the V-shaped emargination (Benedict, 1957; see primitive condition for phyllostomids. character 4 below; ®g. 16). In contrast, the Straney (1980: character F1) ®rst used this scales are completely appressed to the ®la- character in a component analysis. Our char- ment in the desmodontines (®g. 16); there is acter states are identical to his, although our no difference in degree of divergence be- scoring of desmodontines is different. Ben- tween the sides of each scale (Benedict, edict (1957) described Diphylla as lacking 1957). In Mormoops, consecutive cuticular the basal bulb that is present in the other two scales have margins that are entirely diver- desmodontine genera. Although Straney gent from the shaft; there is no difference in (1980: 35) accurately described the taxonom- degree of divergence between the sides of ic distribution of this character in his text, he each scale (Benedict, 1957; ®g. 16). In Pter- scored desmodontines ``ϩ'' in his matrix, in- onotus and Noctilio, consecutive cuticular dicating that all desmodontines are derived scales are appressed to the shaft (Benedict, in having a basal bulb. Our scoring re¯ects 1957). This distribution of character states Benedict's (1957) description. suggests that the appressed scale pattern seen Character 3: Consecutive cuticular scales in Pteronotus and Noctilio may be the prim- appressed to hair shaft (0); or entire margin itive condition for phyllostomids. equally divergent (1); or consecutive scales Straney (1980: characters F3±F5) ®rst with most divergent margins on alternate used the divergence of the margin of the cu- sides (`àlternate annular'' condition) (2). In ticular scales as a character in a component most phyllostomids, each cuticular scale en- analysis, based on Benedict's (1957) descrip- circles the entire hair shaft and the bodies of tions. Straney (1980) de®ned three binary successive scales diverge most from the op- characters whose derived conditions are posite sides of the ®lament (Benedict, 1957). equivalent to our character states. However, This type of arrangement, known as `àlter- Straney (1980) scored taxa for the alternate nate annular,'' makes every other scale ap- annular character (F5) as well as one of the pear ``hastate'' (with a slight V-shaped emar- other two characters (F3 or F4, scales either 48 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 divergent or appressed). Benedict (1957) individual hairs (Benedict, 1957). We scored characterized the degree of divergence of al- this taxon with state 3 in the matrix. Phyl- ternate annular scales from the shaft along lonycteris aphylla has irregular scale margins what appears to be a continuum (appressed, while Phyllonycteris poeyi has entire scale slightly divergent, divergent, divaricate). Be- margins. We scored the genus Phyllonycteris cause of the continuous nature of this de- with states 0 and 1 in the matrix. In Mor- scription, and the fact that the alternate an- moops and Pteronotus quadridens, the scale nular condition describes a particular kind of margins are toothed (Benedict, 1957), where- divergence of the cuticular scales from the as Pteronotus parnellii has both entire and shaft of the hair, we did not include the ad- hastate scale margins on a single ®lament ditional information Straney (1980) used in (Benedict, 1957; note that this is not the al- his characters F3 and F4 in our version of ternate annular condition described in char- this character. acter 3 above). In Pteronotus gymnotus and Straney (1980) was not clear about which Noctilio the scale margins are entire (Bene- type of hair or which part of it he used to dict, 1957). We scored Pteronotus with states make his comparisons. Benedict (1957) not- 0, 2, and 4 in the matrix. The primitive con- ed that cuticular scale type changes over the dition for phyllostomids cannot be assessed length of the ®lament, and may be different a priori given the distribution of these states on over hair and under hair, when these can in the outgroup taxa. be identi®ed. Although Straney (1980) stated Benedict (1957) ®rst described the mor- that all phyllostomids have entire, irregular phology of the scale margins on bat hairs; emarginate, or denticulate scales alternating this character has not appeared in previous with hastate cuticular scales, this `àlternate phylogenetic analyses. Benedict (1957) de- annular'' condition occurs only on the mid- scribed considerably more variation than we dle part of the under hairs in Phylloderma have recognized. For example, Benedict (Benedict, 1957). On the over hairs and the (1957) described our `ìrregular'' scale mar- tip of the under hairs, the cuticular scales are gins (state 1) as being ``repand'' (slightly ir- sinuate to erose coronal. Benedict (1957: regular at some points along the margin), 287) noted that only those scale types in the ``sinuate'' (irregular along the entire scale midregion of the hair shaft are deemed to be margin with deeper notches), or `èrose'' ``mature'' cuticular scale types. For this rea- (very irregular, with some sides of the scale son we used her descriptions of the midre- being longer proximodistally than others). gion of all hairs, or the under hairs (when These scale types seem to illustrate states on these could be identi®ed; see character 1 a continuum. Thus, they may not represent above), to score this character and character 4. discrete character states. For this reason, we Character 4: Majority of scale margins decided to pool such states under the condi- on each hair entire (0); or irregular (1); or tion `ìrregular.'' Similarly, we included both toothed (2); or entire and irregular (3); or ``dentate'' and ``denticulate'' scale types entire and hastate (4). In most phyllosto- within the character state ``toothed'' (state 2). mids, cuticular scale margins are entire We also scored taxa in which some scale (scales with a smooth straight margin; Ben- margins occasionally differed from the most edict, 1957). However, in Desmodus, Diae- common margin type as having a single state mus, Phylloderma, Phyllostomus, and Ar- because it seemed likely that occasional de- dops, the scale margins are irregular (Bene- viations may have resulted from damage to dict, 1957; ®g. 16). In Lonchorhina, Macro- the scale margin (particularly because these phyllum, Anoura, Leptonycteris, Lionycteris, deviations were almost always `ìrregular'' in and many stenodermatines (e.g., Ariteus, nature). Platyrrhinus), the scales have toothed mar- Character 5: Dorsal fur unicolored (0); gins (Benedict, 1957; ®g. 16). Chrotopterus or distinctly bicolored, hairs with pale bases seems to be unique among phyllostomids in and dark tips (1); or tricolored, hairs with consistently possessing scales of two differ- distinct dark bases, a pale median band, and ent morphologies on a single ®lament: both dark tips (2). In most phyllostomids, the dor- entire and irregular shaped scales appear on sal fur is distinctly bicolored, with dark tips 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 49 and paler bases. Especially in phyllostomines re¯ects a condition that Ectophylla may po- with this feature, the pale bases of the hairs tentially share with other phyllostomids. Fi- tend to be most evident over the shoulders nally, Owen (1987) noted that Chiroderma and upper back and appear less distinct cau- improvisum has dorsal fur with a pale base dally. The white base may occupy more than and dark tip (in contrast to other Chiroderma one-half of the hair shaft (e.g., Macrotus)or species which have tricolored fur). We were be con®ned to a narrow stripe at the base of unable to examine this species; however, the fur (e.g., Lonchorhina). In contrast, Mi- Jones and Baker (1980: 1) described Chiro- cronycteris nicefori, M.sylvestris, Hylonyc- derma improvisum as having individual hairs teris, Lichonycteris, Carollia, and many sten- that are ``pale brownish gray throughout odermatines (e.g., Ametrida, Mesophylla) most of their length, slightly darker basally, have dorsal fur that is tricolored. A dark bas- and tipped with a dark, rich brown.'' A ju- al band is succeeded by a pale stripe and a venile individual examined by the same au- dark tip. Micronycteris brachyotis, Phyllos- thors had distinctly tricolored fur (Jones and tomus hastatus, Monophyllus, and Lionycter- Baker, 1980). We therefore score Chiroder- is are the only phyllostomids that have uni- ma as possessing tricolored fur. Owen's colored fur (dark from tip to base). Because (1991) observations are identical to those of other species of Phyllostomus have bicolored Owen (1987). Both Simmons (1996: char- fur, we scored this genus 0 and 1 in the ma- acter 1) and Marques-Aguiar (1994: charac- trix. The dorsal fur in Mormoops has a pale ter 4) employed a similar character in their base and darker tip, but in Pteronotus and analyses of Micronycteris and large-bodied Noctilio the dorsal fur is unicolored. This Artibeus, respectively; our observations distribution of character states leads us to agree with theirs. conclude that unicolored fur is primitive for Character 6: Facial fur of uniform color phyllostomids. Although we have not scored (0); or facial stripes present (1). In most Ectophylla for most of the characters related phyllostomids, facial fur is of uniform color to pelage color, we did score this taxon for and no facial stripes are present. In contrast, this character. In Ectophylla, the posterior many stenodermatines (e.g., Artibeus, Uro- half of the dorsum is a grayish-brown color, derma) have facial stripes. There are two and the individual hairs in this region have stripes on each side of the face in most taxa: pale bases and a darker tip. a more intensely colored dorsomedial stripe, Owen (1987: character 3) used an ordered and a lighter ventrolateral stripe. The dorso- character with states identical to ours, later medial stripe begins at the posterolateral modifying it slightly when he restricted his edge of the noseleaf and runs superior to the analysis to a subset of original taxa surveyed eye, generally terminating at some point (Owen, 1991: character 7). Our observations along the medial external surface of the pin- agree with those of Owen (1987) with a few na. The ventrolateral stripe begins along the exceptions. Owen (1987) described the dor- corner of the mouth and generally runs to- sal pelage of Platyrrhinus dorsalis, which we ward the tragus, ending at its base. Stripes observed to have a pale base and dark tip vary in length, width, and color intensity like other Platyrrhinus species, as tricolored. within and between genera. Although the Owen (1987) also scored Ectophylla as hav- presence of dorsomedial stripes is invariant ing unicolored fur. Ectophylla is unique in genera, ventrolateral stripes are not always among phyllostomids in having a white head present in all species of a genus. For exam- and shoulders and a brownish lower body. ple, Chiroderma trinitatum has two bright The hairs on the anterior of the dorsum are pairs of facial stripes, whereas Chiroderma white from base to tip, while the brownish villosum lacks the ventrolateral pair and has fur on the lower body has a brown tip and lightly colored bands in the region of the dor- pale base. We suggest that the presence of somedial stripes. Nevertheless, all species of `ùnicolored'' white fur on the anterior of the Chiroderma have at least dorsomedial dorsum of Ectophylla alba is an autapomor- stripes. Other taxa that exhibit similar vari- phy of this species, and that the hair on the ation in the presence of ventrolateral facial lower body that has a pale base and dark tip stripes are Artibeus, Dermanura, Platyrrhin- 50 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 us, Uroderma. Mormoopids and Noctilio 267193) did not reveal any facial striping. lack facial stripes, suggesting that this is the Frosted patches of fur are located on the primitive condition for phyllostomids. We forehead, face, neck, and shoulders of Me- scored Ectophylla ``Ϫ'' for this character sophylla, and some specimens of Koopmania (see below). (e.g., AMNH 266267) also have frosted Owen (1987: character 4) used facial patches between the eyes, posterior to the stripes as part of a complex multistate char- noseleaf. These frosted patches may have acter that he termed ``pelage pattern.'' This been interpreted by Lim (1993) as facial character incorporated facial striping, dorsal stripes, but we do not believe they are ho- striping (see character 7 below), shoulder mologous. Thus, we agree with Owen's patches (see character 8 below), and striping (1987, 1991) assessment that Mesophylla and on the wing (an autapomorphy of Centurio). Koopmania lack facial stripes. Owen (1987) ordered the character to pre- Lim (1993) scored Ectophylla as having serve potential homologies. In general (see facial and dorsal stripes and lacking a shoul- below) we agree with Owen's (1987, 1991) der spot, whereas Owen (1987) scored the scoring of facial striping; however, Owen's taxon as having uniformly colored pelage. (1987) description of character state 2 sug- We disagree with both scoring schemes. gests that all taxa with facial stripes have More than three quarters of the body of Ec- paired stripes, rather than the situation we tophylla is grayish white; only the lower described above. One of our disagreements back is brown. Lim (1993) may have scored with Owen's (1987) scoring of facial stripes Ectophylla as possessing facial stripes be- concerns Artibeus; Owen (1987) reported cause of a brown patch that extends from the that A. fuliginosus (ϭ A. obscurus) lacks fa- lateral part of the noseleaf over the eye, but cial stripes whereas we observed faint dor- does not reach the ear. This brown fur visu- somedial facial stripes in all A. obscurus we ally separates the continuous white of the examined. We also scored Ectophylla as head into two parts that correspond to the ``Ϫ'', whereas Owen (1987) regarded the areas in which the dorsomedial and ventro- pelage as uniformly colored. Owen (1991: lateral stripes appear in other taxa. However, character 8) considered facial stripes as a this color pattern is not strictly comparable separate character for an analysis of a subset to that seen in other phyllostomids with fa- of stenodermatine taxa; however, scoring did cial stripes, and we found it impossible to not differ from his earlier work. determine if stripes are unambiguously pres- Lim (1993: character 1) used facial strip- ent. Accordingly, we scored Ectophylla ``Ϫ'' ing in his analysis of stenodermatine rela- for this character. tionships. Although our descriptions of char- Character 7: Dorsal stripe absent (0); or acter states are identical to Lim's (1993), we always present (1); or sometimes present; do not agree with all his assessments con- polymorphic within species (2). A pale dorsal cerning the taxonomic distribution of facial stripe is present in the fur of one phyllosto- stripes. Lim (1993) noted that live specimens mine (Mimon crenulatum) and several sten- of Koopmania and Mesophylla have faint fa- odermatines (e.g., Platyrrhinus, Uroderma). cial stripes similar to those in Artibeus ob- In Micronycteris nicefori, a very pale dorsal scurus. We could distinguish no trace of fa- stripe is present in most individuals (e.g., cial stripes in Mesophylla, either in color AMNH 266019) whereas it appears to be ab- slides of live bats (AMNH 267281), or in sent in the fur of others (e.g., AMNH well-preserved alcohol and skin preparations 266015). We scored this species with state 2 (e.g., AMNH 267281, AMNH 267563). Al- in the matrix. Dorsal stripes usually begin at though our examinations of some color slides a point between the shoulder and the anterior of live Koopmania suggested medial facial margin of the medial pinnae, extend down stripes were present in this genus, we believe the center of the back, and always reach the that these ``stripes'' are the pale bases of the base of the tail. Length and color intensity of fur in this region, or frosted patches, because the dorsal stripe may vary among and within our examinations of skins of the bats ap- genera. For example, within Vampyressa, V. pearing in these photos (e.g., AMNH pusilla has no dorsal stripe, V. brocki has a 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 51 faint stripe present on the lower back to the dermatines (e.g., Ametrida, Sphaeronycteris) base of the tail, and V. bidens has a distinct have a small, clearly de®ned patch of white stripe present from a point between the pin- fur on the shoulder. No white spot is present nae to the base of the tail. Such differences in mormoopids and Noctilio, suggesting that also exist among species of Chiroderma and this is the primitive condition for phyllos- Platyrrhinus, although no species in the latter tomids. We scored Ectophylla ``Ϫ'' for this genus completely lacks a dorsal stripe. Be- character because the shoulders in this spe- cause some species of Chiroderma lack a cies are completely white (see discussion un- dorsal stripe, whereas in others this feature der character 5 above). is present, this genus is scored 0 and 1 in the Owen (1987: character 4; 1991: character matrix. As noted above (see character 5), the 9) included shoulder spots in his phyloge- fur of the lower back is brown in Ectophylla netic analyses. Our character states and scor- and would therefore be expected show evi- ing are identical to those of Owen (1991) and dence of dorsal striping (if present) since Lim (1993: character 2). dorsal stripes always reach the base of the Character 9: Fur on neck of uniform col- tail. There is no evidence of a dorsal stripe or, no white neck spot (0); or distinct white in Ectophylla. In mormoopids, dorsal stripes spot sometimes or always present on neck are absent; however, some individuals of (1). In most phyllostomids, the area of the both Noctilio species have a pale to strongly neck ventral and caudal to the lateral part of colored dorsal stripe. Noctilio is scored with the pinna is roughly the same color as the fur state 2 in the matrix. The primitive state for on the neck and shoulders, and no white spot phyllostomids cannot be assessed a priori. is present. In contrast, a white spot is present Owen (1987: character 4) originally used in Ametrida, Centurio, Phyllops, Pygoderma, this feature in his external character ``pelage Stenoderma, and Sphaeronycteris.InPygo- pattern,'' although he included several addi- derma, the spot appears to be absent in some tional states to accommodate variation in specimens (e.g., AMNH 234296). In Centu- stripe length. Because stripe length varies rio, the spot is present on the face mask (see among species within genera, we choose to character 38 below). In the outgroups, mor- omit this information from our study. Our moopids and Noctilio, no neck spot is pres- scoring of this character agrees with Owen's ent, suggesting that this is the primitive con- (1987) observations, with the exception of dition for phyllostomids. We scored Ecto- Ectophylla, which Owen (1987) scored as phylla, which has a completely white neck, possessing unicolored fur. ``Ϫ'' (see character 5 above). Character 8: Fur on shoulder of uniform This character has not been used in pre- color, no white shoulder spot (0); or distinct vious studies. We have chosen to treat poly- white shoulder spot present (1). In most morphism in this character conservatively phyllostomids, the shoulder region is gener- (i.e., the derived condition is sometimes or ally uniform in color and there is no distinct always present) because Peterson (1965: 8), white spot. Although some specimens may who was able to examine many more repre- appear to have pale patches on the shoulders, sentatives of Ametrida than we did, found these patches do not have well-de®ned edges that this spot is not always present in this and are either extensions of the lighter ven- species. Our observations indicate that the tral fur into the shoulder region (e.g., Chro- spot varies in intensity, leading us to suspect topterus AMNH 267131), or the pale bases that the spot may be polymorphic in more of bicolored fur that are visible when the fur taxa than we report here. is slightly parted (e.g., Tonatia silvicola Character 10: Uropatagium without AMNH 267108). In Sturnira, many adults of fringe of hair along trailing edge (0); or with both sexes have shoulder glands which pro- distinct fringe of hair along trailing edge (1). duce a waxy secretion that may stain the fur The phyllostomid uropatagium usually does in the shoulder region, producing what are not have a distinct fringe of hair that extends termed `èpaulettes'' (Davis et al., 1964), but beyond the trailing edge of membrane. In these patches are usually bright orange or contrast, a distinct fringe is present along brown, not white. In contrast, some steno- most of the caudal edge of the uropatagium 52 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 in desmodontines, Macrophyllum, Mimon mologous to the superciliary vibrissae of oth- crenulatum, Anoura, Leptonycteris, and er taxa. Superciliary vibrissae are absent in many stenodermatines (e.g., Ardops, Platyr- mormoopids, but present in Noctilio (®g. rhinus). In Rhinophylla ®scherae, Artibeus 17F); accordingly the primitive condition for hirsutus, Dermanura azteca, and D. tolteca, phyllostomids cannot be reconstructed a distinct fringes are present, whereas other priori. species in these genera lack a fringe. We Although this character has not been used scored these genera with states 0 and 1 in the in previous phylogenetic analyses, there is matrix. The uropatagium lacks a fringe in much evidence to suggest that homologies mormoopids and Noctilio, suggesting that may be drawn among vibrissal groups, and this is the primitive condition for phyllos- potentially among individual vibrissae. The tomids. locations of various groups of vibrissae on Lim (1993: character 13) used the pres- the face is highly conservative among mam- ence of a fringe on the uropatagium as a mals, and presumably homologous groups or character, erecting two states: `Èdge of uro- individual vibrissae may be identi®ed on the patagium naked or sparsely and irregularly basis of their relative position (Pocock, 1914; haired (0); edge of uropatagium densely and Brown, 1971; Wineski, 1985). Many and evenly haired (1).'' Only Platyrrhinus and perhaps all facial vibrissae have individual Vampyrodes were scored with the derived representations in both the spinal trigeminal condition. In contrast, we found dense and nucleus and the somatosensory cortex (Zuck- evenly haired fringes in many additional er and Welker, 1969; Waite, 1973a, 1973b, stenodermatine taxa. Woolsey, 1978), further supporting the idea Character 11: Superciliary vibrissae ab- that homologies may be drawn among indi- sent (0); or always present (1). Most phyl- vidual vibrissae and vibrissal clusters in dif- lostomids lack superciliary (ϭ supraorbital) ferent taxa, as we do here. vibrissae (®g. 17D±E), a group of one or Character 12: Genal vibrissae absent (0); more vibrissae that occurs above each eye in or one vibrissa present in each cluster (1); most mammalian groups (Pocock, 1914; or two genal vibrissae present in each cluster Brown, 1971; ®g. 17A). However, a single (2). Genal vibrissae occur on the sides of the superciliary vibrissa is present above each face ventral and/or posterior to the eye (Po- eye in Desmodus (®g. 17B), Diaemus, some cock, 1914; Brown, 1971; ®g. 17A). Genal phyllostomines (e.g., Micronycteris megalo- vibrissae are present in most phyllostomids tis, Vampyrum; ®g. 17C), and one stenoder- (®g. 17B±C, E). However, these vibrissae are matine (Centurio). Taxonomic polymor- absent in some phyllostomines (e.g., Macro- phism exists in two genera: Lonchorhina phyllum, Mimon), Platalina, and several (present in L. marinkellei, absent in other glossophagines (e.g., Glossophaga, Mono- species), and Tonatia (present in T. silvicola phyllus). In most phyllostomids with genal and T. schulzi, absent in other species). We vibrissae, two occur in each cluster. How- scored these genera with states 0 and 1 in the ever, Brachyphylla, Erophylla, most glosso- matrix. In one specimen of Chrotopterus phagines, some lonchophyllines, and Pygo- (AMNH 267443), the superciliary vibrissae derma have a single genal vibrissa on each are absent, but these vibrissae are present in cheek. Taxonomic polymorphism occurs in all other specimens of Chrotopterus we ex- Lonchorhina (absent in L. fernandezi, two in amined. We consider this an anomaly and L. aurita), Lonchophylla (®g. 17D; absent in score Chrotopterus with state 1 in the matrix. L. mordax and L. thomasi, two in L. robus- In some taxa (e.g., Micronycteris megalotis, ta), Anoura (one in A. caudifer and A. cul- Tonatia silvicola) the superciliary vibrissae trata, two in A. geoffroyi), Choeroniscus (ab- are anterior to their position in other species. sent in C. minor [ϭ C. intermedius], one in Rather than directly superior to the eye, they C. periosus), Carollia (absent in C. perspi- are anterior and superior to it. However, they cillata, one in C. subrufa), and Sturnira (ab- occur in the same position relative to the oth- sent in S. bidens, one in S. bogotensis, two er groups of facial vibrissae (®g. 17C). We in other species we examined). We scored therefore consider these vibrissae to be ho- these taxa with the appropriate combinations 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 53

Fig. 17. Lateral view of the head (left) and ventral view of the chin (right) of selected taxa illustrating the position and number of vibrissae found in clusters. A. Generalized mammal (after Brown, 1971: ®g. 6) B. Desmodus rotundus (AMNH 267503). C. Micronycteris megalotis (AMNH 267411). D. Lonchophylla thomasi (AMNH 266107). E. Uroderma bilobatum (AMNH 268564). F. Noctilio leporinus (AMNH 267408). Scale bar ϭ 10 mm. of character states in the matrix (e.g., Lon- gle vibrissa on each cheek and Noctilio has chorhina is scored 0 and 2). In one specimen two genal vibrissae in each cluster (®g. 17F). of Vampyrodes caraccioli (AMNH 239254), The primitive state for phyllostomids cannot we found three genal vibrissae present in be reconstructed a priori. each cluster, whereas all other individuals we This character has not appeared in previ- examined had two. We view this as an anom- ous phylogenetic analyses. aly as all other phyllostomids we examined Character 13: Interramal vibrissae al- had no more than two of these vibrissae; con- ways absent (0); or none or one interramal sequently, we score Vampyrodes with state 2 vibrissa present, polymorphic within species in the matrix. The genal vibrissae are absent (1); or one interramal vibrissa always pres- in Mormoops, whereas Pteronotus has a sin- ent (2); or one or two interramal vibrissae 54 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 present, polymorphic within species (3); or sent in D. cinerea, whereas D. anderseni has two interramal vibrissae always present (4); one and D. tolteca has two. We scored this or none or two interramal vibrissae present, genus with states 0, 2, and 4 in the matrix. polymorphic within species (5); or three in- Two interramal vibrissae are present in Mor- terramal vibrissae always present (6). Inter- moops and Noctilio (®g. 17F), whereas Pter- ramal vibrissae occur between the rami of the onotus has one. This distribution of character lower jaws well posterior to the mandibular states suggests that two interramal vibrissae symphysis in most mammals (Pocock, 1914; are primitive for phyllostomids. Brown, 1971; ®g. 17A). Two interramal vi- This character has not been used in pre- brissae are present in most phyllostomids vious phylogenetic analyses. (®g. 17B±C). However, interramal vibrissae Character 14: Vibrissae lateral to nose/ are apparently uniformly absent in some noseleaf arranged in two columns; medial phyllostomines (e.g., Chrotopterus, Vampyr- column with three or more vibrissae, lateral um) and many stenodermatines (e.g., Pygo- column with two vibrissae (0); or single col- derma, Uroderma; ®g. 17E). The three lon- umn with three or more vibrissae present, chophylline genera appear to be unique lateral column absent (1). In desmodontines, among phyllostomids in possessing three in- many phyllostomines (e.g., Chrotopterus, terramal vibrissae (®g. 17D). Choeronycteris Phylloderma), Brachyphylla, phyllonycteri- and Musonycteris have a single interramal nes, glossophagines, and lonchophyllines, vibrissa. Among other genera, there is a large the vibrissae lateral to the noseleaf are ar- amount of both taxonomic and intraspeci®c ranged in two columns: a medial column polymorphism. Several species exhibit intra- (nearest the noseleaf), consisting of three or speci®c variation: Lonchorhina aurita (ab- more vibrissae (see character 15), and a lat- sent or two), Macrophyllum macrophyllum eral column, composed of only two vibrissae (absent or one), Phyllostomus hastatus (ab- (®g. 17B±D). The two vibrissae in the lateral sent or one), Trachops cirrhosus (absent or column occur on the dorsum of the snout one interramal vibrissae), Choeroniscus in- more or less posterior to the ®rst and second termedius (absent or one), Chiroderma vil- vibrissae in the medial column (numbering losum (absent or one), and Koopmania con- from dorsal to ventral). The lateral column color (one or two), Platyrrhinus aurarius of vibrissae is absent in several phyllosto- (absent or one). Taxonomic polymorphism mines (e.g., Macrophyllum, Phyllostomus), occurs frequently; interramal vibrissae ap- carolliines, and stenodermatines (®gs. 17E, pear to be uniformly absent in Tonatia sau- 18A±C). The lateral column is present in rophila, Sturnira bidens, S. nana, and S. til- mormoopids and Noctilio (®gs. 17F, 18D), dae whereas the congeners of these species suggesting that this is the primitive condition have two interramal vibrissae. We scored for phyllostomids. these two genera with states 0 and 4 in the This character has not appeared in previ- matrix. In Phyllostomus discolor, Chiroder- ous phylogenetic analyses. ma trinitatum, Platyrrhinus dorsalis and P. Character 15: Vibrissae in column adja- helleri, the interramal vibrissae are absent cent to noseleaf/nose number two (0); or whereas other species in these genera are three (1); or more than 3 (2). In most phyl- polymorphic (absent or one vibrissa). We lostomids there are more than three vibrissae scored these three genera with states 0 and 1 in the column immediately adjacent to the in the matrix. In Lonchorhina, L. fernandezi noseleaf. These vibrissae are extremely long has two interramal vibrissae, whereas L. au- and well developed, with swollen bases (or rita is polymorphic (absent or two). We pads as described in character 16 below; ®g. scored this genus with states 4 and 5 in the 18A±B). In contrast, in Ametrida, Centurio, matrix. In Choeroniscus minor these vibris- and Sphaeronycteris there are only three of sae are either absent or a single vibrissa is these well-developed vibrissae surrounding present, whereas in C. periosus a single vi- the noseleaf (®g. 18C). In mormoopids and brissa is present. We scored the genus Cher- Noctilio, there are two vibrissae in this col- oniscus with states 1 and 2 in the matrix. In umn surrounding the nose (due to the mor- Dermanura the interramal vibrissae are ab- phology of the face in mormoopids these vi- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 55

Fig. 18. Close-up view of the lateral column(s) of vibrissae and vibrissal papillae in A. Phyllostomus hastatus (AMNH 202308). B. Rhinophylla pumilio (AMNH 267163). C. Sphaeronycteris toxophyllum (AMNH 194213). D. Noctilio leporinus (AMNH 267408). Scale bar ϭ 5 mm. brissae are posterolateral to the nose; ®g. brissal papillae (0); or vibrissal papillae 18D). The presence of two vibrissae in the connected, form padlike structure (1); or vi- column lateral to the noseleaf appears to be brissal papillae connected, form free skin primitive for phyllostomids. ¯ap (2); or vibrissal papillae separate, form This is the ®rst use of this character in a elongate, ¯eshy, cylindrical projections (3). phylogenetic analysis. In most phyllostomids, the papillae of the in- Character 16: Vibrissae surrounding the dividual vibrissae located immediately adja- nose/noseleaf arise from small, separate vi- cent to the noseleaf are connected with each 56 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 other, forming two padlike structures, one on well-de®ned cream patches present on prox- each side of the noseleaf (®g. 18A). These imal parts of the lateral lobes and edges of pads incorporate the vibrissal papillae in the horseshoe, center of horseshoe and spear medial and, when present, the lateral col- brown (2). In most phyllostomids the skin of umns described in character 14. In contrast, the noseleaf is a shade of pale to dark brown, the vibrissal papillae form ¯eshy skin ¯aps generally the same color as the lips and face. in a several taxa: Rhinophylla, Ametrida, Ar- Although the noseleaf appears lighter in col- dops, Ariteus, Phyllops, and Stenoderma (®g. or in some taxa (e.g., Chrotopterus), the 18B). In Centurio and Sphaeronycteris, the range of colors seen in different species and vibrissal papillae are not connected with each within populations of a single species varies other, and the papillae instead form elongate, to such a degree that they cannot be coded ¯eshy cylinders that protrude from the face as discrete character states. Mimon crenula- (®g. 18C). Chrotopterus, Tonatia, Trachops, tum presents an excellent example of the and Vampyrum also have separate vibrissal range of variation seen within species. The papillae, but although these are often swol- noseleaf in AMNH 267112 is a pale, creamy len, they are never elongate and cylindrical, pink color; however, other individuals from nor do they form pads. In mormoopids and the same locality have noseleaves that range Noctilio, the vibrissal papillae are small and from light brown around the edges of the do not form padlike, ¯aplike, or cylindrical spear with pale creamy pink in the center and structures (®g. 18D); this distribution of on the horseshoe (AMNH 267883) to entire- character states suggests that the presence of ly brown (AMNH 267888). We thus recog- small, separate vibrissal papillae is the prim- nize ``pale cream to dark brown'' as a single itive condition for phyllostomids. character state. Ectophylla and Mesophylla This is the ®rst use of this character in a are unique among phyllostomids in having a phylogenetic analysis. noseleaf that is yellow in life (this generally Character 17: Padlike or ¯aplike vibris- fades to a much paler yellow color in pre- sal papillae not in contact across dorsum of served specimens). Uroderma, Vampyressa snout (0); or pads touch, or are con¯uent bidens, V. nymphaea, V. pusilla, and Vam- across dorsum of snout (1). Most taxa with pyrodes exhibit yet another pattern, a consis- padlike or ¯aplike vibrissal papillae have two tently bicolored noseleaf. In these taxa the discrete, clearly separated, pads or ¯aps on central rib of the noseleaf is always brown, either side of the noseleaf. These structures but the lateral lobes and horseshoe edges are not in contact, nor are they continuous, have distinct patches of cream or yellow. In across the dorsum of the snout. In contrast, some species of Chiroderma (e.g., C. trini- in Brachyphylla, Erophylla, and Phyllonyc- tatum), Dermanura (e.g., D. cinereus), and teris the padlike structures meet, or are con- Platyrrhinus (e.g., P. helleri) the noseleaf is tinuous, posterior to the noseleaf. In these also distinctly bicolored, but congeners of three genera, the vibrissae that are generally these species have a brown or cream nose- more laterally placed in other phyllostomids leaf. Thus, these three genera are scored 0 due to the wider separation of these pads, are and 2 in the matrix. We scored mormoopids located somewhat more posteriorly. Mor- and Noctilio, which clearly lack a noseleaf, moopids and Noctilio do not have padlike or ``Ϫ'' for this character; thus, the primitive ¯aplike like vibrissal papillae; therefore, we state for phyllostomids cannot be assessed a scored these taxa ``Ϫ'' for this character. The priori. We also scored Centurio ``Ϫ'' for this primitive condition for phyllostomids cannot character (see below). be reconstructed a priori. Taxa scored 0 and This is the ®rst use of this character in a 3 in character 16 are also scored ``Ϫ'' for phylogenetic analysis. Although one might this character. imagine that state 2 might be correlated with This is the ®rst use of this character in a presence of strong facial striping (see char- phylogenetic analysis. acter 6), our observations suggest that this is Character 18: Noseleaf pale cream to not the case. In Artibeus lituratus, for ex- dark brown, not strongly bicolored (0); or ample, strong facial stripes are present but noseleaf yellow (1); or noseleaf bicolored, the noseleaf is uniformly brown. However, 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 57 we did not use Lim's (1993: character 9) than half the height of the horseshoe. We character for yellow coloration of the thumb scored mormoopids and Noctilio, which because it has an identical distribution with clearly lack a noseleaf, ``Ϫ'' for this char- the yellow noseleaf (found only in Ectophyl- acter; thus, the primitive condition for phylla and Mesophylla); this suggests that these lostomids cannot be determined a priori. We characters may not be independent. also scored Centurio ``Ϫ'' for this character. Although presence of a noseleaf is gener- Owen (1987: character 2, 1991: characters ally regarded as a diagnostic feature of Phyl- 3±5) and Marques-Aguiar (1994: character 9) lostomidae, a noseleaf with a well-de®ned both described the length of the noseleaf in horseshoe and spear may not be present in relation to its width, within stenodermatines all phyllostomids. Although some authors and Artibeus (Artibeus), respectively. We have indicated that a noseleaf is present in have rede®ned this character to account for Centurio (e.g., Hill and Smith, 1984; Lim, the variation we observed among all phyl- 1993), others have suggested that this taxon lostomids. lacks a true noseleaf (Miller, 1907; Paradiso, Character 20: Spear of noseleaf with 1967; Snow et al., 1980). We found that a pointed or rounded distal tip (0); or with U- spearlike structure is present between the shaped notch in distal tip (1). The spear of nostrils in Centurio, but it is not con¯uent the noseleaf has a pointed or rounded distal internarially with a horseshoe. Instead, this tip in most phyllostomids (®gs. 19B±C, structure forms the medial part of the upper 20A±C, 21A±C). In contrast, a U-shaped lip. The horseshoe portion of the noseleaf (if notch is present in the distal tip of the nose- present) surrounds only the lateral portions leaf in desmodontines and Brachyphylla (®g. of the nostrils. One way to settle the question 19A). Presence of this notch may be depen- of the homology of these structures might be dent upon truncation of the spear (see char- to examine ontogeny of the narial region, but acter 19 above), but these features are not this has not yet been attempted. Pending such fully correlated. Phyllonycterines, which a study, we chose to score Centurio ``Ϫ'' for have a truncated spear, lack a notch in the all characters related to the noseleaf. distal tip (®g. 20A). We scored mormoopids Character 19: Noseleaf spear long, great- and Noctilio, which clearly lack a noseleaf, er than twice the height of the horseshoe (0); ``Ϫ'' for this character; thus, the primitive or spear truncated, equal to or less than the condition for phyllostomids cannot be deter- height of the horseshoe (1). The phyllosto- mined a priori. We also scored Centurio ``Ϫ'' mid noseleaf is divided into two parts, the for this character. horseshoe and the spear. The horseshoe sur- This character has not appeared in previ- rounds the lateral and inferior aspects of the ous phylogenetic analyses. nostrils. The spear is a ¯eshy protuberance Character 21: Central rib absent (0); or that rises off the surface of the snout from rib restricted to proximal part of spear (1); the superior aspect of the nostrils. The spear, or rib extends to distal tip of spear (2). A which is continuous with the horseshoe both thick, ¯eshy central rib is a prominent feature internarially and laterally, is generally delin- of the noseleaf of many phyllostomids. How- eated in some way from the alae of the horse- ever, the central rib is absent and the noseleaf shoe. The spear of the noseleaf is long and is ¯at in desmodontines, Macrotus, Brachy- tapers to a point in most phyllostomids. In phylla, phyllonycterines, Platalina, and all terms of relative proportions, the spear (as glossophagines (®gs., 19A, 20A, 20C). When measured from the inferior margin of the the rib is present, it originates between the nostril to the tip of the spear) is greater than nostrils and extends distally into the spear. twice the height of the horseshoe (as mea- The rib extends into the proximal part of the sured from inferior to superior horseshoe spear but does not reach the tip in species of border; e.g., ®g. 19B). In contrast, the spear Micronycteris, Lionycteris, Lonchophylla, is much shorter (truncated) in desmodonti- carolliines, Sphaeronycteris, and Sturnira nes, Brachyphylla, and phyllonycterines (®gs., 20B, 21A). In these taxa, the rib is (®gs. 19A, 20A). In these taxa, relative well de®ned laterally but not distally, where height of the spear ranges from equal to less it loses thickness and grades into the proxi- 58 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 19. Anterior view of the noseleaf in A. Desmodus rotundus (AMNH 267503). B. Vampyrum spectrum (AMNH 202292). C. Phyllostomus hastatus (AMNH 233176). Scale bar ϭ 5 mm.

Fig. 20. Anterior view of the noseleaf in A. Erophylla sezekorni (AMNH 194202) B. Lonchophylla robusta (AMNH 267452) C. Choeroniscus intermedius (AMNH 266122). Scale bar ϭ 5 mm. 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 59

Fig. 21. Anterior view of the noseleaf in A. Carollia perspicillata (AMNH 266144) B. Uroderma bilobatum (AMNH 268564) C. Ariteus ¯avescens (AMNH 214944). Scale bar ϭ 2 mm. mal surface of the spear. In contrast, the rib Platalina). In contrast, a well-delineated, extends to the tip of the spear in most sten- narrow, ¯eshy ridge or line of papillae is odermatines and phyllostomines (®gs. 19B± consistently present along the midsagittal C, 21B±C). We scored mormoopids and Noc- line of the internarial region in some phyl- tilio, which clearly lack a noseleaf, ``Ϫ'' for lostomines (e.g., Lonchorhina, Vampyrum), this character; thus, the primitive condition Lionycteris, Lonchophylla, glossophagines, for phyllostomids cannot be determined a and Carollia (®gs. 19B, 20B±C, 21A). In one priori. We also scored Centurio ``Ϫ'' for this phyllostomine species that we examined (Mi- character. cronycteris hirsuta), two out of 10 specimens This is the ®rst use of this character in a exhibited an internarial ridge; consequently, phylogenetic analysis. we score this taxon with state 2 in the matrix. Character 22: Internarial region smooth, We did not detect any other within-species no midsagittal ridge or papillae (0); or nar- polymorphism in any other taxon. We scored row ¯eshy ridge or line of papillae always mormoopids and Noctilio, which clearly lack present along midsagittal line (1); or inter- a noseleaf, ``Ϫ'' for this character; thus, the narial ridge or papillae variably present, primitive condition for phyllostomids cannot polymorphic within species (2). A variety of be determined a priori. We also scored Cen- structures occur between the nostrils in phyl- turio ``Ϫ'' for this character. lostomids. The central rib of the spear con- Presence/absence of an internarial ridge or tinues into the internarial region in all taxa line of papillae is not correlated with pres- that have a rib (see character 21 above; ®gs., ence/absence of the central rib; both are pres- 19B±C, 20B, 21A±C). In most, the rib is the ent in some taxa (e.g., Vampyrum) and ab- only structure present between the nostrils sent in others (e.g., Platalina), whereas only and it presents a smooth, slightly convex sur- a rib is present in some forms (e.g., Phyllos- face in this region. The internarial region is tomus) and only an internarial ridge or line generally smooth (although ¯atter) in taxa of papillae is present in others (e.g., Glos- that lack a rib on the spear (desmodontines, sophaga). When both rib and ridge are pre- Macrotus, brachyphyllines, phyllonycterines, sent, the former is distinguished by its great- 60 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 er width and continuity with the rib of the on the edge of the horseshoe (®g. 22B). Be- spear; the internarial ridge is located on the cause the horseshoe is completely bounded rib between the nostrils in these taxa (e.g., inferiorly by an extensive ¯ap of skin in Vampyrum). Chrotopterus and Vampyrum (see characters The form of the internarial structures 25 and 26), the sella appears to occur in the varies among taxa in which they are present middle of the horseshoe; however, it main- (taxa scored 1 and 2 for this character). For tains the same position relative to the nostrils example, Chrotopterus and Vampyrum al- as seen in Lonchorhina. We scored mor- ways have a narrow, undivided internarial moopids and Noctilio, which clearly lack a ridge, whereas a line of distinct papillae is noseleaf, ``Ϫ'' for this character; thus, the present in the same location in other taxa primitive condition for phyllostomids cannot (e.g., Macrotus, Trachops). Although we ini- be determined a priori. We also scored Cen- tially considered these to represent distinct turio and those phyllostomids that are scored conditions, further examination of glosso- with state 0 of character 22 (no internarial phagines and lonchophyllines led us to dis- ridge or line of papillae) with ``Ϫ'' for this cover a series of intermediates between the character. ``ridge'' and ``papillae'' conditions. Many Straney (1980: character G15) described glossophagines have a ridge that is subdivid- the morphology of the noseleaf in Loncho- ed by small grooves (e.g., Glossophaga), rhina and Macrophyllum as having ``lateral whereas in others the internarial structure and medial projections on basal plate.'' How- more closely resembles a series of partly ever, our observations of this region in Mac- fused papillae (e.g., Lionycteris). The latter rophyllum indicate that most individuals do condition is also seen in some phyllosto- not have a trilobed structure (e.g., AMNH mines (e.g., Micronycteris hirsuta). Mor- 222040). Therefore we do not consider dif- phology of internarial structures also appears ferences in sella morphology as a separate to vary within species. For example, within character because we observed the trilobed Lonchophylla robusta, some individuals sella in only one taxon. (e.g., AMNH 235760) have a solid internar- Character 24: Lateral edges of horseshoe ial ridge, whereas others (e.g., AMNH thin and free (0); or superior portion of 235761) have an internarial ridge that con- swollen edge of horseshoe forms free, ¯ap- sists of segments (three in this case) that aplike edge (1); or swollen lateral edges of pear to be elongate papillae. This type of var- horseshoe ridgelike, fused to face along en- iation leads us to conclude that the ridge and tire length with no free edge (2). The lateral line of papillae represent endpoints of a con- edges of the horseshoe consist of thin, free tinuum. As a result, we have not scored ¯aps of skin in most phyllostomids (®gs. ``ridge'' and ``papillae'' as separate character 19A±C, 20A, 21 A±C). In contrast, in some states. glossophagines (e.g., Choeronycteris, Hylo- This character has not been used by other nycteris) only the superior part of the lateral authors. edge of the horseshoe forms a free, thickened Character 23: Sella absent (0); or present ¯ap, while the inferior part of the horseshoe (1). The sella (sensu Hernandez and Cadena, is fused to the face (®g. 20C). In loncho- 1978) is a ¯eshy globular or trilobed struc- phyllines and some glossophagines (e.g., An- ture that occurs on the horseshoe just inferior oura, Glossophaga), both the superior and (or anterior) to the proximal end of the in- inferior parts of the horseshoe are fused to ternarial ridge or line of papillae, with which the face and there is no free edge (®g. 20B). it is continuous. Most phyllostomids lack a In some species of Tonatia (e.g., T. silvico- sella; it is present only in Chrotopterus, Lon- la), both the superior and inferior portions of chorhina, and Vampyrum.InChrotopterus the horseshoe are fused to the face, whereas and Vampyrum (®g. 22A), the sella is a sim- in other species (e.g., T. saurophila) the en- ple spherical structure. In contrast, in Lon- tire lateral edge of the horseshoe is a thin, chorhina the sella is a complex, trilobed free ¯ap. Consequently, we score Tonatia structure with one anterior and two lateral with states 0 and 2 in the matrix. We scored lobes. It is located just inferior to the nostrils mormoopids and Noctilio, which clearly lack 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 61

Fig. 22. Anterodorsal view of the sella in A. Vampyrum spectrum (AMNH 202292) and B. Lon- chorhina aurita (AMNH 199218). Note the position of this structure relative to the nostrils in both taxa. Scale bar ϭ 5 mm. a noseleaf, ``Ϫ'' for this character; thus, the 19A, 20A±C, 21A). There is no clear bound- primitive condition for phyllostomids cannot ary between the horseshoe and the upper lip be determined a priori. We also scored Cen- in these taxa. We scored mormoopids and turio ``Ϫ'' for this character. Noctilio, which clearly lack a noseleaf, ``Ϫ'' This is the ®rst use of this character in a for this character; thus, the primitive condi- phylogenetic analysis. tion for phyllostomids cannot be determined Character 25: Inferior border of horse- a priori. We also scored Centurio ``Ϫ'' for shoe is thin, free ¯ap of skin (0); or inferior this character. horseshoe is thickened ridge with no free Straney (1980: characters G16±17) de- edge (1); or inferior horseshoe grades scribed two alternative conditions for the la- smoothly into upper lip, no distinct boundary bial part of the horseshoe: fused to the upper between lip and horseshoe (2). The inferior lip (G16) or free (G17). We disagree with border of the horseshoe is de®ned by a thin, Straney's (1980) assessments of the labial free ¯ap of skin in many phyllostomines horseshoe in Mimon bennettii, Phylloderma, (e.g., Phyllostomus, Vampyrum) and most glossophagines, and carolliines. Straney re- stenodermatines (e.g., Ametrida, Uroderma; ported that in all of these taxa the ventral ®gs. 19B -C, 21B). In other phyllostomines margin of the horseshoe is a free ¯ap, contra (e.g., Micronycteris megalotis, Mimon benet- our observations. Simmons (1996: character tii) and stenodermatines (e.g., Ardops, Py- 7) also used a character similar to this one, goderma), the inferior border of horseshoe is although she did not consider our state 0 be- formed by thickened ridge with no free edge cause it did not occur among members of the (®g. 21C). In contrast, the inferior border of genus Micronycteris. Our scoring of this the horseshoe grades smoothly into the upper character in Micronycteris species is identi- lip in desmodontines, some phyllostomines cal to hers. Marques-Aguiar (1994: character (e.g., Phylloderma, Tonatia, Trachops), 10) also used this character, describing three Brachyphylla, phyllonycterines, glossophag- conditions of the labial horseshoe (free, ines, lonchophyllines, carolliines, Enchisthe- bound down and rimmed, and bound down nes, Sphaeronycteris, and Sturnira (®gs. and not rimmed), which she ordered 0 ↔ 1 62 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

↔ 2. Although we agree with her observa- moopids and Noctilio, which lack a noseleaf, tions concerning most species, we found that ``Ϫ'' for this character. The primitive con- the conditions she described for both Arti- dition for phyllostomids thus cannot be de- beus ®mbriatus (bound down and rimmed) termined a priori. We scored Centurio and and Artibeus amplus (bound down and not those phyllostomids that do not have a thick- rimmed) appeared very similar to the con- ened labial border (states 0 and 2 for char- ditions we observed in all other species of acter 25) ``Ϫ'' for this character. Artibeus (s.s.; ¯aplike: character state 0); This character has not been used in pre- thus we do not score Artibeus as polymor- vious phylogenetic analyses. phic for this character. Character 28: Ridge of skin absent on Character 26: Thin, free edge of inferior dorsum of snout (0); or present (1). No spe- horseshoe lies ¯at over upper lip (0); or cial structures occur behind the spear of the forms upright cup around nostrils (1). noseleaf in most phyllostomids (although see Among those phyllostomids that have a characters 17 and 29). However, in desmo- horseshoe with a thin, free inferior edge dontines a simple ridge of skin runs across (most phyllostomines and stenodermatines; the dorsum of the snout just posterior to the see character 25), this ¯ap lies ¯at over the spear. The ridge extends between the supe- upper lip (®gs. 19C, 21B). In contrast, the rior margins of the lines of vibrissae de- ¯ap forms an upright cup around the nostrils scribed in character 14, but is not associated in Chrotopterus and Vampyrum (®g. 19B). with any vibrissae. In Desmodus and Diae- We scored mormoopids and Noctilio, which mus, this ridge of skin is connected to the lack a noseleaf, ``Ϫ'' for this character. The posterior part of the noseleaf by two thin primitive condition for phyllostomids thus ridges, one from each of the lateral edges of cannot be determined a priori. We also the U-shaped notch of the spear. However, scored Centurio and those phyllostomids that this connection is usually absent in Diphylla lack a thin, free ¯ap of skin along the inferior (one individual, AMNH 91283, had a single border of the horseshoe (states 1 and 2 of central connection running between the nose- character 25) ``Ϫ'' for this character. leaf and the ridge). In Diaemus and Diphylla, This character has not been used in pre- the ridge is generally relatively thin, and in vious phylogenetic analyses. Desmodus it is typically thicker and some- Character 27: Inferior border of thick- what higher. However, there is signi®cant ened horseshoe smoothly curved or straight, within-species variation in ridge morpholo- no V-shaped notch or projection (0); V- gy, and the conditions seen in former taxa shaped notch sometimes present (1); or V- overlap that seen Desmodus.InMormoops shaped notch always present (2); or long, and Noctilio, there is no ridge of skin on the broad, V-shaped projection present (3). The dorsum of the snout; however, in Pteronotus, form of the inferior edge of the horseshoe a ridge is present on the dorsum of the snout. varies among those taxa with a thickened la- Based on this distribution of character states, bial border (see character 25 above). The in- it seems that absence of the ridge is the prim- ferior edge of the horseshoe generally forms itive condition for phyllostomids. a smoothly rounded U-shape or a straight line This is the ®rst use of this character in a under the nostrils. However, a V-shaped phylogenetic analysis. notch is sometimes present in the center of Character 29: No outgrowth from poste- the inferior border of the horseshoe in Mi- rior spear (0); or small scalloped outgrowth cronycteris megalotis (e.g., AMNH 267411) present, connected to base of spear by small and Micronycteris minuta (e.g., AMNH ridge (1); or large sexually dimorphic out- 233221) and is always present in Micronyc- growth from base of spear present (2). Al- teris hirsuta. An alternative condition is seen though most phyllostomids do not have in Ardops and Ariteus, which have a long, structures directly posterior to the spear (but broad, V-shaped projection that extends to- see characters 17 and 28), an outgrowth of ward the lip from the center of the labial bor- the posterior base of the rib is present in Me- der of the horseshoe and gradually grades sophylla and Vampyressa pusilla. There is into the upper lip (®g. 21C). We scored mor- within-species variation in morphology of 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 63 this ¯ap, which always has a scalloped ap- from that of other phyllostomids with chin pearance, but may appear bi®d or tri®d in pads (®g. 23D). The chin in these genera is different individuals. This outgrowth is con- slightly (e.g., Erophylla) to deeply cleft (e.g., nected to the posterior base of the rib by a Choeroniscus) at the midline and the surfac- small central connective ridge. The only oth- es of the cleft are lined with smooth, hairless er stenodermatine which has an outgrowth skin. The anterior borders of the cleft are from the posterior spear is Sphaeronycteris. ¯eshy and scalloped; these small projections In this taxon, the outgrowth is large and sex- resemble papillae when the cleft is closed ually dimorphic (better developed in males). and the bat is viewed anteriorly. Muscular In Sphaeronycteris, there are also two parts action apparently keeps the cleft closed much to the outgrowth. The ®rst is directly poste- of the time in living bats (personal obs.), but rior to the spear, and attaches to almost the the cleft is often open in preserved speci- entire posterior face of spear, leaving a free mens. In such cases, the ¯eshy walls of the border around the spear's edge. This primary cleft can be seen to resemble the chin pads attachment is a small, rounded ridge that typical of other bats with these structures, connects the spear with the larger outgrowth, leading us to conclude that they are homol- commonly termed the ``visor'' due to its ap- ogous. This hypothesis is further supported pearance in male Sphaeronycteris. The ``vi- by the morphology of the chin in loncho- sor'' connects with the entire posterior face phyllines, which is not cleft. In some indi- of the ®rst ridge in both sexes, but has a dif- viduals, only the superior margins of the ferent morphology in each. In female Sphae- pads appear to be scalloped as in glosso- ronycteris, the ``visor'' is small and has its phagines and phyllonycterines (e.g., Lionyc- origin over the center of each eye. In males, teris AMNH 145504); however, substantial the origin is from the lateral corner of the variation in edge ornamentation exists and eye and the structure is about four times the the entire edge is scalloped in many individ- size of that present in females. We scored uals (e.g., Lionycteris AMNH 202295). The mormoopids and Noctilio, which lack a no- chin pads in lonchophyllines are arranged in seleaf, ``Ϫ'' for this character. Therefore, the the same U-orV-shaped pattern seen in des- primitive state for phyllostomids cannot be modontines, phyllostomines, and Rhinophyl- assessed a priori. We also scored Centurio la. Based on these observations, we score ``Ϫ'' for this character. phyllonycterines, glossophagines, and lon- This is the ®rst use of this character in a chophyllines as having a pair of chin pads; phylogenetic analysis. differences in chin morphology among taxa Character 30: Chin with pair of dermal with chin pads are treated separately in char- pads, one present on each side of midline acters 31 and 32. (0); or chin with multiple, well-developed Many phyllostomines, Brachyphylla, Car- dermal papillae (1); or chin smooth or with ollia, and most stenodermatines exhibit a a few poorly developed papillae (2); or chin markedly different pattern in which a U-or partly or completely covered with skin ¯aps V-shaped row of small dermal papillae oc- (3). In desmodontines, many phyllostomines curs in place of paired chin pads (®gs. 23B, (e.g., Macrotus, Vampyrum), phyllonycter- F). These papillae differ from the padlike ines, glossophagines, lonchophyllines, and structures described above because they arise Rhinophylla, a pair of naked dermal pads are from the ¯at anterior surface of the chin rath- present on the chin (®gs. 23A, 23C±E). er than the scalloping of the free edge of a These pads are typically oblong and are ar- chin pad. The space between the two arms ranged in a U or V pattern just ventral to the of the U or V may be bare, or may contain lower lip. The space between the pads is al- some small papillae whose position and de- ways hairless and may lack other dermal velopment varies from individual to individ- structures, although small or large dermal pa- ual. However, one large central papilla is pillae may be present in this region (e.g., consistently present in some taxa (e.g., most Rhinophylla, see character 33 below). stenodermatines; see character 33). Centurio The chin morphology of phyllonycterines lacks chin papillae and pads altogether, and and glossophagines is super®cially different instead has a smooth chin. Most specimens 64 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 23. Anterior views of the chins of representative phyllostomids and outgroup taxa. A. Desmodus rotundus (AMNH 267503). B. Phyllostomus hastatus (AMNH 202308). C. Lonchophylla robusta (AMNH 267452). D. Choeroniscus intermedius (AMNH 266122). E. Rhinophylla pumilio (AMNH 267163). F. Uroderma bilobatum (AMNH 268564). G. Pteronotus davyi (AMNH 214413). Scale bar ϭ 1 mm. of Sphaeronycteris also have a completely mouth there is a low ridge of skin that is smooth chin, but one individual (USNM continuous with the pinna and with the lip 494538) that we examined has an incomplete pad. On the ventral surface of the chin, cau- row of very small, poorly developed papillae dal to the lip pad is a caudal chin ¯apÐa on the chin. However, we consider the con- simple skin ¯ap that encompasses the area dition in both Centurio and Sphaeronycteris between the canines. In Mormoops, the lip to be potentially homologous and score both pad and caudal chin ¯ap are similar to those with state 2 in the matrix. in Pteronotus, although they are slightly Mormoops, Pteronotus, and Noctilio have smaller, but there are more complicated fo- chins that are partly or completely covered liations connecting these elements. Two sets with complex folds of skin (®g. 23G). These of lip ¯aps are present. The superior lip ¯aps ¯aps and folds do not appear to be compa- are broad, ruf¯ed foliations that course al- rable with the chin pads or papillae seen in most from the corner of the mouth to the phyllostomids, but instead seem to be out- posterior face of the lip pad, which they join. growths of the lower lip. Pteronotus has a The caudal chin ¯ap runs between the pos- papillated lip pad that covers the area along terior surfaces of the superior lip ¯aps, en- the lip and is continuous with the pinnae. The compassing the area just below the lip pad. papillation is con®ned to the broader central Narrower inferior lip ¯aps, which are con- area of the lip pad. The labial border of the nected to the pinnae, course to just lateral of pad is arched, and the mental border has a the point of attachment of the caudal ¯ap to U-shaped notch. At the lateral corner of the the superior lip ¯aps, where they too join the 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 65 posterior surface of the superior lip ¯aps. chin is slightly to deeply cleft at the midline Noctilio lacks many of the foliations seen in in phyllonycterines and glossophagines (®g., mormoopids. A caudal chin ¯ap runs be- 23D). Depth of the cleft appears to vary con- tween the ¯aplike lips. There are no addi- tinuously however, and is therefore not con- tional foliations on surface of the chin. Be- sidered as a separate character. Mormoopids cause the unique conditions in the outgroup and Noctilio have a chin that lacks a central taxa do not occur among the ingroup, the cleft, suggesting that this is the primitive primitive condition for phyllostomids cannot condition for phyllostomids. be assessed a priori. This is the ®rst use of this character in a Straney (1980: characters G18±19) used phylogenetic analysis. this character, describing two conditions: Character 33: Central papilla absent presence of simple chin pads (derived state from chin (0); or central dermal papilla of G18), roughly equivalent to our state 0, or present on chin just ventral to midline of complex chin pads (derived state of G19), lower lip (1). A central papilla is absent from roughly equivalent to our state 1. Our char- the chin in desmodontines, most phyllostom- acter is not identical to Straney's (1980), as ines, Brachyphylla, and lonchophyllines (®g., he considered the chin ¯aps of mormoopids 23A±C). In contrast, this papilla is present and noctilionids to be similar to chin papil- just ventral to the midline of the lower lip in lae. Also, Straney (1980) scored Lonchorhi- Macrophyllum, carolliines, and stenoderma- na, Brachyphylla, and Rhinophylla as pos- tines (®g., 23E±F). The central papilla is sur- sessing multiple papillae on the chin, where- rounded laterally and inferiorly by either as we consider these taxa to have chin pads. chin pads (e.g., Rhinophylla) or rows of der- This difference may be due to the presence mal papillae (e.g., Carollia). We scored mor- of a large central papilla on the chins of these moopids and Noctilio, which have a unique genera and others. We consider the presence chin morphology involving chin ¯aps (see of this papilla as a separate character (see character 30), ``Ϫ'' for this character. Thus, character 33). the primitive condition for phyllostomids Character 31: Chin pads simple, not scal- cannot be determined a priori. We scored loped (0); or chin pads with scalloped lateral taxa with smooth chins (state 2 in character edges (1). As described above, the chin pads, 30) and those with a chin cleft (state 1 of when present, are simple (not scalloped) in character 32) ``Ϫ'' for this character. desmodontines, phyllostomines, and Rhino- Although Straney (1980) discussed the phylla (®gs., 23A, E). The lateral edges of distribution of the central papilla in phyllos- the pads are smooth in these taxa. In contrast, tomids, he did not use this as a character in the lateral edges of the chin pads are partly his analysis. Our observations are largely in or completely scalloped in phyllonycterines, agreement with Straney (1980), with the ex- glossophagines, and lonchophyllines giving ception of Phylloderma, Phyllostomus, and the pad edge a papillated appearance (see Brachyphylla. Straney (1980) described the character 30 above; ®gs. 23C±D). We scored central pad as doubled in Phylloderma and mormoopids and Noctilio, which lack paired Phyllostomus. We did not observe a doubled chin pads, ``Ϫ'' for this character. The prim- central pad in these taxa, rather we found that itive condition for phyllostomids thus cannot these two genera, and many others, showed be determined a priori. We scored phyllos- evidence of having two lateral central pads tomids that have either multiple chin papillae (see ®g. 23B, F). However, due to marked or smooth chins (states 1 and 2 of character within- and between-species differences, we 30 above) ``Ϫ'' for this character. were unable to use this as a character at this This character has not been used in pre- hierarchical level. In Brachyphylla, we did vious phylogenetic analyses. not observe a central papilla, while Straney Character 32: Chin without central cleft (1980) indicated that this papilla was present. (0); or with slight to deep central cleft (1). Character 34: Internal labial papillae ab- As described above (see character 30), the sent (0); or internal labial papillae limited chin is relatively ¯at (not cleft) in most phyl- to lip line (1); or internal labial papillae lostomids (®g., 23A±C, E±F). In contrast, the more widely distributed, cover most of inside 66 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 of cheeks (2). In most phyllostomids, the in- has been designed to avoid dealing with the ner surface of the lips and cheek are smooth, continuous nature of pinnal emargination but in stenodermatines these areas are stud- while still describing relevant variation. Al- ded with numerous small projections which though we concur with Owen's (1987, 1991) Silva-Taboada and Pine (1969) called `ìnter- observation of an emarginate lateral border nal labial papillae.'' These ¯eshy papillae in all stenodermatines and carollines, we dis- range in shape from somewhat ¯at, triangular agree on the morphology of the pinna in projections to more conical structures. The Macrotus. Owen (1987: 58) described the extent to which internal labial papillae cover pinna in Macrotus as ``Completely emargin- the inside of the mouth varies among taxa in ated; pronounced indentation as in state 1.'' which these structures are present. All sten- We did not use Owen's (1991: character 1) odermatines have internal labial papillae character `Ànterioproximal lobe on pinna,'' along the upper lip, but the pattern of distri- because this feature is an autapomorphy of bution is variable within species and no clear Centurio. Simmons (1996) used presence or distinctions can be made among taxa. This is absence of a concavity in the lateral edge of not true of the lower lip and cheeks, where the pinna in an analysis of relationships of the internal labial papillae are distributed in Micronycteris, and our scoring agrees with one of two patterns. Most stenodermatines hers. have papillae located only along the lip line, Character 36: Interauricular band absent and lack internal labial papillae on the inside (0); or present (1). In most phyllostomids, of the cheeks. In contrast, Ametrida, Ardops, the ears are not connected by an interauric- Ariteus, Centurio, Phyllops, Pygoderma, ular band. In contrast, a band of skin extends Sphaeronycteris, and Stenoderma have inter- between the anterior external surfaces of the nal labial papillae that more or less cover the medial pinnae in Macrotus, Micronycteris inside of the cheeks. Mormoopids and Noc- hirsuta, M. megalotis, and M. minuta. All of tilio lack internal labial papillae, suggesting these species have ears that, for phyllosto- that absence of these structures is the prim- mids, are large in proportion to the size of itive condition for phyllostomids. the skull: when folded forward they reach This is the ®rst use of this character in a approximately to the end of the snout. The phylogenetic analysis. skin band in Macrotus and these species of Character 35: Pinna margin not smoothly Micronycteris is slightly less than one quarter rounded, concavity located on lateral border the height of the ears. This band is ¯exible (0); or pinna with smoothly rounded margin, and is typically folded back against the brain- no concavity on lateral border (1). In most case in preserved specimens. The band in phyllostomids, a concavity is present at some each taxon originates from the external sur- point along the lateral margin of the pinna. face of each pinna just posterior to the me- However, in Chrotopterus, Macrotus, Mi- dial edge, and runs transversely across the cronycteris hirsuta, M. megalotis, M. minuta, skull just anterior to the slope of the brain- Tonatia, Trachops, and Vampyrum, there is case. Although Koopman (1994) reported no concavity on the lateral margin of the pin- that Tonatia silvicola has these skin ¯aps, na and the margin of the pinna is smoothly our examination of this species revealed that rounded. In mormoopids and Noctilio, a con- these structures do not appear similar to cavity is present on the lateral margin of the those present in Macrotus and Micronycteris. pinna, suggesting that this is the primitive Instead, we interpret the ``¯aps'' in Tonatia condition for phyllostomids. as slightly larger versions of the ridges of Owen (1987: character 1; 1991: characters skin which cover the auricular muscles in all 1±2) described the emargination of both the species (these are particularly visible if the medial and lateral edges of the pinna. Al- pinna is pulled forward). We therefore score though we agree with Owen (1987, 1991) Tonatia as lacking an interauricular band. Al- that the pinna presents variation that may be though there are extensive foliations between useful in a phylogenetic context, our survey the ears in Mormoops (see below), Pterono- of this character suggests that Owen (1987) tus and Noctilio lack an interauricular band. subdivided a continuous trait. Our character The distribution of character states in the out- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 67 group taxa suggests that absence of the in- lar band into two triangular ¯aps, one asso- terauricular band is primitive for phyllosto- ciated with each ear. We scored mormoopids mids. and Noctilio, which lack an interauricular In Mormoops the extensive foliations be- band, ``Ϫ'' for this character. The primitive tween the ears differ signi®cantly from the condition of this feature thus cannot be de- morphologies described above. A ¯ap of skin termined a priori. We scored other phyllos- originates from each ear, but these ¯aps are tomids that lack an interauricular band (state much more extensive and complex than 0 of character 36) ``Ϫ'' for this character. those described above. Each ¯ap arises from Simmons (1996: character 6) previously the medial external surface of the pinna and used depth of the interauricular band as a courses anteromedially toward the midsagit- character in an analysis of relationships of tal line. Near the midpoint, each ¯ap divides, Micronycteris species. Due to the inclusion giving rise to a poorly de®ned branch that of several species of Micronycteris that do connects with its counterpart across the top not appear in our analysis, Simmons (1996) of the cranium and a much larger ¯ap that subdivided the character more ®nely than we courses anteriorly until it terminates on the have done here, recognizing three states dorsum of the nose halfway between the tip (shallow, moderate, and deep). However, our of the snout and the slope of the braincase. observations agree with hers. These large right and left ¯aps are separated Character 38: Facial hood absent in both throughout their course along the rostrum. At sexes (0); or present in males (1). The skin the point of termination there is a roughly of the neck is relatively taut and smooth in triangular lateral expansion in each of these most phyllostomids. In contrast, folds of skin ¯aps. Because of the differences between the are present around the neck in Centurio and simple skin band seen in some phyllostomids Sphaeronycteris. These skin bands are small- and the complex branched ¯aps found in er in females than in males, and they reach Mormoops, we suspect that these structures their greatest development in male Centurio. are not homologous. Pteronotus and Noctilio In males of both Centurio and Sphaeronyc- lack skin ¯aps between the ears, and pres- teris, the throat folds apparently operate as a ence of these structures is generally inter- hood, or mask, which drops over the face preted as autapomorphy of the genus Mor- when the bat is roosting (Hill and Smith, moops (Smith, 1972). Accordingly, we score 1984; Paradiso, 1967; Goodwin and Green- Mormoops ``Ϫ'' for this character. hall, 1961; Nowak, 1991: photograph, p. Straney (1980: character G20) used pres- 314). The facial hood is absent in mormoop- ence of an interauricular band in his com- ids and Noctilio, suggesting that this is the ponent analysis. However, he observed a primitive condition for phyllostomids. band between the ears in Lonchorhina and Lim (1993: character 3) ®rst used this several species of Tonatia. As we noted character in a cladistic analysis. Our char- above, we interpret these ¯aps as being acter states and scoring are identical to his. slightly larger versions of the ridges of the auricular muscles that connect the pinna with SKULL AND DENTITION the fascia covering the skull. Simmons (1996: character 5) previously used presence Characters described in this section are of an interauricular band as a character in an based largely on features originally noted by analysis of relationships of Micronycteris Straney (1980: character numbers from ap- species; our assessments agree with hers. pendix 1), Owen (1987: character numbers Character 37: Notch in interauricular from his appendix 2; 1991: character num- band shallow (0); or deep (1). The interau- bers from the appendix), Lim (1993: char- ricular band is characterized by a shallow, V- acter numbers from appendix 2), Marques- shaped midsagittal notch in Micronycteris Aguiar (1994: character numbers from ap- hirsuta and M. megalotis. In contrast, the pendix 4) and Wible and Bhatnagar (1996). midsagittal notch is much deeper in Macro- Taxa examined in these studies are listed in tus and Micronycteris minuta. In these taxa, table 4. We have revised many of the char- the notch effectively divides the interauricu- acter descriptions given by these authors to 68 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 better describe variation within phyllosto- in most phyllostomids (Wible and Bhatnagar, mids as a whole, and have scored most char- 1996). Because the curved vomeronasal car- acters for all of the 63 taxa included in our tilage changes shape over its length, more analysis. In addition, we developed several than one of these shapes may occur in a sin- new characters of potential phylogenetic sig- gle individual. In contrast, Brachyphylla pos- ni®cance; some of these ®rst appeared in sesses a bar-shaped vomeronasal cartilage Simmons' (1996) analysis of relationships of (Wible and Bhatnagar, 1996). Among the Micronycteris species. outgroup taxa, a curved vomeronasal carti- In the descriptions we present, we refer to lage is present in Mormoops, whereas a bar- paired structures (e.g., premaxillae) as sin- shaped vomeronasal cartilage is present in gular (e.g., premaxilla), in effect describing Pteronotus and Noctilio (Wible and Bhatna- half of the cranium or dentition. We do note gar, 1996). Given this distribution of char- cases where the two halves of the skull or acter states, the presence of a bar-shaped dentition differ. vomeronasal cartilage appears to be primitive Character 39: Vomeronasal tube well de- for phyllostomids. veloped with neuroepithelium present (0); or Wible and Bhatnagar (1996) de®ned and rudimentary, neuroepithelium absent (1); or mapped this character onto preexisting phy- absent (2). The vomeronasal epithelial tube logenies of bats. Our scoring is identical to (VET) is `àn elongate, cigar-shaped structure theirs. In the context of the phylogenies Wi- lined with epithelium'' that is associated with ble and Bhatnager (1996) examined, they de- nerve ®bers, vomeronasal glands, blood ves- scribed the primitive state as our state 1. sels, and venous sinuses (Wible and Bhat- Their state 2 (vomeronasal cartilage absent) nagar, 1996: 294). In desmodontines, Macro- does not occur in any of our taxa and we tus, glossophagines, Carollia, and stenoder- therefore excluded it from consideration. matines the VET is well-developed and has Character 41: Nasopalatine duct present a neuroepithelial lining medially (Wible and (0); or absent (1). The nasopalatine duct, Bhatnagar, 1996). In contrast, in Brachy- which connects the nasal and oral cavities phylla the VET occurs in a rudimentary form through the incisive foramen in the palate, is and lacks the neuroepithelium (Wible and present in most phyllostomids (Wible and Bhatnagar, 1996). In mormoopids, the VET Bhatnagar, 1996). In contrast, the nasopala- is rudimentary in Mormoops, well developed tine duct is absent in Macrotus (Wible and in Pteronotus parnellii, and completely ab- Bhatnagar, 1996). The nasopalatine duct is sent in Pteronotus personatus (Wible and present in Mormoops and Pteronotus person- Bhatnagar, 1996). We scored Pteronotus with atus, but is absent in Pteronotus parnellii and states 0, and 2 in the matrix. The VET is Noctilio (Wible and Bhatnagar, 1996). We absent in Noctilio (Wible and Bhatnagar, scored Pteronotus with states 0 and 1 in the 1996). Thus, the primitive state for phyllos- matrix. Due to this distribution of character tomids cannot be assessed a priori. states among the outgroup taxa, the primitive Based primarily on the work of Bhatnager state for phyllostomids cannot be recon- and his colleagues (e.g., Bhatnager and Kal- structed a priori. len, 1974; Cooper and Bhatnager, 1976; Wible and Bhatnagar (1996) ®rst used this Bhatnager 1980, 1985; Frahm and Bhatnager, character. Our character states and scoring 1980; Frahm 1981; Bhatnager et al., 1982), are identical to theirs. Wible and Bhatnagar (1996) de®ned this Character 42: Zygomatic arch always character (and characters 40, 41, and 135), complete (0); or always incomplete (1); or mapping it onto preexisting hypotheses of polymorphic within species (2). Although chiropteran relationships. Our character most phyllostomids have a complete zygo- states and scoring are identical to theirs. matic arch, this structure is incomplete in Character 40: Vomeronasal cartilage Phyllonycteris, many glossophagines (e.g., bar-shaped (0); or curved, J-, C-, U-, or O- Choeroniscus, Lichonycteris), lonchophyl- shaped in cross section (1). The vomeronasal lines, and carolliines. In all taxa that lack the cartilage, which supports the VET, is present zygomatic arch, a distinct zygomatic process and J-, C-, U-, or O-shaped in cross section projects anteriorly from the lateral margin of 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 69 the mandibular fossa and posteriorly from Carollia). However, in species where the dorsal or slightly posterior to the last molar. posterior process of the zygoma does not ex- Both processes taper and have rounded tips. tend past the level of the mastoid, we did not Among brachyphyllines, phyllonycterines, attempt to score this character; these taxa are glossophagines, and lonchophyllines, a thin scored ``Ϫ'' in the matrix. Mastoid breadth but complete arch is present only in Brachy- is less than zygomatic breadth in mormoo- phylla, Erophylla, Glossophaga, Leptonyc- pids and Noctilio, suggesting that this is the teris, and Monophyllus. Anoura is the only primitive condition for phyllostomids. phyllostomid that exhibits polymorphism in Simmons (1996: character 14) ®rst used zygomatic development. Some specimens of mastoid versus zygomatic breadth as a char- Anoura geoffroyi have two complete zygo- acter in a phylogenetic analysis of relation- matic arches whereas others lack both, and ships of Micronycteris species; our assess- one specimen (AMNH 78288) has a comments agree with hers. plete zygomatic arch on the right side and an Character 44: Hard palate long, extends incomplete arch (that does not appear to have into interpterygoid space (0); or no palatal been broken) on the left side. However, in extension or emargination (1); or palate with Anoura caudifera it appears that both zygo- shallow posterior emargination that extends matic arches may be complete in all individ- maximally to middle of upper second molar uals (Miller, 1907; personal obs.). Those (2); or with deep emargination that extends specimens without a complete arch show minimally to posterior border of ®rst molar clear evidence of breakage (e.g., AMNH (3). The posterior part of the hard palate ex- 176347). Based on the states de®ned above, tends into the interpterygoid space in most we scored Anoura with states 0 and 2. Mor- phyllostomids (®g. 24A). In some taxa (e.g., moopids and Noctilio have a complete zy- Pygoderma) there is no extension of the hard gomatic arch, suggesting that this condition palate into the interpterygoid space, nor is is primitive for phyllostomids. the palate emarginate (®g., 24B). However, Lim (1993: character 19) ®rst used this in some stenodermatines the hard palate character in a cladistic analysis. We have fol- lacks a posterior extension and is emarginate. lowed his character state de®nitions but have In Centurio and Sphaeronycteris, the palatal added state 2 (polymorphic). Although emargination is shallow, extending to ap- Koopman (1994: 81, 86) expressed some res- proximately the middle of M2 measured at ervation about the status of the zygomatic the labial margin of the tooth (one specimen arch in Lichonycteris (``more or less com- [AMNH 175651] of Centurio in which the plete'') and in a subgenus of Sturnira (Cor- emargination appears deeper has clearly been vira: ``weak or incomplete''), our examina- damaged; ®g. 24C). The emargination is tions indicate that the zygomatic arch is nev- deeper in Ardops, Ariteus, Phyllops, and er complete in Lichonycteris, but is always Stenoderma ending somewhere along the la- complete in Sturnira (Corvira). bial border of M1 (®g. 24D). In Ametrida, Character 43: Mastoid breadth less than the pterygoids are very small (compressed in zygomatic breadth (0); or greater than zy- the anterior-posterior dimension) and the in- gomatic breadth (1). In most phyllostomids, ternal nasal aperture is reduced by bony lam- breadth of the skull measured across the mas- inae extending medially from the pterygoids toid region is less than breadth measured in both the coronal and transverse planes (®g. across the zygoma. In contrast, mastoid 24E). This is a unique condition among phyl- breadth is greater than zygomatic breadth in lostomids (Miller, 1907; personal obs.), and Lonchorhina, Micronycteris minuta, and does not appear to be homologous with other some species of Tonatia (e.g., T. silvicola, T. conditions described above; we therefore evotis). We scored Tonatia with states 0 and scored Ametrida ``Ϫ'' for this character. In 1 in the matrix. We scored this character in Mormoops blanvillii, Pteronotus, and Nocti- some taxa with an incomplete zygomatic lio, the hard palate extends into the inter- arch (see character 42) when the posterior pterygoid space; however, there is no palatal process of the zygoma clearly extended lat- emargination or extension in Mormoops me- erally beyond the level of the mastoids (e.g., galophylla. This distribution of character 70 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 24. Ventral view of the palate and mesopterygoid fossa in selected taxa illustrating palate length and shape of the palatal emargination. A. Carollia perspicillata (AMNH 130722) B. Pygoderma bilobatum (AMNH 248339) C. Sphaeronycteris toxophyllum (AMNH 209741) D. Stenoderma rufum (AMNH 208982) E. Ametrida centurio (AMNH 187224). Scale bar ϭ 5 mm. states suggests that possession of a hard pal- palate'' (1), ``Does not extend caudad to an- ate that extends into the interpterygoid space terior border of orbit'' (2). This character was may be primitive for phyllostomids. ordered (0 ↔ 1 ↔ 2). Subsequently, Owen Owen (1987: character 9) measured the (1991: characters 21±22) split his (1987) length of the posterior margin of the hard states 1 and 2 into separate characters. Al- palate from the anterior border of the orbit, though Owen's (1991) scoring remained un- and de®ned three states: `Èxtends farther changed, he transformed the multistate char- caudad than least width of palate between or- acter he had previously described (Owen, bits'' (0), `Èxtends caudad to anterior border 1987: appendix 2) into several characters us- of orbits, but no farther than least width of ing additive binary coding. We have de- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 71 scribed this character in different terms than straight to barely curved (®g. 24B). The those used by Owen (1987, 1991) to accom- shape of the medial posterior border of the modate some of the variation his character hard palate appears to be potentially infor- did not account for. Thus, our character is not mative in stenodermatines with emarginate directly comparable to his. hard palates (states 2 and 3 in character 44). Lim (1993) recognized that Owen's (1987) In Ardops, Ariteus, and Stenoderma, the pal- state 2 encompassed more variation than atal emargination is U-shaped (®g. 24D). The Owen (1987) had described. Accordingly, emargination is V-shaped in Centurio, Phyl- Lim (1993) erected several states to account lops (we examined both P. falcatus and P. for this variation. However, Lim (1993: char- haitiensis; see below), and Sphaeronycteris acter 4) incorporated shape and size in a sin- (®g. 24C). The primitive condition for phyl- gle ordered (0 → 1 → 2 → 3 → 4 → 5) lostomids cannot be assessed a priori because character: ``Hard palate extending into in- mormoopids and noctilionids do not have terpterygoidal space (0); no palatal extension emarginate palates. We scored taxa that do or emargination (1); palatal emargination not have an emarginate palate (states 0 and shallow and V-shaped (2); palatal emargina- 1 of character 44 above) ``Ϫ'' for this char- tion moderate and converging anteriorly (3); acter. palatal emargination deep and narrow (4); Owen (1987: character 10) ®rst described palatal emargination very deep and wide this character using four states, including the (5).'' We have chosen to score shape sepa- `Ù-'' and ``V-shaped'' conditions we recog- rately from length (see character 45). In ad- nize. The other two states, ``square'' (his dition, we do not recognize moderate emar- state 0) and ``median projection'' (his state gination (Lim's state 3) as a distinct character 3) do not apply to taxa with an emarginate state. Lim (1993) stated that this condition palate. Owen (1987) scored his palatal char- appears in Phyllops, but our observations acter for all taxa, including those with long suggest that the ``moderate'' condition, palates, and ordered it in a complex fashion. which would be an autapomorphy of Phyl- Subsequently, Owen (1991: character 23) lops and therefore uninformative in a phy- again used a similar character but only rec- logenetic analysis, more closely resembles ognized two states: ``square'' and ``V- the conditions seen in Ardops, Ariteus, and shaped.'' Our observations often do not Stenoderma. We also do not score the palate agree with those of Owen (1987, 1991). of Ametrida for this character; Lim (1993) Owen (1987) considered Ametrida to have a considered the palate in this genus to lack an V-shaped posterior border of the hard palate extension or emargination. and described this border as U-shaped in Py- Character 45: Posterior border of hard goderma and Phyllops falcatus. Owen (1991) palate always U-shaped (0); or always V- grouped taxa into two categories. ``Square'' shaped (1). The shape of the caudal border included Ametrida, Centurio, Phyllops hai- of the palate does not appear to be a useful tiensis, and Sphaeronycteris.``V-shaped'' in- character among phyllostomids with long cluded the remaining taxa with emarginate palates (state 0 in character 44) because there palates and Pygoderma. is a continuous range of variation among the Lim (1993: character 4) considered palatal U-, V-, and W-shaped conditions both within shape and length as a single ordered char- genera and individual species. For example, acter, but scored the shape of the caudal bor- Lonchophylla thomasi has a V-shaped emar- der only in taxa with palatal emargination. gination, whereas other Lonchophylla spe- Marques-Aguiar (1994: character 21), who cies, such as L. robusta, have a U shape also considered whether the ``mesopterygoid (Koopman, 1994; personal obs.). In Carollia fossa'' had U-orV-shaped borders, scored subrufa, the posterior palate is V shaped in this character for taxa with long palates. We some specimens (e.g., AMNH 164023), but have followed Lim (1993), because posterior is U shaped in others (e.g., AMNH 186375). border shape is variable within species with In those taxa that do not have a palatal ex- long palates, and because the posterior bor- tension or emargination (state 1 in character der of the palate may not be homologous in 44), the posterior edge of the palate is taxa with emarginate palates and those with 72 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 25. Ventral view of the basicranium in A. Glossophaga soricina and B. Choeroniscus intermedius. The difference in in¯ation of the pterygoids is indicated by the arrows. Redrawn from Phillips (1971: ®g. 45). longer palates. We have described the states basicranium. In most phyllostomids, the pter- of this character in slightly different terms, ygoid may curve slightly at its posterior and recognizing only two general shape catego- distal end, but over most of its course it di- ries, ``V-'' and ``U-shaped,'' rather than the verges from the midline of the basicranium. four states Lim (1993) recognized (V-shaped; In most phyllostomids, the pterygoid lamina moderate and converging anteriorly; deep is not in¯ated or ballonlike (®g. 25A). In and narrow; very deep and wide). Two of the contrast, in Choeroniscus, Choeronycteris, four conditions Lim (1993) identi®ed are au- and Musonycteris, the pterygoid lamina is tapomorphies. Phyllops was the only genus greatly in¯ated, ballooning laterally posterior Lim (1993) scored with the ``moderate and to the anterior margin of the mandibular, or converging anteriorly'' condition, while only glenoid, fossa (®g. 25B). Anterior to this Ardops was scored as ``palatal emargination point, the pterygoid diverges from the mid- very deep and wide.'' Because these are au- line of the basicranium. In mormoopids and tapomorphies, they are not informative in a Noctilio the pterygoid does not balloon lat- phylogenetic analysis at this level. In addi- erally, suggesting that this is the primitive tion, we observed that these two taxa share condition for phyllostomids. more general conditions (either ``V-''or``U- This character has not been used in pre- shaped'') with other taxa. Consequently, we vious cladistic analyses, although it has been have recognized only these two conditions. described by other authors (e.g., Miller, We did not score genera with a palate that 1907; Phillips, 1971; Koopman, 1993). has no extension or emargination (state 1 in Character 47: Rostrum sloping; facial character 44) because a straight to slightly processes of premaxilla and maxilla above curved posterior border of the hard palate teeth lie perpendicular to palatal processes and state 1 in character 44 have identical dis- (0); or facial processes of premaxilla and tributions. maxilla oriented horizontally above teeth (1). Character 46: Pterygoid lamina sheetlike The rostrum slopes gently from the forehead (0); or pterygoid lamina greatly in¯ated pos- to the external nares in most phyllostomids. terior to anterior margin of mandibular fossa Just above the teeth, the facial processes of (1). The pterygoid lamina is a relatively thin the premaxilla and maxilla lie roughly per- sheet of bone that projects ventrally from the pendicular to the palatal processes of these 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 73

Fig. 26. Lateral view of skull contrasting the gently sloping rostrum of A. Mesophylla macconnelli (AMNH 268539) to the more foreshortened `àpelike'' rostrum in B. Ametrida centurio (AMNH 267274). bones in these taxa (®g. 26A). In contrast, this character with two states (``gradually the facial bones that compose the normally sloping rostral pro®le'' and ``dorso-ventral sloping rostrum in most phyllostomids (nasal compression of the rostrum''), which corre- and maxilla) are upturned and largely ori- spond to our character states; our observa- ented vertically in Ametrida and Sphaeron- tions agree with his. ycteris. The nasal aperture in these two taxa Character 48: I2 present (0); or absent is located at the base of the cranium, giving (1). No bat has more than two upper incisors the skull a distinctly `àpelike'' appearance. (Slaughter, 1970). Although there has been Just above the teeth, the facial processes of some debate regarding the homologies of the premaxilla and maxilla lie in roughly the these teeth to those in other mammals, most same plane as the palatal processes (rather workers agree that these teeth probably rep- than being roughly perpendicular to the pal- resent I1 and I2 (Thomas, 1908; Anderson, atal processes), an arrangement that makes 1912; Slaughter, 1970). Most phyllostomids this area appear dorsoventrally compressed retain both I1 and I2. In contrast, Desmodus (®g. 26B). In mormoopids and Noctilio, the and Diaemus lack I2. Mormoopids and Noc- rostrum is gently sloping and the facial sur- tilio retain I2, suggesting that this condition faces of the maxilla and premaxilla just is primitive for phyllostomids. above the teeth are roughly perpendicular to Owen (1987: character 21) used complete the palatal processes, suggesting that this ar- dental formula as a character in his exami- rangement represents the primitive condition nation of stenodermatine relationships. Un- for phyllostomids. der Owen's (1987) coding scheme, different Lim (1993: character 7) originally scored dental formulae were scored as different 74 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 character states and the transformation series clusal margin evenly bi®d (2); or main cusp was ordered in a complex fashion. The result on occlusal margin pointed (3); or mesially of this coding was that homologies of indi- and distally projecting lobes present (4). I1 vidual teeth and tooth regions were not al- in most phyllostomids has a roughly straight ways preserved. For example, Owen's (1987) or somewhat rounded occlusal margin (®g. coding precluded the possibility that pres- 27B±C). In desmodontines, I1 has a strongly ence of a single lower incisor might be a concave (C-shaped) occlusal margin (®g. shared, derived feature of two taxa with dif- 27A). Yet another pattern appears in some ferent overall dental formulae. Owen (1991) stenodermatines (e.g., Artibeus, Uroderma), recoded these data, using characters consid- which have an I1 with a well-developed ering speci®c tooth regions separately. Lim notch in the occlusal margin near the center (1993) did not use dental formula in his anal- of the tooth (®g. 27D). Presence of this notch ysis of stenodermatine relationships. We con- gives the tooth a bi®d or bilobed appearance. sider dental formula to be potentially infor- Although the notch may be slightly offset mative, but prefer to score presence/absence from the center of the tooth, both lobes are of teeth at individual tooth loci to better pre- approximately the same size. Finally, other serve initial assessments of putative homol- stenodermatines (e.g., Ametrida, Stenoder- ogies. ma) have a pointed main cusp on I1 (®g. Character 49: I1 distinctly larger than I2 27E±G). Rhinophylla pumilio appears to (0); or I1 and I2 approximately equal in size have a unique I1 morphology. In this species, (1). In most phyllostomids that retain both I1 the occlusal margin of the tooth has two pro- and I2, the inner tooth (I1) is noticeably larg- jecting lobes, one which projects mesially er than the outer tooth (I2; ®g. 27B, D±G). and one distally, with a slight depression be- In contrast, I1 and I2 are approximately tween them. However, both R. alethina and equal in size in glossophagines (®g. 27C). In R. ®scherae have a straight occlusal margin mormoopids and Noctilio, I1 is noticeably on I1; thus we have scored Rhinophylla with larger than I2, suggesting that this is the states 0 and 4 for this character. I1 is bi®d in primitive condition for phyllostomids. We mormoopids, but has a straight or somewhat scored taxa that lack I2 (state 1 in character rounded occlusal margin in Noctilio. The 48) ``Ϫ'' for this character. primitive state for phyllostomids thus cannot Lim (1993: character 10) described a sin- be reconstructed a priori. gle ordered character for size and shape of Lim (1993: character 10) described a mul- I1. The criterion Lim (1993) used to judge tistate ordered character based on the size the size of I1 appears to have been arbitrary. and shape of I1. We do not agree with Lim's Many of the taxa that Lim (1993) scored as (1993) coding of this character. Mormoops possessing ``reduced'' teeth of various mor- and Pteronotus, which both have a bi®d I1, phologies (his states 0, 2, and 3 of character were coded solely on the basis of tooth size 10) have an I1 roughly two times as large as by Lim (1993), who apparently ignored the I2 (e.g., Ametrida; ®g. 27E). We prefer to similarity between the condition in these taxa de®ne two characters that describe size and and those seen in other taxa with bi®d inci- shape separately (this character and character sors (e.g., Artibeus). Lim (1993) also de- 50) because this method allows clear de®ni- scribed two conditions for I1 of the ``short- tion of potential homologies. Phillips (1971: faced'' stenodermatines (see discussion in 26), who commented at length on the denti- character 51). In this character, we have cho- tion of glossophagine (s.l.) bats, considered sen to emphasize the similarity in the shape the I1 of Lonchophylla to be `àbout as large of the main cusps of I1 in all ``short-faced'' as'' I2. The representatives of Lonchophylla stenodermatines, while describing certain we have examined (see appendix 1 for spec- differences among them in the characters be- imens) all possessed an I1 that was notice- low (51 and 52). Dividing the characters in ably larger (often 2X as large) than I2. this way best re¯ects our hypotheses of ho- Character 50: I1 occlusal margin gener- mology. ally straight or slightly rounded (0); or oc- Marques-Aguiar (1994: character 24), in clusal margin concave, C-shaped (1); or oc- her analysis of relationships among species 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 75

Fig. 27. Anterior view of the upper central and lateral incisors in A. Desmodus rotundus (AMNH 174303). Desmodus has only one pair of incisors. B. Phyllostomus hastatus (AMNH 267905). C. Glos- sophaga soricina (AMNH 209354). D. Artibeus jamaicensis (AMNH 266337). E. Ametrida centurio (AMNH 187225). F. Centurio senex (AMNH 256846) G. Stenoderma rufum (AMNH 208982). Scale bar ϭ 5 mm. 76 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 of the subgenus Artibeus, described a char- Centurio and Pygoderma as similar to those acter based on morphology of I1. The two of Ametrida and Sphaeronycteris, when both character states were ``(0) bilobed (bi®d) and Centurio and Pygoderma clearly had mesi- not pointed,'' and ``(1) simple (not bi®d) and ally offset main cusps. pointed.'' Although we agree with her char- Although Miller (1907) reported that the acterization of all members of the subgenus main cusp in Centurio is in the center of the Artibeus, Dermanura gnoma, and Koopman- tooth, our observations indicate that it is ia as possessing a bilobed I1, we do not placed towards the mesial edge of the tooth. agree that I1 in Enchisthenes is pointed. Finally, Miller (1907) observed, as we did, Character 51: I1 with pointed main cusp an often poorly developed accessory cusp to- offset mesially, lateral part of tooth forms ward the tip of the main cusp in Pygoderma. poorly to well-developed cusp (0); or, point- This condition appears to be an autapomor- ed cusp roughly in center of tooth (1). phy of this genus. In the other taxa with a Among those phyllostomids with a pointed lateral cusp, as in Pygoderma, the lateral main cusp on I1, there appear to be two dis- cusp arises from the base of the tooth, it does tinct morphologies. In Ardops, Ariteus, Cen- not arise from the main cusp. turio, Pygoderma, Phyllops, and Stenoder- Character 52: I1 with pointed main cusp ma, the most well-developed part of I1 is the only slightly longer than lateral cusp (0); or mesial part of the tooth. This part of the tooth main cusp 2ϫ the length of the lateral cusp forms a distinct, pointed cusp in all these (1). In those stenodermatines with a lateral taxa (®gs. 27F±G). Thus, the main cusp ap- cusp, there are two patterns of main cusp de- pears offset from the center (longitudinal velopment. In Stenoderma and Pygoderma, axis) of the tooth. The lateral part of the the main cusp is very long, roughly twice the tooth may form a well-developed (e.g., Ar- size of the smaller lateral cusp (®g. 27G). In dops) or poorly developed (e.g., Pygoderma) the remaining taxa with a mesially displaced cusp. In contrast, the main cusp comes to a main cusp, this cusp is only slightly longer point in roughly the center of the tooth in than the lateral cusp (®g. 27F). We scored Ametrida and Sphaeronycteris (®g. 27E). In mormoopids and Noctilio, which do not have one specimen (AMNH 187225) of Ametrida a pointed I1, ``Ϫ'' for this character. The out of 10 we examined for this character, we primitive state for phyllostomids thus cannot found an exceptionally well-developed ac- be reconstructed a priori. We scored all other cessory cusp near the tip of the main cusp. We interpret this as an anomaly, and score taxa lacking pointed incisors (states 0, 1, 2, Ametrida with state 1. Mormoopids and Noc- and 4 in character 50) and well-developed tilio do not have pointed inner upper incisors mesial cusps (state 1 in character 51) ``Ϫ'' and are therefore scored ``Ϫ'' for this char- for this character. acter, as are all other taxa lacking pointed Although this character has not appeared incisors (states 0, 1, 2, and 4 in character 50). in previous cladistic analyses, this feature of The primitive state for phyllostomids thus the teeth was noted by previous authors (e.g., cannot be reconstructed a priori. Miller, 1907). Lim (1993: character 10) originally scored Character 53: i1 present (0); or absent morphology of the inner upper incisors in his (1). Most phyllostomids have two lower in- analysis of stenodermatine relationships. He cisors, which we identify as i1 and i2 follow- scored Ametrida, Centurio, Pygoderma and ing Slaughter (1970; although see Thomas, Sphaeronycteris as having an I1 that is tri- 1908 for a different view). However, i1 is angular in shape. He described the remaining absent in many glossophagines (e.g., Anoura, ``short-faced'' taxa as having an I1 with a Choeronycteris, Hylonycteris). Mormoopids small secondary outer cusp (we call this cusp and Noctilio retain i1, suggesting that pres- the ``lateral cusp'' to avoid confusion with ence of i1 is the primitive condition for phyl- the ``accessory cusp''; see below). We ob- lostomids. served the lateral cusp in most ``short-faced'' This character has not been used in pre- taxa, including Pygoderma. In addition, we vious phylogenetic analyses, although nu- found it dif®cult to characterize the teeth of merous authors have described dental for- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 77 mulae of glossophagines (e.g., Miller, 1907; where I1 occludes with a broad cingular shelf Phillips, 1971; Koopman 1994). that encircles the posterolingual base of the Character 54: i2 present (0); or absent lower canine. Thus, I1 lies well posterior to (1). Although most phyllostomids retain i2, i1 when the canines and postcanine teeth are this tooth is absent in some phyllostomines in occlusion. Similarly, in desmodontines, (e.g., Chrotopterus, Tonatia), many glosso- the I1 lies posterior to the i1 when the ca- phagines (e.g., Anoura, Choeronycteris), and nines and postcanine teeth occlude. However, some stenodermatines (Sturnira bidens, S. in these taxa I1 rests in a fossa located in the nana, and Vampyressa bidens). We scored mandible posterior to i1. In mormoopids and Sturnira with states 0 and 1 in the matrix. Noctilio, the I1 and I1 occlude or are sepa- Mormoopids retain both lower incisors, but rated by a very small gap when the canines i2 is absent in Noctilio. The primitive con- and postcanine teeth occlude, suggesting that dition for phyllostomids cannot be assessed this is the primitive condition for phyllos- a priori. tomids. We scored taxa lacking both lower Straney (1980: character K3) based a bi- incisors (state 1 of characters 53 and 54) nary character on presence or absence of i2. ``Ϫ'' for this character. Our character states are equivalent to his, as To our knowledge, patterns of incisor oc- is our scoring with the exception of our ob- clusion have not been previously used as servation that not all stenodermatines have characters in phylogenetic analyses of phyl- i2. Owen (1991: character 44) de®ned a char- lostomid taxa, although the occlusal pattern acter for loss of a lower incisor (from two and fossae of desmodontines have been de- incisors to one) in some stenodermatines, but scribed before (e.g., Miller, 1907). our observations are not consistent with his Character 56: P3 present (0); or absent coding of this character. Owen's (1987) char- (1). No bat is known to have more than three acter for dental formula correctly scored the upper premolars (Slaughter, 1970). Although two Sturnira species and Vampyressa bidens there has been continuing debate regarding as possessing a single lower incisor. How- the homologies of these teeth to those in oth- ever, Owen (1991) scored Ariteus and some er mammals (e.g., Miller, 1907; Thomas, species of Dermanura as having only one 1908; see review in Slaughter, 1970), most lower incisor; all specimens of these taxa that recent workers have settled on numbering the we examined retained both i1 and i2. premolars ``P2-P3-P4'' in bats (e.g., Slaugh- Character 55: I1 and i1 in contact or sep- ter, 1970), and we follow this usage mainly arated by small gap when cheek teeth oc- 4 clude (0); or I1 and i1 separated by marked for convenience . Of the three upper pre- gap (1); or I1 occludes with cingular shelf molars found primitively in bats, most phyl- on lower canine well posterior to i1 (2); or lostomids retain two, which are apparently I1 occludes posterior to i1 in mandibular homologous to P3 and P4 of other bats fossae (3). When the jaw is closed and the (Slaughter, 1970). Only one noctilionoid, An- canines and postcanine teeth are in occlusion, oura, has P2 in addition to P3 and P4; this the I1 and i1 contact each other in most phyl- appears to be an autapomorphy. In contrast, lostomids, though in some taxa these teeth desmodontines have only one upper premo- may be separated by a very small gap. In lar, having lost P3 (Slaughter, 1970). Mor- contrast, I1 and i1 are separated by an ex- moopids retain both P3 and P4, whereas tremely large gap when the canines and post- Noctilio has apparently lost P3. The primitive canine teeth are in occlusion in most stenodermatines. The only stenodermatines with a ``normal'' occlusal pattern (state 0) are two 4 Thomas (1908) argued that the missing premolar in of the three species of Chiroderma we ex- both upper and lower jaws is the second. This argument has been accepted by some authors (e.g., Handley, 1959; amined (C. salvini and C. trinitatum), Ecto- Simmons and Handley, 1998). However, the identity of phylla, Mesophylla, Sturnira, and Vampyres- the missing tooth has no bearing on phyllostomid rela- sa species. We scored Chiroderma with tionships. Given the widespread usage of P2-P3-P4, we states 0 and 3 in the matrix. Another pattern adopt this system to avoid confusion, but recognize that occurs in Chrotopterus and Vampyrum, P2 may be P1. 78 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 state for phyllostomids thus cannot be recon- ly small (less than one quarter the overall structed a priori. size of p2 and p4) and peglike with cusps Straney (1980: character K6) described a that are poorly developed or absent. In two binary character based on presence of P3; specimens of Lonchorhina (AMNH 149233 our character states and scoring are identical and AMNH 183850) we suspect that p3 may to his. be absent; however, dried tissue is present in Character 57: Second distal accessory this region, making it impossible for us to cusp on P4 absent (0); or second distal ac- unambiguously determine whether p3 is cessory cusp sometimes present (1); or sec- present in these two individuals. We there- ond distal cusp always present (2). P4 lacks fore score p3 as present in Lonchorhina. This a second distal accessory cusp in most phyl- tooth is absent from both sides in one spec- lostomids. In contrast, two distal accessory imen of Macrophyllum (AMNH 209320), but cusps are often present in the stenodermati- we consider this an anomaly, and score Mac- nes Ardops, Ariteus, Artibeus, Dermanura, rophyllum with state 1. In some phyllostom- Phyllops, Platyrrhinus, and Pygoderma. The ines (e.g., Phylloderma, Tonatia), the crown second distal accessory cusp is always pres- of p3 is reduced in size, but the height of p3 ent only in Uroderma and Vampyrodes. Mor- is less than or equal to one third the height moopids and Noctilio do not have a second of p2 and p4. However, the tooth is still large distal accessory cusp, suggesting that this is and has well-developed cusps. In all remain- the primitive state for phyllostomids. ing phyllostomids in which this tooth is pres- Lim (1993: character 6) ®rst described this ent, the lower premolars have crowns that are character; however, he scored only Platyr- subequal in height. In mormoopids, p3 is rhinus and Uroderma as having two distal moderately reduced in Mormoops and is accessory cusps, and did not note the poly- greatly reduced and peglike in Pteronotus. morphism we have discovered in almost ev- Noctilio has only two premolars. Although it ery species in which this feature is present. is not clear which tooth has been lost, we Despite our consideration of the ``sometimes score Noctilio as having state 0 to leave open present'' and ``always present'' conditions as the possibility that this is a derived trait separate character states, we suspect that the shared with some phyllostomids. The primi- second distal accessory cusp is not uniformly tive condition for phyllostomids thus cannot present in any species (this suspicion is be reconstructed a priori. strengthened because Lim [1993] did not re- Straney (1980: characters K8, 10±12) ®rst port the presence of this cusp in Vampyro- used premolar loss and reduction in a phy- des). logenetic analysis. His character K8 dealt Character 58: p3 absent (0); or p3 great- with the presence or absence of this tooth. ly reduced, tooth peglike, cusps poorly or not Our observations for this character are iden- developed (1); or moderately reduced com- tical with his, although Straney (1980: 68) pared to p2 and p4, cusps present and well scored Noctilio as possessing p3. However, developed (2); or p3 crown height roughly he noted that this tooth was absent in this equivalent to p2 and p4 (3). As with the up- taxon in his text. To evaluate degree of re- per premolar dentition (see above), the three duction of p3, Straney (1980) used the height lower premolars of bats are typically num- of the cingula on p2 and p4; thus, our results bered ``p2-p3-p4'' (Slaughter, 1970), and we are somewhat different from his. If we con- follow this usage. Only two premolars (pre- sider our character states 3, 2, and 1, equiv- sumed to be p2 and p4) are present in des- alent to Straney's (1980) K10 (subequal), modontines, some phyllostomines (e.g., Mi- K11 (p3 lower than p2, 4 but above cingu- mon, Phyllostomus), Brachyphylla, phyllo- lum), and K12 (p3 lower than cingulum on nycterines, carolliines, and stenodermatines. p2,4), respectively, the only taxa that we dis- The hypothesis that p3 is absent in these taxa agree on are Phylloderma, Tonatia, and is supported by the observation that p3 is re- Trachops. Straney (1980) considered both duced in size relative to p2 and p4 in many Phylloderma and Trachops to have the de- phyllostomids. In these taxa (e.g., Loncho- rived state of K12 (p3 lower than cingula on rhina, Macrophyllum), this tooth is extreme- p2, 4), and found that different species of 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 79

Tonatia had different states for this character quently p2 and p4 are not in contact in this (derived for K10, K11, and K12). Our dis- genus. This tooth is absent in Noctilio. Thus agreement with Straney (1980) may be due the primitive state for phyllostomids cannot to the presence of a better developed lingual be reconstructed a priori. It seems likely that cingulum in these three genera. As noted lingual displacement can only occur in taxa above, we consider p3 to be moderately re- in which p3 is greatly reduced in size (e.g., duced in these three taxa. taxa scored 1 for character 58). We are un- Owen (1991: character 43) used a char- aware of any examples from other microchi- acter whose derived state was de®ned as the ropteran families where a tooth of normal loss of a lower premolar (reduction in num- size is displaced, but there are examples of ber from three to two premolars). Among his lingual displacement of greatly reduced teeth outgroups, Owen (1987) correctly scored (e.g., Myotis seabrai and M. lesueuri; Koop- only Macrotus as possessing three lower pre- man, 1994). Therefore, we score this char- molars. However, in his 1991 paper, Owen acter only in those taxa that have a greatly incorrectly scored the following taxa as hav- reduced p3; we scored genera in which p3 is ing three lower premolars (state 0 for char- either absent, slightly reduced, not reduced, acter 43): Carollia perspicillata, Ametrida, or (states 0, 2, and 3 of character 58) ``Ϫ'' Ardops, Enchisthenes, Koopmania, Phyllops, for this character. Sphaeronycteris, Stenoderma, and Sturnira Straney (1980: character K13) previously ludovici. used this feature as a binary character and Simmons (1996: character 18) described scored it in all phyllostomines in his analysis. an ordered character similar to this in her Although we have found that this tooth is analysis of relationships among Micronycter- sometimes lingually displaced in Loncho- is species. Our character state 0, p3 absent, rhina, our other observations agree with was not used by Simmons (1996) because all Straney's (1980) scoring of this character. Micronycteris retain p3. We have further However, we found that this feature is often modi®ed this character by comparing the variable, and have taken this into account in crown heights of the premolar teeth, rather constructing our character states. We dis- than the overall size of tooth. Consequently, agree with Smith (1972: 56) who stated that our character is not directly comparable to p3 is ``markedly reduced to a small peg-like hers. unicuspid tooth and is almost always exclud- Character 59: Postcanine teeth including ed (lingually), or nearly so, from the tooth- p3 aligned in row roughly parallel to long row'' in the genus Pteronotus. axis of mandible (0); or p3 sometimes lin- Character 60: W-shaped ectoloph present gually displaced from toothrow (1); or p3 al- on M1±M2 (0); or absent (1). A W-shaped ways lingually displaced from toothrow (2). ectoloph is prominent on the upper molar In most phyllostomids, all of the postcanine teeth of most microchiropterans (Slaughter, teeth, including p3, are aligned in a row that 1970; Koopman and MacIntyre, 1980; Koop- runs roughly parallel to the long axis of the man, 1994), and many phyllostomids retain jaw. However, p3 is always lingually dis- this characteristic (e.g., ®g 28B, E). The W- placed from the toothrow in Chrotopterus.In shaped ectoloph is best developed on M1 and Lonchorhina, Macrophyllum, and Trachops, M2, but is generally present, although often some specimens retain p3 in the toothrow altered, on M3 when this tooth is present (see (e.g., Lonchorhina: AMNH 184701; Macro- character 64). However, desmodontines, bra- phyllum: AMNH 94551; Trachops: AMNH chyphyllines, phyllonycterines, several glos- 266081), whereas in others p3 is lingually sophagines (e.g., Lichonycteris, Musonycter- displaced (e.g., Lonchorhina: AMNH is), Platalina, carollines, and stenodermati- 230122; Macrophyllum: AMNH 262424; nes do not have a W-shaped ectoloph on Trachops: AMNH 267442). In those speci- M1±M2 (®g. 28A, C±D, F). A W-shaped ec- mens with lingually displaced p3s, p2 and p4 toloph is present on M1±M2 of mormoopids are in contact to the labial side of this tooth. and Noctilio, suggesting that this is the prim- This tooth is not reduced in Mormoops. Pter- itive condition for phyllostomids. onotus retains p3 in the toothrow, conse- Although the presence of a W-shaped ec- 80 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 28. Occlusal view of M1-M3 in selected phyllostomids. A. Desmodus rotundus (AMNH 174303). Only M1 is present. B. Chrotopterus auritus (AMNH 267852). C. Brachyphylla cavernarum (AMNH 208181). D. Phyllonycteris poeyi (USNM 103542). E. Monophyllus redmani (AMNH 236662). F. Artibeus jamaicensis (AMNH 266331). Scale bar ϭ 1 mm. toloph has not been formally de®ned as a Phillips (1971). Slaughter (1970) viewed the character in a cladistic analysis, the phylo- primitive condition as retention of the W- genetic implications of the presence/absence shaped ectoloph and noted various condi- of the W-shaped ectoloph in phyllostomids tions which evolved from a dilambdodont have been discussed by Slaughter (1970) and condition. Phillips (1971) divided glosso- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 81 phagine (s.l.) bats into three groups based on veloped hypocone is present in the well-de- molar morphology, primarily on the presence veloped hypocone basin. In many phyllos- or absence of the W-shaped ectoloph, and tomines this cusp is incorporated into a crest noted that presence of a W-shaped ectoloph that runs into the hypocone region and may was more primitive. We agree with most of encircle it; thus the cusp itself is often indis- Phillips' (1971) observations, but have found tinct (®g. 28B). In stenodermatines, the cusp that Hylonycteris has an ectoloph, whereas is usually not incorporated into a crest and is Lichonycteris does not, contra Phillips more distinctly conical than in the phyllos- (1971). Finally, in considering the homolo- tomines (®g. 28F). Monophyllus appears to gies of the cusps of the molar teeth of Brach- be unique among phyllostomids in having yphylla, Grif®ths (1985: 546) commented what Phillips (1971) termed a hypoconal that ``In every stenodermatine genus but two, wing, a buttresslike structure that slopes to- the W-shaped ectoloph pattern was clearly wards the palate rather than connecting with evident . . . displaced laterally up against the the metacone (®g. 28E). In mormoopids and stylar cusps of the tooth ....''Weconsider Noctilio, a distinct hypocone is present, sug- a strongly W-shaped ectoloph pattern to be gesting that this is the primitive condition for absent in all stenodermatine genera. Al- phyllostomids. though there are crests that are displaced lat- Straney (1980: characters K19±23) scored erally against the stylar cusps, these do not hypocone development in what we interpret form a W shape (compare ®g. 28B and F). as two series of characters. The ®rst series Although we are aware that we may be (character K19 and K20) dealt with whether lumping several conditions together under this cusp was indistinct (the derived condi- our state 1 (i.e., the W-shaped ectoloph may tion of character K19) or distinct (the derived have been lost in different taxa due to the condition of character K20). Straney (1980) loss or movement of different crests or scored most taxa, but not all, for these two cusps), we de®ned the character in this man- characters. The second series (characters K21 ner due to the problems associated with to K23) dealt with the height of the hypo- drawing homologies among the cusps of the cone: very low (derived condition of char- upper molar teeth in taxa with unique molar acter K21), low (derived condition of char- morphologies. Often, homologies among acter K22), or relatively high (derived con- cusps on the upper molars are drawn in condition of character K23). Straney scored all sideration of the taxonomic af®nities of the taxa for this second series of characters. genus in question (e.g., Grif®ths' [1985] in- Owen (1987: character 14) described hy- terpretation of molar morphology in Brachy- pocone development on the ®rst upper molar phylla), an approach that is not suited for cla- in a ®ve-state ordered character. Marques- distic analysis. Aguiar (1994: character 26) also described Character 61: Hypocone basin distinct on the development of the hypocone on M1, de- M1, hypocone indistinct to well developed ®ning three states. Although the ®ne grade (0); or basin and cusp both indistinct or ab- distinction between a moderately developed sent (1); or hypocone wing present (2). The hypocone and a well-developed hypocone hypocone is conspicuously absent on the up- might be appropriate for a species level anal- per ®rst molar of many phyllostomids, in- ysis in some taxa (e.g., Marques-Aguiar, cluding desmodontines, brachyphyllines, 1994), the character states de®ned by Owen phyllonycterines, glossophagines, loncho- (1987) appear to subdivide a continuous se- phyllines, carollines, and some stenoderma- ries of hypocone development. Our obser- tines (e.g., Mesophylla, Platyrrhinus;®g vations of the presence or absence of this 28A, C±D). In most of these taxa, the hy- cusp con¯ict with several of Owen's (1987) pocone basin is poorly developed and the observations. Owen (1987) found that some tooth triangular, rather than square, in occlu- species of Chiroderma, Platyrrhinus, Stur- sal outline (some stenodermatines like Pla- nira, and Vampyrodes have at least a mod- tyrrhinus and Sturnira are exceptions, but in erately developed hypocone; our observa- these genera there is no cusp in the hypocone tions suggest that this cusp is absent in these region). In all other genera, a variably de- taxa. Owen (1991: character 30, 31, 32) 82 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 transformed his original multistate character elongate (®g 28A, C±F). In contrast, in Chro- into three separate binary characters. One of topterus, Trachops, and Vampyrum, an ex- Owen's (1991) characters (30) is equivalent ceptionally long metastyle is present at the to ours. Our observations of the presence/ab- distolabial border of these teeth (®g. 28B). sence of this cusp agree with those of Owen Although the metastylar region of M1±M2 is (1991). elongate in some glossophagines (e.g., Glos- Lim (1993: character 5) rede®ned this sophaga), the loph present on this region of character as a discrete state ordered (0 ← 1 the tooth (the postmetacrista) is approximate- → 2) character by scoring the presence of ly equal in length to the other lophs present either three or four cusps on the ®rst upper on the tooth. This situation is unlike the con- molar (the fourth cusp being the hypocone). dition seen in Chrotopterus, Trachops, and Our states and scoring are identical to his Vampyrum, where the postmetacrista is much with one exception. Lim (1993) described longer than any other loph on the tooth. Thus Sturnira as having an autapomorphic state in we do not consider these two conditions ho- which the protocone and hypocone are ob- mologous and score taxa like Glossophaga scured by a longitudinal groove. Although with state 0. The labial margin of M1±M2 in we observed a longitudinal trough in Stur- mormoopids and Noctilio does not have an nira between the buccal and lingual cusps, exceptionally elongate metastylar region, this trough does not obscure the protocone suggesting that this is the primitive state for and hypocone regions. Sturnira appears to phyllostomids. We scored this character on lack a hypocone; accordingly, we score this the morphology of M1 alone in taxa lacking taxon 1. an M2 (e.g., Desmodus). Phillips (1971) considered a hypocone ba- Straney (1980: character K26) used this sin to be present on the upper molars of An- character previously in a component analy- oura, Glossophaga, Leptonycteris, and Lon- sis, and our scoring is identical to his. chophylla. Although this region of the tooth Slaughter (1970) also described this feature may be present in these taxa, it is more poor- of the teeth in phyllostomids. ly developed in these genera than this region Character 63: M3 present (0); or some- is in phyllostomines and stenodermatines. times or always absent (1). Most bats have The outline of the ®rst upper molar in these three upper molars (``M1-M2-M3'') as do genera remains much more triangular in out- most other mammals. The third upper molar line than those of the stenodermatines and (M3) is present in most phyllostomids (®g. phyllostomines we examined. We also dis- 28B±E). In contrast, M3 is absent in des- agree with Slaughter (1970), who observed modontines, Leptonycteris, Lichonycteris, that all stenodermatines have a hypocone. and many stenodermatines (e.g., Centurio, Considerable disagreement exists concern- Ectophylla; ®g 28A, F). In both Artibeus and ing the presence and location of many cusps Chiroderma, M3 is almost always absent in (especially the paracone and metacone) on some species (e.g., A. lituratus, C. villosum), the upper ®rst molars of some taxa (e.g., Bra- whereas it is consistently present in others chyphylla, Erophylla, and Phyllonycteris). (e.g., A. obscurus, C. trinitatum). Conse- However most authors (e.g., Miller, 1907; quently, we scored these two taxa with both Slaughter, 1970; Koopman and MacIntyre, 0 and 1 in the matrix. In addition, we found 1980; Grif®ths, 1985) have considered the that M3 was present on both sides in one hypocone to be lacking in these genera. Our specimen each of Enchisthenes (AMNH observations support this interpretation, con- 233791; we examined 6 for this character), sequently we have scored these taxa 1 for and Mesophylla (AMNH 262540; we exam- this character. ined 10). We scored these taxa with state 1. Character 62: M1±M2 lacking extensive In Sphaeronycteris, we found that M3 was elongation in metastylar region (0); or with frequently absent from one side of the den- exceptionally long metastyle present (1). In tition (e.g., AMNH 24379), although all most phyllostomids, the metastylar region of specimens we examined had M3 present on the upper ®rst two molars is poorly to mod- at least one side. Consequently, we scored erately developed and is not exceptionally Sphaeronycteris with state 0 for this char- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 83 acter. Mormoopids and Noctilio have three Straney (1980: character K24) ®rst used upper molars, suggesting that presence of M3 this character in a component analysis. Al- is the primitive condition for phyllostomids. though our scoring and Straney's are identi- Owen (1987: character 15) ®rst used a cal, we have described this character differ- character to describe the development of M3. ently. Straney (1980) described his character In this character, Owen (1987) considered as the presence or absence of only the post- not only the presence or absence of this tooth metacrista on M3. Although we suspect that but its size (well developed with cusps pre- the premetacrista, metacone, and postmeta- sent 0; or poorly developed, small, and peg- crista have been lost, a view proposed by like 1). Our observations of the size and cusp Slaughter (1970), we have avoided erecting development of M3 suggest that there is no any character (except see character 61) that simple way to characterize tooth size and de- would force us to draw homologies among velopment in phyllostomids; these features the cusps on the upper molars. There is too appear to vary continuously. Subsequently much ongoing debate concerning the homol- Owen (1991: character 34) used a character ogy of the cusps on the upper molars in identical to ours. Our observations of the many genera (e.g., Brachyphylla, Phyllon- presence/absence of this tooth agree with ycteris, and Erophylla). those of Owen (1987, 1991), with the excep- Phillips (1971) indicated that the W- tion of our observation that this tooth is shaped ectoloph is absent from M3 in An- sometimes present in Artibeus jamaicensis, oura, Glossophaga, Lonchophylla, and Mon- Enchisthenes, and Mesophylla, and may be ophyllus, usually because one of the styles absent in Sphaeronycteris. Although Owen (e.g., mesostyle, metastyle) or cristae (e.g., (1991: 19) noted that ``the condition of both metacrista ϭ postmetacrista) are lacking. M3 and m3 was miscoded in the table [1987: However, our observations of M3 in these table 1], even though properly coded in the taxa indicate that the W-shaped lophs are still analyses,'' our observations do not differ distinctly visible on M3 in these taxa. Thus, greatly from Owen's (1987) table 1. our observations agree with Slaughter (1970: Although we have usually erected a 68), who noted that ``glossophagines have a unique character state for polymorphic taxa, well developed premetacrista and a short me- we have not done so here because we suspect tacrista.'' that polymorphism is more widespread than Character 65: m2 present (0); or absent we have reported, particularly with the very (1). As in the upper dentition, most phyllos- reduced teeth present in some stenodermati- tomids have three lower molars (designated nes. Thus, we have been conservative, pend- ``m1-m2-m3''). Most phyllostomids retain ing further evaluation of polymorphism in m2; however, this tooth is absent in Desmo- these taxa. dus and Diaemus. Mormoopids and Noctilio Character 64: Ectoloph on M3 W-shaped have three lower molars, suggesting that (0); or V-shaped (1). Among phyllostomids presence of m2 is the primitive condition for that have a W-shaped ectoloph on M1 and phyllostomids. M2 and a third upper molar (see characters This character has not been used in pre- 60 and 63), there are two distinct morphol- vious phylogenetic analyses, although nu- ogies of the M3 ectoloph. In glossophagines merous authors have described dental for- and lonchophyllines with a W-shaped ecto- mulae in phyllostomids (e.g., Miller, 1907; loph on M3, a virtually complete W-shape is Phillips, 1971; Owen, 1987; Koopman, present (®g. 28E). In contrast, in phyllostom- 1994). ines the ectoloph appears V-shaped on M3 Character 66: m3 present (0); or some- (®g. 28B). In mormoopids and Noctilio,a times or always absent (1). Most phyllostom- complete W-shaped ectoloph is present on ids retain m3. In contrast, m3 is absent in M3, suggesting that this is the primitive con- desmodontines, Leptonycteris, Lichonycteris, dition for phyllostomids. We scored taxa and many stenodermatines (e.g., Centurio, lacking both a W-shaped ectoloph and an M3 Ectophylla). In Chiroderma, Dermanura, (state 1 of character 60 and state 1 of char- and Sturnira, m3 is present in some species acter 63) ``Ϫ'' for this character. (e.g., C. doriae, D. glauca, S. lilium) and is 84 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 absent in others (e.g., C. villosum, D. cine- most bats, most phyllostomids have lower rea, S. thomasi). Consequently, we score molars that do not form a straight, continu- these taxa with both 0 and 1 for this char- ous, shearing ridge with the anterior cheek acter. In Pygoderma, a single m3 is present teeth. In contrast, desmodontines have later- on the left side in two individuals (AMNH ally compressed mandibular cheek teeth that 261761, AMNH 246408), one specimen has are incorporated into a continuous, straight both the left and right m3 present (AMNH (i.e., parallelling the long axis of the jaw) 234295), and the remaining individuals lack shearing ridge that includes the canines, pre- m3 entirely (a total of seven). We score Py- molars, and any additional molar teeth (pres- goderma with state 1. In Sphaeronycteris,we ent in Diphylla). Mormoopids and Noctilio found that one m3 was frequently absent have molars that do not form a straight, con- (e.g., AMNH 76251), although all specimens tinuous, shearing ridge, suggesting that this we examined had an m3 on at least one side. is the primitive condition for phyllostomids. Consequently we score Sphaeronycteris with This character has not been used in pre- state 0 for this character. Mormoopids and vious cladistic analyses, although many au- Noctilio have three lower molars, suggesting thors have described the morphology of the that presence of m3 is the primitive condition teeth in desmodontines (e.g., Miller, 1907; for phyllostomids. Slaughter 1970). Owen (1987: character 22) ®rst used a Character 68: Paraconid always present character to describe the development of m3. on m1 (0); or sometimes absent (1); or al- In this character Owen (1987) considered not ways absent (2). A paraconid is present on only the presence or absence of this tooth but m1 in most phyllostomids (®g. 29B, E). In its size (well developed with cusps present contrast, the paraconid is always absent from 0; or poorly developed, small, and peglike 1). m1 in Brachyphylla, phyllonycterines, and Our observations of the size and cusp devel- most stenodermatines, including some spe- opment of m3 suggest that there is no simple cies of Sturnira (e.g., S. erythromos; other way to characterize tooth size and develop- Sturnira species, like S. oporophilum, have a ment in phyllostomids; these features appear distinct paraconid; ®g. 29C±D, G±I). In some vary in a continuous fashion. Owen (1991: individuals of Platyrrhinus helleri (e.g., character 46) and Marques-Aguiar (1994: AMNH 263616; appears in six of 10 indi- character 29) both used the presence of the viduals examined for this character) a small third lower molar as a character. With the paraconid is occasionally present but is ab- exception of our observations of the vari- sent, or appears less frequently, in other spe- ability of this feature in Pygoderma and cies (e.g., P. dorsalis, where it was absent in Sphaeronycteris, we agree with Owen (1987, four of four specimens). This situation also 1991) and Marques-Aguiar (1994). occurs in Vampyrodes (e.g., AMNH 186381; Although absence of M3 (see character 63 appears in one of 10 individuals examined), above) and absence of m3 are clearly corre- and some species of Dermanura (e.g., D. ci- lated in some taxa, this is not true of all taxa. nerea AMNH 97075; appears in one of 10 For example, some species of Artibeus, Ari- individuals examined for this character). We teus, some species of Dermanura, Meso- scored Dermanura and Platyrrhinus with phylla, Vampyressa bidens, and Vampyrodes states 1 and 2 in the matrix. Due to dif®cul- lack M3 but retain m3. Accordingly, we ties in drawing homologies among the cusps score presence/absence of upper and lower of the lower molars in Rhinophylla and the third molars as separate characters. We also desmodontines, we scored these taxa ``?'' in score polymorphism in this character conser- the matrix for this character and characters vatively (see discussion in character 61). 69±70 (®g. 29A, F). The paraconid is present Character 67: m1 does not form high on m1 in mormoopids and Noctilio, suggest- shearing ridge, lingual aspect of tooth well ing that this is the primitive condition for developed (0); or m1 laterally compressed, phyllostomids. lingual aspect of tooth poorly developed, Lim (1993; character 9) ®rst introduced tooth incorporated into continuous shearing this character. However, Lim's (1993) origi- ridge with anterior cheek teeth (1). Like nal description did not draw any homologies 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 85

Fig. 29. Occlusal view of m1 in selected phyllostomids. A. Desmodus rotundus (AMNH 174303). B. Phyllostomus hastatus (AMNH 267905). C. Brachyphylla cavernarum (AMNH 208181). D. Phyl- lonycteris poeyi (AMNH 103542). E. Glossophaga soricina (AMNH 209354). F. Rhinophylla pumilio (AMNH 266192). G. Chiroderma villosum (AMNH 267191). H. Ardops nicholsi (AMNH 213954). I. Sphaeronycteris toxophyllum (AMNH 262637). Scale bar ϭ 1 mm. between cusps, describing the character derived condition of this character (three states as four prominent cusps (0), or three prominent cusps) in only four taxa: Chiro- prominent cusps (1). We ®nd substantial dis- derma, Ectophylla, Mesophylla, and Vam- agreement between our observations and pyressa. Our observations indicate that there those reported by Lim (1993), who found the are two prominent cusps in these taxa, an 86 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 enormous protoconid and a smaller hypoco- phyllonycterines, glossophagines, loncho- nid. In the remaining taxa Lim (1993) scored phyllines, Rhinophylla, and many stenoder- with the primitive condition (four prominent matines (e.g., Chiroderma, Mesophylla), a cusps), we observed from three to ®ve dis- diastema is sometimes or always present be- tinct cusps. We prefer to divide observed var- tween P3 and P4. In Mormoops a diastema iation in m1 cusp patterns into three separate separates these teeth, but in Pteronotus and characters (68±70), in order to account for Noctilio the upper third and fourth premolars this variation and putative cusp homologies. are in contact, suggesting that this is the Although Slaughter (1970) hypothesized primitive condition for phyllostomids. We that it is the lingual cusps that are lost in scored taxa in which P3 is absent (state 1 in Rhinophylla and desmodontines, much of character 56) ``Ϫ'' in the matrix. this argument rests on the relationships of Lim (1993: character 11) de®ned a single these taxa to other phyllostomids. Therefore, ordered (0 → 1 → 2) character based on di- we have chosen to score these taxa ``Ϫ'' until astemata in the lower jaw: ``No gaps between more detailed information is available con- mandibular cheek teeth (0); space between cerning the homologies of the cusps in these mandibular premolars (1); additional space taxa. between mandibular premolar and molar Character 69: Metaconid present on m1 (2).'' We apply this character to the maxil- (0); or absent (1). A metaconid is present on lary, rather than mandibular, tooth row to the lower m1 of most phyllostomids (®g. avoid problems stemming from the loss of 29B±C, E, I). In contrast, the metaconid is p3 in some phyllostomids (see character 58). absent from m1 in phyllonycterines and We have chosen to split Lim's (1993) origi- many stenodermatines (e.g., Ardops, Uro- nal character into two separate characters derma; ®g. 29D, G±H). In some species of (and one new character we developed after Sturnira (e.g., S. lilium) a metaconid is pres- including glossophagines in our analysis) to ent, whereas in others this cusp is absent preserve our ideas about the homology of the (e.g., S. bidens). A metaconid is present on various gaps. With few exceptions, our scor- m1 in mormoopids and Noctilio, suggesting ing of this character is identical to Lim's that this is the primitive condition for phyl- (1993). The only difference is the appearance lostomids. We scored desmodontines and of a diastema between the third and fourth Rhinophylla ``?'' for this character. upper premolars in Platyrrhinus and Uro- Character 70: Entoconid present on m1 derma, which does not occur in the lower (0); or absent (1). An entoconid is present tooth row. on the lower m1 of most phyllostomids (®g. Character 72: P4 and M1 in contact, no 29B±C, H±I). In contrast, the entoconid is diastema present (0); or diastema sometimes absent from m1 in phyllonycterines and or always present between P4 and M1 (1). many stenodermatines (e.g., Chiroderma, In most phyllostomids, the upper fourth pre- Mesophylla; ®g. 29D, G). In Sturnira, an en- molar and ®rst molar are not separated by a toconid is present in some species (e.g., S. diastema. In contrast, in phyllonycterines, tildae) and absent in others (e.g., S. erythry- lonchophyllines, most glossophagines (e.g., omos). An entoconid is present on m1 in Choeronycteris, Hylonycteris), Rhinophylla, mormoopids and Noctilio, suggesting that and some stenodermatines (e.g., Chiroderma this is the primitive condition for phyllos- and Mesophylla), a diastema is sometimes or tomids. We scored desmodontines and Rhin- always present between P4 and M1. In mor- ophylla ``?'' for this character. moopids and Noctilio, the upper fourth pre- Character 71: P3 and P4 in contact, no molar and ®rst molar are in contact, sug- diastema present (0); or diastema sometimes gesting that this is the primitive condition for or always present between P3 and P4 (1). In phyllostomids. most phyllostomids, the upper third and As noted above, we broke Lim's (1993: fourth premolars are in contact and there is character 11) original character, which had no gap (diastema) present between these been based on diastemata in the lower jaw, teeth. In contrast, in some phyllostomines into two separate characters. Our observa- (e.g., Micronycteris hirsuta, Vampyrum), tions differ little from Lim's (1993). The only 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 87 difference is the appearance of a P4±M1 di- and acromion processes (1). The distal tip of astema in all Vampyressa species included in the clavicle is attached to the coracoid pro- our analysis. These gaps do not occur con- cess of the scapula by ligaments in Desmo- sistently in the lower tooth row. We also dis- dus, Diphylla, Macrotus, Phyllostomus, Lep- agree with Phillips (1971) concerning this tonycteris, Choeronycteris, and Carollia character in Lonchophylla. Phillips (1971) (Vaughan, 1959; Strickler, 1978). There is no noted that a cingular style of P4 contacts the ligamentous connection to the acromion pro- parastyle of M1 in Lonchophylla. However, cess in these taxa. In contrast, the distal tip we have found that there is no contact be- of the clavicle is attached by ligaments to tween these two teeth in at least some indi- both the coracoid and acromion processes of viduals of Lonchophylla. the scapula in Glossophaga and Centurio Character 73: M1 and M2 in contact, no (Strickler, 1978). The distal tip of the clavicle diastema present (0); or diastema sometimes is attached to the coracoid only in mormoo- or always present between M1 and M2 (1). pids, but is attached to both the coracoid and In most phyllostomids, the upper ®rst and acromion in Noctilio (Strickler, 1978). The second molars are in contact and there is no primitive condition for phyllostomids cannot gap (diastema) present between these two be assessed a priori. teeth. However, many glossophagines (e.g., This feature has not been used previously Choeronycteris, Hylonycteris), Platalina, in a phylogenetic analysis. Rhinophylla, Ectophylla, Mesophylla, and all Character 75: M. occipitopollicalus with Vampyressa species, a diastema is sometimes single cranial muscle belly (0); or with a or always present between M1 and M2. In cranial and a distal muscle belly (1).M.oc- mormoopids and Noctilio, the upper ®rst and cipitopollicalus is a complex muscle system second molars are in contact, suggesting that that originates on the occiput of the skull and this is the primitive condition for phyllos- inserts into the metacarpal or ®rst phalanx of tomids. Desmodus lacks an M2 , consequent- the pollex (Strickler, 1978). In some species, ly we scored this character ``Ϫ'' in this tax- a second tendon may insert onto the ventral on. metacarpal of digit one. This muscle com- This character has not been used in pre- plex includes a variable number of muscle vious phylogenetic analyses of phyllosto- bellies, segments of tendon, and segments of mids, but was described by other authors elastic tissue (Strickler, 1978). M. occipito- (e.g., Phillips, 1971). pollicalus has a single cranial muscle belly in phyllostomines, Brachyphylla, Choero- POSTCRANIUM nycteris, Glossophaga, Leptonycteris, Car- ollia, Centurio, Sturnira, and Uroderma The following section includes characters (Vaughan, 1959; Strickler, 1978; Straney, of the postcranial musculoskeletal system ex- 1980; ®g. 30A±B). This cranial muscle belly cluding the hyoid region. Many of these is also present in Desmodus, Diphylla, Arti- characters are based on descriptions by beus, and Chiroderma; however, in these Vaughan (1959), Vaughan and Bateman taxa an additional muscle belly is present dis- (1970), Strickler (1978), Altenbach (1979), tal to the ®rst (Strickler, 1978; Straney, 1980; and Straney (1980). Taxa examined in these ®g. 30C). In these taxa, the two muscle bel- studies are listed in table 4. We have revised lies are separated by a small proximal tendon many of the character descriptions given by and a distal stretch of elastic tissue (Straney, Straney (1980: character numbers from ap- 1980; Strickler, 1978). Only the cranial belly pendix 1). Although we provide new data for is present in mormoopids; however, both cra- some osteological characters, we did not col- nial and distal muscle bellies are present in lect new observational data for any of the Noctilio (Strickler, 1978). Thus, the primitive myological characters drawn from prior stud- state for phyllostomids cannot be assessed a ies. priori. Character 74: Distal tip of clavicle at- Straney (1980: characters C1±2) devel- tached by ligaments to coracoid process of oped two binary characters whose derived scapula (0); or attached to both coracoid states are equivalent to our states 1 and 0, 88 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 30. Diagramatic representation of the general conditions of the occipitopollicalus muscle complex with a key below (redrawn from Strickler, 1978: ®g. 31). The solid line represents tendon, the dashed line is muscle, and the wavy line is elastic tissue. A. Condition in most phyllostomines, glossophagines, Carollia, and many stenodermatines. B. Condition in Phyllostomus and mormoopids. C. Condition in desmodontines, Artibeus, Chiroderma, and Noctilio. respectively. Our scoring is identical to Stra- nates from vertebral border of scapula only ney's (1980). (0); or from vertebral border of scapula and Character 76: M. occipitopollicalus has transverse scapular ligament (1). In bats, m. one connection to ventral ¯ight musculature deltoideus comprises three separate muscles: (0), or two attachments to ventral ¯ight mus- m. clavodeltoideus, m. acromiodeltoideus, culature (1). In most phyllostomids, a single and m. spinodeltoideus (Strickler, 1978). M. tendinous connection exists between the spinodeltoideus arises from the vertebral bor- proximal tendon of m. occipitopollicalus and der of the scapula in Choeronycteris, Glos- the ventral ¯ight musculature (Vaughan, sophaga, and Leptonycteris (Strickler, 1978). 1959; Strickler, 1978; Straney, 1980; ®g. However, m. spinodeltoideus has an enlarged 30A, C). The single exception is Phyllosto- origin in Desmodus, Diphylla, Macrotus, mus, which has two tendinous attachments to Phyllostomus, Carollia, and Centurio, where the ventral ¯ight musculature (Strickler, it arises from the vertebral border of the 1978; Straney, 1980; ®g. 30B). In mormoo- scapula and part of the transverse scapular pids, there is a double tendinous attachment ligament (Vaughan, 1959; Strickler, 1978). In to the ventral ¯ight musculature (Strickler, mormoopids, m. spinodeltoideus arises from 1978; Straney, 1980). Noctilio has a single the vertebral border of the scapula and the connection to the ventral ¯ight musculature transverse scapular ligament (Strickler, (Strickler, 1978; Straney, 1980). Due to this 1978). In Noctilio, m. spinodeltoideus arises distribution of character states, the primitive from the vertebral border of the scapula only state for phyllostomids cannot be assessed a (Strickler, 1978). Thus, the primitive condi- priori. tion for phyllostomids cannot be reconstruct- Straney (1980: characters C3±4) devel- ed a priori. oped two binary characters whose derived This feature has not been previously used conditions are equivalent to our states 1 and in a phylogenetic analysis. 0, respectively. Our scoring is identical to Character 78: Caput mediale of m. tri- Straney's (1980). ceps brachii inserts on both elbow sesamoid Character 77: M. spinodeltoideus origi- and caput laterale tendon (0); or on caput 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 89 laterale tendon only (1); or on elbow sesa- onotus parnellii, m. palmaris longus inserts moid only (2). In bats, as in most mammals, on digit II. In Pteronotus quadridens, P. dav- the m. triceps brachii is composed of three yi, and P. personatus the distal tendons of heads: caput longum, caput laterale, and ca- this muscle are small and dif®cult to trace put mediale (Strickler, 1978). In Diphylla, (Vaughan and Bateman, 1970); therefore, we Phyllostomus, and Carollia the caput medi- have scored Pteronotus with the state ale of m. triceps brachii inserts on the elbow Vaughan and Bateman (1970) described for sesamoid (Strickler, 1978). In contrast, the Pteronotus parnellii. Based on the distribu- caput mediale of m. triceps brachii inserts on tion of character states among outgroup taxa, the ventral edge of the tendon of the caput presence of an insertion of m. palmaris lon- laterale in Desmodus, Choeronycteris, Glos- gus on digit II appears to be the primitive sophaga, Leptonycteris, and Centurio condition for phyllostomids. (Strickler, 1978). In Mormoops, the caput Straney (1980: characters C5±7) based mediale of m. triceps brachii is absent three characters on his descriptions of the in- (Vaughan and Bateman, 1970, Strickler, sertion of m. palmaris longus: ``Palmaris lon- 1978), and we scored Mormoops ``Ϫ'' for gus with insertion on digits one and two;'' this character. In Pteronotus and Noctilio, the ``Palmaris longus with insertion on digits one caput mediale inserts on both the elbow ses- and three;'' and ``Palmaris longus with in- amoid and the ventral edge of the caput la- sertion on digits one, three, four, and ®ve.'' terale tendon (Vaughan and Bateman, 1970; However, Straney (1980) appears to have Strickler, 1978). This distribution of charac- been unaware of the earlier study by ter states suggests that the primitive condi- Vaughan and Bateman (1970), and did not tion for phyllostomids is the insertion of the discuss how his results compared to theirs. caput mediale of m. triceps brachii on the Although Straney (1980) described the dis- elbow sesamoid and the caput laterale ten- tribution of this character in many more don. phyllostomids than those examined by This character has not been used previ- Vaughan and Bateman (1970), his descrip- ously in a phylogenetic analysis. Vaughan tions frequently con¯ict with theirs, particu- (1959), who described the postcranial mus- larly the description of the insertion of m. culature of Macrotus, noted only that the ca- palmaris longus onto digit II. Straney (1980) put mediale of the m. triceps brachii inserted found that m. palmaris longus inserts on digit on the olecranon process of the ulna. This II in stenodermatines, Phyllostomus, and description is not detailed enough for us to Macrotus, in contrast to Vaughan's (1959) score Macrotus for this character. Vaughan and Vaughan and Bateman's (1970) reports and Bateman (1970) described the origin and that these taxa (or representatives of them, insertion of this muscle after examining all e.g., Sturnira) do not have an insertion on mormoopid species and 18 phyllostomid spe- digit II. Further, Straney (1980) found that cies, but did not describe the different inser- Glossophaga and mormoopids lack an insertions noted by Strickler (1978). tion on digit II, whereas Altenbach (1979) Character 79: M. palmaris longus inserts and Vaughan and Bateman (1970) reported on digit II (0); or does not insert on digit II that an insertion on digit II was present in (1). M. palmaris longus originates on the dis- these taxa. Because Straney's (1980) descrip- tal tip of the spinous process of the humerus, tions differ so greatly from the reports by divides into several tendons at the wrist, and Vaughan (1959), Vaughan and Bateman inserts on various digits and tendons in the (1970), and Altenbach (1979), we did not use manus (Vaughan and Bateman, 1970). This Straney's (1980) descriptions to score this muscle inserts on digit II in Desmodus, Er- character. ophylla, Glossophaga, and Artibeus (Alten- Character 80: M. palmaris longus does bach, 1979; Vaughan and Bateman, 1970). In not insert on digit III (0); or inserts on digit contrast, m. palmaris longus has no insertion III (1). M. palmaris longus inserts on digit on digit II in Macrotus, Phyllostomus, Car- III in Desmodus, most phyllostomines (e.g., ollia, and Sturnira (Vaughan, 1959; Vaughan Macrotus, Trachops), Erophylla, Anoura, and Bateman, 1970). In Mormoops and Pter- Glossophaga, Lonchophylla, and Carollia 90 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

(Vaughan, 1959; Vaughan and Bateman, V (1). M. palmaris longus does not insert on 1970; Altenbach, 1979; Straney, 1980). In digit V in Macrotus, Phyllostomus, Artibeus, contrast, there is no insertion on digit III in and Sturnira (Vaughan, 1959; Vaughan and Phyllostomus and stenodermatines (Vaughan Bateman, 1970). However, in Desmodus, Er- and Bateman, 1970; Straney, 1980). M. pal- ophylla, Glossophaga, and Carollia, m. palmaris longus does not insert on digit III in maris longus inserts on digit V (Vaughan and Mormoops, but does insert on this digit in Bateman, 1970; Altenbach, 1979). M. pal- Pteronotus parnellii (Vaughan and Bateman, maris longus does not insert on digit V in 1970). In Noctilio there is no insertion on Mormoops and Pteronotus parnellii digit III (Straney, 1980). This distribution of (Vaughan and Bateman, 1970), suggesting character states among the outgroup taxa that this is the primitive condition for phyl- suggests that the absence of an insertion on lostomids. digit III is the primitive condition for phyl- This character has not been used previ- lostomids. ously in a phylogenetic analysis. Straney Unlike Straney's (1980) descriptions of the (1980) did not describe the insertion of m. insertion of m. palmaris longus onto digit II palmaris longus on digit V. (see character 79), his descriptions of the in- Character 83: M. ¯exor digitorum pro- sertion of m. palmaris longus onto digit III fundus inserts on digit IV (0); or does not agree, for the most part, with those of pre- insert on digit IV (1). M. ¯exor digitorum vious authors. The only differences are found profundus, which originates on the distal tip in descriptions of this insertion in Macrotus of the spinous process of the humerus, di- and mormoopids. In Macrotus, Straney vides into multiple tendons at the carpus (1980) found no insertion on digit III, where- (Vaughan and Bateman, 1970). These ten- as Vaughan (1959) reported that an insertion dons insert on digits I and III in all phyllos- was present on this digit. Straney (1980) re- tomids (Vaughan, 1959; Vaughan and Bate- ported that m. palmaris longus inserted on man, 1970; Altenbach, 1979). M. ¯exor dig- digit III in all mormoopids, whereas Vaughan itorum profundus has an additional insertion and Bateman (1970) found that it did not in- on digit IV in Phyllostomus (Vaughan and sert on this digit in Mormoops. Although Bateman, 1970). In other phyllostomids, these disagreements do exist, we decided to there is no insertion on digit IV (Vaughan, utilize Straney's (1980: 53) text descriptions 1959; Vaughan and Bateman, 1970; Alten- to score this character in taxa that were not bach, 1979). Mormoops and Pteronotus have described by previous authors. an insertion on digit IV, in addition to the Character 81: M. palmaris longus does insertions on digits I, III, and V (Vaughan not insert on digit IV (0); or inserts on digit and Bateman, 1970). Presence of an insertion IV (1). There is no insertion of m. palmaris on digit IV appears to be the primitive con- longus on digit IV in Desmodus, Macrotus, dition for phyllostomids. Phyllostomus, Carollia, Artibeus, and Stur- This is the ®rst use of this character in a nira (Vaughan, 1959; Vaughan and Bateman, phylogenetic analysis, although Straney 1970; Altenbach, 1979). In contrast, m. pal- (1980: characters C8±9) based two characters maris longus inserts on digit IV in Erophylla on the insertion of this muscle into digits I, and Glossophaga (Vaughan and Bateman, II, and III. Straney (1980), unlike other pre- 1970). Both Mormoops and Pteronotus par- vious authors, did not ®nd that all phyllos- nellii lack an insertion of m. palmaris longus tomids had insertions on digits I and III. In- on digit IV (Vaughan and Bateman, 1970), stead, Straney (1980: 53) reported that four suggesting that this is the primitive condition taxa, Lonchorhina, Macrophyllum, Macro- for phyllostomids. tus, and the outgroup taxon Noctilio, had in- This character has not been used previ- sertions on digits I and II. Because Straney's ously in a phylogenetic analysis. Straney (1980) descriptions differ so greatly from the (1980) did not describe the insertion of m. reports by Vaughan (1959), Vaughan and palmaris longus on digit IV. Bateman (1970), and Altenbach (1979), we Character 82: M. palmaris longus does did not use Straney's (1980) descriptions to not insert on digit V (1); or inserts on digit score this character. 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 91

Character 84: Third metacarpal longer lostomids, the ®rst (proximal) phalanx of than fourth or ®fth (0); or third and fourth digit III is noticeably shorter than the second metacarpals subequal in length, both longer phalanx. In contrast, the ®rst and second pha- than ®fth (1); or fourth metacarpal longest langes are subequal in Micronycteris hirsuta, (2); or fourth and ®fth metacarpals subequal M. megalotis, M. minuta, Tonatia, and phyl- in length, both longer than third (3); or ®fth lonycterines. In mormoopids and Noctilio, metacarpal longest (4); or third and ®fth the ®rst phalanx is shorter than the second, metacarpals subequal in length, both longer suggesting that this is the primitive condition than fourth (5); or third, fourth, and ®fth for phyllostomids. metacarpals all subequal in length (6). In Simmons (1996: character 9) used this Lonchorhina, Micronycteris nicefori, Phyl- character in her study of relationships among lostomus, glossophagines, lonchophyllines, species of Micronycteris; our character states Ametrida, and Centurio, the third metacarpal and scoring are identical to hers. is the longest of metacarpals three, four, and Character 86: First phalanx of digit IV of ®ve. In Macrophyllum, Mimon crenulatum, wing shorter than second phalanx (0); or Brachyphylla, phyllonycterines, Carollia, subequal to second phalanx (1); or longer and many stenodermatines (e.g., Ariteus, than second phalanx (2). In most phyllos- Uroderma) the third and ®fth metacarpals are tomids, the ®rst (proximal) phalanx of digit subequal in length and both are always lon- IV is noticeably shorter than the second pha- ger than the fourth metacarpal. In contrast, lanx. In contrast, the ®rst and second phalan- the third and fourth metacarpals are subequal ges are subequal in many phyllostomines in length and both are always longer than the (e.g., Chrotopterus, Mimon bennettii), car- ®fth metacarpal in desmodontines and Mi- olliines, Ectophylla, Mesophylla, and Vam- cronycteris brachyotis. Finally, in most phyl- pyressa pusilla. The ®rst phalanx is distinctly lostomines, Rhinophylla, and several sten- longer than the second in some phyllosto- odermatines (e.g., Sturnira, Phyllops) the mines (e.g., Micronycteris hirsuta, Macro- ®fth metacarpal is the longest. In only one tus), Erophylla, Phyllops, and Centurio. Tax- taxon, Enchisthenes, did we observe the onomic polymorphism occurs in three gen- third, fourth, and ®fth to all be subequal in era. In Tonatia, some species (e.g., T. schul- length. In mormoopids, the third metacarpal zi) have the ®rst and second phalanges is longer than the fourth and ®fth, while in subequal, whereas others (e.g., T. saurophi- Noctilio, the fourth metacarpal is the longest. la) have a ®rst phalanx that is longer than Thus, the primitive condition for phyllosto- the second. We scored Tonatia with states 1 mids cannot be determined a priori. and 2 in the matrix. In Chiroderma and Pla- Simmons (1996: character 8) scored the tyrrhinus, some species have a ®rst phalanx metacarpal formula of species of Micronyc- shorter than the second (e.g., C. salvini, P. teris. We had originally attempted to code helleri), whereas others have the ®rst and this character in a similar fashion to that de- second phalanges subequal (e.g., C. villosum, veloped by Simmons (1996; the complete P. infuscus). We scored Chiroderma and Pla- metacarpal formula, e.g., 3 Ͼ 4 Ͼ 5); how- tyrrhinus with states 0 and 1 in the matrix. ever, due to the number of unique metacarpal In Mormoops, the ®rst and second phalanges formulas we discovered (more than nine), we are subequal in length. In Pteronotus and found it more useful to express this diversity Noctilio, the ®rst phalanx is shorter than the by creating a character for the longest meta- second. This distribution of character states carpal(s). We cannot create an additional suggests that a ®rst phalanx that is shorter character for the shortest metacarpal because than the second is the primitive condition for these characters would not be independent in phyllostomids. taxa like Brachyphylla, which has two meta- Simmons (1996: character 10) used this carpals that are equal in length and longer character in her study of relationships among than a third. species of Micronycteris. Although our char- Character 85: First phalanx of digit III of acter states and scoring are identical to hers, wing shorter than second phalanx (0); or we have not ordered the character as she did ®rst and second subequal (1). In most phyl- (0 ↔ 1 ↔ 2). 92 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Character 87: Calcar present, equal to or sent'' that was not used previously by Stra- longer than foot (0); or shorter than foot (1); ney (1980) or Simmons (1996). We con- or vestigial or absent (2). The length of the ®rmed these observations with measure- calcar is equal to or greater than the length ments on pickled specimens where neces- of the foot (including claws) in most phyl- sary. lostomines. In contrast, the calcar is notice- Character 88: Posterior edge of plagio- ably shorter than the foot in Diphylla, several patagium inserts onto lower leg or ankle re- phyllostomines (e.g., Micronycteris nicefori, gion (0); or onto calcar (1); or onto lateral Phylloderma), Erophylla, glossophagines, surface of ®rst metatarsal (2); or onto dorsal lonchophyllines, carolliines, and most sten- surface of ®rst metatarsal (3). The posterior odermatines (e.g., Artibeus, Uroderma). The edge of the plagiopatagium inserts at, or calcar is either vestigial or entirely absent in proximal to, the ankle region in desmodon- Desmodus, Diaemus, Brachyphylla, Phyllo- tines, many phyllostomines (e.g., Macro- nycteris, and Sturnira. In mormoopids and phyllum, Macrotus), Brachyphylla, phyllon- Noctilio, the calcar is equal to or longer than ycterines, glossophagines, lonchophyllines, the foot, suggesting that this is the primitive Carollia, and Sturnira. A unique condition state for phyllostomids. among phyllostomids occurs in Lonchorhina, Straney (1980: G6±8) developed three bi- where the plagiopatagium is attached to the nary characters based on the length of the proximal part of the calcar creating a small calcar: ``Calcar longer than foot,'' ``Calcar ``pocket'' between the plagiopatagium and subequal to foot,'' and ``Calcar shorter than uropatagium. The posterior edge of the pla- foot.'' We found that Straney's (1980) second giopatagium inserts onto the lateral metatar- character (G7) included what we consider to sal in Micronycteris hirsuta, Phylloderma, be two distinct conditions. Straney (1980) Phyllostomus, Rhinophylla, and stenoderma- described the calcar as subequal to the foot tines (except Sturnira). The plagiopatagium in Chrotopterus, Macrotus, some species of is attached to the dorsal surface of the ®rst Micronycteris, Phylloderma, some species of metatarsal in Chrotopterus, Tonatia, and Phyllostomus, some species of Tonatia, Vampyrum. In mormoopids, the plagiopata- Trachops, Vampyrum, Choeronycteris, Pla- gium is attached to the proximal part of the talina, and unidenti®ed stenodermatines. Al- calcar creating a small ``pocket'' between the though it may be reasonable to describe the calcar as subequal to the foot in Choeron- plagiopatagium and the uropatagium; how- ycteris and Platalina, in all glossophagines ever, the plagiopatagium is attached proximal that we examined, the calcar was consistently to the ankle on the lower leg in Noctilio. shorter than the foot (the same is true of the Thus, the primitive condition for phyllostom- stenodermatines we examined). The calcar in ids cannot be assessed a priori. many phyllostomines is apparently subequal Straney (1980: characters G9±13) scored to the foot; however, in several taxa the cal- attachment site of the plagiopatagium in a car is clearly longer than the foot (e.g., Mac- series of binary characters. We have sub- rophyllum), and in at least some species the sumed Straney's (1980) characters G9 calcar varies from equal to slightly longer (``wing attaches to lateral side of ankle'') and than the foot (e.g., Micronycteris brachyotis; G13 (``wing attaches on lower leg'') into a Simmons, 1996). As a result, we recognize single state (our state 0) because we were ``calcar equal to or longer than foot'' as a unable to divide the continuous range of var- single character state. Due to our revision of iation of ankle/lower leg attachments that we character states, our scoring of this character observed into these discrete states. In addi- differs from Straney's (1980) for those taxa tion to changing the character coding, we did that he considered to have a calcar subequal not observe the attachment of the plagiopa- to the foot. Our observations agree with tagium to the calcar (scored as a vento-me- those of Simmons (1996: 11) who used this dial ankle attachment of the plagiopatagium character in an analysis of relationships in Straney's [1980] matrix) in Macrotus and among species of Micronycteris. We have Mimon bennettii, and found that Tonatia added a new state ``calcar vestigial or ab- shares the same condition seen in Chrotop- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 93 terus and Vampyrum, an attachment to the gines, lonchophyllines, and Carollia, the tail dorsal surface of the ®rst metatarsal. is of moderate length and is shorter than the Owen (1987: character 6) also used this hind legs. A very long tail, approximately character, recognizing four states: attachment equal in length to the hind legs, is present in to the tibial region 0, attachment to the tarsal Lonchorhina, Macrophyllum, and Macrotus. region 1, attachment to the metatarsal region The tail is effectively absent in desmodonti- 2, and attachment to the metatarsal-phalan- nes, Vampyrum, Brachyphylla, Leptonycter- geal joint 3 (ordered 0 ↔ 1 ↔ 2 ↔ 3). We is, Rhinophylla, and stenodermatines. Taxo- have combined his ®rst two states into our nomic polymorphism occurs in Anoura state 0, and the latter two conditions into our where some species have a medium length state 2, because of our ®nding that these con- tail (e.g., A. caudifera), whereas others lack ditions were variable within genera, or our a tail (e.g., A. geoffroyi). We scored Anoura inability to separate the variation we ob- with states 0 and 2 in the matrix. Examina- served into these discrete states. However, tion of caudal vertebrae reveals that one or our observations largely agree with his, with two small vertebrae may extend beyond the the exception of Sturnira nana. Owen (1987) ischia in some individuals of these taxa. For observed that the plagiopatagium in S. nana example, some specimens of Anoura geof- is attached to the metatarsal region. However, froyi (e.g., USNM 49356) lack postischial we found that it inserts at the ankle, as in vertebrae, whereas others have two vestigial other Sturnira species. Owen (1987) also re- postischial vertebrae (e.g., USNM 385799). ported that in Chiroderma doriae, the attach- As originally noted by Miller (1907), all ment of the plagiopatagium is to the ankle specimens of Brachyphylla that we examined region rather than the metatarsals as is the had three long, cylindrical postischial verte- case in other Chiroderma. Unfortunately, we brae similar to those found in all taxa with were unable to examine specimens of C. do- external tails; however, these vertebrae do riae and therefore score Chiroderma with not extend into the uropatagium. Because the state 2 in the matrix. Owen (1991: characters condition in Brachyphylla externally appears 14, 15, 16) used additive binary coding, but identical to the condition in other taxa lack- both his characters and scoring are equiva- ing tails, we scored this genus with state 2 lent to his earlier work. in the matrix. In mormoopids and Noctilio, Marques-Aguiar (1994: character 7) also the tail is of moderate length, suggesting that used this character in her analysis of rela- this is the primitive condition for phyllos- tionships among large-bodied Artibeus spe- tomids. cies. Marques-Aguiar (1994) recognized Straney (1980: G1±3) developed three bi- three states: plagiopatagium attaches at side nary characters devoted to tail morphology: of foot, attaches at ankle, or attaches at base ``Tail present,'' ``Tail short,'' and ``Tail to of toes. We combined her state 0 (side of edge of uropatagium,'' whose derived con- foot) with 1 (base of toes), because both ditions correspond to our character states. states are often present within genera. Our Our scoring agrees with that of Straney observations agree with those of Marques- (1980) with the exception of Rhinophylla. Aguiar (1994) with one exception. Marques- Straney (1980) scored all carolliines as pos- Aguiar (1994) reported that the plagiopata- sessing a medium-length tail; however, an gium attaches to ankle in Enchisthenes hartii, externally visible tail is absent in Rhinophyl- whereas we observed (as did Owen, 1987) la. We have chosen to de®ne tail length in that in this species the attachment is to the relation to the hind legs rather than the uro- metatarsals. patagium (as Straney [1980] did) because the Character 89: Tail of medium length, ex- uropatagium is more variable in length. tends into uropatagium but is shorter than hind legs (0); or tail long, approximately HYOID APPARATUS equal to length of hind legs (1); or tail effectively absent, caudal vertebrae do not ex- The following characters are based on de- tend into uropatagium (2). In most phyllos- scriptions by Sprague (1943) and Grif®ths tomines, phyllonycterines, most glossopha- (1982, 1983b) for taxa appearing in table 4. 94 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Because Grif®ths (1982) repeats earlier ney's (1980) character E10, which was based (Grif®ths, 1978) observations for Macrotus, on Sprague's (1943) description of the entire Glossophaga, Monophyllus, Artibeus, and anterior portion of the m. mylohyoideus as Phyllops, we cite only the later paper. We aponeurotic in Phyllonycteris and Glosso- have revised many of Grif®ths' (1982) char- phaga (a condition that is not equivalent to acter descriptions, but have not collected new having two muscle masses separated by a observational data. We follow Grif®ths' ¯eshy aponeurosis). Grif®ths (1982) ob- (1982, 1983b) terminology for osteological served that the mylohyoid was sometimes and myological features, which is identical thin enough to reveal the geniohyoid deep to to that used by Sprague (1943) with three it, but that this condition was variable within exceptions: (1) Grif®ths (1982, 1983b) iden- glossophagine species. Morphology of m. ti®ed the element termed the hypohyal by mylohyoideus was not included in Grif®ths' Sprague (1943) as the ceratohyal; (2) Grif- (1982) character analysis, or in that of Gi- ®ths (1982, 1983b) identi®ed the element menez et al. (1996). termed the ceratohyal by Sprague (1943) as Character 91: Medial ®bers of m. ster- the epihyal; and (3) Grif®ths (1982, 1983b) nohyoideus originate from medial manubri- identi®ed the muscle termed m. constrictor um (0); or from mesosternum (1); or from pharyngeus inferior by Sprague (1943) as m. xiphoid process of sternum (2). The medial cricopharyngeus. Character numbers from ®bers of m. sternohyoideus originate from Straney (1980) are from appendix 1; those the medial manubrium in phyllostomines, from Grif®ths (1982) are from table 1; and Carollia, and stenodermatines (including those from Gimenez et al. (1996) are from Sturnira; Sprague, 1943; Grif®ths, 1982). table 4. Straney (1980) based his descriptions The origin is shifted posteriorly to the me- on the work of Sprague (1943). sosternum in Desmodus, Brachyphylla, and Character 90: M. mylohyoideus undivid- phyllonycterines (Grif®ths, 1982). In glos- ed (0); or partly divided into anterior and sophagines and lonchophyllines, the medial posterior parts by a ¯eshy aponeurosis (1); ®bers of m. sternohyoideus originate entirely or with a pronounced break, clearly divided from the xiphoid process (Grif®ths, 1982). into distinct anterior and posterior parts (2). Among the outgroups, the manubrium is the M. mylohyoideus consists of a single, undi- origin of the medial ®bers of m. sternohy- vided sheet of muscle in Desmodus, glosso- oideus in Pteronotus (Sprague, 1943). In phagines, lonchophyllines, and Carollia Noctilio, m. sternohyoideus is not clearly dif- (Wille, 1954; Grif®ths, 1982; ®g. 31A). Stur- ferentiated into groups of medial and lateral nira also has a single muscle belly (Sprague, muscle ®bers. Instead, the single origin of all 1943). In Brachyphylla and phyllonycterines, ®bers in this taxon is the manubrium m. mylohyoideus is partly divided into an- (Sprague, 1943). Nevertheless, the homologs terior and posterior parts by a ¯eshy aponeu- of the medial and lateral ®bers found in other rosis (Grif®ths, 1982). In contrast, m. mylo- phyllostomids are presumably present in hyoideus exhibits a pronounced break and is Noctilio, thus we have scored Noctilio for clearly divided into distinct anterior and pos- this character and character 92. The distri- terior parts in phyllostomines and most sten- bution of character states in the outgroups odermatines (Wille, 1954; Grif®ths, 1982; suggests that a manubrial origin is the prim- ®g. 31C). M. mylohyoideus is undivided in itive condition for phyllostomids. Pteronotus and Noctilio (Sprague, 1943), Grif®ths (1982: characters 1±2) scored suggesting that this is the primitive condition variation in the origin of m. sternohyoideus for phyllostomids. as two binary characters: ``posterior shift of Straney (1980: characters E9±11) used sternohyoid origin,'' and ``xiphoid origin of three binary characters to describe variation sternohyoid.'' Gimenez et al. (1996) recently in the morphology of m. mylohyoideus. Our used this character and apparently followed character states 0 and 2 correspond to the Smith and Hood (1984) in regarding Grif- derived conditions of Straney's (1980) char- ®ths' (1982) characters 1 and 2 as states of acters E9 and E11, respectively. We did not the same character. Gimenez et al.'s (1996: include the information contained in Stra- character 17) single multistate character is 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 95

Fig. 31. A. Super®cial and B. deep views of the hyoid musculature of Glossophaga soricina. C. Super®cial and D. deep views of the hyoid musculature of Vampyressa pusilla (redrawn from Grif®ths, 1982: ®gs. 3, 4, 18, 19). identical to ours in construction and coding. 92) which separately describe the origin of Grif®ths' (1982) characters (which are sub- the two sets of ®bers. sumed under states 1 and 2 above) did not Character 92: Lateral ®bers of m. ster- account for differences in the origins of me- nohyoideus originate from manubrium (0); dial versus lateral ®bers of m. sternohyoide- or originate from manubrium and clavicle us. Accordingly, we chose to score variation (1); or originate from clavicle and ®rst rib in the origin of m. sternohyoideus as two (2); or originate from xiphoid process (3). multistate characters (our characters 91 and The lateral ®bers of m. sternohyoideus orig- 96 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 inate from the manubrium in Sturnira derived condition of their character (``con- (Sprague, 1943). These ®bers originate from nection of sternohyoideus muscle to hyoid the lateral manubrium and clavicle in phyl- bone'') as ``dissociated.'' Straney (1980: lostomines, Erophylla, Carollia, and sten- characters E19±20) erected two binary char- odermatines (except Sturnira; Grif®ths, acters whose derived conditions are equiva- 1982). The lateral ®bers of m. sternohyoi- lent to our character states: ``Sternohyoid in- deus originate from the clavicle and the prox- sertion on basihyal,'' and ``Sternohyoid in- imal head of the ®rst rib in Desmodus, Bra- sertion on raphe ventral [to] basihyal.'' chyphylla, and Phyllonycteris (Grif®ths, Although our scoring of this character 1982). In glossophagines and lonchophyl- agrees with the work of Grif®ths (1982) and lines, the lateral ®bers of m. sternohyoideus Giminez et al. (1996), Straney (1980) scored originate from the xiphoid process (Grif®ths, Desmodus and glossophagines as possessing 1982). Sprague (1943) noted that the lateral an insertion on the basihyal element, while ®bers of m. sternohyoideus originate from he scored all other taxa for which data was the manubrium in Pteronotus and Noctilio; available as possessing an insertion on the thus, a manubrial origin appears to be prim- basihyal raphe. In the text, however, Straney itive for phyllostomids. (1980: 32) noted that Desmodus was unique Character 93: M. sternohyoideus inserts among noctilionoid bats in possessing the in- via tendon on basihyal (0); or via raphe into sertion on the basihyal. Straney's (1980) de- the ®bers of m. hyoglossus and m. genio- scription agrees with Sprague's (1943) dis- glossus (1). M. sternohyoideus inserts via cussion of this character. Sprague (1943) de- tendon to the basihyal in Desmodus, phyl- scribed an insertion on the basihyal raphe for lostomines, Brachyphylla, phyllonycterines, Phyllostomus, Phyllonycteris, Glossophaga, Carollia, and stenodermatines (Grif®ths, Carollia, Artibeus, Sturnira, Pteronotus, and 1982). In contrast, this muscle inserts via a Noctilio. Desmodus possessed an insertion raphe into the ®bers of m. hyoglossus and m. on the basihyal by means of a tendon genioglossus in glossophagines and loncho- (Sprague, 1943). Although Grif®ths' (1982) phyllines (Grif®ths, 1982). Typically, there is examination of the hyoid apparatus demon- no muscular connection to the hyoid appa- strated that Sprague's (1943) description of ratus in these taxa. A remnant of a basihyal an insertion in the basihyal raphe is inaccu- tendon was found by Grif®ths (1982) in one rate for most taxa, no additional data are individual each of Lionycteris spurelli and available for Pteronotus, Noctilio,orSturni- Lonchophylla robusta, but the principal in- ra. Because it seems unlikely that Sprague's sertion was identical to that of other loncho- (1943) description of this character is correct, phyllines. Accordingly, we scored these taxa we have chosen to score these taxa ``?'' in with state 1 in the matrix. Although Sprague the matrix. (1943) reported the insertion of m. sterno- Wille (1954) reported that this muscle hyoideus in Pteronotus, Noctilio, and Stur- originates on the xiphoid process of the ster- nira, these three taxa are scored ``?'' in the num and inserts into the tongue in glosso- matrix (see discussion below). Thus, the phagines and lonchophyllines. As Grif®ths primitive condition for phyllostomids cannot (1978) noted, Wille (1954) seemed unaware be reconstructed a priori. that m. hyoglossus is greatly elongated in Straney (1980), Grif®ths (1982), and Gi- glossophagines and lonchophyllines. The menez et al. (1996) previously used this muscle segment that Wille (1954) considered character. Our character states are equivalent part of the sternohyoideus is actually part of to those recognized by these authors, al- m. hyoglossus (Grif®ths, 1978). though Grif®ths (1982) and Gimenez et al. Character 94: Part of m. ceratohyoideus (1996) described what they believed to be the inserts on ceratohyal (0); or m. ceratohyoi- derived condition (our state 1) differently. deus does not insert on ceratohyal (1). Part Grif®ths (1982: character 3) described the of m. ceratohyoideus inserts on the cerato- derived condition as ``loss of sternohyoid's hyal element in Micronycteris nicefori, Bra- connection to hyoid bone,'' whereas Gime- chyphylla, phyllonycterines, glossophagines, nez et al. (1996: character 18) described the lonchophyllines, Carollia, and stenoderma- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 97 tines (including Sturnira; Sprague, 1943; direct connection of m. hyoglossus to the Grif®ths, 1982). In contrast, this muscle does bones of the hyoid apparatus in these taxa. not insert on the ceratohyal in Desmodus, Although Sprague (1943) reported the origin Macrotus, and Phyllostomus (Grif®ths, of m. hyoglossus in Pteronotus, Noctilio, and 1982). Part of m. ceratohyoideus inserts on Sturnira, these three taxa are scored ``?'' in the ceratohyal element in Pteronotus the matrix (see discussion below). The prim- (Sprague, 1943; Grif®ths, 1983b) and Noc- itive condition for phyllostomids cannot be tilio (Sprague, 1943), suggesting that this reconstructed a priori. condition is primitive for phyllostomids. Straney (1980: characters E21±22) used Straney (1980), Grif®ths (1982), and Gi- origin of m. hyoglossus as two binary char- menez et al. (1996) did not include any characters, whose derived conditions are identical acters based on m. ceratohyoideus in their to our character states. However, Straney analyses, although Grif®ths (1982) clearly (1980) scored the raphe origin (our character described the variation that we have scored state 1) as occurring in phyllostomines, brain our characters 94 and 95. Although chyphyllines, carolliines, stenodermatines, Sprague (1943) described the attachment of Pteronotus, and Noctilio. Straney (1980) re- m. ceratohyoideus to the ceratohyal in both ported an origin from the basihyal (our state Pteronotus and Noctilio, he referred to this 0) in glossophagines and desmodontines. connection as an origin, and noted that the Based on Sprague's (1943) descriptions, only insertion of this muscle in these taxa was on Straney's coding of Noctilio, which Sprague the thyrohyal. described as having an origin from the ba- Character 95: M. ceratohyoideus does sihyal and thyrohyal, was inaccurate. Al- not insert on stylohyal (0); or part of m. cer- though Grif®ths' (1982) examination of the atohyoideus inserts on stylohyal (1). M. cer- hyoid apparatus demonstrated that Sprague's atohyoideus does not insert on the stylohyal (1943) description of the origin of m. hyo- element in Brachyphylla, phyllonycterines, glossus from the basihyal raphe was inac- many glossophagines (e.g., Choeroniscus, curate for most taxa, no additional data are Choeronycteris), lonchophyllines, Carollia, available for Pteronotus, Noctilio,orSturni- and some stenodermatines (e.g., Artibeus, ra. Because it seems unlikely that Sprague's Uroderma; Grif®ths, 1982; Sturnira; (1943) description of this character is correct, Sprague, 1943). In contrast, part of m. cera- we have chosen to score these taxa ``?'' in tohyoideus inserts on the stylohyal in Des- the matrix. Our scoring of this character was modus, phyllostomines, some glossophagines identical to that of Grif®ths (1982: character (e.g., Glossophaga, Leptonycteris), Platyr- 4), although he described the derived condi- rhinus, and Vampyressa pusilla (Grif®ths, tion (our state 1) as ``hyoglossus elongated 1982). M. ceratohyoideus does not insert on and loses connection to hyoid bone.'' Our the stylohyal element in Pteronotus (Grif- character states are also equivalent to those ®ths, 1983b) and Noctilio (Sprague, 1943), used by Gimenez et al. (1996: character 19), suggesting that this is the primitive condition who described the derived condition in terms for phyllostomids. similar to those used by Grif®ths (1982). Character 96: M. hyoglossus originates Wille (1954: 317) reported that this muscle via tendon from basihyal bone (0); or from extends ``from basihyal to basihyal raphe in raphe which forms insertion of m. sternohy- geniohyoideus'' in glossophagines and lon- oideus (1). M. hyoglossus originates via ten- chophyllines. Given the observations of Grif- don from the basihyal bone in Desmodus, ®ths (1982), this description appears to be phyllostomines, Brachyphylla, phyllonycter- incorrect. ines, Carollia, and stenodermatines (Grif- Character 97: M. geniohyoideus has sin- ®ths, 1982). In contrast, m. hyoglossus orig- gle insertion via tendon to basihyal or basi- inates from the raphe, which forms the in- hyal raphe (0); or muscle splits near inser- sertion of m. sternohyoideus (the former ba- tion, deep ®bers insert directly on anterior sihyal raphe, now disconnected from the surface of basihyal, super®cial ®bers insert basihyal bone) in glossophagines and lon- in association with m. hyoglossus and m. chophyllines (Grif®ths, 1982). There is no sternohyoideus (1). M. geniohyoideus has a 98 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 single insertion via tendon to the basihyal Giminez et al. (1996: character 20), although bone in Desmodus, phyllostomines, Brachy- both described the derived condition (our phylla, phyllonycterines, Carollia, and sten- state 1) as ``double insertion of geniohyoid.'' odermatines (Grif®ths, 1982). In contrast, m. Wille (1954) also discussed the morphology geniohyoideus splits as it passes posteriorly of m. geniohyoideus in glossophagine and in lonchophyllines and glossophagines (Grif- lonchophylline bats, describing an anterior ®ths, 1982). The deep ®bers of m. geniohy- and posterior part to this muscle. For the an- oideus insert directly on the anterior surface terior part of the muscle, Wille (1954) only of the basihyal bone, while the super®cial ®- noted an insertion into the basihyal raphe; he bers form an insertion in association with m. was apparently unaware that there was a hyoglossus and m. sternohyoideus (Grif®ths, deep insertion of this muscle to the basihyal 1982). Although Sprague (1943) reported the bone. origin of m. geniohyoideus in Pteronotus, Character 98: Super®cial ®bers of m. ge- Noctilio, and Sturnira, these three taxa are niohyoideus pass ventral to basihyal and in- scored ``?'' in the matrix (see discussion be- sert into ®bers of m. hyoglossus and m. ster- low). The primitive condition for phyllos- nohyoideus via raphe (0); or super®cial ®- tomids cannot be reconstructed a priori. bers insert in well-developed loop around Straney (1980: characters E17±18) de®ned ventral and dorsal surfaces of the intersec- two binary characters to describe the insertion of m. hyoglossus and m. sternohyoideus tion of m. geniohyoideus: ``Geniohyoid in- (1). For taxa with a split insertion of m. ge- sertion on basihyal'' and ``Geniohyoid inser- niohyoideus, two conditions are possible for tion on raphe ventral to basihyal.'' The de- the insertion of the super®cial ®bers of this rived conditions of these characters are es- muscle. In some glossophagines (e.g., Glos- sentially equivalent to our character states, sophaga, Lichonycteris) and lonchophyl- although Straney (1980), and Sprague lines, the super®cial ®bers pass ventral to the (1943), who was his source, were apparently basihyal to form a weak insertion into ®bers unaware that deep ®bers of m. geniohyoideus of m. hyoglossus and m. sternohyoideus via retain their attachment to the basihyal in the former basihyal raphe (Grif®ths, 1982). forms with the ``raphe'' insertion of the su- In the remaining glossophagines (e.g., Cho- per®cial ®bers. Additionally, Straney (1980) eronycteris, Hylonycteris), the super®cial ®- scored all noctilionoids, except Phyllostomus bers of m. geniohyoideus insert in a well- and Noctilio, as having the insertion on the developed loop around ventral and dorsal basihyal, which con¯icts with Sprague's surfaces of the intersection of m. hyoglossus (1943) description. Sprague (1943) found and m. sternohyoideus. This ``tunnel inser- that in Phyllostomus, Phyllonycteris, Carol- tion'' is unique among mammals (Grif®ths, lia, Artibeus, Sturnira, Pteronotus, and Noc- 1982). The primitive state for phyllostomids tilio, the insertion of m. geniohyoideus was cannot be determined a priori because we on the basihyal raphe, whereas in Desmodus scored Pteronotus and Noctilio ``?'' for char- and Glossophaga the insertion was on the acter 97. We scored all taxa lacking a split entoglossal process of the basihyal. Sprague insertion of m. geniohyoideus (state 0 of (1943) found no evidence of split insertions character 97), ``Ϫ'' for this character. in these taxa. Although Grif®ths' (1982) ex- Grif®ths (1982: character 6) scored this amination of the hyoid apparatus demonstrat- feature as ``tunnel insertion of geniohyoid'' ed that Sprague's (1943) description of the (present/absent). Grif®ths (1982) scored taxa insertion of m. geniohyoideus was inaccurate lacking a split insertion and those with a split for most taxa, no additional data are avail- insertion but lacking a ``tunnel'' insertion as able for Pteronotus, Noctilio,orSturnira. ``absent.'' Gimenez et al. (1996: character Because it seems unlikely that Sprague's 21) scored this character similarly; their char- (1943) description of this character is correct, acter states correspond to ours. Sprague we have chosen to score these taxa ``?'' in (1943) did not describe the tunnel insertion, the matrix. consequently this character was not used by Our scoring of this feature was identical Straney (1980). to that of Grif®ths (1982: character 5) and Wille (1954) described the posterior part 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 99 of m. geniohyoideus as forming a muscular ines, Brachyphylla, phyllonycterines, loncho- tube that surrounds the sternohyoideus (ϭ m. phyllines, Carollia, and stenodermatines hyoglossus; Grif®ths, 1978). Wille (1954) re- (Grif®ths, 1982; including Sturnira; Sprague, ported that this tube was present in all taxa 1943; ®g. 31C±D). In contrast, m. styloglos- he examined except Glossophaga and Mon- sus inserts on the posterolateral ``corner'' of ophyllus, where the posterior part of m. ge- the tongue in glossophagines (Grif®ths, niohyoideus formed a median belly between 1982; ®g. 31A±B). M. styloglossus inserts on the two bellies of the sternohyoideus (ϭ m. the lateral surface of the tongue along much hyoglossus; Grif®ths, 1982). Wille (1954) of its length in Pteronotus and Noctilio therefore does not agree with Grif®ths (1982) (Sprague, 1943), suggesting that this is the on the taxonomic distribution of this char- primitive condition for phyllostomids. acter, because Grif®ths (1982) reported that Our scoring of this character is identical the tunnel insertion does not occur in Lon- to that of Grif®ths (1982: character 7) and chophylla, whereas Wille (1954) reported the Gimenez et al. (1996: character 22), although presence of the tunnel insertion in this taxon. Grif®ths (1982) described the derived con- We have chosen to follow Grif®ths (1982) in dition (our state 1) as ``posterior shift of sty- this instance. loglossus insertion.'' Character 99: Right and left m. genio- Character 101: M. genioglossus inserts hyoideus muscles partly or completely fused into ventral surface of tongue along more across midline (0); or muscles not fused (1). than half of its length (0); or inserts into pos- M. geniohyoideus is partly or completely terior half to third of ventral surface of fused with its counterpart across the midline tongue (1); or inserts into posterior quarter in Desmodus and Glossophaga (Sprague, of ventral surface of tongue (2). M. genio- 1943). In contrast, the right and left m. ge- glossus inserts into the ventral surface of the niohyoideus are not fused in Phyllostomus, tongue along much of its length in Desmo- Phyllonycteris, Carollia, Sturnira, and Arti- dus, phyllostomines, Brachyphylla, phyllon- beus (Sprague, 1943). These muscles are ycterines, Carollia, and stenodermatines fused in Pteronotus and Noctilio (Sprague, (Grif®ths, 1982). In contrast, this muscle in- 1943), suggesting that this condition is prim- serts into the posterior half to third of the itive for phyllostomids. ventral surface of tongue in lonchophyllines Straney (1980: characters E12±14) de- (Grif®ths, 1982). In glossophagines, this scribed fusion of m. geniohyoideus with muscle inserts into the posterior quarter of three binary characters. We combined two of the ventral surface of tongue (Grif®ths, his characters, ``Geniohyoids fused into sin- 1982). Although Sprague (1943) provided a gle muscle'' and ``Geniohyoids unfused an- description of the insertion of m. genioglos- teriorly,'' into a single character state be- sus in the bats he surveyed, the description cause Sprague (1943) found that these mus- is not detailed enough to permit us to score cles are never fused near their origins. Pteronotus, Noctilio, and Sturnira. Because Sprague's (1943) taxonomic descriptions dis- we cannot score our outgroups, the primitive agree with the distributional account of this state for phyllostomids cannot be recon- character appearing in his discussion section. structed a priori. We have used the description given in the In Grif®ths' (1982: character 8) matrix the discussion section (Sprague, 1943: 463) to derived condition of this feature was ``pos- score this character. Grif®ths (1982) did not terior shift of genioglossus insertion'' (pres- mention fusion of m. geniohyoideus with its ent/absent). Grif®ths' (1982) descriptions, counterpart and consequently this character however, suggest that multiple conditions ex- was not used by Gimenez et al. (1996). ist. Therefore, we followed Gimenez et al. Character 100: M. styloglossus inserts on (1996: character 23) in recognizing the three lateral surface of tongue along much of its character states de®ned above. length (0); or inserts on posterolateral ``cor- Character 102: M. stylohyoideus absent ner'' of tongue (1). M. styloglossus inserts (0); or present (1); or sometimes present; on the lateral surface of the tongue along polymorphic within species (2). M. stylohy- much of its length in Desmodus, phyllostom- oideus is consistently absent in Desmodus, 100 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 phyllostomines, Brachyphylla, most glosso- attempted to score the variation described by phagines (e.g., Anoura, Choeronycteris), lon- Grif®ths (1982) using four character states: chophyllines, and stenodermatines (including ``(0) well-developed, with two (or three) Sturnira; Sprague, 1943; Wille, 1954; Grif- slips; (1) reduced to a few ®bers (originated ®ths, 1982). In contrast, m. stylohyoideus is from the fascia of posterior milohyoid [sic] present in phyllonycterines, some glosso- region); (2) almost complete reduction (even- phagines (e.g., Glossophaga, Lichonycteris), tually vestigial); (3) completely absent.'' We and Carollia (Grif®ths, 1982). This muscle have chosen to divide the character some- is variably present within species of Lepto- what differently and treat presence or ab- nycteris (Grif®ths, 1982). M. stylohyoideus sence of anterolateral and lateral slips in m. is present, but reduced and fused to m. sty- sphincter colli profundus as two characters lopharyngeus, in Pteronotus parnellii (Grif- (characters 103 and 104). ®ths, 1983b). This muscle is absent in Noc- Character 104: Lateral slip of m. sphinc- tilio (Sprague, 1943). Thus, the primitive ter colli profundus present (0); or absent (1). condition for phyllostomids cannot be estab- The lateral slip of m. sphincter colli profun- lished a priori. dus is present in phyllostomines, Brachy- Although our character states 0 and 1 are phylla, many glossophagines (e.g., Glosso- identical to those used by Grif®ths (1982: phaga, Leptonycteris), lonchophyllines Car- character 9) and Gimenez et al. (1996: char- ollia, and stenodermatines (Grif®ths, 1982), acter 24), we have erected a separate char- except Sturnira where this slip is absent acter state to deal with the variation found (Sprague, 1943). This slip is also absent in within the species of Leptonycteris that Grif- Desmodus, phyllonycterines, Choeroniscus, ®ths (1982) examined. Both Grif®ths (1982) and Choeronycteris (Grif®ths, 1982). These and Gimenez et al. (1996) scored Leptonyc- slips are absent in Pteronotus and Noctilio teris as 0/1, which is not an appropriate cod- (Sprague, 1943), suggesting that this condi- ing for situations of within-species variation tion is primitive for phyllostomids. (see Simmons, 1993). Character 105: Lateral slip of m. sphinc- Character 103: Anterolateral slip of m. ter colli profundus passes laterally (0); or sphincter colli profundus present (0); or ab- passes anterolaterally to insert on skin of sent (1). In bats, m. sphincter colli profundus cervical region (1). The lateral slip of m. originates from the fascia of the mylohyoid sphincter colli profundus passes laterally region, forms a variable number of slips, and from its origin in the fascia of the mylohyoid inserts on the inner surface of the skin of the region to insert on the inner surface of the cervical region (Sprague, 1943; Grif®ths, skin of the cervical region in phyllostomines, 1982). An anterolateral slip is present in Des- Brachyphylla, glossophagines that have this modus, phyllostomines, Brachyphylla, lon- slip, lonchophyllines, and Carollia (Grif®ths, chophyllines, Carollia, and stenodermatines 1982). In contrast, the lateral slip of m. (Grif®ths, 1982; including Sturnira; Sprague, sphincter colli profundus passes anteriorly 1943). In contrast, this slip is absent in phyl- and laterally in stenodermatines that have lonycterines and glossophagines (Grif®ths, this slip (Grif®ths, 1982). The primitive con- 1982). Among the outgroups, an anterolat- dition of this character for phyllostomids eral slip is absent in Pteronotus, but present could not be established a priori because both in Noctilio (Sprague, 1943), so the primitive Pteronotus and Noctilio lack the lateral slip condition for phyllostomids cannot be deter- of this muscle (Sprague, 1943). We scored mined a priori. taxa lacking the lateral slip of m. sphincter Grif®ths (1982: character 10) treated the colli profundus (state 1 of character 104) condition of m. sphincter colli profundus as ``Ϫ'' for this character. one character, ``reduction of m. sphincter col- This feature was not treated in Grif®ths' li profundus'' (present/absent). However, this (1982) or Gimenez et al.'s (1996) character arrangement did not accommodate the vari- analyses because stenodermatines, the only ation Grif®ths (1982) described in this mus- taxa with state 1, were not the focus of their cle (see discussion in Smith and Hood, studies. 1984). Gimenez et al. (1996: character 25) Character 106: M. cricopharyngeus con- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 101 sists of a single large slip (0), or two slips the midline (®gs. 33A, 34A±C). In contrast, (1), or three slips (2), or more than three the MVPs are absent in desmodontines and slips (3). M. cricopharyngeus consists of a some glossophagines (e.g., Anoura, Cho- single large slip in Desmodus.InBrachy- eroniscus; ®g. 32C). Mormoopids and Noc- phylla, phyllonycterines, and lonchophyl- tilio have two MVPs on the dorsum of the lines, this muscle is composed of two distinct tongue (®g. 32A±B), suggesting that this is slips or bellies (Grif®ths, 1982). M. crico- the primitive condition for phyllostomids. pharyngeus consists of three slips in phyl- Grif®ths (1982: character 15) ®rst used lostomines, Carollia, and stenodermatines this character in a cladistic analysis, erecting (including Sturnira; Sprague, 1943; Grif®ths, states equivalent to ours. However, Grif®ths 1982). In glossophagines, the muscle consists (1982) did not score lonchophyllines with of ``several slips'' (although Grif®ths [1982: any state for this character (``see text'' ap- 18] is not clear, we interpret this as more than peared in his matrix instead of ``ϩ'' or 0) to three slips). A single undivided slip is pres- indicate that lonchophyllines lack both me- ent in Pteronotus (Grif®ths, 1983b) and Noc- dial and lateral circumvallate papillae (see tilio (Sprague, 1943), suggesting that this below). Gimenez (1993: characters 3.3.2.1, condition is primitive for phyllostomids. 3.3.5.1) and Gimenez et al. (1996: character Grif®ths (1982: character 11) scored the 4) also used this character. Gimenez (1993) morphology of m. cricopharyngeus as ``cri- described the derived condition (absence of copharyngeus reduced to two bellies'' (pres- MVPs, her state 1) as MVPs vestigial or ab- ent/absent), apparently assuming three slips sent, whereas Gimenez et al. (1996) erected was the primitive condition. Gimenez et al. states identical to ours. (1996: character 26) recognized three states Grif®ths (1982) reported the absence of of this character that are equivalent to our MVPs not only in the taxa noted above, but states 1, 2, and 3. Although Grif®ths (1982: also in Lionycteris, Lonchophylla, and Pla- 18) noted that this muscle consists of ``sev- talina. Grif®ths and Criley (1989) also noted eral slips'' in glossophagines (s.s.), he did not that Lonchophylla lacks all vallate papillae. give the number of slips present. The single- In contrast, we observed MVPs in every slip condition seen in Desmodus and the out- specimen of Lionycteris, Lonchophylla, and groups was not included in Grif®ths' (1982) Platalina that we examined (see appendix 1 or Gimenez et al.'s (1996) character analysis. for species and specimen numbers). Our results agree with the ®ndings of Smith and TONGUE Hood (1984), who reported the presence of these structures in Lionycteris spurelli and Characters described in this section are Lonchophylla robusta, and with those of Gi- based largely on features originally noted by menez (1993) and Gimenez et al. (1996), Grif®ths (1982; character numbers from his who observed MVPs in Lionycteris spurelli, table 1), Gimenez (1993), and Gimenez et al. Lonchophylla bokermanni, L. dekeyseri, and (1996; character numbers from their table 1). L. mordax, although Gimenez (1993) had Taxa examined in these studies are listed in scored these papillae as ``vestigial or absent'' table 4. We have introduced several new in L. bokermanni. characters, modi®ed previous descriptions, Grif®ths (1982) reported that Lichonycter- and scored tongue morphology for all taxa is has MVPs, whereas Gimenez (1993) and included in this study except Scleronycteris Gimenez et al. (1996) did not ®nd MVPs in and Musonycteris, for which specimens were the specimens that they examined. Giminez unavailable. Nomenclature for papillae and (1993) coded Lichonycteris 0/1 to account other tongue structures follows Sonntag for her observations and those of Grif®ths (1920), except where noted. (1982). We agree with Gimenez (1993) and Character 107: Medial circumvallate pa- Gimenez et al. (1996) that Lichonycteris pillae present (0); or absent (1). In most lacks MVPs. Although Smith and Hood phyllostomids, two medial circumvallate pa- (1984) noted that Centurio lacks MVPs, the pillae (MVPs) are present proximally on the specimens that we examined all possess very dorsum of the tongue, one on either side of small MVPs that are located at the proximal 102 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 32. Dorsal surface of the tongue in selected noctilionoids. A. Pteronotus davyi (AMNH 175276). Insets from top to bottom: basketlike papilla, basin-shaped medial posterior mechanical papilla, lateral circumvallate papilla. B. Noctilio leporinus (AMNH 175534) C. Desmodus rotundus (AMNH 210962). Scale bar ϭ 2 mm. extreme of the tongue. In addition, the pap- Based on our observations, we have chosen illary bodies of the MVPs are partly fused to to score both Leptonycteris and Lonchophyl- the valla (see character 108), which makes la as having MVPs (state 0). observation very dif®cult and may account Character 108: Papillary bodies of me- for the report of Smith and Hood (1984). dial circumvallate papilla separate from val- Gimenez (1993) and Gimenez et al. (1996) la, fossae complete (0); or papillary bodies reported that MVPs were variably present in fused anterolaterally and posteromedially to the specimens of Lonchophylla thomasi they valla, fossae incomplete (1). In most phyl- examined (present in only three of the six lostomids, the fossa of each MVP is com- specimens; Gimenez et al., 1996). We ex- plete and the papillary body is not fused to amined six specimens of L. thomasi and the vallum (®g. 35A). In contrast, in Centu- found MVPs present in all. Grif®ths (1982: rio and Sphaeronycteris each MVP papillary table 1) scored Leptonycteris as ``variable'' for this character, though in the text he noted body is fused with the vallum anterolaterally that only one specimen of ®ve L. sanborni and posteromedially (®g. 35B). Consequent- (ϭ L. curasoae) lacked one MVP. Our ob- ly, the fossae are not circular furrows, but servations of Leptonycteris curasoae and L. appear as paired slits on each side of the pap- nivalis agree with those of Park and Hall illary body. In mormoopids and Noctilio, the (1951), Greenbaum and Phillips (1974), MVP bodies are not fused to the valla and Smith and Hood (1984), Gimenez (1993) and the fossae are complete, suggesting that this Gimenez et al. (1996), who reported that the condition is primitive for phyllostomids. We MVPs were always present in Leptonycteris. scored the eight taxa lacking medial vallate 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 103

Fig. 33. Dorsal surface of the tongue in selected phyllostomids. Lowest inset is of a lateral circumvallate papilla. A. Phyllonycteris poeyi (AMNH 23762). B. Glossophaga soricina (AMNH 237911). Lonchophylla thomasi (AMNH 266107). Upper inset: basketlike papilla. Scale bar ϭ 2 mm. papillae (state 1 of character 107) ``Ϫ'' for Grif®ths (1982) did not use presence of this character. LVPs as a character, but did note the loss of Smith and Hood (1984) examined Centu- all circumvallate papillae in lonchophyllines rio, but incorrectly reported that the MVPs in his matrix (see discussion under character were absent in this genus (see character 107). 108). Gimenez (1993: characters 3.3.2.2, No previous study of tongue morphology has 3.3.5.3) and Gimenez et al. (1996: character included Sphaeronycteris. Consequently, this 6) used presence of LVPs as a character and character has not been used in previous clade®ned states identical to ours. distic analyses. Grif®ths (1982) indicated that lonchophyl- Character 109: Lateral circumvallate pa- lines, Brachyphylla, and desmodontines pillae present (0); or absent (1). Most phyl- lacked LVPs, and Grif®ths and Criley (1989) lostomids have a pair of circumvallate papil- noted that Lonchophylla lacks all vallate pa- lae on the proximal part of the lateral surfac- pillae. However, our observations agree with es of the tongue (®gs. 33A±C, 34A±C). Des- those of Smith and Hood (1984), Gimenez modontines, Brachyphylla, and Centurio lack (1993), and Gimenez et al. (1996) who found these lateral circumvallate papillae (LVPs; that LVPs are present in all Lionycteris and ®g. 32C). LVPs are present in mormoopids, Lonchophylla specimens they examined. Be- but absent in Noctilio (®g. 32A±B). Thus, the cause specimens of Lonchophylla robusta primitive condition for phyllostomids cannot and Platalina were not available for study, be reconstructed a priori. Giminez (1993) accepted Grif®ths' (1982) 104 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 34. Dorsal surface of the tongue in selected phyllostomids. Lowest inset is of a lateral circumvallate papilla. A. Phyllostomus hastatus (AMNH 233176). Upper inset: basketlike papilla. B. Uroderma bilobatum (AMNH 171294). C. Ametrida centurio (AMNH 247645). Scale bar ϭ 2 mm. conclusion that these papillae appeared to be sent in lonchophyllines. Unlike Greenbaum absent in these taxa. Our observations of and Phillips (1974), who reported the ab- LVPs in all specimens of Lionycteris, Lon- sence of one of the two LVPs in one speci- chophylla, and Platalina that we examined men each of Leptonycteris curasoae and L. (see appendix 1 for species and specimen nivalis, we did not ®nd missing LVPs in any numbers) con®rm that these papillae are pre- representative of Leptonycteris we examined.

Fig. 35. Close-up dorsal views of a medial circumvallate papilla in A. Noctilio leporinus (AMNH 243905) and B. Centurio senex (AMNH 99645). 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 105

Park and Hall (1951) reported the presence the other Vampyressa species that we ex- of only two vallate papillae in most taxa they amined, these papillae are located in a dor- examined (including Leptonycteris); howev- solateral position. In Artibeus, Dermanura, er, as Greenbaum and Phillips (1974) noted, and Koopmania, Gimenez (1993) described it appears these authors simply overlooked the LVPs as being dorsolateral. In contrast, the lateral vallate papillae, reporting only on we found that the LVPs were located on the the presence of medial vallate papillae. Fi- dorsum of the tongue all representatives of nally, Gimenez et al. (1996) indicated that all these taxa that we examined (see appendix 1 stenodermatines possess lateral vallate papil- for specimens examined). lae. This study, however, did not include Character 111: Pharyngeal region of Centurio, which lacks these papillae (Smith tongue completely covered with papillae (0); and Hood, 1984; this study). or covered laterally with papillae and me- Character 110: Lateral circumvallate pa- dially bare (1); or wholly unpapillated (2). pillae located on lateral surface of tongue The pharyngeal region of the tongue lies (0); or in dorsolateral position, at border be- proximal to the circumvallate papillae (Sonn- tween lateral and dorsal surfaces of tongue tag, 1925). In taxa lacking circumvallate pa- (1); or on dorsal surface of tongue (2). In pillae (e.g., desmodontines), this region ap- most phyllostomids, LVPs are located on the parently lacks fungiform papillae (Sonntag, lateral surface of the tongue (®gs. 33A±C, 1925), providing a means of identi®cation. 34A). However, in some stenodermatines Most phyllostomids have ¯eshy mechanical (e.g., Uroderma, Vampyrodes), the LVPs are papillae covering the lateral sections of the located at the steeply curved ``border'' be- pharyngeal region of the tongue (®gs. 32C, tween the dorsal and lateral surfaces of the 33A±C, 34B). In these taxa, a smooth, un- tongue (®g. 34B). In other stenodermatines papillated bare patch is present medially, of- (e.g., Ametrida, Artibeus), the LVPs are lo- ten extending distally between the MVPs. In cated on the dorsal surface of the tongue (®g. some phyllostomines (e.g., Chrotopterus, 34C). In some of these taxa (e.g., Pygoder- Phyllostomus) and some stenodermatines ma), the LVPs are almost directly adjacent to (e.g., Mesophylla, Vampyrodes) the tongue the MVPs. Whereas mormoopids have LVPs lacks this medial bare patch and has a carpet that are located on the lateral surface of the of mechanical papillae over the entire pha- tongue (®g. 32A), LVPs are absent in Noc- ryngeal region (®g. 34A). In contrast, many tilio; therefore, the primitive condition for stenodermatines (e.g., Ametrida) have an en- phyllostomids appears to be a lateral position tirely bare pharyngeal region (®g. 34C). One for the LVPs. We scored taxa lacking LVPs individual of Pygoderma (AMNH 248334) (state 1 of character 109) ``Ϫ'' for this char- has a single medial line of papillae bisecting acter. the otherwise bare pharyngeal region, where- Gimenez (1993: character 3.3.4.2) ®rst as all other specimens we examined have a used the position of the LVPs as a character completely bare pharyngeal region. One in- in cladistic analysis, and we follow her state dividual of Phyllops (AMNH 176190) has de®nitions. Smith and Hood (1984: 450) approximately 14 papillae present laterally characterized the LVPs as ``dorsal'' in all on both sides in the pharyngeal region, stenodermatines after examining Ametrida, whereas all other specimens that we exam- Artibeus, Centurio (which lacks LVPs), and ined do not have any papillae present in the Sturnira. Smith and Hood's (1984) assess- pharyngeal region. Similarly, a single speci- ment of LVP position in Ametrida and Arti- men (USNM 522707) of Stenoderma has beus appears to be supported by our obser- several papillae present laterally on both vations; however, we agree with Gimenez sides of the pharyngeal region, whereas all (1993) that Sturnira has dorsolateral LVPs. other specimens do not. We interpret these Gimenez (1993) characterized the LVPs as conditions as anomalies and score these gen- being located laterally in Vampyressa bidens era with state 2 in the matrix. Mormoopids and V. pusilla. Our observations of LVPs in have a pharyngeal region that is bare (®g. these species and V. nymphaea indicate that 32A), whereas Noctilio has ¯eshy mechani- only V. pusilla has laterally located LVPs. In cal papillae surrounding a central bare patch 106 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

(®g. 32B). The primitive state for phyllos- Character 112: Lingual sulci absent (0); tomids cannot be reconstructed a priori. or lateral lingual sulci present (1); or ventral Gimenez (1993: characters 3.3.2.3; lingual sulci present (2). Lingual sulci are 3.3.3.3; 3.3.4.7±8) ®rst used the location of present in Desmodus, Diaemus, and loncho- papillae in the pharyngeal region in a cladis- phyllines, but are absent in all other phyllos- tic analysis. One character (Gimenez, 1993: tomids. In Desmodus and Diaemus, the sulci character 3.3.4.7) accounted for presence or are located on the ventral aspect of the absence of the pharyngeal papillae, and ap- tongue and run anteriorly from the frenulum peared only in the analysis of Stenodermatini (a median vertical fold present on the ¯oor genera (see Historical Background for details of the mouth) along the distal half of the of Gimenez's [1993] analyses). The other tongue (®g. 36C). The sulci do not meet at character (Gimenez, 1993: character 3.3.2.3, the tongue tip. In lonchophyllines, sulci are 3.3.3.3, 3.3.4.8) described the two patterns of found on the lateral surface of the tongue, papillation (pharyngeal region completely beginning just distal to the LVPs and cours- covered or covered only laterally), and ap- ing to the tongue tip, where the right and left peared in most analyses, except the analysis sulci meet (®g. 36D). The sulci in loncho- of Glossophagini genera. We have chosen to phyllines appear to be relatively deeper than combine the two characters recognized by those of Desmodus and Diaemus and are bor- Gimenez (1993) to avoid the problems of dered by a fringe of hairlike papillae (see missing data often associated with hierarchi- character 113), which is absent in the des- cal characters. modontine taxa. Mormoopids and Noctilio In her analyses, Gimenez (1993) scored lack lingual sulci, suggesting that this is the those Stenodermatini genera with bare pha- primitive condition for phyllostomids. ryngeal regions (character 3.3.4.7: pharyn- Grif®ths (1982: character 12) de®ned a geal papillae absent) as having the laterally character for presence or absence of a papil- papillated state in the second character (char- la-lined lingual sulcus. Our character states 0 acter 3.3.4.8), rather than ``Ϫ'' as appropriate and 1 are roughly equivalent to those of Grif- for a hierarchical character. In Gimenez's ®ths (1982), although we have added an ex- (1993) analysis of desmodontines, basal tra state to include the variation introduced phyllostomines, and the tribes Phyllostomini, by the desmodontines, which Grif®ths (1982) Glossophagini, and Stenodermatini, she described but did not consider in his analysis. scored Phyllostomini and Stenodermatini as Unlike Grif®ths (1982), we consider the hair- having only the laterally papillated condition. like papillae lining the sulcus in lonchophyl- In fact, these taxa are polymorphic, with both lines in the context of a different character laterally papillated and totally papillated con- (see character 113). Wille (1954) reported the ditions occurring in Phyllostomini and Sten- presence of a groove along the lateral mar- odermatini genera. Although mormoopids gins of half the length of the tongue in Lep- have a completely bare pharyngeal region, tonycteris nivalis. This groove does not ap- Gimenez's (1993) character did not include pear to be present in Leptonycteris, as was this condition as a state, and she scored mor- previously noted by Park and Hall (1951), moopids as having a medially bare pharyn- Winklemann (1971), and Greenbaum and geal region. Phillips (1974). Gimenez (1993) described the pharyngeal Grif®ths and Criley (1989) found that the region of Phylloderma as being fully papil- sulci of Desmodus and Diaemus are ventral lated. We observed a smooth medial pharyn- to the lingual nerve and the transverse mus- geal region laterally ¯anked by ¯eshy, ®li- cle mass (®g. 36A), but in lonchophyllines form papillae in all specimens we examined the sulci are dorsal to these structures (®g. (see appendix 1 for specimens examined). In 36B). On this basis, these authors suggested Mesophylla and Vampyrodes, Gimenez that the sulci in desmodontines and loncho- (1993) described the pharyngeal region as phyllines are not homologous. Gimenez being laterally papillated and medially bare. (1993: characters 3.3.2.8, 3.3.5.14) erected We observed a completely papillated pharyn- two different lingual sulcus characters, one geal region in both genera. for the presence of a ventral sulcus, the other 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 107

Fig. 36. Cross sections of the tongues of A. Desmodus rotundus and B. Lonchophylla robusta illustrating differences between the lingual sulci in these species (redrawn from Grif®ths and Criley, 1989: ®g. 2). Note that the sulci of Desmodus are ventral to the lingual nerve (ln), but in Lonchophylla the sulci are dorsal to this structure. for the presence of a lateral sulcus. Her char- papillae,'' and ``brush tip formed by hairlike acter 3.3.2.8 is equivalent to our state 2 as a papillae.'' He coded phyllonycterines and binary (present/absent) character. Gimenez's glossophagines ``absent'' for the former and (1993) second character (3.3.5.14), ``pres- ``present'' for the latter, whereas the reverse ence of a lateral lingual sulcus bordered by was true for lonchophyllines. Smith and ®liform papillae,'' is equivalent to our char- Hood (1984: 448±449) objected to these acter state 1. Gimenez et al. (1996: character characters, which appeared to score presence 2) de®ned and scored a character for the of ``Glossophaga-type'' and ``lonchophyl- presence of a lateral lingual sulcus, following line-type'' tongues, obscuring the potential Grif®ths and Criley's (1989) and Gimenez's homology of the hairlike papillae, which are (1993) assessment that the lateral and ventral not found on the tongues of other phyllos- lingual sulci are not homologous. We prefer tomids. to make no such a priori assessments and in- Gimenez (1993: character 3.3.5.11) used a cluded both types of sulci as states in a single single unordered multistate character with character to allow all possible transforma- three states (absent, 0; present but short and tions. few in number, 1; present and abundant, 2) Character 113: Brush of hairlike papillae to describe variation of hairlike papillae. Gi- around the distal margin of tongue absent menez et al. (1996: character 14) retained (0); or present (1). Long, thin, ®liform pa- states 0 and 1 (Gimenez, 1993), but added pillae, known as ``hairlike'' papillae (Park two new states: brushlike tip present, with and Hall, 1951), are present laterally on the abundant hairlike papillae 2, and brushlike distal part of the tongue in phyllonycterines, tip extremely well developed, with abundant glossophagines, and lonchophyllines (®g. and very long hairlike papillae 3 (see char- 33A±C). The hairlike papillae form a brush- acter 115). Gimenez et al. (1996) ordered the like tongue tip in these taxa. Hairlike papillae transformations in this character (0 ↔ 1 ↔ are absent in all other phyllostomids (®gs. 2 ↔ 3). In addition to this character, Gime- 32C, 34A±C). Hairlike papillae are also ab- nez et al. (1996: character 3) included a sep- sent in mormoopids and Noctilio (®g. 32A± arate character describing the line of ®liform B), suggesting that absence of these papillae papillae that runs along the lateral lingual is the primitive condition for phyllostomids. sulcus in lonchophyllines. Grif®ths (1982: characters 12, 14) de®ned We prefer to recognize the presence of two binary (present/absent) characters based hairlike papillae, derived relative to the out- on hairlike papilla morphology and distri- groups, as a single state, with additional hi- bution: ``groove in tongue lined with hairlike erarchically dependent characters (characters 108 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 37. Dorsal (left) and lateral (right) views of the tongues of representative species with hairlike papillae. Insets feature close-ups of individual hairlike papillae. A. Glossophaga soricina (AMNH 237911). B. Lonchophylla thomasi (AMNH 267452). C. Phyllonycteris poeyi (AMNH 236698). Scale bar ϭ 2 mm.

114 and 115) to accommodate variation in that lacks hairlike papillae runs down the the distribution and shape of the papillae. midline of the dorsum of the tongue. Mor- This approach best re¯ects our views of po- moopids and Noctilio lack hairlike papillae, tential homologies. so the primitive state for phyllostomids can- Character 114: Hairlike papillae con®ned not be assessed a priori. We scored taxa that to lateral margin of distal third of tongue, lack hairlike papillae (state 0 of character with a single line of papillae that extends 113) ``Ϫ'' for this character. roughly to LVPs (0); or hairlike papillae dis- As noted above, Grif®ths (1982: charac- tributed around lateral margin and dorsum ters 12, 14) was the ®rst to de®ne characters of distal third of tongue, not arranged in a based on distribution of hairlike papillae. Gi- single line (1). The distribution of hairlike menez's (1993: character 3.3.5.11) character papillae differs substantially among taxa in states 1 (present but short and few in num- which they are present. In lonchophyllines, ber) and 2 (present, long and abundant) are the papillae are restricted to the distal part of roughly equivalent to our states 0 and 1, re- the lateral surface of the tongue, with a single spectively and our scoring is identical to line of hairlike papillae running just ventral hers. Gimenez et al.'s (1996: character 14) to the lingual sulci, extending posteriorly character incorporated aspects of both distri- roughly to the LVPs (®g. 37B). These single bution pattern and shape of individual papil- lines of papillae are not developed in phylla, which we have chosen to treat as two sep- lonycterines and glossophagines (®g. 37A, arate characters (this character and character C). In these taxa, the papillae are found on 115). the lateral and dorsal surfaces of the tongue, Character 115: Hairlike papillae ¯eshy and a narrow channel (``median trough,'' or and conical (0); or ¯eshy and conical with ``anterior groove'' of Park and Hall, 1951) ®lamentous tips (1); or cylindrical with el- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 109 lipse-shaped distal end (2). Phyllonycterines, are not inclined). In Noctilio, the posterior- glossophagines, and lonchophyllines are each most medial-posterior mechanical papillae characterized by a different shape of hairlike are on ridges that curve anteriorly towards papilla. The individual hairlike papillae of the lateral margins of the tongue (®g. 32B). lonchophyllines are short, ¯eshy, and conical The ridges disappear as the papillae on them (®g. 37B, inset). In glossophagines, the pa- become more sharply pointed (roughly one pillae are longer and thinner than in loncho- third of the area of the medial-posterior me- phyllines, and have ®lamentous tips (see chanical papillae is ridged). The individual scanning electron micrographs in Vaughan, medial-posterior mechanical papillae are di- 1986; ®g. 37A, inset). These papillae most rected toward the midline of the tongue. closely resemble hairs. Phyllonycterines have Thus, the primitive condition for phyllostom- cylindrical hairlike papillae (®g. 37C, inset). ids cannot be reconstructed a priori. The distal ends of these papillae appear as Park and Hall (1951) noted that the me- though they have been sliced obliquely, pro- dial-posterior mechanical papillae incline an- ducing ellipsoid ends. Mormoopids and Noc- teriorly in Artibeus among the phyllostomids tilio lack this papilla type; thus the primitive they studied, but Gimenez (1993: character state for phyllostomids cannot be established 3.3.4.11) was the ®rst to use this information a priori. We scored taxa without hairlike pa- in a cladistic analysis. Our character states pillae (state 0 of character 113) ``Ϫ'' for this and scoring are equivalent to hers. character. Character 117: Small patch of anteriorly Although previous workers have not used directed medial-posterior mechanical papil- shape of the hairlike papillae as a separate lae always absent, all papillae oriented to- character, Gimenez et al. (1996: character 14) ward pharyngeal region (0); or medial patch incorporated papilla presence, morphology, present in some individuals; polymorphic and distribution into a single character. Gi- within species (1); or medial patch always menez et al. (1996) recognized three condi- present (2). Among those phyllostomids in tions, roughly corresponding to our character which the medial-posterior mechanical pa- states (see character 113). pillae are principally directed toward the Character 116: All or most medial-pos- pharyngeal region (state 0 of character 116), terior mechanical papillae inclined toward a small patch of papillae near the midline of the pharyngeal region of the tongue (0); or tongue may be oriented toward the tip of the all oriented toward tongue tip (1); or all ori- tongue. This patch is always present in Mi- ented toward midline of tongue (2); or pa- cronycteris sylvestris, Lionycteris, and Pla- pillae not inclined (3). Gimenez (1993) de- talina, and is sometimes present in Loncho- ®ned the medial-posterior mechanical papil- phylla (e.g., Lonchophylla thomasi AMNH lae as those mechanical papillae that lie be- 266107; ®g. 33C). In Mormoops, the small tween the MVPs and the anterior mechanical patch of anteriorly directed papillae is not papillae. In taxa that lack MVPs (e.g., des- present. The medial-posterior mechanical pa- modontines), we de®ne this region as rough- pillae are not directed toward the pharyngeal ly the middle third of the tongue. All or most region in Noctilio and Pteronotus, and we of the medial-posterior mechanical papillae scored these genera ``Ϫ'' for this character. are inclined toward the pharyngeal region of Thus the primtive state for phyllostomids ap- the tongue in most phyllostomids (®gs. 32C, pears to be the absence of the anteriorly di- 33A±C, 34A). In contrast, all of the medial- rected patch of papillae. We scored taxa in posterior mechanical papillae are inclined to- which the medial-posterior mechanical pa- ward the tip of the tongue in all stenoder- pillae are not principally directed toward the matines with the exception of Sturnira (®g. pharyngeal region (states ``1,'' ``2,'' or ``3'' 34B±C). In Mormoops these papillae are di- for character 116) ``Ϫ'' for this character. rected pharyngeally. In Pteronotus, the me- Gimenez (1993: characters 3.3.5.9) ®rst dial-posterior mechanical papillae are unlike used this feature in a cladistic analysis, and the ¯eshy, ®liform papillae in other nocti- our character states are equivalent to hers. lionoids (®g. 32A). Instead, these papillae are Gimenez (1993) left Lonchophylla unscored basin-shaped and directed dorsally (i.e., they because the medial-posterior mechanical pa- 110 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 pillae were too small and rudimentary to per- anterior region medially. We found that the mit con®dent observation. Gimenez et al. shape of the juncture between the anterior (1996: 49) indicated that a patch of distally and medial-posterior mechanical papillae directed papillae may be seen not only in shows continuous variation from transverse Lionycteris, but in ``some of the best pre- or slightly V-shaped to strongly V-shaped. served tongues of Lonchophylla species.'' Accordingly, we have scored only the pres- This description corroborates our observa- ence of these papillae, not the shape of the tion of such a patch in Lonchophylla tho- zone in which they occur. We also found masi. well-differentiated, long-tipped, bi®d anterior Character 118: Long-tipped, bi®d ante- mechanical papillae in several additional taxa rior mechanical papillae absent (0); or pres- (see above). ent at border between anterior mechanical Character 119: Basketlike medial-poste- papillae and medial-posterior mechanical rior mechanical papillae absent (0); or pres- papillae (1). In most phyllostomids, the ent (1). In most phyllostomids, the medial- small, keratinized anterior mechanical papil- posterior mechanical papillae are thick, lae and the larger, ¯eshy medial-posterior ¯eshy, and have one or more rounded points. mechanical papillae meet near the middle of However, some of the medial-posterior me- the tongue (®gs. 32C, 33A±C, 34A, C). No chanical papillae are what Park and Hall other papilla type is present at this juncture. (1951) termed ``basketlike.'' Basketlike pa- In contrast, this juncture is covered by a nar- pillae occur in phyllostomines, Brachyphylla, row, transverse band of keratinized bi®d pa- some glossophagines (e.g., Hylonycteris, Li- pillae in many stenodermatines (e.g., Arti- chonycteris), Lionycteris, Lonchophylla, car- beus, Uroderma; ®g. 34B). The individual olliines, and Sturnira (®gs. 33C, 34A). These papillae are erect and directed toward the papillae have central concavities and ¯eshy, pharyngeal region. Each papilla tip has a cylindrical bases. The rim of each basketlike long, thin, corni®ed ®lament. These papillae papilla is usually uneven, with the posterior seem to be uniformly present in the species part of the rim higher than the anterior. The that possess them, although they may be ab- entire rim, or just the posterior part, may be sent in one poorly preserved specimen of adorned with from one to more than 10 Koopmania (AMNH 80340) of the ®ve we pointed projections, which appear to form a examined. We interpret this possible condi- fringe along the edge of the rim. Basketlike tion as an aberration, and therefore score papillae are present in Pteronotus and Mor- Koopmania as having long-tipped, bi®d an- moops (®g. 32A), but are absent in Noctilio terior mechanical papillae. Long-tipped, bi®d (®g. 32B), making the state primitive for anterior mechanical papillae are absent in phyllostomids impossible to reconstruct a mormoopids and Noctilio (®g. 32A±B), sug- priori. gesting that this is the primitive condition for Gimenez (1993: characters 3.3.2.5, phyllostomids. 3.3.3.5, 3.3.4.9, 3.3.5.7) erected four separate Gimenez (1993: character 3.3.4.10) ®rst characters to describe morphology of the me- used presence of long-tipped, bi®d anterior dial-posterior mechanical papillae in her mechanical papillae as a character. Gimenez analyses of various subsets of taxa. Gimenez (1993) noted that in most taxa, these papillae et al. (1996: character 10) scored presence/ are not well differentiated, and the junction absence of basketlike papillae, which they between these papillae and the ¯eshy medial- called ``posteromedian conical papillae,'' in posterior mechanical papillae is more or less a separate binary character. The multiple transverse. In contrast, Gimenez (1993) in- characters used by Gimenez (1993) included dicated that Mesophylla macconelli, Vampy- eight states describing the basketlike papillae ressa bidens, and V. pusilla have well dif- and other papilla types, many of which are ferentiated long-tipped, bi®d anterior me- apparently autapomorphic. For example, Gi- chanical papillae, and that the junction be- menez (1993: character 3.3.2.5) erected a tween the anterior and medial-posterior state for the ``craterlike'' papillae of Pter- mechanical papillae is V-shaped with the me- onotus, which she considered to be autapo- dial-posterior papillae penetrating into the morphic, describing them as ``circular and 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 111 differentiated in the form of a crater . . . with Lonchophylla dekeyseri, but our observations a depression and fringed border.'' However, of L. robusta and L. thomasi agree with Gi- Gimenez (1993: character 3.3.3.5) reported menez et al.'s (1996) reports of tiny basket- similar papillae, which she described as ``cir- like papillae in L. mordax and L. bokerman- cular, in the form of a crater . . . with a re- ni; thus, we decided to score Lonchophylla duced fringe,'' in Lonchorhina, Mimon cren- with state 1. In addition, we found that Stur- ulatum, and unspeci®ed species of Tonatia. nira has proximally fringed basketlike papil- Gimenez (1993) scored the ``craterlike'' pa- lae similar to those of phyllostomines, rather pillae of Lonchorhina and Tonatia with a than the anteriorly fringed, posteriorly separate state from those of Mimon crenu- smooth papillae described for this taxon by latum, in which she called the ``craterlike'' Gimenez (1993). Park and Hall (1951) ®rst papillae ``huge and irregular.'' We treat all described the presence of basketlike papillae ``basketlike'' papillae as potentially homol- in Carollia, but did not observe these papil- ogous, including Gimenez's (1993) ``crater- lae in any other phyllostomid they examined like'' papillae, because they are very similar (including Macrotus, Leptonycteris, Glosso- in appearance (e.g., presence of fringe, and phaga, Choeronycteris, Artibeus, and Des- central concavity), and position on the modus), whereas we found these papillae to tongue, despite some differences in relative be present in many of these taxa. size (e.g., as in Mimon crenulatum). Further- Character 120: Cluster of horny papillae more, because each tongue includes a variety located near tip of tongue (0); or located sig- of different basketlike types ranging from ni®cantly proximal to tongue tip (1). In most smooth to fringed, we chose to subsume all phyllostomids, the horny papillae (sensu of the varieties in a single state rather than Park and Hall, 1951) are located near the tip attempting to subdivide what appears to be a of the tongue; less than 20% of tongue length range of variation in the number of points in is distal to the horny papillae (®gs. 32C, 33A, the fringe. C, 34 A±C). In contrast, the horny papillae Despite the differences in scoring we not- are located in a more proximal position on ed above, we agree with virtually all of Gi- the tongue in glossophagines (®g. 33B). In menez's (1993) assessments, except her ®nd- these taxa, more than 25% of the tongue ing that basketlike papillae are present in length is distal to the horny papillae. In mor- Glossophaga and are present but reduced in moopids and Noctilio the horny papillae are size in Platalina (data apparently taken from located near the tip of the tongue (®g. 32A± Grif®ths, 1982). We ®nd that these papillae B), suggesting that this condition is primitive are absent in both genera. Gimenez et al. for phyllostomids. (1996) also scored basketlike papillae as Grif®ths (1982) indicated that lonchophyl- present in Glossophaga, as well as Mono- lines lack horny papillae, but Smith and phyllus, which was not mentioned in their Hood (1984), Gimenez (1993), and Gimenez text or list of specimens examined. Our ob- et al. (1996) noted the presence of distinct servations indicate that Monophyllus lacks apical horny papillae in Lionycteris and Lon- basketlike papillae. Giminez et al. (1996: 49) chophylla; our results con®rm that they are scored Lichonycteris ``1/v,'' indicating that indeed present in all lonchophyllines. Gi- this character is polymorphic in this taxon. menez (1993: character 3.3.5.12) was the The reason for this scoring given in their text ®rst to use this character, and our states are was that Lichonycteris has other kinds of pa- equivalent to hers. However, Gimenez (1993) pillae in this region in addition to basketlike considered the horny papillae of phyllonyc- papillae, a factor that does not come into terines to be located in a relatively proximal consideration in our version of this character. position from the lingual apex, and scored Gimenez et al. (1996) scored Lonchophylla them with the same state as glossophagines. as ``0/±,'' indicating that basketlike papillae Gimenez et al. (1996: character 13) de®ned are present in some species, but that the char- and scored the character the same way. In acter is not applicable in others because these contrast, we found that the horny papillae are papillae are so reduced (e.g., L. dekeyseri). located near the tongue tip in phyllonycteri- Unfortunately, we were unable to examine nes, whereas only glossophagines have the 112 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 more proximally located horny papillae. pillae are all equal in size, suggesting that Rough quanti®cation supports this scoring, this is the primitive condition for phyllos- as the largest gap in the distribution of values tomids. We scored taxa that have a V-shaped for percentage of tongue length distal to the cluster of papillae (state 1 of character 121) horny papillae falls between the phyllonyc- ``Ϫ'' for this character. terines and the glossophagines (data avail- This character has no direct counterpart in able on request). previous phylogenetic analyses (see discus- Character 121: Horny papillae arranged sion under character 124). in lines or elliptical cluster (0); or large V- Character 123: Horny papillae all large shaped cluster (1). In most phyllostomids, (0); or all small (1). Among taxa with horny the horny papillae form an elliptical cluster papillae that are all roughly equal in size, (which may be reduced to a few rows of Vampyrum and Rhinophylla have more than horny papillae) near the tip of the tongue 15 very large papillae in the center of the (®gs. 33A±C, 34A±C). In contrast, desmo- cluster, which are surrounded by additional dontines have a large V-shaped arrangement smaller horny papillae (®g. 38A). In contrast, of horny papillae which follows the general Lonchorhina and Macrophyllum have an el- outline of the tongue tip, spreading laterally liptical cluster composed of equally sized and proximally from the near apical position horny papillae, all of which are very small to cover the tongue's width near the tip (®g. 38B). Mormoopids have very small (31C). In mormoopids and Noctilio, the horny papillae. In Noctilio, the papillae in the patch of horny papillae forms an elliptical center of the cluster, more than 15 in number, cluster (®g. 32A±B), suggesting that this is are all equally large. Because of this distri- the primitive condition for phyllostomids. bution of character states, the primitive con- Gimenez (1993: character 3.3.2.7) ®rst dition for phyllostomids cannot be assessed used the V-shaped distribution of horny pa- a priori. We scored taxa that have a V-shaped pillae as a binary character. Grif®ths (1982) cluster of horny papillae (state 1 of character and Gimenez et al. (1996) did not use this 121) or those with papillae of different sizes character, because desmodontines were not in the cluster (state 1 of character 122) ``Ϫ'' the focus of their studies. Although Park and for this character. Hall (1951), who described an elliptical rath- This character has no direct counterpart in er than V-shaped cluster of horny papillae in previous phylogenetic analyses (see discus- desmodontines, and Vierhaus (1983) includ- sion in character 124). To assess the relative ed desmodontine papillae under the descrip- sizes of these papillae, we drew representation ``horny papillae,'' Smith and Hood tive tongues to the same scale for compari- (1984) suggested that the keratinized papillae son. in desmodontines are not comparable to the Character 124: Single large horny papil- horny papillae of other phyllostomids. Based la present in center of elliptical cluster (0); on our observations, the only difference be- or two large horny papillae present in center tween these papillae in desmodontines and of elliptical cluster (1). A single large papilla those in other phyllostomids is the pattern of is present in the elliptical cluster of horny distribution, not the actual structure of the papillae in most phyllostomines (e.g., Phyl- individual papillae. lostomus, Tonatia), Brachyphylla, phyllo- Character 122: All horny papillae present nycterines, Anoura, lonchophyllines, Carol- in elliptical cluster equal in size (0); or some lia, and several stenodermatines (e.g., Arti- papillae markedly larger than others (1). In beus, Sturnira; ®g. 38C±E). This large pa- most phyllostomids, there are one or two pilla is located approximately on the horny papillae that are larger than the others midsagittal line of the tongue. In Micronyc- (®g. 38C±G). In contrast, Lonchorhina, Mac- teris minuta, Mimon crenulatum, most glos- rophyllum, Vampyrum, and Rhinophylla do sophagines (®g. 38F), and some stenoder- not have one or two horny papillae that are matines (e.g., Ardops, Pygoderma; ®g. 38G), larger than others; rather, all the horny pa- there are two large papillae in the cluster, one pillae are roughly equal in size (®g. 38A±B). on either side of the midline of the tongue. In mormoopids and Noctilio, the horny pa- In mormoopids and Noctilio, all horny pa- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 113

Fig. 38. Close-up dorsal views of the clusters of horny papillae, including central, anterior, ¯anking, and posterior, in selected phyllostomids. A. Vampyrum spectrum (AMNH 202292). B. Lonchorhina aurita (AMNH 149218). C. Phyllostomus hastatus (AMNH 233176). D. Erophylla sezekorni (AMNH 186977). E. Platalina genovensium (AMNH 257108). F. Glossophaga soricina (AMNH 214417). G. Ametrida centurio (AMNH 267375). Scale bar ϭ 1 mm. pillae are roughly equal in size. Consequent- Grif®ths (1982) initially noted, these patterns ly, these taxa are scored ``Ϫ'' for this char- are potentially phylogenetically informative. acter and the primitive state for phyllostom- It is evident that the number and distribution ids cannot be determined a priori. We scored of horny papillae are consistent within spe- taxa that have a V-shaped cluster of horny cies/genera and vary among them. Other au- papillae (state 1 of character 121) or those thors have also found consistent patterns in with papillae of uniform size in the cluster phyllostomid horny papillae. For example, (state 0 of character 122) ``Ϫ'' for this char- Greenbaum and Phillips (1974) noted that all acter. specimens of Leptonycteris they examined Grif®ths (1982; character 16) was the ®rst had two large central horny papillae. Park to use arrangement of the horny papillae in and Hall (1951: 65) reported that there were a cladistic analysis, although Smith and ``Generally . . . seven papillae in a group, Hood (1984: 449) later dismissed this effort two to four of which are always larger than as a ``Rorschach analysis'' of horny papillae the others and horny instead of ¯eshy.'' Al- patterns. Our observations suggest that, as though, we found that there was no differ- 114 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 ence in keratinization for the large and small tern of three to four papillae in a transverse horny papillae, Park and Hall's (1951) com- row with smaller papillae anterior to this ments demonstrate that they also observed row. However, the Anoura that we examined some consistent pattern. Even Smith and have a single large central papilla surrounded Hood (1984) indicated that a consistent pat- by smaller papillae. Gimenez et al. (1996) tern appeared to be present in Lionycteris and scored Choeroniscus, Choeronycteris, Hy- Lonchophylla. These authors observed a lonycteris, and Lichonycteris as having the 3/ large anterior papilla with two large papillae 4 pattern (state 2), three small papillae an- posterior to it. terior to four larger papillae, the central two Our de®nitions of characters and character of which are largest. Finally, they described states dealing with these papillae differ sub- all other phyllostomids as having a poorly stantially from Grif®ths' (1982) treatment. de®ned state (0): ``8±15 main papillae ar- Grif®ths (1982: character 16) identi®ed a sin- ranged in a circular to nearly square cluster'' gle pattern of interest, the ``3/4 pattern'' (one (Gimenez et al., 1996: character 12). row of 3 large papillae and a second row of We divided the horny papillae pattern into 4 large papillae), which he scored as a binary several characters (124 to 127) in such a way character. Thus, Grif®ths (1982) grouped as to allow scoring of the consistent variation several unique patterns (see characters 122 to that we observed. The ®rst of these (this 127) into a single, plesiomorphic state de- character) describes the number of large pa- ®ned by absence of a recognizable 3/4 pat- pillae (what we term the ``main papillae''), tern (Smith and Hood, 1984). We found that which provide a landmark for descriptions of many of the taxa in which Grif®ths (1982: the remaining horny papillae (anterior, pos- 31) recognized a ``less orderly'' patch of terior, or lateral to the main papillae). We be- horny papillae (e.g., Macrotus, many sten- lieve, based on position and overall similar- odermatines) are characterized by distinctive ity, that each of the character states de®ned patterns. Grif®ths (1982) observed two large here describes a potentially homologous pat- papillae in Carollia, whereas we found one. tern. Gimenez (1993: character 3.3.5.13) followed Character 125: Three small papillae Grif®ths (1982) in recognizing only the 3/4 present anterior to main papilla(e) (0); or pattern, again scoring this as a binary char- one papilla present (1); or no papillae pre- acter. She scored only Choeroniscus, Cho- sent (2). Among phyllostomids with an ellip- eronycteris, and Hylonycteris with the de- tical cluster of horny papillae, many have rived condition (the presence of the 3/4 pat- one or more papillae located anterior to the tern). In all other taxa, she reported that this main papilla(e). Three small papillae are pre- pattern was absentÐclearly at odds with our sent anterior to the main papilla(e) in description of horny papillae patterns, and phyllostomines (®g. 38C), Brachyphylla (ex- the previous report by Grif®ths (1982). cept AMNH 213731 in which the anterior- Gimenez et al. (1996: character 12) sub- most appears to be missing), some glosso- divided the patterns observed among nectar- phagines (e.g., Choeroniscus, Lichonycteris), feeding phyllostomids and their allies into Carollia, and all stenodermatines (®g. 38G). four states. They described state 1 as a tri- The middle of the three papillae is usually angular arrangement of three papillae, one slightly anterior to the other two and is cen- small papilla anterior to two larger ones (Gi- tered roughly on the midline of the tongue. menez et al., 1996). They scored Glossopha- We considered Brachyphylla AMNH 213731 ga, Monophyllus, Leptonycteris, and loncho- aberrant, and scored this taxon with state 0 phyllines as having this state. Based on Grif- in the matrix. In contrast, in some glosso- ®ths' (1982) descriptions, Gimenez et al. phagines (e.g, Glossophaga, Leptonycteris) (1996) also scored Erophylla and Phyllon- there is only a single papilla anterior to the ycteris with this state, although these taxa main papilla (®g. 38F). No papillae are pre- have one large papilla anterior to two smaller sent anterior to the main papilla(e) in phyl- ones. Gimenez et al. (1996) erected a sepa- lonycterines and lonchophyllines (®g 38D± rate state for Anoura (state 3), which they E). Mormoopids and Noctilio have horny pa- described as having an autapomorphic pat- pillae of uniform size, so the primitive state 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 115 for phyllostomids cannot be determined a sophaga, Leptonycteris, Monophyllus, and priori. We scored taxa that have a V-shaped lonchophyllines (®g. 38E±F). In some spec- cluster of horny papillae (state 1 of character imens of Phyllonycteris (e.g., AMNH 23762) 121) or those with papillae of uniform size these papillae are occasionally absent. Mor- in the cluster (state 0 of character 122) ``Ϫ'' moopids and Noctilio have horny papillae of for this character. uniform size; thus, the primitive state for The ``3/4'' pattern used by Grif®ths (1982: phyllostomids cannot be determined a priori. character 16) and Gimenez (1993; character We scored taxa that have a V-shaped cluster 3.3.5.13) included the presence of three small of horny papillae (state 1 of character 121) anterior papillae. Gimenez et al. (1996: table or those with papillae of different sizes in the 2) scored Erophylla and Phyllonycteris as cluster (state 1 of character 122) ``Ϫ'' for this having one anterior papilla. However, the character. `ànterior'' papilla in phyllonycterines is Grif®ths' (1982: character 16) 3/4 char- much larger than other surrounding horny acter included the presence of four papillae papillae, so we interpret it as the main papilla in the posterior row: two central large papil- of the cluster (see character 124). Conse- lae and two smaller ¯anking papillae. Simi- quently, we score phyllonycterines as lacking larly, the state described by Gimenez et al. small anterior papillae. Gimenez et al.'s (1996: character 12) as ``8±15 main papillae (1996; character 12) state ``8±15 main papil- arranged in a circular to nearly square clus- lae arranged in a circular to nearly square ter,'' apparently included a ¯anking pair. Our cluster,'' scored as present in phyllostomines, observations agree with those of previous au- Brachyphylla, and stenodermatines, appar- thors. ently included three papillae that we ob- Character 128: Paired lingual arteries served anterior to the central largest papil- present, lingual veins not enlarged (0); or la(e) in these taxa. If this is the case, our single, midline lingual artery present, lingual observations are largely in agreement with veins enlarged (1). A pair of lingual arteries those of Gimenez et al.(1996). is present in the tongue of Desmodus, phyl- Character 126: Two or more small horny lostomines, Brachyphylla, lonchophyllines, papillae present posterior to main papilla(e) Carollia, and stenodermatines (Grif®ths, (0); or absent (1). Two or more small horny 1982). The lingual veins are not enlarged in papillae are present posterior to the main pa- these taxa (Grif®ths, 1982). A single, midline pilla(e) in most phyllostomids (®g. 38C±E). lingual artery and a pair of enlarged lingual These are typically aligned on either side of veins are present in phyllonycterines and the midline of the tongue. In contrast, there glossophagines (Grif®th, 1982). The condi- are no horny papillae posterior to the main tion of the lingual vascular system has not papilla(e) in all glossophagines (®g. 38F), been reported for mormoopids or noctilion- except Anoura, and many stenodermatines ids, but most mammals have paired lingual (e.g., Uroderma, Sphaeronycteris; ®g. 38G). arteries (Grif®ths, 1982), suggesting that this Mormoopids and Noctilio have horny papil- condition may be primitive for phyllostom- lae of uniform size, so the primitive state for ids. phyllostomids cannot be determined a priori. Grif®ths (1982: characters 17±18) treated We scored taxa that have a V-shaped cluster the lingual vascular system in two characters: of horny papillae (state 1 of character 121) ``single midline lingual artery'' (present/ab- or those with papillae of uniform size in the sent) and `ènlarged lingual veins'' (present/ cluster (state 0 of character 122) ``Ϫ'' for this absent). However, these features may be character. functionally correlated, and they have iden- Character 127: Main horny papilla(e) tical taxonomic distributions. Accordingly, ¯anked by a pair of smaller horny papillae, we chose to treat the lingual arteries and one on each side (0); or no papillae present veins as part of a single character. lateral to main papilla(e) (1). Most phyllostomids have one small horny papillae located DIGESTIVE TRACT on each side of the main papilla(e) (®g. 38C± Forman (1971, 1973) and Forman et al. D, G). In contrast, these are absent in Glos- (1979) described the morphology of the di- 116 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 gestive tract in phyllostomids; taxa included manura species have ``numerous'' Brunner's in these studies are listed in table 4. Straney glands at the gastroduodenal juncture (For- (1980: character numbers are taken from his man et al., 1979: 225). Because it is dif®cult appendix 1) used these data to construct a to determine exactly which states should be character for use in a phylogenetic analysis considered homologous, and it seems likely (see discussion below). We did not collect to us that such descriptions may actually de- any new data observational data for this scribe states on a continuum, we have chosen character. to disregard such information until such data Character 129: Brunner's glands present can be more readily quanti®ed. at gastroduodenal juncture (0); or absent (1). Brunner's glands, which secrete mucus, REPRODUCTIVE TRACT are present at the gastroduodenal juncture in phyllostomines, glossophagines, and Sturni- The following characters are based on ra (Forman, 1973; Forman et al., 1979). Hood and Smith's (1982, 1983) descriptions Brunner's glands are absent in this region of and analysis of variation in the female repro- the stomach in Brachyphylla and the remain- ductive tract, and Straney's (1980: character ing stenodermatines (Forman, 1973; Forman numbers from appendix 1) descriptions of et al., 1979). In some species of Artibeus, the male reproductive tract (see table 4). We Dermanura, and Platyrrhinus, Brunner's have made only minor changes in the char- glands are present, whereas in other species acter descriptions presented by these authors in these genera these glands are absent. We and did not collect any additional data. score these genera with both states 0 and 1 Character 130: Male accessory gland in- in the matrix. Mormoopids and Noctilio have cludes compact white body situated antero- `ùnusually'' abundant Brunner's glands at medially (0); or compact white body absent the gastroduodenal juncture (Forman, 1971), (1). All phyllostomids have an accessory suggesting that presence of Brunner's glands gland (seminal vesicles and prostate) com- in this region may be the primitive state for posed of two parts: a pink, vesiculate, ante- phyllostomids. rior portion, and a white, compact, posterior Straney (1980: character A1) de®ned a portion (Straney, 1980). However, in some single binary character based on the number phyllostomines (e.g., Phyllostomus, Trach- of Brunner's glands distributed in the proxi- ops), Glossophaga, Carollia, and Sturnira, mal part of the duodenum, describing the de- the accessory gland includes a third part: a rived state as ``Brunner's glands present but compact white body situated anteromedially not numerous.'' Straney (1980) scored the (this body may be the seminal vesicles; Stra- taxon ``Phyllostomatinae'' with the primitive ney, 1980). In mormoopids, the accessory condition of this character (i.e., Brunner's gland is also tripartite with a pink, vesiculate, glands present and numerous). Straney anterior part; a compact, white, posterior (1980) scored all glossophagines, stenoder- part; and a compact, white anteromedial part. matines, carolliines, and brachyphyllines as The accessory bulb in Noctilio is not tripar- having Brunner's glands that were less nu- tite and is primarily pink and vesiculate. merous than those in the phyllostomines. However, a compact white body is present However, it is not easy to determine how and it is situated anteromedially (Straney, abundance of these glands should be treated. 1980). This distribution of character states Forman (1971: 274) described Brunner's suggests that presence of a compact white glands in mormoopids and noctilionids as anteromedial body is the primitive condition `ùnusually abundant,'' whereas glands in for phyllostomids. four phyllostomines were described by For- Straney (1980: characters B1±3, and 239) man et al. (1979: 225) as ``relatively numer- developed three binary characters based on ous.'' Anoura and Glossophaga have ``rela- the morphology of the male accessory gland: tively more abundant'' glands than Choero- ``(1) Prostate of two parts; caudal part dark, niscus and Lichonycteris do (Forman, 1971: vesiculate; cranial part small, white, com- 277), whereas some Artibeus species have pact; (2) Prostate three parted; as in (1) but ``sparse'' Brunner's glands and some Der- caudal part with a large, compact white re- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 117 gion; (3) Prostate of two parts; caudal part reduced to tubular intramural cornua (1); or white, compact; cranial part dark vesicu- completely fused to form a single common late.'' We did not use information pertaining lumen, no remnant of cornual lumina (2). to other parts of the male accessory gland The cornual lumina are distinct and join the because they do not appear to be informative common uterine lumen within the common at this taxonomic level (i.e., within phyllos- uterine body in Desmodus and some phyl- tomids). We followed Straney's (1980: 23± lostomines (e.g., Macrotus, Trachops; Hood 28) text descriptions to score taxa rather than and Smith, 1982, 1983; ®g. 39A±B). In con- his data matrix, because he seems to have trast, the cornual lumina are reduced to tu- scored mormoopids incorrectly in the matrix bular intramural uterine cornua (IUC) in oth- (indicating they have only a bipartite, not a er phyllostomines (e.g., Phyllostomus, Phyl- tripartite accessory gland). Although Straney loderma), Brachyphylla, Phyllonycteris, Lon- (1980) scored Macrophyllum in his matrix chophylla, glossophagines, and carolliines for this character, he did not describe or list (Hood and Smith, 1982, 1983; ®g. 39C). specimen numbers for this taxon; therefore, There is a single common uterine lumen and we did not score Macrophyllum for this char- no remnants of the cornual lumina in sten- acter. odermatines (Hood and Smith, 1982, 1983; Character 131: Uterine horns approxi- ®g. 39D). The cornual lumina are distinct mately one half length of common uterine and join the common uterine lumen within body (0); or one quarter length of common the common uterine body in mormoopids uterine body (1), or uterus fully simplex, and Noctilio (Hood and Smith, 1982, 1983), uterine horns not distinct from common body suggesting that this condition is primitive for (2). The length of the uterine horns is ap- phyllostomids. proximately one half the length of common Our scoring of this character generally fol- uterine body in Desmodus (Hood and Smith, lows that proposed by Hood and Smith 1982, 1983; ®g. 39A). External uterine fu- (1982: character 3), although we combined sion is somewhat greater in some phyllos- two character states de®ned by those authors tomines (e.g., Macrotus, Trachops), which (state 0: ``short common uterine lumen, cor- have uterine horns that are one quarter the nual lamina join within the common uterine length of the common uterine body (Hood body'' and state 1: ``large common uterine and Smith, 1982, 1983; ®g. 39B). In contrast, lumen, cornual lumina join immediately the uterus is simplex (the uterine horns are within the common uterine body'') because not distinct from common body of the uter- the former condition is found only in one us) in other phyllostomines (e.g., Phylloder- outgroup taxon (Noctilio). Our state 0 in- ma, Phyllostomus), Brachyphylla, Phyllo- cludes these two conditions. Hood and Smith nycteris, Lonchophylla, glossophagines, car- (1982) ordered transformations in this char- olliines, and stenodermatines (Hood and acter. Smith, 1982, 1983; ®g. 39C±D). The length Character 133: Oviducts enter lateral of the uterine horns is approximately one half border of uterine horns or body (0); or enter the length of common uterine body in mor- fundic border of uterine body (1); or enter moopids and Noctilio (Hood and Smith, fundic border of uterine body near the mid- 1982, 1983; ®g. 39A), suggesting that this sagittal line (2). The oviducts enter the lat- condition is primitive for phyllostomids. eral border of the uterine horns or body in Our scoring of this character follows that Desmodus and phyllostomines (Hood and proposed by Hood and Smith (1982: char- Smith, 1982, 1983; ®g. 39A±B). The location acter 1); however, we have eliminated their of this intersection is shifted anteromedially character state 0 (`ùterine horns 3/4 length so that the oviducts enter the fundic border of body''), because this state does not appear of the uterine body in Brachyphylla, Phyl- in any of the taxa in our study. Hood and lonycteris, glossophagines, Lonchophylla, Smith (1982) had ordered transformations in and carolliines (Hood and Smith, 1982, this character. 1983; ®g. 39C). In stenodermatines, the ovi- Character 132: Cornual lumina distinct, ducts enter the uterus more medially, near the join within the common uterine body (0); or midsagittal line on the fundic border (Hood 118 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 39. Semidiagrammatic, frontal sections of the female reproductive tract in three phyllostomids and one outgroup taxon representing the range of variation among noctilionoids (redrawn from Hood and Smith, 1983: ®gs. 3, 4, 6). A. Noctilio albiventris. B. Macrotus californicus. C. Leptonycteris curasoae. D. Artibeus jamaicensis. Abbreviations: cl: common uterine lumen; iuc: intramural uterine cornua; ov: ovary; ovd: oviduct; utj: uterotubal junction. Abbreviations follow those used by Hood and Smith (1982, 1983). and Smith, 1982, 1983; ®g. 39D). The ovi- or extends from ovary to lateral border of ducts enter the lateral border of the uterine uterus (1). The ovarian ligament extends horns or body in mormoopids and Noctilio from the ovary to the external entry of the (Hood and Smith, 1982, 1983; ®g. 39A), oviduct into the uterus in Desmodus, phyl- suggesting that this is the primitive condition lostomines, Brachyphylla, Phyllonycteris, for phyllostomids. glossophagines, and Lonchophylla (Hood This character was previously used in a and Smith, 1982, 1983; ®g. 40A). In contrast, phylogenetic analysis by Hood and Smith the ovarian ligament extends from the ovary (1982: character 5); our character is identical to the lateral border of uterus in carolliines to theirs, except we have not ordered trans- and stenodermatines (Hood and Smith, 1982, formations. 1983; ®g. 40B). The ovarian ligament ex- Character 134: Ovarian ligament extends tends from the ovary to the external entry of from ovary to external entry of oviduct (0); the oviduct into the uterus in mormoopids 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 119

Fig. 40. Diagram illustrating the two different types of attachment of the ovary to the uterus via the ovarian ligament: A. to the external oviductal entry, or B. to the lateral uterine wall (redrawn from Hood and Smith, 1983: ®g. 15). and Noctilio (Hood and Smith, 1982, 1983; is also absent in Mormoops, Pteronotus per- ®g. 40A), suggesting that this is the primitive sonatus, and P. gymnotus, and Noctilio (Wi- condition for phyllostomids. ble and Bhatnagar, 1996). The accessory ol- This character is identical to that used by factory bulb is present in Pteronotus parnel- Hood and Smith (1982: character 6) in their lii (Wible and Bhatnagar, 1996). Therefore, phylogenetic analysis. the primitive state for phyllostomids appears to be absence of the accessory olfactory BRAIN bulb. The following characters are based on de- Wible and Bhatnagar (1996) ®rst de®ned scriptions of brain morphology provided by and mapped this character onto preexisting McDaniel (1976), Bhatnager (1988), Baron phylogenies of bats; however, they expressed et al. (1996a, 1996b), and Wible and Bhat- some reservations concerning its use because nagar (1996; see table 4). Straney (1980: the absence of the accessory olfactory bulb character numbers from his appendix 1) used is perfectly correlated with a reduced or ab- features of the brain in an analysis of phyl- sent vomeronasal epithelial tube (VET) in lostomine relationships, but did not present mammals (i.e., a VET that lacks neuroepi- new observational data. thelium; see character 39). However, because Character 135: Accessory olfactory bulb reduced VET have structures that indicate a absent (0); or present (1). In most phyllos- sensory function (e.g., vomeronasal nerves, tomids, the accessory olfactory bulb, the part paravomeronasal ganglia, and receptorlike of the forebrain receiving sensory informa- cells), Wible and Bhatnagar (1996) consid- tion from the vomeronasal organ, is well de- ered this character separately from the VET veloped (Wible and Bhatnagar, 1996). In character. We do the same here, adding that contrast, the accessory olfactory bulb is ab- the accessory olfactory bulb has been de- sent in Brachyphylla (Wible and Bhatnagar, scribed in many more taxa than has the vom- 1996), Choeroniscus, and Rhinophylla (Bar- eronasal organ. Because we are not certain on et al., 1996). The accessory olfactory bulb of the state of the VET in these taxa (i.e., the 120 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

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 Wi- ble 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 ®ssure or inferior colliculi (0); or cerebellar vermis completely covers longitudinal ®ssure 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; ®g. 41A; McDaniel, 1976; Baron et al., 1996a, 1996b). The medial longitudinal ®s- sure 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 ®ssure between the inferior colliculi in Diphylla, some phyllostomines (e.g., species of Micronycteris), phyllonycterines, and many stenodermatines (e.g., Centurio, Me- sophylla; McDaniel, 1976; Bhatnager, 1988; Baron et al., 1996a, 1996b; ®g. 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), Brachyphyl- la, Lonchophylla, glossophagines, carolli- Fig. 41. Semidiagrammatic illustrations of the ines, and many stenodermatines (e.g., En- dorsal brain in three phyllostomids illustrating the chisthenes, Uroderma; McDaniel, 1976; range of variation in coverage of the inferior col- Bhatnager, 1988; Baron et al., 1996a, 1996b; liculi (ic) by the cerebellar vermis. A. Mimon ®g. 41C). No portion of the inferior colliculi crenulatum. B. Mesophylla macconnelli. C. Li- chonycteris obscura (drawn from McDaniel, is visible in these taxa. Taxonomic polymor- 1976: ®gs. 2, 20, 42). phism occurs in some genera. In Phyllosto- mus, only P. elongatus has dorsally exposed inferior colliculi whereas all other species have complete coverage (McDaniel, 1976). 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 121

Fig. 42. Diagramatic dorsal view (left) and longitudinal section of one hemisphere (right) of the chiropteran brain illustrating the progressive coverage of the median longitudinal ®ssure and inferior colliculi (ic). A. Complete exposure of the median longitudinal ®ssure and inferior colliculi. B. Partial coverage of both the ®ssure and inferior colliculi. C. Complete coverage of the ®ssure and colliculi. Note that the cerebellar vermis covers the inferior colliculi, but the relationship of the inferior colliculi to each other does not change (redrawn from Schober and Brauer, 1974: ®g. 99)

Accordingly, we score Phyllostomus with contiguous at midline'' and ``Inferior collic- states 0 and 2 in the matrix. In Artibeus, Der- uli separate at midline.'' Although the latter manura, Platyrrhinus, and Sturnira some is misleading (the inferior colliculi are not species in each genus have partly exposed actually separated, just covered medially; ®g. inferior colliculi, whereas their congeners 42), the derived conditions of these charac- have completely covered inferior colliculi ters correspond to our states 0 and 1. Straney (McDaniel, 1976); we score these taxa with (1980) did not score complete coverage of states 1 and 2 in the matrix. The inferior col- the inferior colliculi by cerebral and cerebel- liculi are only partly exposed in Mormoops, lar tissues. However, Straney (1980) left Pteronotus, and Noctilio albiventris, but are blank spaces in his matrix to indicate that the fully exposed dorsally in Noctilio leporinus condition of the inferior colliculi was un- (Baron et al., 1996a, 1996b). This distribu- known in glossophagines and carolliines be- tion of character states suggests that partial cause they were completely covered by the exposure of the inferior colliculi is the prim- cerebellar vermis. In his matrix, Straney itive condition for phyllostomids. (1980) appears to have mistakenly scored Straney (1980: characters D1±2) treated desmodontines, brachyphyllines, and sten- dorsal exposure of the inferior colliculi in odermatines for the derived conditions of two binary characters: ``Inferior colliculi both of his characters even though he cor- 122 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 rectly notes in his text (Straney, 1980: 39) of Carollia, and many stenodermatines (e.g., that the inferior colliculi, when they can be Artibeus, Ametrida) have an XX/XY1Y2 sex viewed in these taxa, are separated. system (Baker, 1967, 1970, 1973, 1979; Hsu Our scoring of this character agrees with et al., 1968; Baker and Hsu, 1970; Baker and McDaniel's (1976) descriptions, with the ex- Lopez, 1970; Baker and Bleier, 1971; Green- ception of Artibeus jamaicensis, which we baum et al., 1975; Stock, 1975; Gardner, have scored as state 1, and McDaniel de- 1977a; Baker et al., 1979, 1981b, 1982; scribed as having state 2. We scored Artibeus Johnson, 1979; Myers, 1981). In these taxa, jamaicensis, and several other taxa, based on the X chromosomes are translocated to au- ®gures (e.g., McDaniel, 1976; Bhatnager, tosomes. Thus males have an unpaired Y 1988) when text descriptions were not clear chromosome (because the X is translocated about the exposure and contiguity of the in- to an autosome), and the homolog of the au- ferior colliculi. Our scoring of Artibeus ja- tosome carrying the X is referred to as a sec- maicensis agrees with the description of this ond Y. Some individuals of Carollia casta- taxon by Schober and Brauer (1974). nea, and C. brevicauda, C. perspicillata, and

C. subrufa have an XX/XY1Y2 system (Bak- SEX CHROMOSOMES er, 1967, 1979; Hsu et al., 1968; Baker and Bleier, 1971; Patton and Gardner, 1971; The following character is based on de- Stock, 1975; Baker et al., 1982). Choeron- scriptions of chromosome morphology pro- iscus godmani has an XX/XY1Y2 system, vided by Baker (1967, 1970, 1973, 1979), whereas C. minor (ϭ C. intermedius) does Forman et al. (1968), Hsu et al.(1968), Baker not (Baker, 1967, 1970, 1979; Patton and and Lopez (1970), Baker and Bleier (1971), Gardner, 1971; Stock, 1975; Baker et al., Patton and Gardner (1971), Stock (1975), 1982). Dermanura cinerea has an XX/

Cadena and Baker (1976), Gardner (1977a), XY1Y2 system, whereas other species in this Bass (1978), Patton and Baker, (1978), Baker genus do not. Thus, we score Carollia with and Bass (1979), Baker et al. (1979, 1981b, states 1 and 2 in the matrix, and score Cho- 1982), Johnson (1979), Myers (1981) (for a eroniscus and Dermanura with states 0 and complete list of taxa covered by karyological 1. Mormoopids (Baker, 1967; Baker and studies see Baker et al., 1982). We have not Hsu, 1970; Baker and Lopez, 1970; Patton collected any additional observational data and Baker, 1978; Sites et al., 1981) and Noc- for this character. tilio (Patton and Baker, 1978) have XX/XY Character 137: Sex-determining system sex chromosomes, suggesting that this is the XX/XY; X chromosomes never translocated primitive condition for phyllostomids. to autosomes (0); or sex determining system Although several authors have drawn con-

XX/XY1Y2; X chromosomes translocated to clusions about phylogenetic relationships us- autosomes (1); or X chromosomes sometimes ing the sex chromosomes (e.g., Baker and translocated to autosomes; polymorphic Lopez, 1970; Greenbaum et al., 1975), this within species (2). Most phyllostomids, like is the ®rst use of this feature as a character most other mammals, have an XX/XY sex- in a cladistic analysis. determining system (Baker, 1967, 1970, 1973, 1979; Hsu et al., 1968; Forman et al., RESTRICTION SITES 1968; Baker and Hsu, 1970; Baker and Lo- pez, 1970; Baker and Bleier, 1971; Baker et The following characters are based on a al. 1973; Nagorsen and Peterson, 1975; Cad- study of restriction sites present in the tran- ena and Baker, 1976; Gardner, 1977a; Bass, scribed portion of the rDNA complex con- 1978; Patton and Baker, 1978; Baker and ducted by Van Den Bussche (1991, 1992; see Bass, 1979; Baker et al., 1979; Johnson, table 4; see ®g. 43 for all sites). We have not 1979; Honeycutt et al., 1980; Baker et al., collected any additional data for these char- 1981a, 1981b, 1982). In species with this acters. Restriction site numbers are those es- system, males and females have an equiva- tablished by Van Den Bussche (1991, 1992) lent number of chromosomes. However, and are used to facilitate discussion. some species of Choeroniscus, some species Character 138: Restriction site 28 present 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 123

Fig. 43. Restriction site map of the transcribed portion of the rDNA complex, including all sites from Van Den Bussche's (1991) original study. Restriction sites variable within noctilionoids and Homo are below the line, invariant (those found in all taxa in the study) are above the line (redrawn from Van Den Bussche, 1991: ®g. 2).

(0); or absent (1). In most phyllostomids, an in the ETS (Van Den Bussche, 1991, 1992). Xba I restriction site (Van Den Bussche's However, in Choeroniscus, Choeronycteris, character 28) is present in the external tran- Hylonycteris, and Musonycteris, this restric- scribed spacer (ETS; Van Den Bussche, tion site is absent (Van Den Bussche, 1991, 1991, 1992). However, in Platyrrhinus, 1992). This site is present in Mormoops and Uroderma, and Vampyrodes, this restriction Noctilio (Van Den Bussche, 1991, 1992), site is absent (Van Den Bussche, 1991, suggesting that this is the primitive condition 1992). This site is present in Mormoops and for phyllostomids. Noctilio (Van Den Bussche, 1991, 1992), Character 141: Restriction site 38 present suggesting that this is the primitive condition (0); or absent (1). An Sst I restriction site for phyllostomids. (character 38) is present within the 18S gene Our scoring of this character and those that in most phyllostomids (Van Den Bussche, follow are identical to that of Van Den 1991, 1992). In contrast, this restriction site Bussche (1991, 1992: character 28); howev- is absent in Ametrida, Ardops, Ariteus, Cen- er, Van Den Bussche (1991, 1992) scored turio, Pygoderma, and Stenoderma (Van Den Hylonycteris ``9'' for this character and char- Bussche, 1991, 1992). This site is present in acters 139, 141, 142, 143, 146, and 148 be- Mormoops and Noctilio (Van Den Bussche, low to indicate that there was missing data 1991, 1992), suggesting that this is the prim- for this taxon, whereas we use ``?'' for this itive condition for phyllostomids. purpose. Character 142: Restriction site 43 present Character 139: Restriction site 32 present (0); or absent (1). In most phyllostomids, an (0); or absent (1). A Hinc II restriction site Nco I restriction site (character 43) is present (character 32) is present in the ETS of most in the internal transcribed spacer (ITS; Van phyllostomids (Van Den Bussche, 1991, Den Bussche, 1991, 1992). However, in 1992). However, in Chrotopterus, Mimon Chrotopterus and Trachops, this restriction crenulatum, Artibeus, Chiroderma, and Stur- site is absent (Van Den Bussche, 1991, nira this restriction site is absent (Van Den 1992). This site is present in Mormoops and Bussche, 1991, 1992). The Hinc II (character Noctilio (Van Den Bussche, 1991, 1992), 32) restriction site is present in Mormoops suggesting that this is the primitive condition and Noctilio (Van Den Bussche, 1991, 1992), for phyllostomids. suggesting that this is the primitive condition Character 143: Restriction site 46 present for phyllostomids. (0); or absent (1). Most phyllostomids have Character 140: Restriction site 36 present an Nco I restriction site (character 46) in the (0); or absent (1). In most phyllostomids a ITS (Van Den Bussche, 1991, 1992). In con- Xho I restriction site (character 36) is present trast, this restriction site is absent in Ariteus, 124 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Pygoderma, and Stenoderma (Van Den Bus- suggesting that this is the primitive condition sche, 1991, 1992). This site is present in for phyllostomids. Mormoops and Noctilio (Van Den Bussche, Character 148: Restriction site 53 present 1991, 1992), suggesting that this is the prim- (0); or absent (1). In most phyllostomids, a itive condition for phyllostomids. Bgl II restriction site (character 53) is present Character 144: Restriction site 47 present in the ITS (Van Den Bussche, 1991, 1992). (0); or absent (1). In most phyllostomines, a In contrast, this restriction site is absent in Stu I restriction site (character 47) is present Lonchophylla and Lionycteris (Van Den Bus- in the ITS (Van Den Bussche, 1991, 1992). sche, 1991, 1992). This restriction site is However, this restriction site is absent in present in Mormoops and Noctilio (Van Den Macrophyllum, Brachyphylla, Phyllonycter- Bussche, 1991, 1992), suggesting that this is is, lonchophyllines, glossophagines, carol- the primitive condition for phyllostomids. lines, and stenodermatines (Van Den Bus- Character 149: Restriction site 54 present sche, 1991, 1992). This site is present in (0); or absent (1). In glossophagines, lon- Mormoops and Noctilio (Van Den Bussche, chophyllines, and stenodermatines except 1991, 1992), suggesting that this is the prim- Sturnira,aXho I restriction site (character itive condition for phyllostomids. 54) is present in the ITS (Van Den Bussche, Character 145: Restriction site 49 present 1991, 1992). However, in desmodontines, (0); or absent (1). In most phyllostomids, a phyllostomines, Brachyphylla, Phyllonycter- Pvu II restriction site (character 49) is pres- is, and Sturnira, this restriction site is absent ent in the ITS (Van Den Bussche, 1991, (Van Den Bussche, 1991, 1992). This restric- 1992). However, in Choeroniscus, Choeron- tion site is present in Mormoops and Noctilio ycteris, Hylonycteris, and Musonycteris, this (Van Den Bussche, 1991, 1992), suggesting restriction site is absent (Van Den Bussche, that this is the primitive condition for phyl- 1991, 1992). This site is present in Mor- lostomids. moops and Noctilio (Van Den Bussche, 1991, 1992), suggesting that this is the primitive condition for phyllostomids. ECORI-DEFINED DNA REPEAT Character 146: Restriction site 50 present (0); or absent (1). A BamH I restriction site Van Den Bussche et al. (1993) described (character 50) is present in the ITS in most the structure of an EcoRI-de®ned nuclear sat- phyllostomids (Van Den Bussche, 1991, ellite DNA repeat in phyllostomid bats; taxa 1992). In contrast, in many phyllostomines included in this study are listed in table 4. (e.g., Phyllostomus, Trachops), Leptonycter- We did not collect any new data for this char- is, Monophyllus, and lonchophyllines this re- acter. striction site is absent (Van Den Bussche, Character 150: 900 base pair EcoRI-de- 1991, 1992). This restriction site is present ®ned nuclear satellite DNA repeat absent in Mormoops and Noctilio (Van Den Bus- (0); or present (1).AnEcoRI-de®ned nuclear sche, 1991, 1992), suggesting that this is the satellite DNA repeat is present in Artibeus, primitive condition for phyllostomids. Dermanura, and Koopmania (Van Den Bus- Character 147: Restriction site 52 present sche et al., 1993). However, this repeat is ab- (0); or absent (1). Most phyllostomids have sent in all other phyllostomids (Van Den a Stu I restriction site (character 52) present Bussche et al., 1993). The repeat is absent in in the ITS (Van Den Bussche, 1991, 1992). Mormoops and Noctilio, suggesting that this However, this restriction site is absent in is the primitive condition for phyllostomids. Choeroniscus, Choeronycteris, Hylonycteris, Van Den Bussche et al. (1993) discussed and Musonycteris (Van Den Bussche, 1991, the phylogenetic implications of the taxo- 1992). This site is present in Mormoops and nomic distribution of this feature, but did not Noctilio (Van Den Bussche, 1991, 1992), formally de®ne and score it as a character. 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 125

RESULTS TAXONOMIC CONGRUENCE for all taxa resulted in more than 30,000 most ANALYSIS parsimonious trees (119 steps each, CI ϭ 0.555, RI ϭ 0.832). A strict consensus of As described in the Materials and Methods these trees is shown in ®gure 45. In this tree, section, we divided our data set into a series Phyllostominae is monophyletic and appears of partitions to facilitate comparisons with as the sister taxon of a clade comprising the previous studies and to evaluate the utility of remaining phyllostomids. The basal node of each partition for resolving relationships at the latter clade is a polytomy of Brachy- different hierarchical levels and in different phylla, Sturnira, a clade that includes des- parts of the tree. Results of the separate anal- modontines and numerous stenodermatines, yses of the partitions are presented below, and another clade that includes phyllonycter- followed by a summary. The entire data ma- ines, glossophagines, lonchophyllines, carol- trix is presented in appendix 2; data on poly- liines, and the remaining stenodermatines. morphism and completeness is presented in The desmodontine-stenodermatine group table 5. includes three clades: (1) the desmodontines, Pelage and Integument: A heuristic with Desmodus and Diaemus as sister taxa, search using 38 unordered characters for all (2) Platyrrhinus and Uroderma, and (3) taxa resulted in more than 30,000 most par- Ametrida and Sphaeronycteris. Relationships simonious trees (183 steps each, CI ϭ 0.503, among these groups and the other genera in RI ϭ 0.789). A strict consensus of these trees this clade are unresolved (see ®g. 45). is shown in ®gure 44. Desmodontinae ap- Carollia occupies the ®rst branch of the pears as a monophyletic group in this tree; remaining large clade. The positions of sev- Desmodus and Diaemus form a clade. The desmodontines are the sister taxon of a po- eral taxa within the larger group are unre- lytomy including Brachyphylla, Erophylla, solved: Erophylla, Phyllonycteris, Platalina, Phyllonycteris, and a clade of the remaining Rhinophylla, and Chiroderma are part of a phyllostomids. Within this latter group the polytomy that also includes a clade of Ec- positions of many taxa are unresolved. How- tophylla, Mesophylla, and the three Vampy- ever, four clades occur in all 30,000 trees: ressa species (no resolution in this clade), (1) a clade of glossophagines and loncho- and another clade including Lonchophylla, phyllines, (2) a group composed of two pairs Lionycteris, and all glossophagines. Loncho- of sister taxa, Chrotopterus and Vampyrum, phylla and Lionycteris are basally unresolved and Tonatia and Trachops, (3) a clade in within this group. At the next node, Glos- which Micronycteris minuta, Macrotus, and sophaga, Leptonycteris and Monophyllus M. brachyotis are the successive sister taxa form a polytomy with a monophyletic group of M. hirsuta and M. megalotis, and (4) a of the remaining glossophagines. Within this clade of ``short-faced'' stenodermatines in group, Choeroniscus, Choeronycteris, and which Centurio and Sphaeronycteris are sis- Musonycteris form a clade. Lichonycteris ter taxa. and Scleronycteris form a polytomy with the Glossophagines and lonchophyllines are Choeroniscus group. Hylonycteris and An- each monophyletic. Within the glossopha- oura are successive sister taxa to the Licho- gine clade, Glossophaga, Monophyllus, and nycteris-Choeroniscus group. a clade of the remaining genera form a po- Postcranium: A heuristic search using 16 lytomy. This last group is largely unresolved, unordered postcranial characters for all taxa with the exception of two pairs of sister taxa: resulted in more than 30,000 most parsimo- Anoura and Leptonycteris, and Hylonycteris nious trees (68 steps each, CI ϭ 0.441, RI ϭ and Lichonycteris. Within the lonchophylline 0.764). The strict consensus of these trees clade, Lionycteris and Lonchophylla are sis- (not shown) is completely unresolved. ter taxa. Hyoid Apparatus: A heuristic search us- Skull and Dentition: A heuristic search ing 17 unordered hyoid characters for 27 taxa using 35 unordered craniodental characters (see table 4) resulted in 24 most parsimoni- 126 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 127

ous trees (40 steps each, CI ϭ 0.625, RI ϭ Platalina, and a clade of glossophagines also 0.893); a strict consensus of these trees is form a polytomy. Within the glossophagine shown in ®gure 46. In this tree, there are two clade, there is a sister group relationship be- large clades of phyllostomids, one composed tween Monophyllus and Glossophaga, which of Desmodus, Brachyphylla, phyllonycter- form a clade with Lichonycteris. A second ines, glossophagines, and lonchophyllines, clade contains the sister taxa Anoura and and the other composed of phyllostomines, Leptonycteris, the sister taxa Choeroniscus carolliines, and stenodermatines. In the for- and Choeronycteris, and Hylonycteris. mer clade, Desmodus occupies the basal In the other large clade, Carollia and Stur- branch. Brachyphylla, the sister taxa Ero- nira are successive sister taxa to the group phylla and Phyllonycteris, and a clade com- that includes phyllostomines and other sten- prising glossophagines and lonchophyllines odermatines. In the phyllostomine-stenoder- form a polytomy. Lionycteris, Lonchophylla, matine clade, Macrotus and Phyllostomus are 128 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 44. Results of a heuristic search of 38 pelage and integument characters for all 63 taxa. The tree shown here is a strict consensus of more than 30,000 most parsimonious trees, each of 183 steps (CI ϭ 0.503, RI ϭ 0.789). 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 129

Fig. 46. Results of a heuristic search of 17 hyoid characters for 27 taxa. The tree shown here is a strict consensus of 24 most parsimonious trees, each of 40 steps (CI ϭ 0.625, RI ϭ 0.893).

sister taxa. Relationships among the remaining genera are unresolved. Tongue: A heuristic search using 22 unordered tongue characters for 61 taxa (Mu- sonycteris and Scleronycteris were excluded; see Character Descriptions: Tongue) resulted in more than 30,000 most parsimonious trees (55 steps each, CI ϭ 0.564, RI ϭ 0.876); a strict consensus of these trees is shown in ®gure 47. Although the positions of many taxa in this tree are unresolved, a few clades appear in all 30,000 trees. Desmodontinae, Glossophaginae, and Lonchophyllinae are each monophyletic. The relationships among these groups and the genera of phyllostomines, brachyphyllines, phyllonycterines, carolliines, Sturnira, and a clade consisting of the remaining stenodermatines could not be resolved using these data. Within the monophyletic subfamilies, several additional relationships are apparent. Within Desmodontinae, Desmodus and Diae- Fig. 45. Results of a heuristic search of 35 mus are sister taxa. Anoura appears as the craniodental characters for all 63 taxa. The tree sister taxon of the remaining glossophagines. shown here is a strict consensus of more than Within this glossophagine clade, the relation- 30,000 most parsimonious trees, each of 119 steps ships of Choeroniscus and Choeronycteris (CI ϭ 0.555, RI ϭ 0.832). 130 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 47. Results of a heuristic search of 22 tongue characters for 61 taxa. The tree shown here is a strict consensus of more than 30,000 most parsimonious trees, each of 55 steps (CI ϭ 0.564, RI ϭ 0.876). 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 131 are unresolved. These two genera form a polytomy with (1) a clade comprising Glosso- phaga, Leptonycteris, and Monophyllus, and (2) the sister taxa Hylonycteris and Licho- nycteris. All stenodermatines, except Sturnira, form a clade. Within this large group, the relationships of most taxa remain unresolved. How- ever, all ``short-faced'' stenodermatines form a clade and within this group Centurio and Sphaeronycteris are sister taxa. Successive sister taxa to the ``short-faced'' clade are Ec- tophylla, Platyrrhinus, Uroderma, and Vam- pyrodes. Digestive Tract: A heuristic search using 1 unordered character for 24 taxa (see table 4) resulted in 51 most parsimonious trees (4 steps each, CI ϭ 1.000, RI ϭ 1.000). The strict consensus of these trees (not shown) is completely unresolved. Reproductive Tract: A heuristic search using 5 unordered characters for 33 taxa (see table 4) resulted in more than 30,000 most parsimonious trees (10 steps each, CI ϭ 0.800, RI ϭ 0.966). The strict consensus of these trees (not shown) is completely unresolved. Brain: A heuristic search using 2 unordered characters for 52 taxa (see table 4) resulted in more than 30,000 most parsimonious trees (11 steps each, CI ϭ 0.909, RI ϭ 0.960). The strict consensus of these trees (not shown) is completely unresolved. Fig. 48. Results of a heuristic search of 12 Sex Chromosomes: A heuristic search us- restriction site characters for 44 taxa. The tree ing 1 unordered character for 56 taxa (see shown here is a strict consensus of 120 most par- table 4) resulted in 51 most parsimonious simonious trees, each of 17 steps (CI ϭ 0.706, RI trees (4 steps each, CI ϭ 1.000, RI ϭ 1.000). ϭ 0.917). The strict consensus of these trees (not shown) is completely unresolved. Restriction Sites: A heuristic search using Lonchophylla. Choeroniscus, Choeronycter- 12 unordered characters for 44 taxa (see table is, Hylonycteris, and Musonycteris form a 4) resulted in 120 most parsimonious trees polytomy. Uroderma, Platyrrhinus, and (17 steps, CI ϭ 0.706, RI ϭ 0.917); a strict Vampyrodes form a clade, as do all ``short- consensus of these trees is shown in ®gure faced'' stenodermatines. Within the ``short- 48. Although the relationships of many taxa faced'' group, Ariteus, Pygoderma, and Sten- could not be resolved using these data, sev- oderma form a clade. eral clades occur in all 120 most parsimoni- EcoRI-De®ned DNA Repeat: A heuristic ous trees. A clade of phyllostomines in- search using 1 unordered character for 30 cludes: Chrotopterus, Lonchorhina, Micro- taxa (see table 4) resulted in 1 most parsi- nycteris minuta, Mimon crenulatum, Phyllos- monious tree (1 step, CI ϭ 1.000, RI ϭ tomus, Tonatia, and Trachops. In another 1.000). The strict consensus of these trees clade, Monophyllus and Leptonycteris form a (not shown) includes a single clade of Arti- polytomy with the sister taxa Lionycteris and beus, Dermanura, and Koopmania. 132 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Summary: In the strict consensus trees of CHARACTER CONGRUENCE four data partitions, Glossophaginae appears ANALYSIS as monophyletic group (see table 6 for this and other clades). Glossophaginae is found We used the entire data matrix in our char- in the strict consensus trees from the pelage acter congruence analysis. The data matrix is and integument, craniodental, hyoid, and presented in appendix 2; data on polymorphism and completeness is presented in table tongue analyses. Monophyly of Desmodon- 5. A heuristic search using all 150 unordered tinae is supported by analyses of the pelage characters resulted in 18 optimal trees (613 and integument, craniodental, and tongue steps; CI ϭ 0.462; RI ϭ 0.765). The strict partitions. Monophyly of Lonchophyllinae is consensus of these trees is shown in ®gure supported by three data partitions (pelage 49. Monophyly of Desmodontinae (decay and integument, tongue, restriction sites). value: 8; bootstrap: 100) is strongly support- Both Phyllonycterinae and Phyllostominae ed. Within Desmodontinae, Desmodus and appear as monophyletic groups in the strict Diaemus are sister taxa; this relationship is consensus trees of a single data partition (hy- strongly supported with a decay value of 6 oid and craniodental, respectively). Surpris- and a bootstrap value of 100. ingly, Carolliinae and Stenodermatinae are Desmodontinae is the ®rst (most basal) not monophyletic in the strict consensus trees branch within Phyllostomidae. At the next of any of these data partitions. However, node, there is a polytomy consisting of (1) ``short-faced'' stenodermatines appear as a Brachyphylla, (2) a clade comprising phyl- monophyletic group in the strict consensus lonycterines, lonchophyllines, and glosso- trees of three data partitions. A number of phagines, and (3) a clade that includes phyl- other higher-level clades appear in the strict lostomines, carollines, and stenodermatines. consensus trees of single data partitions (see This large clade has a decay value of 2 and table 6). a bootstrap value of 36. In nine of the most There is some disagreement between the parsimonious trees, Brachyphylla appears as hyoid partition and the pelage and integu- the sister taxon of the phyllostomine, carol- ment partition concerning relationships liine, and stenodermatine group. In the other among the nectar-feeding taxa. The hyoid nine trees, Brachyphylla is the sister taxon of partition unites brachyphyllines, phyllonyc- a more inclusive clade that includes the phyl- terines, glossophagines, and lonchophyllines lonycterines, lonchophyllines, glossopha- in a single clade, and unites this clade with gines, phyllostomines, carollines, and steno- desmodontines. In contrast, glossophagines dermatines. and lonchophyllines form a clade with car- Phyllonycterinae (decay: 7; bootstrap: olliines, phyllostomines, and stenodermatines 100), Lonchophyllinae (decay: 3; bootstrap: 92), and Glossophaginae (decay: 4; boot- in the pelage and integument tree (see ®g. strap: 95) are each monophyletic. Phyllon- 44). Brachyphyllines and phyllonycterines ycterinae is the sister taxon of the very are the sister taxa of this large clade. The strongly supported lonchophylline-glosso- pelage and integument partition and the cran- phagine clade (decay value: 7; bootstrap val- iodental partition also place different taxa as ue: 96). Within Lonchophyllinae, Loncho- the most basal branch of the family (Des- phylla and Lionycteris are sister taxa, a re- modontinae and Phyllostominae, respective- lationship that is moderately supported with ly). Most relationships at lower taxonomic a decay value of 2 and a bootstrap value of levels (e.g., between genera) are supported 80. Within Glossophaginae, Scleronycteris, by only a single partition, but are not contra- the sister taxa Hylonycteris and Lichonycter- dicted by others (compare ®gs. 44, 45, 46, is, and an unresolved clade including Cho- 47, and 48). However, both the pelage and eroniscus, Choeronycteris, and Musonycteris integument and tongue partitions recover the form a strongly supported polytomy (decay: clade comprising Hylonycteris and Lichon- 3; bootstrap: 93). Successive sister taxa to ycteris, as well as the clade comprising Cen- this group are Anoura, Leptonycteris, and the turio and Sphaeronycteris. sister taxa Glossophaga and Monophyllus 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 133

(see ®g. 49 for additional decay and boot- weakly supported (decay: 2; bootstrap: 40). strap values not mentioned in the text). In six Phyllostominae includes four clades: (1) of the most parsimonious trees, Musonycteris Phyllostomus and Phylloderma, (2) two pairs appears as the sister taxon of Choeroniscus. of sister taxa: Lonchorhina and Macrophyl- In another six trees, Musonycteris is the sister lum, and Mimon bennettii and M. crenula- taxon of a clade containing Choeroniscus tum, (3) Trachops, Tonatia, Chrotopterus, and Choeronycteris. In the remaining six and Vampyrum, and (4) all Micronycteris trees, relationships within the clade contain- species and Macrotus (see ®g. 49 for addi- ing Musonycteris, Choeroniscus and Cho- tional decay and bootstrap values not men- eronycteris are unresolved. Scleronycteris tioned in the text). The Micronycteris and appears in three positions (six trees each): as Trachops groups form a weakly supported the sister taxon of a clade including Hylonyc- clade, and the Lonchorhina and Phyllosto- teris and Lichonycteris; as the sister taxon of mus groups are successive sister taxa to this Choeronycteris, Choeroniscus, and Muso- group. The subgenus Micronycteris is mono- nycteris; and as the sister taxon of a clade phyletic, with M. hirsuta and M. megalotis including the Choeronycteris group and as sister taxa; this sister group relationship is Hylonycteris and Lichonycteris. very strongly supported with a decay value Phyllostominae is monophyletic, but is of 3 and a bootstrap value of 85. Successive 134 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 49. Results of a heuristic search using all 150 characters for all 63 taxa. The tree shown here is a strict consensus of 18 most parsimonious trees, each of 613 steps (CI ϭ 0.463; RI ϭ 0.765). Numbers appearing above the lines are decay values, below the lines are bootstrap values. 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 135 sister taxa to the subgenus Micronycteris are Within the Centurio group, Centurio and Macrotus, Micronycteris brachyotis, and the Sphaeronycteris are sister taxa. Within the sister taxa M. sylvestris and M. nicefori. Pygoderma group, Stenoderma, Phyllops, Within the Trachops group, Tonatia and and Pygoderma are successive sister taxa to Trachops are successive sister taxa to Chro- Ardops and Ariteus (see ®g. 49 for decay and topterus and Vampyrum. The sister group re- bootstrap values). A second stenodermatine lationship between Chrotopterus and Vam- clade is composed of three smaller clades: pyrum is very strongly supported with a de- (1) Koopmania with the sister taxa Artibeus cay value of 4 and a bootstrap value of 89. and Dermanura, (2) Uroderma with the sis- Carolliinae (decay value: 1; bootstrap: 33) ter taxa Platyrrhinus and Vampyrodes, and and Stenodermatinae (decay value: 1; boot- (3) the sister taxa Mesophylla and Ectophylla strap: 54) are each monophyletic and togeth- with the successive sister taxa Vampyressa er form a weakly supported clade. Sturnira pusilla, V. nymphaea, V. bidens, and Chiro- derma. The clade including groups 2 and 3 is the sister taxon of the remaining stenod- above is moderately supported with a decay ermatines. The clade of stenodermatines value of 3 and a bootstrap value of 53. Other without Sturnira is strongly supported with moderately supported groups include: Vam- high decay and bootstrap values (6 and 97 pyressa pusilla, Ectophylla, and Mesophylla respectively). This group is itself composed (decay: 1; bootstrap: 51), and this group plus of two large clades. One stenodermatine all other Vampyressa species and Chiroder- clade includes two smaller groups: (1) Ame- ma (decay value: 1; bootstrap value: 70). trida, Centurio, and Sphaeronycteris, and (2) Other relationships within this larger clade Pygoderma, Phyllops, Stenoderma, Ardops, are weakly supported (see ®g. 49 for decay and Ariteus (see ®g. 49 for addtional decay and bootstrap values). The Koopmania clade and bootstrap values). Monophyly of this en- and the genus Enchisthenes are successive tire large clade is strongly supported with de- sister taxa to the Uroderma and Chiroderma cay value of 9 and a bootstrap value of 98. groups.

DISCUSSION COMPARISON OF RESULTS OF characters together may allow weak phylo- TAXONOMIC AND CHARACTER genetic signal in some partitions to over- CONGRUENCE ANALYSES whelm ``noise'' in the data (Barrett et al., 1991; de Queiroz, 1993); this seems to be the The trees derived from analysis of the sep- case with our data set. arate data partitions (our taxonomic congru- Our analyses highlight another problem ence analyses), when compared with the results of our character congruence analysis, with the taxonomic congruence approach, demonstrate the problems with relying exclu- namely, that data partitions are often infor- sively on a taxonomic congruence approach. mative only in part of a tree (Hillis, 1987; de Many data partitions that are informative Queiroz, 1993). For example, most of our within the character congruence analysis data partitions appeared to be most infor- (e.g., reproductive tract morphology) result mative at a speci®c taxonomic level. We in bushlike trees when analyzed separately. found that tongue data delimited clades that This is due in part to the small number of roughly correspond to subfamilies (e.g., des- characters compared to the large number of modontines, glossophagines, lonchophyl- taxa, but even the data partitions with the lines, and stenodermatines). However, rela- largest number of characters relative to the tionships within these higher-level taxa re- number of taxa appear unable to resolve remained unresolved. Furthermore, we found lationships adequately (e.g., pelage and in- that data sets useful in resolving relationships tegument partition with 38 characters for 63 within some clades were uninformative in taxa, and hyoid partition with 17 characters other clades. The craniodental data appeared for 27 taxa). As we noted earlier, analyzing to be useful for resolving relationships 136 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 among glossophagine genera, but were not as (see Simmons and Geisler [1998] for an useful in Phyllostominae or Stenodermatinae. overview of these issues). We prefer to retain Interestingly, the craniodental data suggest most traditional subfamilial names (e.g., novel relationships for many taxa (e.g., Des- those used by Koopman, 1993; 1994), be- modontinae in a clade of stenodermatines), cause these names have been extensively leading us to conclude that craniodental data used in the past by systematists and research- may be more useful at lower taxonomic lev- ers in ®elds such as ecology, behavior, and els (e.g., within subfamilies) than at higher physiology. levels. Beginning at the base of the phyllostomid Although we have not used any statistical radiation (see ®g. 49), we recognize Des- tests to determine whether our data sets are modontinae as the clade arising from the last ``heterogenous,'' a look at the separate trees common ancestor of Desmodus, Diaemus, produced for the three data sets in which at and Diphylla. Desmodontinae is the basal least 61 of the 63 taxa were scored for most branch of the family Phyllostomidae. Al- characters (e.g., pelage and integument, cran- though it is possible to name the clade com- iodental, and tongue) suggests that there is prising the remaining phyllostomid taxa some disagreement among these data sets. (e.g., Baker et al.'s [1989] use of Phyllostom- For example, the desmodontine clade ap- inae; see table 7), this group is weakly sup- pears in all three data sets, but its position ported. Our experience suggests that the bas- changes in trees derived from different par- al branching pattern of the family may titions (e.g., basal [pelage and integument], change in future analyses as additional char- or with stenodermatines [craniodental]). acters are added to the data matrix. Thus, we Some relationships within Glossophaginae believe it would be inadvisable to apply a (e.g., the position of Anoura in craniodental name to this group until basal relationships vs. pelage and integument trees) and Sten- among phyllostomids are better supported. odermatinae (e.g., Sphaeronycteris in cranio- Continuing up the tree (see ®g. 49), Bra- dental vs. pelage and integument) are not chyphyllinae includes only Brachyphylla, compatible when these three trees are com- pending further investigation of the relation- pared. As we noted previously, we ®nd that ships of this genus. Although most early in- there is no reliable way to divide morpho- vestigators considered Brachyphylla a unique logical characters into groups that re¯ect and primitive member of Stenodermatinae some underlying biological reality. The in- (e.g., Dobson, 1875; Miller, 1907; Hall and congruent areas of these partitions may Kelson, 1959; Slaughter, 1970), later studies merely represent an arbitrary division of ev- documented the similarity of Brachyphylla idence. and Phyllonycterinae. Silva Taboada and Pine (1969) and Baker et al. (1981a) con- CLASSIFICATION OF PHYLLOSTOMID cluded that Brachyphylla should be placed in BATS the subfamily Phyllonycterinae (correctly called Brachyphyllinae, which is the older Higher-level Classi®cation: Our phylo- name; Gray 1866d). Other workers have con- genetic results indicate that many groupings tinued to maintain Brachyphylla in its own of phyllostomid taxa previously recognized subfamily (e.g., Koopman, 1994), and based in formal classi®cations (e.g., Desmodonti- on our results, we recommend following this nae, Phyllostominae, Phyllonycterinae, Glos- usage. sophaginae, Lonchophyllinae, Carolliinae, We recognize the clade arising from the Stenodermatinae) are monophyletic. How- last common ancestor of Phyllonycterinae ever, we suggest a somewhat different solu- and Glossophaginae (see below for the de®- tion for taxonomic problems than those re- nitions of these taxa) as Hirsutaglossa (hir- cently proposed (see table 7). We classify suta ϭ hairy [Latin]; glossa ϭ tongue taxa phylogenetically (see e.g., de Queiroz [Greek]). We propose this as an unranked and Gauthier, 1990, 1992, 1994), name only taxonomic name because no rank currently monophyletic groups, and use both ranked exists between that of subfamily and family. and unranked names in this classi®cation Baker et al. (1989) identi®ed this clade plus 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 137

Brachyphylla as Glossophagini within the Lionycteris, Lonchophylla, and Platalina.We subfamily Phyllostominae (see table 7). Mc- recognize Glossophagini as the clade arising Kenna and Bell (1997) applied the name from the last common ancestor of Anoura, Glossophaginae to Hirsutaglossa plus Bra- Choeroniscus, Choeronycteris, Glossophaga, chyphylla, and recognized three tribal taxa Hylonycteris, Leptonycteris, Lichonycteris, within Glossophaginae (Phyllonycterini, Monophyllus, Musonycteris, and Scleronyc- Lonchophyllini, and Glossophagini; see table teris. 7). However, this arrangement leaves no We recognize Phyllostominae as the clade name that can be used for the clade tradi- arising from the last common ancestor of the tionally called Glossophaginae (i.e., the clade tribes Phyllostomini (®rst used as a tribal including glossophagines and lonchophyl- name by Baker et al., 1989), Lonchorhinini lines), a clade that is characterized by nu- (Gray, 1866d), Vampyrini (Bonaparte, 1838), merous synapomorphies (see appendix 4) and Micronycterini (Van Den Bussche, and merits recognition. In our estimation, ap- 1992). None of these tribal names is new, plication of a traditional subfamilial name although the last represents a change in rank (e.g., Glossophaginae) to a larger, more var- from subfamilial to tribal level. We de®ne iable group is unnecessarily confusing. Phyllostomini as the clade arising from the Therefore, we recommend using Hirsutaglos- last common ancestor of Phylloderma and sa for the clade comprising most nectar feed- Phyllostomus (see below for comments on ers, and continue with more established us- this relationship). Phyllostomini is therefore age for names like Glossophaginae (see be- restricted from the de®nition of Baker et al. low). (1989), who applied this name to a larger Within Hirsutaglossa, we de®ne Phyllo- clade that included Lonchorhina, Macro- nycterinae as the clade arising from the last phyllum, Mimon, Phylloderma, Phyllosto- common ancestor of Erophylla and Phyllo- mus, and Tonatia (see table 7). We de®ne nycteris. Continued use of this name at the Lonchorhinini as the clade arising from the subfamilial level facilitates discussion and last common ancestor of Lonchorhina, Mac- recognizes the ecological and morphological rophyllum, and Mimon. This group name was distinctiveness of these two genera. We pro- previously only applied to the nominate ge- pose that Glossophaginae be de®ned as the nus by Gray (1866d). We de®ne Vampyrini clade arising from the last common ancestor as the clade arising from the last common of Lonchophyllini and Glossophagini (see ancestor of Chrotopterus, Tonatia, Trachops, below for de®nition). This clade was consis- and Vampyrum; inclusion of Tonatia in this tently recognized by most authors until Grif- group is new. Finally, we recognize Micro- ®ths (1982) recognized Lonchophyllinae. nycterini as the clade arising from the last Since that time, there has been no name ap- common ancestor of Glyphonycteris, Lam- plied to the clade that includes lonchophyl- pronycteris, Macrotus, Micronycteris, and lines and glossophagines. Application of the Trinycteris. Previously, Van Den Bussche name Glossophaginae to this group and rec- (1992) named two new subfamilies (Macro- ognition of the two clades it contains as tinae and Micronycterinae) for these genera. tribes (Lonchophyllini and Glossophagini) Micronycterini is the appropriate name to ap- facilitates discussion of the ecology and mor- ply to the clade that includes both Macrotus phology of these bats using names recogniz- and Micronycteris because Micronycteris is able to nonsystematists. Both tribes rely the older generic name (Gray, 1866d). heavily on nectar and pollen for nutrients The clade including Phyllostominae, Car- (Gardner, 1977b; Ferrarezzi and Gimenez, olliinae, and Stenodermatinae is weakly sup- 1996) and share numerous synapomorphies ported and members share only a few derived including features of the pelage, face, teeth, features (see appendix 4). Given the possi- wing, tongue, hyoid, and restriction sites. bility that additional data may change rela- Some of these features are not found in any tionships among these taxa, we believe it other phyllostomids (see appendix 4). would be inadvisable to apply a name to this We de®ne Lonchophyllini as the clade clade until relationships among these groups arising from the last common ancestor of are better supported. 138 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 139

Although some recent classi®cations in- no [Latin]; cauda ϭ tail [Latin]), a new and cluded a clade comprised of carolliines and unranked name. stenodermatines named Stenodermatinae We de®ne Carolliinae as the clade arising (McKenna and Bell, 1997) or Stenodermatini from the last common ancestor of Carollia (Baker et al., 1989; see table 7), we have and Rhinophylla. This usage is entirely con- again chosen to conserve traditional subfam- sistent with traditional classi®cations (e.g., ily names to avoid confusion and highlight Koopman, 1993, 1994). the morphological and ecological distinctive- We de®ne Stenodermatinae as the clade ness of taxa. Therefore, we recognize the arising from the last common ancestor of clade arising form the last common ancestor Sturnirini and Stenodermatini, two tribal lev- of Carolliinae and Stenodermatinae (see be- el taxa that have been recognized by previous low for de®nition) as Nullicauda (nullus ϭ authors (e.g., Koopman 1994). We recognize 140 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Sturnirnini as including only the genus Stur- this classi®cation, stating ``. . . I do not rec- nira. We recognize Stenodermatini as arising ommend adoption of the classi®cation in Ap- from the last common ancestor of Stenoder- pendix IV.'' matina and Ectophyllina (see below). Generic Considerations: Previous au- ``Long-faced'' and ``short-faced'' steno- thors have questioned the monophyly of sev- dermatines have been recognized as unique eral phyllostomid genera including Micron- assemblages by many authors and our anal- ycteris, Mimon, Phyllostomus (sensu Baker ysis provides support for both these groups et al., 1988), Artibeus, and Vampyressa (An- (e.g., Miller, 1907; de la Torre, 1961; Smith, derson, 1906; Miller, 1907; Straney et al., 1976; Lim, 1993; see appendix 4). We there- 1979; Honeycutt, 1981; Straney 1980; Koop fore propose recognition of two new subtri- and Baker, 1983; Owen, 1987, 1991; Baker bal taxa within Stenodermatini, Ectophyllina et al., 1988a; Van Den Bussche, 1992; Lim, and Stenodermatina, for the taxa formerly 1993; Van Den Bussche and Baker, 1993; known colloquially as the ``long-faced'' and Van Den Bussche et al., 1993, 1998; Sim- the ``short-faced'' stenodermatines. We de- mons, 1996; Jassal and Simmons, 1996). We ®ne these taxa using a stem-based approach. tested the monophyly of each of these taxa. We de®ne Ectophyllina as the clade of gen- Our results for several generic groupings do era within Stenodermatini that share a more not conform to currently accepted taxonomy recent common ancestor with Ectophylla (e.g., Koopman, 1993, 1994). We therefore than with Centurio (these genera are Arti- recommend adopting a revised classi®cation beus, Chiroderma, Dermanura, Ectophylla, that better re¯ects phylogenetic relationships. Enchisthenes, Koopmania, Mesophylla, Pla- Although we identi®ed additional clades that tyrrhinus, Uroderma, Vampyressa, and Vam- share numerous synapomorphies and are pyrodes). We de®ne Stenodermatina as those strongly supported (e.g., the clade compris- genera within Stenodermatini that share a ing Chrotopterus and Vampyrum), we ad- more recent common ancestor with Centurio dress only controversial relationships. For than with Ectophylla (these genera are Ame- taxa whose classi®cation has not been ques- trida, Ardops, Ariteus, Centurio, Phyllops, tioned, we retain traditionally used names to Pygoderma, Stenoderma, and Sphaeronyc- avoid confusion. teris). We use a stem-based approach here Micronycteris currently comprises ®ve ap- because the positions of the taxa Artibeus, parently monophyletic subgenera: Glypho- Dermanura, Enchisthenes, and Koopmania nycteris, Lampronycteris, Micronycteris, are weakly supported. Use of a stem-based Neonycteris, and Trinycteris (Simmons, approach will allow the names Stenoderma- 1996). A ``total evidence'' analysis that in- tina and Ectophyllina to be used in the future cluded allozymic, karyological, and morpho- even if the relative positions of these four logical data supported the following relation- taxa should change. ships: (1) Neonycteris and Trinycteris were The names Artibeini and Vampyressini, successive sister taxa to Glyphonycteris, (2) which appear in a tree in a paper by Ferrarezi the Neonycteris clade was the sister taxon of and Gimenez (1996), are not avaliable. These Micronycteris, and (3) Lampronycteris was names and several others appear in a classi- the basal branch within the genus (Simmons, ®cation by Owen (1987: appendix 4). Con- 1996). Two karyological characters support- fusingly, Owen (1987) used Stenodermatini ed the monophyly of Micronycteris (Patton and Artibeini for both tribes and subtribes and Baker, 1978). However, the karyological (the same name, with the same ending), mak- study presumed that Macrotus exhibited the ing it dif®cult to interpret papers that use this primitive karyotype, an assumption we con- classi®cation. Owen (1987) also did not pro- sider ¯awed (see below). Our analysis indi- vide a list of characters differentiating these cates that Micronycteris as currently recog- taxa, making these names unavailable under nized is not monophyletic. Although we article 13 of the International Code of Zoo- could have rede®ned Micronycteris to in- logical Nomenclature (International Com- clude Macrotus, we do not consider this a mission on Zoological Nomenclature, 1985). reasonable approach because the subgenera Furthermore, Owen (1987: 33) discounted of Micronycteris are morphologically dis- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 141 tinct, and all differ from Macrotus (see ap- man, 1993, 1994) and recognize Phylloder- pendix 4). Based on the work of Simmons ma as a genus distinct from Phyllostomus. (1996), which supported monophyly of Mi- Koopman (1994) recognized three subgen- cronycteris and Glyphonycteris (Lampronyc- era of Artibeus: (1) the subgenus Artibeus, teris, Neonycteris and Trinycteris are mono- which includes large-bodied species, (2) the typic), we suggest that all subgenera of the subgenus Dermanura, which includes small- genus Micronycteris be raised to generic bodied species and Dermanura concolor, and rank. This approach renders all genera mono- (3) the subgenus Enchisthenes. There was phyletic without requiring any new names, substantial disagreement concerning the re- and recognizes the substantial differences in lationships among these taxa and the mono- morphology among these groups. phyly of the group as a whole (e.g., Owen, Handley (1960: 460) synonomized An- 1987; 1991; Van Den Bussche, 1992; Lim, thorhina and Mimon suggesting that the two 1993; Van Den Bussche et al., 1993, 1998). taxa were ``not distinguishable even as sub- Some authors raised each of these subgenera genera.'' Although we found facial, pelage, to generic rank (e.g., Enchisthenes: Miller, and wing characters that distinguish the two 1907; Baker, 1973; Gardner, 1977a; Hall, subgenera (see appendix 4), our results sup- 1981; Artibeus: Owen, 1987; Dermanura: port Handley's (1960) contention that M. Owen, 1987). Additionally, Owen (1991) bennettii and M. crenulatum are sister taxa. proposed a new genus, Koopmania, for Der- Several characters support this relationship manura concolor. We found that Artibeus including dental, molecular, and vibrissal (s.l.) is paraphyletic because Enchisthenes features (see appendix 4). does not group with the other three represen- Some authors have considered Phylloder- tatives of Artibeus, which form a clade (see ma a junior synonym of Phyllostomus (e.g., ®g. 49). We recommend that Enchisthenes be Anderson, 1997) since Baker et al. (1988a: recognized as a genus distinct from Artibeus and that Artibeus, Dermanura, and Koop- 12) concluded that ``from the protein point mania be recognized as subgenera of Arti- of view, the inclusion of Phylloderma in the beus pending further research into the rela- genus Phyllostomus would not add signi®- tionships of these three taxa. Molecular evi- cant genetic variation to that already present dence from the cytochrome b gene also sup- in the genus.'' In a phylogenetic analysis of ports these relationships (Van Den Bussche cytochrome b sequence data, Van Den Bus- et al., 1993, 1998). Although we could rec- sche and Baker (1993) con®rmed that Phyl- ognize Artibeus, Dermanura, and Koopman- loderma and Phyllostomus are closely relat- ia at the generic level as Owen (1987, 1991) ed. However, examination of Van Den Bus- recommended, there would be no convenient sche and Baker's (1993) shortest trees re- way to refer to these taxa as a group. Thus, vealed that in all cases, Phylloderma did not we believe that the best classi®cation option nest within Phyllostomus, but appeared as the is to recognize three subgenera within Arti- sister group of the clade including all tradi- beus. tionally recognized Phyllostomus species (P. Koopman (1994) recognized ®ve species discolor, P. elongatus, P. hastatus, and P. and three subgenera of Vampyressa: (1) the latifolius). Although Phylloderma could still subgenus Vampyressa includes V. melissa be referred to Phyllostomus based on these and V. pusilla, (2) the subgenus Metavam- data and our results, Phylloderma differs pyressa includes V. nymphaea and V. brocki, morphologically from the species of Phyllos- and (3) the subgenus Vampyriscus includes tomus, which share a distinct dental formula, only V. bidens. Recently, Owen (1987) found skull shape, and noseleaf structure (see ap- that these subgenera did not appear to form pendix 4). Thus, inclusion of Phylloderma in a monophyletic group (e.g., Vampyressa pus- Phyllostomus would signi®cantly alter the di- illa and Mesophylla formed a clade), but was agnosis of the genus Phyllostomus, but unable to resolve relationships among most would not improve our understanding of clades that included Vampyressa species. group monophyly. Consequently, we follow Owen (1987) recommended the following traditional usage (e.g., Miller, 1907; Koop- classi®catory changes: (1) placing Vampy- 142 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 ressa melissa in a new unnamed subtribe, (2) clade of Ectophylla alba and E. macconnelli, placing Vampyressa nymphaea in the genus which is distinct and merits recognition. Mesophylla and recognizing the two taxa as Throughout the remainder of the discus- the subtribe Mesophyllatini, (3) placing all sion section we employ the new classi®ca- remaining Vampyressa species in the sub- tion (see table 7). tribe Vampyressatini. More recent studies of stenodermatines (e.g., Lim, 1993) have failed INTERPRETATION OF to sample within Vampyressa and conse- CHROMOSOMAL DATA quently did not provide a test of Owen's (1987) hypothesis. Although our results are In the 1960s and 1970s, several studies not generally congruent with Owen's (1987), used chromosomal data to address relation- we found that Vampyressa is not monophy- ships among phyllostomid bats. Attempts to letic because V. pusilla, V. nymphaea, and V. produce a consensus tree of phyllostomid re- bidens are successive sister taxa to Ecto- lationships (e.g., Honeycutt and Sarich, phylla and Mesophylla. However, we do not 1987a; Baker et al., 1989) assessed congru- recommend nomenclatural changes for this ence between chromosomal data and other taxon because there has not been a recent types of data (immunological and morpho- review of the species relationships within logical). Therefore, we felt it was important this genus. We did not include all species of to evaluate these data in the context of our Vampyressa in our analysis, so we have no phylogeny. way of knowing if, as Owen (1987) found, We agree with Gardner (1977a) who hy- the subgenera are not monophyletic. We rec- pothesized that Desmodontinae, Phyllostom- ommend further study of the relationships inae, Phyllonycterinae, Carolliinae, and Sten- within Vampyressa and between this genus odermatinae were each monophyletic, as and other stenodermatines before changes are were the subgroups we identify as Micron- ycterini (Macrotus, Micronycteris, and Tri- made in the classi®cation of species currently nycteris) and Phyllostomini (Phylloderma referred to Vampyressa. and Phyllostomus). Our analyses also agree Although most previous authors recog- with Patton and Baker's (1978) suggestion nized Mesophylla as a monotypic genus in- that Lampronycteris, Micronycteris, and Tri- cluding only M. macconnelli (e.g., Starrett nycteris form a clade. However, unlike Pat- and Casebeer, 1968; Koopman, 1994), this ton and Baker (1978) we found that Macro- genus has included members of the genus tus is also a member of this group. Vampyressa (Owen, 1987, see above). Me- Gardner (1977a) indicated that Brachy- sophylla has also been placed in the genus phylla is part of a hirsutaglossan group and Ectophylla (Laurie, 1955; Goodwin and forms a clade with Phyllonycterinae. Al- Greenhall, 1961; Anderson et al., 1982). We though the position of Brachyphylla is un- found that Mesophylla and Ectophylla are resolved in our tree, the karyotypes of Bra- sister taxa that share several synapomorphies chyphylla, Phyllonycterinae, Glossophaga, including features of the dentition, noseleaf, Monophyllus, and Leptonycteris could be a and tongue (see appendix 4). Lim (1993) also derived feature of Hirsutaglossa plus Brachy- found support for a sister-taxon relationship phylla. However, if Brachyphylla is the sister between these genera, with two synapomor- taxon of Phyllostominae and Nullicauda, an phies (dental and external) uniting this node. alternative interpretation is that this karyo- We therefore propose that Mesophylla be type evolved at the node joining Hirsutag- considered a junior synonym of Ectophylla, lossa to the Brachyphylla, Phyllostominae, the oldest name available for this clade (Al- and Nullicauda clade. Therefore, this karyo- len, 1892b). We do not recommend revising type would represent a primitive feature re- Ectophylla to include some or all members tained by Brachyphylla. Several hirsutaglos- of Vampyressa because relationships among san clades that we identi®ed have also been Vampyressa species remain unclear (see supported by chromosomal data: Glossopha- above). Revising the usage of Ectophylla in ga and Monophyllus (Gardner, 1977a), and this way would also leave no name for the Choeroniscus and Choeronycteris (karyo- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 143 types of Musonycteris are unknown; Baker, posed primitive karyotype for phyllostomids 1967; Gardner, 1977a). [Patton and Baker, 1978]) that we can say Most authors working with karyotypes that M. nicefori has ``Rearrangements of have identi®ed Stenodermatinae as a mono- biarmed waterhousi chromosomes 1/2, 23/ phyletic group (e.g., Baker, 1973; Gardner, 24, [and] 26/25 . . .'' (Patton and Baker, 1977a), a ®nding with which we agree. With- 1978: 453). We found that Macrotus nests in Stenodermatinae, all authors have found well within Phyllostomidae, suggesting that Stenodermatina to be monophyletic (Baker, chromosomal structure of Macrotus may 1973; Greenbaum et al., 1975; Gardner, have converged on the state seen in Pteron- 1977a). We also agree with the ®ndings of otus and Noctilio, the outgroups Patton and Baker (1973), Greenbaum et al. (1975), and Baker (1978) used in their analysis. This pos- Gardner (1977a) that Chiroderma, Ectophyl- sibility affects the interpretation of all chro- la macconnelli, and Vampyressa are closely mosomal data reported thus far, as Macrotus related. We also found, as these authors sug- was the primitive reference taxon for most gested, that Ectophylla maconnelli and Vam- studies. pyressa represent the most derived group in Examination of areas where our results this lineage. Additionally, Johnson (1979) disagree with karyological studies suggests found that Ectophylla macconnelli is karyo- that these differences may also be, in part, logically very similar to Vampyressa pusilla. caused by additional methodological prob- Our results agree with these ®ndings. lems. For example, Haiduk and Baker (1982) There were a number of differences be- implicated limited sampling as one possible tween our conclusions and those of previous reason for their failure to support monophyly authors who used chromosomes to investi- of Lonchophyllini. Autapomorphic karyo- gate phyllostomid relationships. There are types (i.e., those that have few recognizable several possible methodological explanations arms relative to the primitive karyotype) of- for these differences. Many of the discrep- fer additional problems and very often these ancies between the results of early work us- taxa have been placed at the base of clades ing karyology and later studies were due to because their karyotypes offer few clues to inaccurate homology assessments. With the their relationships (e.g., Anoura: Baker, advent of G- and C-banding, it became easier 1967; Gardner, 1977a; Uroderma: Baker, to discern which chromosomal arms were 1967, 1973; Gardner, 1977a). Because auta- homologs by examining banding patterns. pomorphies may evolve at any point in the Nevertheless, there are still some problems evolutionary history of a lineage, these taxa with the use of karyotypes in phylogenetic need not be basal. Finally, when interpreted reconstruction. Most investigators today use in light of our results, it seems clear that in what is often termed ``global parsimony'' to several cases taxa that retain primitive kar- investigate relationships among taxa. This yotypes have been grouped together based on method involves searching for the shortest shared primitive features rather than shared networks, which are subsequently rooted at derived characters (e.g., Glossophaga, Mon- the node connecting the outgroup to the rest ophyllus, Leptonycteris, Phyllonycterinae, of the tree. Accordingly, this method in- Gardner, 1977a). volves no a priori assumptions about ingroup There are many areas of our tree that differ monophyly or direction of character change. from hypotheses based on chromosomal Global parsimony has not been used with data. For example, Johnson (1979) indicated karyological data in phyllostomids. Instead, that phyllostomines diverged from phyllos- most studies rely on comparisons with pre- tomid stem stock before other subfamilies, viously identi®ed ``primitive'' reference taxa, and subsequently were followed by desmo- typically within the ingroup, to identify ho- dontines and stenodermatines, who split from mologous segments and rearrangements that a glossophagine-brachyphylline group before are synapomorphies (see Materials and either of these subfamilies was distinct. Bass' Methods). For example, it is only by com- (1978) examination of G-banded karyotypes paring karyotypes of Micronycteris nicefori suggested to her that Desmodontinae, Glos- to Macrotus waterhousi (bearer of the pro- sophaginae (s.l.), and Phyllonycterinae 144 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 formed a clade, with Phyllostominae as the ilar to those of phyllostomines, glossophag- sister group. Our results, that desmodontines ines, and stenodermatines. In contrast, we and hirsutaglossans branched off prior to the found support for a sister taxon relationship divergence of phyllostomines and nullicau- between Rhinophylla and Carollia. In the dans, agree with none of these ®ndings. context of this hypothesis, we suggest that These differences may be due to assumptions the similarity between the karyotype of Rhi- about the primitive karyotype for the family. nophylla and other noncarolliine taxa is due Cadena and Baker (1976) suggested that to primitive retention of these karyotypic Diaemus and Diphylla were less derived than traits. Desmodus because they have karyotypes that Baker and Lopez (1970) suggested that the are more similar to the proposed primitive XX/XY1Y2 sex system delineated a group karyotype for phyllostomids. Bass (1978) in- consisting of Ametrida, Artibeus, Enchis- dicated that Diphylla and Desmodus are thenes, and Stenoderma. Greenbaum et al. more closely related to each other than either (1975) suggested that two ``short-faced'' is to Diaemus. Contra Bass (1978), we found groups could be identi®ed: (1) Centurio and that Desmodus and Diaemus form a clade. Sphaeronycteris, both of which have an XX/ Within this context, it is still possible for XY sex system, and (2) the remaining Desmodus to be highly autapomorphic as ``short-faced'' taxa, all of which have an XX/

Cadena and Baker (1976) suggested. XY1Y2 system. Gardner (1977a) included Gardner (1977a) grouped Chrotopterus Ametrida in a clade with Ardops, Ariteus, and Tonatia but excluded two genera, Vam- Phyllops, and Pygoderma. In contrast, we pyrum and Trachops, that we include in found that Ametrida is more closely related Vampyrini. Although Gardner (1977a) iden- to Centurio and Sphaeronycteris and that Ar- ti®ed a clade including Phylloderma and tibeus and Enchisthenes are basal members Phyllostomus (Phyllostomini), he indicated of Ectophyllina. According to our hypothe- that the sister taxon of this group was Mi- sis, not all taxa sharing the XX/XY1Y2 sys- mon, which we found to be a member of tem form a clade. We found that the XX/

Lonchorhinini. Patton and Baker (1978) also XY1Y2 system appears to be derived for identi®ed a group of Mimon, Phyllostomus, Stenodermatini (and primitive for Ectophyl- and Tonatia. Our results suggest that these lina and Stenodermatina). The XX/XY sex taxa may have been grouped together be- system is apparently a reversal in most mem- cause they share primitive karyotypes. bers of Ectophyllina and the sister taxa Cen- Although Haiduk and Baker's (1982) turio and Sphaeronycteris. study failed to ®nd support for Lonchophyl- Gardner (1977a) placed Sturnira within a lini, they suggested that this may have been large clade of Stenodermatini taxa, whereas a sampling problem. Haiduk and Baker we found that this genus was basal within (1982) placed Hylonycteris as a sister taxon Stenodermatinae. Baker (1967, 1973) and of Choeronycteris and Musonycteris, with Gardner (1977a) were both unable to resolve Choeroniscus as the sister of this clade. We the relationships of Uroderma due to its disagree with this ®nding, although all four chromosomal uniqueness. We found that genera appear to be closely related in our tree Uroderma clearly belongs in the Ectophyl- (see ®g. 49). Gardner (1977a) identi®ed Hy- lina clade. Within this context, Uroderma re- lonycteris, Lichonycteris, Platalina, and mains highly autapomorphic. We also dis- Scleronycteris as forming a separate clade, agree with both Baker (1973) and Gardner contra our results. Gardner (1977a) also sug- (1977a), who indicated that Enchisthenes gested that the sister taxa Glossophaga and was most closely related to Artibeus. Our re- Monophyllus formed a clade with Leptonyc- sults indicate that Enchisthenes is a basal off- teris, and that this group and Phyllonycteri- shoot of Ectophyllina and is not closely re- nae (s.l.) formed a clade. We suggest that lated to Artibeus. Greenbaum et al. (1975) Gardner (1977a) grouped these four taxa on proposed that Ectophylla alba diverged from the basis of shared primitive karyotypes. the E. macconnelli-Vampyressa line before a Baker and Bleier (1971) found that the reduction in diploid number occurred. Sub- karyotype of Rhinophylla appears most sim- sequently, Gardner (1977a) placed Ectophyl- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 145 la alba in a clade with Artibeus, Enchis- supported by Straney's (1980) and Honey- thenes, Vampyrodes, and Vampyrops (ϭ Pla- cutt's (1981) comparisons. However, in both tyrrhinus). In the context of our phylogeny, of these earlier studies this clade did not in- Ectophylla alba, the sister taxon of E. mac- clude Tonatia, which we found associated connelli, appears to be chromosomally auta- with this group. pomorphic. There were a number of differences between our conclusions and those of previous INTERPRETATION OF authors who used immunology to investigate IMMUNOLOGICAL DATA phyllostomid relationships. There are several possible methodological explanations for Although it is not possible to incorporate these differences. Because distance data is immunological distance data into our study, phenetic in nature, it has been heavily criti- many previous workers based phylogenetic cized because, among other problems, it fails hypotheses on these data. Previous attempts to distinguish between ``primitive'' and ``de- to produce a consensus tree of phyllostomid rived'' conditions, instead grouping taxa by relationships (e.g., Honeycutt and Sarich, overall similarity and increasing the number 1987a; Baker et al., 1989) assessed congru- of ad hoc assumptions of homoplasy that ence between immunological data and other must be made (e.g., Farris, 1983; Wiley, types of data (chromosomal and morpholog- 1981). Thus, one possible explanation of the ical). Therefore, we felt it was important to incongruence between our results and those evaluate these data in the context of our phy- of immunological studies is that some taxa logeny. may have slower rates of immunological Our ®nding that desmodontines are a basal evolution or may have retained primitive im- phyllostomid lineage is congruent with some munological features. of Pierson's (1986) analyses, and all analyses Arnold et al. (1982) noted that the im- from Honeycutt's (1981) and Honeycutt and munological distinctiveness of phyllostomid Sarich's (1987a) studies. Other points of albumins relative to other bats (the equiva- agreement include Honeycutt's (1981) and lent of 30 to 40 units of immunological dis- Honeycutt and Sarich's (1987a) ®nding that tance) may be associated with a rate desta- Desmodontinae and Phyllonycterinae were bilization of the albumins in this group. each monophyletic. We also agree with Stra- These authors further suggested that changes ney's (1980) ®nding that Artibeus, his sole in the three-dimensional structure of proteins representative of Stenodermatinae, and Car- (involving a cysteine and a possible reloca- ollia, his sole representative of Carolliinae, tion of disul®de bridges) could lead to non- formed a group, offering support for the equivalence of immunological distances be- clade we named Nullicauda. Honeycutt tween those caused by single amino-acid (1981) and Honeycutt and Sarich (1987a) substitutions and those caused by confor- provided additional support for Nullicauda. mational changes. Although Arnold et al. Gerber (1968) and Gerber and Leone's (1982: 11±12) did not discuss particular in- (1971) results suggested that Stenodermati- stances where this may have occurred, they nae was monophyletic, as did our study. concluded ``We include this as ®nal note for Honeycutt's (1981) and Honeycutt et al.'s consideration of albumin evolution in phyl- (1981) conclusion that Desmodus and Diae- lostomoid bats, where more than the usual mus were sister taxa, Honeycutt's (1981) number of interpretative problems exist.'' ®nding that Phylloderma was strongly asso- Another problem with interpreting results ciated with Phyllostomus hastatus, and Ho- of immunological distance data occurs when neycutt (1981) and Honeycutt and Sarich's only unidirectional comparisons are made (1987a) recognition of a Vampyrum-Gly- between taxa. Such tests allow only an `àp- phonycteris group all agree with our results. proximate'' placement of taxa into the tree, In these earlier studies, however, this clade which may be interpreted as ``species group'' did not include Macrotus, which we found relationships (Maxson and Maxson, 1990: associated with the Vampyrum-Glyphonyc- 154). Furthermore, the relationship between teris group. Our recognition of Vampyrini is sequence difference and immunological re- 146 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 activity has only been determined for micro- lationship between Macrotus and Carollia. complement ®xation, the technique used by We failed to recover any of these relation- Straney (1980), Honeycutt (1981), Honeycutt ships. Pierson (1986) suggested that the as- and Sarich (1987a), and in part by Pierson sociation of Macrotus with Carollia was (1986). Other techniques, including the im- probably due to the conservative nature of munodiffusion and precipitin tests used by the transferrin of Carollia. Honeycutt (1981), Gerber (1968) and Gerber and Leone (1971), and Honeycutt and Sarich (1987a) postulated and immunodiffusion and radioimmunoassay that a slower rate of evolution in desmodon- used by Pierson (1986), have a general cor- tines or equal yet independent divergence in relation with microcomplement ®xation, but albumins from those of other phyllostomids use of these techniques introduces another might have caused the association they found source of error that may compromise phy- between Macrotus and desmodontines. logenetic studies (Maxson and Maxson, The association of Tonatia with Phyllos- 1990). tomus seems strongly supported by immu- All immunological investigations have nological data (Pierson, 1986; Honeycutt, suggested that phyllostomines are paraphy- 1981; Honeycutt and Sarich, 1987a). Place- letic (e.g., Pierson, 1986; Honeycutt, 1981; ment of this genus was made using micro- Honeycutt and Sarich, 1987a; Gerber, 1968; complement ®xation and two-way compari- Gerber and Leone, 1971), a ®nding with sons. Our grouping of Tonatia with Vampyr- which we disagree. Reviewing the results of ini genera (a novel hypothesis) is weakly these studies, it is interesting to note that they supported, and there is weak support for the are not only incongruent with our results, but entire Vampyrini group. This suggests that also often with each other. Pierson (1986), additional data might support a relationship Honeycutt (1981), and Honeycutt and Sarich between Phyllostomus and Tonatia, as indi- (1987a) found that some phyllostomines cated by the immunological data. Alterna- were more closely related to members of tively, Tonatia could be immunologically Nullicauda than to other phyllostomines, and primitive. If Phyllostomus is immunological- that members of Hirsutaglossa were derived ly primitive, as suggested by its position in from within a clade that included phyllos- our tree, this might explain the persistent im- tomines. Gerber (1968) and Gerber and Le- munological clustering of Tonatia with Phyl- one (1971) suggested that the phyllostomines lostomus, a member of Phyllostomini. they surveyed, which did not form a monophyletic group, were more closely related to Although the position of Brachyphylla is desmodontines and some glossophagines. unresolved in the strict consensus tree from Honeycutt (1981) and Honeycutt and Sar- our character congruence analysis, immuno- ich (1987a) failed to recover Lonchorhinini logical results consistently placed this genus and Micronycterini, and Pierson's (1986) with hirsutaglossan taxa (e.g., Baker et al., analysis did not recover a Vampyrini-Mi- 1981a; Honeycutt, 1981; Pierson, 1986; Ho- cronycterini clade. The use of unidirectional neycutt and Sarich, 1987a). Placement of this comparisons to place Lonchorhina, Macro- genus was made using microcomplement ®x- phyllum, Mimon may explain the differences ation and two-way comparisons. Baker et al. between our phylogeny and that of Honey- (1981a), Honeycutt (1981), and Honeycutt cutt (1981) and Honeycutt and Sarich and Sarich (1987a) identi®ed a relationship (1987a). The failure of all these authors to between Brachyphylla and phyllonycterines recover Micronycterini (Honeycutt, 1981; (Pierson [1986] did not include phyllonyc- and Honeycutt and Sarich, 1987a) and a terines in her study), and the latter two stud- Vampyrini-Micronycterini clade (Pierson, ies revealed that this clade nested within 1986) is due to the basal position of Macro- Glossophagini. However, none of our most tus in these trees. parsimonious topologies agree with the im- Straney (1980), Honeycutt (1981), and munological data in this respect, although Honeycutt and Sarich (1987a) found a rela- trees in which Brachyphylla is the sister tax- tionship between Macrotus and the desmo- on of Hirsutaglossa are only two steps longer dontines, whereas Pierson (1986) found a re- than the most parsimonious topology. Bra- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 147 chyphylla, like Tonatia, may be immunolog- Jones, 1917; Wimsatt, 1975, 1979; Moss- ically conservative. man, 1977, 1987; see characters 131 and 132 Gerber (1968) and Gerber and Leone above). (1971) reported that Choeronycteris associ- In most mammalian orders, simplex uteri ated more closely with phyllostomines and are correlated with small litters (Wimsatt, desmodontines than with other glossophagi- 1975, 1979; Mossman 1977). Bats typically nes. We found that Glossophagini is mono- produce single young regardless of uterine phyletic. The relationships we found within structure although multiple births occur in Glossophagini differ from those found by members of a few bat families, (e.g., vesper- Honeycutt (1981). Honeycutt's (1981) uni- tilionids like Lasiurus). Because cases of directional tests suggested that there were multiple births occur in taxa that nest well two major groups within Glossophagini: (1) within the microchiropteran clade, small lit- Glossophaga and Leptonycteris and (2) An- ter size appears to be primitive for Micro- oura, Choeroniscus, and Monophyllus. The chiroptera (and Phyllostomidae) and there- position of Hylonycteris was equivocal with fore does not provide an explanation for the respect to the two reference taxa (Glosso- evolution of a simplex uterus in these chi- phaga and Monophyllus). The use of unidi- ropteran groups (Hood and Smith, 1983). rectional comparisons to place these taxa Previous hypotheses concerning uterine may explain the differences between our evolution in mammals suggested that evolu- phylogeny and that of Honeycutt (1981). tionary changes in fusion of the uterine horns Gerber (1968) and Gerber and Leone and internal uterine spaces were progressive, (1971) found that Carollia had the greatest passing through intermediate states to reach af®nity for Glossophaga species, not for more derived states, and unidirectional stenodermatines as we found. Gerber (1968) (Wood Jones, 1923; Mossman, 1953, 1977; and Gerber and Leone (1971) found that Le Gros Clark, 1959; Lillegraven, 1969, Stenodermatini, Sturnirini, Artibeus (Arti- 1976; Luckett, 1980; Hood and Smith, 1982, beus), and Artibeus (Dermanura) were not 1983). Under this scenario, a duplex uterus monophyletic. They also failed to recover the was considered primitive, the simplex uterus Platyrrhinus-Uroderma clade that we found. was derived, and the bicornuate condition Instead, their results suggested that Chiro- represented an intermediate stage. Reversals derma and Uroderma were closest relatives were not thought to occur at any point in this and formed a clade with a group comprised evolutionary process. of Artibeus phaeotis, Platyrrhinus helleri, Hood and Smith (1982, 1983) tested the and Sturnira lilium. As we noted above, im- progressive fusion hypothesis of uterine evo- munodiffusion and precipitin tests introduce lution for phyllostomid bats and concluded additional sources of error into phylogenetic that it was correct. More primitive phyllos- analysis. This may be responsible for the dis- tomids (e.g., desmodontines) have bicornuate crepencies between our tree and that of Ger- uteri with a small common lumen, whereas ber (1968) and Gerber and Leone (1971). more derived taxa have fully simplex uteri with a single lumen (e.g., carolliines and UTERINE FUSION: PROGRESSIVE stenodermatines). Hood and Smith (1982) AND UNIDIRECTIONAL? found no reversals. Our results suggest a different pattern of Many mammalian orders have uterine uterine evolution (®gs. 50, 51)5. We found morphologies that are shared by all group that, as Hood and Smith (1982, 1983) indi- members (e.g., all edentates have a simplex cated, the last common ancestor of all phyl- uterus, all perissodactyls have a bicornuate lostomids had uterine horns that were rough- uterus; Mossman, 1987). In contrast, chirop- terans show a diversity of uterine morphol- 5 Although we lack data for these characters (see table ogies with duplex, bicornuate, and fully sim- 4) for many taxa included in our analysis, we have used plex uteri, and varying degrees of internal the character state assignments reconstructed in Mac- uterine fusion; most of these morphologies Clade as predictions of these states. Further work is occur in Phyllostomidae (Robin, 1881, Wood needed to be con®dent of these interpretations. 148 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 ly half the length of the common uterine the characters in a manner that re¯ected the body. After the split with the desmodontines, progressive fusion hypothesis appears to have the last common ancestor of the remaining partly biased the outcome of Hood and phyllostomids had a more derived degree of Smith's (1982) analysis. With the collapse of external uterine fusion: a fully simplex uterus the Macrotus-group node, no claim can be (®g. 50). The `ìntermediate'' condition of made that the fusion of the external uterus external uterine fusion (uterine horns ¼ the proceeded in a progressive fashion. length of the common uterine body) only oc- Our character congruence analysis, and the curs within the phyllostomine part of the reanalyzed data of Hood and Smith (1982, tree, as a synapomorphy of Micronycterini 1983) suggest that fusion of the external uter- plus Vampyrini (possibly also including Lon- us was not progressive. In our tree the `ìn- chorhinini, for which data is currently un- termediate'' condition occurs only as a syn- available). Thus, external uterine fusion is apomorphy of Micronycterini plus Vampyr- clearly not progressive in phyllostomids ini. Although patterns of variation in internal when these data are optimized on our tree. uterine fusion appear to be more compatible Patterns of variation in internal uterine fu- with the hypothesis that uterine fusion pro- sion (character 132) appear to be slightly ceeded in a progressive fashion, the reversal more congruent with the progressive fusion seen in Micronycterini and Vampyrini sug- hypothesis (®g. 51). The last common ances- gests that unidirectionality may not always tor of all phyllostomids was characterized by characterize the evolution of internal uterine a distinct cornual lumina. After the split with fusion. the desmodontines, the last common ancestor of the remaining phyllostomids had a more EVOLUTION OF FACIAL FEATURES: derived state: cornual lumina that were re- A PHYLOGENETIC PERSPECTIVE duced to intramural uterine cornua. In stenodermatines, the cornual lumina are absent, Vibrissae: Most mammals possess vibris- an even more derived condition. However, sae, long, stiff hairs that occur in discrete the evolution of internal uterine fusion in groups on the face. The locations of various phyllostomids was not unidirectional as dis- groups of facial vibrissae are highly conser- tinct cornual lumina evolved secondarily in vative (Pocock, 1914; Brown, 1971; Wines- the last common ancestor of Micronycterini ki, 1985; ®g. 17A). Many and perhaps all plus Vampyrini (and possibly Lonchorhinini; facial vibrissae have separate representations data is currently unavailable for this clade). in both the spinal trigeminal nucleus and the This represents a reversal to a more primitive somatosensory cortex (Zucker and Welker, condition. 1969; Waite, 1973a, 1973b; Woolsey, 1978), We explored the possibility that the out- con®rming a tactile role for these hairs. come of Hood and Smith's (1982) study was A single superciliary vibrissa occurs above the result of testing the progressive fusion hy- the eye on each side of the face in several pothesis with characters that had predeter- phyllostomids (see ®g. 17A). This trait ap- mined ordered transformation series (i.e., pas- pears to have evolved independently at least sage from the primitive state to the most de- ®ve times within the family (see character rived state requires passage through the inter- 11; ®g. 52). Unambiguous transformations mediate condition). Our comparison of occurred in: (1) the last common ancestor of analyses of the four female reproductive tract Desmodus and Diaemus, (2) the last common characters included in our study as ordered ancestor of Lampronycteris, Macrotus, and and unordered (characters 131±134) indicates Micronycteris, (3) within Lonchorhina (L. that running characters unordered results in marinkellei), and (4) in Centurio. There are the collapse of the Macrotus-group node that two alternative interpretations of superciliary Hood and Smith (1982) identi®ed (i.e., Ma- vibrissal evolution in Vampyrini. These vi- crotus, Trachops, Micronycteris hirsuta, M. brissae either evolved in the last common an- megalotis and Desmodus form a polytomy cestor of Chrotopterus and Vampyrum with the clade containing the remaining phyl- (DELTRAN), or in the last common ancestor lostomids; see ®g. 8 above). Thus, ordering of Tonatia and these two genera (ACCT- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 149

Fig. 50. Degree of external uterine fusion (character 131) optimized onto the strict consensus tree from our character congruence analysis. The ``intermediate'' state of external uterine fusion (character 131: horns one quarter the length of uterine body) is derived from the simplex condition, suggesting that external uterine fusion is not progressive. The equivocal optimization within Phyllostominae is due to missing data in Lonchorhinini. To prevent an equivocal reconstruction for the base of the clade that includes all phyllostomids except desmodontines, we examined trees in which the position of Brachy- phylla was resolved and ®xed the node at the base of the clade that includes all phyllostomids except desmodontines with the state that occurred under the two alternative placements for this genus. 150 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 51. Degree of internal uterine fusion (character 132) optimized onto the strict consensus tree from our character congruence analysis. There is a reversal to distinct cornual lumina from reduced cornual lumina (character 132) in some phyllostomines, suggesting that internal uterine fusion is not unidirectional. The equivocal optimization within Phyllostominae is due to missing data in Lonchorhin- ini. To prevent an equivocal reconstruction for the base of the clade that includes all phyllostomids except desmodontines, we examined trees in which the position of Brachyphylla was resolved and ®xed the node at the base of the clade that includes all phyllostomids except desmodontines with the state that occurred under the two alternative placements for this genus. 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 151

Fig. 52. Evolution of superciliary vibrissae inferred from optimization of character 11 on the strict consensus tree from our character congruence analysis. The equivocal reconstruction for the clade including Tonatia, Chrotopterus, and Vampyrum is due to the presence of taxonomic polymorphism in Tonatia and has two possible resolutions. The state for Lonchorhina and Tonatia is ``uncertain'' because of taxonomic polymorphism in these genera (see character 11).

RAN). Under the former alternative, the su- tions: under ACCTRAN the reconstruction is perciliary vibrissae present in some species equivocal, whereas under DELTRAN two of Tonatia (e.g., T. silvicola, and T. schulzi) superciliary vibrissae are primitive for Mor- are derived within the genus. moopidae. All desmodontines, most phyllos- Possession of two genal vibrissae ventral tomines, and most members of Nullicauda to and/or posterior to the eye on each side of retain the primitive complement of two genal the face (see ®gure 17A) appears to be the vibrissae. Members of Lonchorhinini share, primitive state in phyllostomid bats (see as a synapomorphy, the absence of these vi- character 12; ®g. 53). The reconstruction for brissae. Pygoderma is characterized by a re- the base of Mormoopidae has two resolu- duction in vibrissal number to one, as are 152 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 many species of Carollia. However, other Vampyrini, the reconstruction of the primi- Carollia species lack these vibrissae, making tive state for the tribe is equivocal as there the reconstruction equivocal. are two possible resolutions of this character. In Hirsutaglossa, many genera also have a Under ACCTRAN, absence of interramal vi- reduced number of genal vibrissae. Although brissae is primitive for the group of Chro- Phyllonycteris retains two genal vibrissae in topterus, Tonatia (presence of these vibrissae each cluster, Erophylla is autapomorphic in T. silvicola is derived within the genus), with a single genal vibrissa on each side of and Vampyrum. The base of Vampyrini re- the face. The genal vibrissae were complete- mains equivocal, however. Under DEL- ly lost in the last common ancestor of glos- TRAN, presence of two interramal vibrissae sophagines. Within Lonchophyllini, Lionyc- is primitive for all of Vampyrini and this con- teris has a single vibrissa on each side of the dition is retained by Tonatia (absence of face. A single genal vibrissa on each side of these vibrissae in T. saurophila is derived the face unites the clade of Anoura, Cho- within the genus). eroniscus, Choeronycteris, Hylonycteris, Within Hirsutaglossa, Lonchophyllini is Leptonycteris, Lichonycteris, Musonycteris, uniquely diagnosed by the presence of three and Scleronycteris. Within this group there is interramal vibrissae. Additional changes additional variation: Choeroniscus species within Hirsutaglossa occurred in the clade either retain the primitive complement (one comprising Choeroniscus, Choeronycteris, vibrissa) or have no genal vibrissae, whereas and Musonycteris, where there has been a re- Musonycteris has no genal vibrissae. The duction to a single interramal vibrissa. possible resolutions of the Musonycteris The interramal vibrissae were lost in the clade affect the reconstruction of the state last common ancestor of Stenodermatina, as changes. A single vibrissa on each cheek is well as in the last common ancestor of all primitive for Choeroniscus under all possible members of Ectophyllina except Artibeus optimizations except when Choeroniscus and and Enchisthenes. Within the Ectophyllina Musonycteris are sister taxa and ACCTRAN group, a reversal to the primitive condition is used. Under this optimization absence of of two vibrissae occurs in Vampyrodes. Ar- these vibrissae is primitive for the Choero- tibeus (Koopmania) has either one or two in- niscus-Musonycteris clade. terramal vibrissae, a condition that is unique Possession of two vibrissae between the to this subgenus. rami of the lower jaws well posterior to the Lateral vibrissal columns, consisting of mandibular symphysis appears to be the two vibrissae, were primitively present in primitive state for phyllostomids (``interra- phyllostomids (character 14; ®g. 55). These mal vibrissae;'' see character 13; ®g. 54). vibrissae have been lost in Phyllostomus, This condition is present basally in all sub- Lonchorhinini, and Nullicauda. However, the families. Within Phyllostominae, consider- reconstruction of this character is equivocal able vibrissal evolution occurred within because there are two alterantive optimiza- tribes, except Micronycterini. Within Phyl- tions of the character. If ACCTRAN is used, lostomini, the reconstruction for Phyllosto- loss of the lateral columns is a synapomor- mus is equivocal because some species have phy of the phyllostomine and nullicaudan a single vibrissa, whereas others are poly- clade. Reversal to the primitive condition morphic (absent or one vibrissa). Within (lateral vibrissal column present) occurs in Lonchorhinini, few genera retain the primi- Phylloderma and the clade comprising Vam- tive number of interramal vibrissae. Mimon pyrini and Micronycterini. The DELTRAN has lost the interramal vibrissae completely, alternative indicates that loss of the lateral whereas Macrophyllum is characterized by a vibrissal column evolved independently three reduction in vibrissal number to either none times (once each in Phyllostomus, Loncho- or one. Within Vampyrini, Trachops is rhinini, and Nullicauda). One unambiguous uniquely characterized by the presence of a character that uniquely diagnoses a steno- polymorphic condition (zero or one interra- dermatine group is the number of vibrissae mal vibrissa). Chrotopterus and Vampyrum in each medial vibrissal column (character have lost the interramals. However, within 15; ®g. 55). A reduction to three medial vi- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 153

Fig. 53. Evolution of the number of genal vibrissae inferred from optimization of character 12 on the strict consensus tree from our character congruence analysis. The equivocal reconstruction for the base of Mormoopidae is due to differences in interpretation under ACCTRAN and DELTRAN. The equivocal reconstructions for both Choeroniscus and Carollia are due to the presence of taxonomic polymorphism, and, in the case of Choeroniscus, different resolutions of the clade including this genus, Choeronycteris, and Musonycteris. The ``uncertain'' state, which appears for several taxa (e.g., Loncho- phylla, Lonchorhina), is due to taxonomic polymorphism (see character 12). To prevent an equivocal reconstruction for the base of Hirsutaglossa, we examined trees in which the position of Brachyphylla was resolved and ®xed the node at the base of Hirsutaglossa with the state that occurred under the two alternative placements for this genus. 154 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 54. Evolution of the number of interramal vibrissae inferred from optimization of character 13 on the strict consensus tree from our character congruence analysis. The equivocal reconstruction for Scleronycteris is due to missing data, while in Phyllostomus it is caused by taxonomic polymorphism. The equivocal reconstruction of this character in Vampyrini is due to the occurence of taxonomic polymorphism in Tonatia; there are two possible resolutions of this character in this clade under ACCT- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 155 brissae in each medial row occurred in the last common ancestor of Ametrida, Centurio, and Sphaeronycteris. Although vibrissal characters have not been previously used in phylogenetic analysis of phyllostomid bats, they appear to be informative at many taxonomic levels. Vi- brissal characters diagnose numerous groups within Phyllostomidae (e.g., Lonchorhinini, Lonchophyllini, Stenodermatina). Even the interramal vibrissae character, which includes much polymorphism and is highly homoplastic (character 13: ci ϭ .417), diagnoses several groups (e.g., Mimon, Lonchophyllini, the Choeroniscus clade, Stenodermatina). The utility of this character is re¯ected in its high retention index (ri ϭ .731). Vibrissal patterns may be related to foraging behavior in phyllostomids. Brown (1971) noted that facial vibrissae are well developed in predatory mammals. Thus, a high number of vibrissae would be expected in insectivorous and carnivorous bats, and a low number in nonpredatory bats. Our reconstructions indicate that, for most groups of vibrissae, the full complement is present primitively. Superciliary vibrissae, which are absent primitively in phyllostomids, evolved in predatory lineages (e.g., some Desmodon- tinae, Vampyrini, and Micronycterini) and Fig. 55. Evolution of the lateral vibrissal col- the genus Centurio, a frugivore. Loss of vi- umn inferred from optimization of character 14 brissae in some clusters (e.g., interramal, the on the strict consensus tree from our character lateral column) occurred most frequently in congruence analysis. The equivocal reconstruc- nullicaudans, suggesting these vibrissae may tion that begins with the last common ancestor of be less important in frugivorous taxa. Curi- Phyllostominae and Nullicauda is due to differ- ously, many of these vibrissae were also lost ences in interpretation of the character under in members of Lonchorhini, a group of pri- ACCTRAN or DELTRAN. The three taxa with marily insectivorous species. asterisks after their names have only three vibris- The Noseleaf: The noseleaf of phyllos- sae in each medial vibrissal column (character 15). tomid bats appears to be a uniquely derived feature of this group, despite the appearance of noseleaves in other bat families (e.g., Rhinolophidae, Megadermatidae; interpreted in the context of Simmons, 1998, and Sim-

← RAN and DELTRAN. The ``uncertain'' state in some taxa (e.g., Choeroniscus, Lonchorhina) is due to taxonomic polymorphism (see character 13). To prevent an equivocal reconstruction for the base of the clade including Choeroniscus, Choeronycteris, and Musonycteris, we examined trees in which the position of Scleronycteris, and the polytomy consisting of Choeroniscus, Choeronycteris, and Musonyc- teris, was resolved and ®xed the node at the base of Choeroniscus, Choeronycteris, and Musonycteris with the state that occurred under all alternative placements for these genera. 156 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 mons and Geisler, 1998). Noseleaves appear noseleaf of phyllonycterines is convergent to function in directing nasally emitted echo- with that in desmodontines and Brachyphyl- location calls (Fenton, 1984, 1985; Novick, la. Under DELTRAN, the shortened noseleaf 1977), but few studies have documented the evolved once in phyllostomids and the longer role that these structures play in echolocation spear of Glossphaginae is convergent with or addressed how differences in noseleaf that of the phyllostomine and nullicaudan structure affect nasal emission. clade. If Brachyphylla is the sister taxon of Hartley and Suthers (1987) investigated Phyllostominae and Nullicauda (®g. 56B), the role of the spear (which they termed the then a short noseleaf is primitive for Hirsu- lancet) in Carollia perspicillata. These au- taglossa. This, in turn, suggests that a long thors found that the spear acts to direct echo- spear evolved twice, once in Glossophaginae location pulses in the vertical dimension, but and once in the clade comprising Phyllosto- does not affect emission in the horizontal di- minae and Nullicauda. mension. Interference between the emission The position of Brachyphylla also in¯u- from each nostril is apparently used to direct ences the interpretation of the evolution of sound in the horizontal dimension (Hartley the U-shaped notch in the tip of the spear. If and Suthers, 1987). These results are similar Brachyphylla is the sister taxon of Hirsutag- to those for other species of bats in which lossa and the clade of Phyllostominae and the role of the noseleaf was studied (e.g., Me- Nullicauda (®g. 56C), the presence of a gaderma and Rhinolophus;MoÈhres and Neu- rounded or pointed spear tip (and absence of wiler, 1966; Strother and Mogus, 1970; So- the U-shaped notch) is a synapomorphy unit- kolov and Makarov, 1971; Schnitzler and ing the latter three taxa. However, if Brachy- Grinnell, 1977). Although other features of phylla is the sister taxon of Phyllostominae the noseleaf and horseshoe have not been and Nullicauda (®g. 56D), the reconstruction studied in relation to their affect on echolo- is equivocal. Under ACCTRAN, the recon- cation pulses, it seems likely that these char- struction suggests that the presence of the U- acters in¯uence some aspects of pulse emis- shaped notch evolved convergently in des- sions. Here, we explore the evolution of fea- modontines and Brachyphylla. Using DEL- tures of the noseleaf used in our phylogenetic TRAN, the reconstruction remains equivocal, analysis and evaluate synapomorphies and exactly as it appears in ®gure 56D. various optimization options for characters of Interestingly, some phyllostomids that both the spear and horseshoe. We also brie¯y have a truncated spear and a U-shaped notch examine the correlation between noseleaf at the tip also have a well-developed skin morphology and mode of foraging. ridge posterior to the spear that could pos- Our phylogenetic analysis indicates that sibly assist in directing echolocation pulses the last common ancestor of phyllostomids in the vertical dimension (see character 28). had a truncated (short) noseleaf (see charac- Although this structure, which evolved in the ters 19, 20; ®g. 56). In desmodontines, the last common ancestor of the three desmo- spear is short and has a U-shaped indenta- dontine genera, is more posterior to the nos- tion. The spear is also short in Brachyphylla. trils than the spear, its function has never Because the position of Brachyphylla is un- been investigated in relation to the emission resolved in the strict consensus tree, there are of echolocation calls. two possible interpretations for the evolution Several phyllostomid taxa have structures of spear length in Hirsutaglossa based on the posterior to the spear that differ from those position of this genus. If Brachyphylla is the in the desmodontines (see character 29). sister taxon of the clade that includes Hir- Sphaeronycteris is characterized by the pres- sutaglossa, Phyllostominae, and Nullicauda ence of a large, visorlike structure, a feature (®g. 56A), the reconstruction for Hirsutag- apparently uniquely evolved in this genus. lossa is equivocal because there are two al- The visor is much larger in males than in ternative optimizations. Under ACCTRAN, females, suggesting that sexual selection may the long noseleaf evolved once in the last have played a role in its evolution. Ecto- common ancestor of Hirsutaglossa, Phyllo- phylla macconnelli and Vampyressa pusilla stominae, and Nullicauda, and the shortened have small outgrowths (``lea¯ets'') posterior 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 157

Fig. 56. Evolution of spear length and spear tip shape inferred from the optimization of characters 19 and 20, respectively, on the strict consensus tree from our character congruence analysis. Optimi- zation of spear length A. with Brachyphylla as the sister taxon of Hirsutaglossa, Phyllostominae, and Nullicauda. The equivocal reconstruction beginning with the last common ancestor the clade including Hirsutaglossa, Phyllostominae, and Nullicauda is due to alternative optimizations under ACCTRAN and DELTRAN. B. with Brachyphylla as the sister taxon of Phyllostominae and Nullicauda. Optimizations of spear tip shape C. with Brachyphylla as the sister taxon of Hirsutaglossa, Phyllostominae, and Nul- licauda, and D. with Brachyphylla as the sister taxon of Phyllostominae and Nullicauda. The equivocal reconstruction beginning with the last common ancestor of Phyllostomidae is due to alternative reconstructions under ACCTRAN and DELTRAN. to the spear that probably do not function in the tip of the spear, evolved at least four directing echolocation pulses because they times in Phyllostomidae. A proximally re- are completely hidden behind the large spear. stricted rib is a synapomorphy of the sister There are two alternative explanations for the taxa Lionycteris and Lonchophylla. Within origin of these lea¯ets. They may be a syn- Phyllostominae, a proximally restricted rib apomorphy of the clade that includes both appears as a synapomorphy of all Micron- species of Ectophylla and Vampyressa pus- ycterini. However, within this clade, Macro- illa, or they may be the result of convergent tus lacks a rib. Similarly, all members of evolution in Ectophylla macconnelli and Stenodermatini have, as a synapomorphy, a Vampyressa pusilla. rib which extends to the tip of the spear, ex- The interpretation of variation in the cen- cept Sphaeronycteris, which has a proximal- tral rib is also somewhat ambiguous (see ly restricted rib. The reconstruction in the character 21, ®g. 57). In the most basal phyl- phyllostomine-nullicaudan clade is equivocal lostomid taxa (e.g., desmodontines, Brachy- because there are two alternative optimiza- phylla, phyllonycterines, and most glosso- tions of this character for this group. Regard- phagines), the rib is undifferentiated. A prox- less of the position of Brachyphylla (as the imally restricted rib, which does not reach sister taxon of Hirsutaglossa, Phyllostomi- 158 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 57. Evolution of length of the central rib inferred from optimization of character 21 on the strict consensus tree from our character congruence analysis. The equivocal reconstruction for the base of the clade including phyllostomines and nullicaudans is due to alternative interpretations under ACCT- RAN and DELTRAN. The reconstruction for Centurio is equivocal due to missing data (see character 18). See text for discussion. nae, and Nullicauda, or as the sister taxon of ACCTRAN reconstruction suggests that hav- Phyllostominae and Nullicauda), the optimi- ing a proximally restricted rib is primitive, zation for this clade is similar. Under ACCT- whereas the DELTRAN reconstruction re- RAN, the resolution for the base of the phyl- mains equivocal. lostomine-nullicaudan clade is equivocal. Other features of the noseleaf have only Under DELTRAN, the absence of a rib is slightly less ambiguous origins. The inter- primitive for the clade uniting Phyllostomi- narial ridge evolved within Phyllostomidae nae and Nullicauda. For Nullicauda, the either ®ve or six times (see character 22, ®g. 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 159

Fig. 58. Evolution of the internarial structures inferred from optimization of character 24 on the strict consensus tree from our character congruence analysis. The equivocal reconstruction within Hir- sutaglossa is due to alternative interpretations under ACCTRAN and DELTRAN. Note that ``polymorphic'' indicates that some individuals in a species have a ridge or papillae, whereas others do not.

58). This feature appears as a synapomorphy sized was homologous with a pommel- of the sister taxa Lonchorhina and Macro- shaped structure in Lonchorhina (see char- phyllum, as well as all members of Vampyr- acter 23). The sella appears to have evolved ini. The presence of an internarial ridge is an twice, once in the last common ancestor of autapomorphy of Macrotus. The appearance Chrotopterus and Vampyrum and a second of this ridge in some individuals of Micro- time in Lonchorhina, and thus is not homol- nycteris hirsuta uniquely characterizes this ogous in all three taxa. species. In addition, depending on the opti- The lateral boundary of the phyllostomid mization used, the internarial ridge is either horseshoe was primitively thin and free (see a synapomorphy of Glossophaginae with a character 24, ®g. 59). Within Hirsutaglossa, reversal in Platalina (ACCTRAN), or it the last common ancestor of all glossophag- evolved convergently, once in the last com- ines apparently evolved a horseshoe that was mon ancestor of Glossophagini and once in fully fused to the face. Subsequently, a the last common ancestor of Lionycteris and horseshoe that was only partly fused to the Lonchophylla (DELTRAN). face (superior part is free, inferior part is A unique feature of the noseleaf appears fused to the skin of the face) probably in three other phyllostomid taxa. Chrotopte- evolved in the last common ancestor of Cho- rus and Vampyrum have a sella, a rounded eroniscus, Choeronycteris, Hylonycteris, Li- globular structure that we initially hypothe- chonycteris, Musonycteris, and Scleronycter- 160 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 59. Evolution of the lateral horseshoe inferred from optimization of character 24 on the strict consensus tree from our character congruence analysis. The equivocal reconstruction for the evolution of a ``partly free edge'' is due to the missing data for Scleronycteris. is. Only when Scleronycteris appears as the thenes, Sphaeronycteris, and Sturnira, (see sister taxon of the other genera that have a ®g. 60). Within Stenodermatina, a thickened partially free horseshoe is the reconstruction boundary evolved in the last common ances- at the base of the clade equivocal due to tor of Ardops, Ariteus, Stenoderma, Phyllops, missing data. and Pygoderma.InStenoderma, a thin, free The labial boundary of the horseshoe ap- ¯ap has evolved. A thin, free ¯ap also appears to have been undifferentiated from the pears as autapomorphy of Ametrida. A thin, upper lip primitively, as this condition occurs free ¯ap also evolved in the last common an- basally in the family as well as all subfami- cestor of all Ectophyllina save Enchisthenes, lies (see character 25, ®g. 60). The presence making this feature a synapomorphy of this of the thin, free edge is highly homoplastic, group. having evolved six times within the family Among taxa that have a thickened labial (see below). In Phyllostominae, the lip and border of the horseshoe, Ardops and Ariteus horseshoe were undifferentiated primitively share, as a synapomorphy, the presence of a within Phyllostomini (a more derived con- V-shaped labial projection (see character 27, dition, a thin, free ¯ap, occurs in Phyllosto- ®g. 60). The presence of a V-shaped notch in mus), Vampyrini (a more derived condition, some individuals appears in Micronycteris a thin, free ¯ap, occurs in Chrotopterus and megalotis and M. minuta, whereas the pres- Vampyrum), and Micronycterini. Within Mi- ence of this feature in all individuals of Mi- cronycterini, the thickened condition of the cronycteris hirsuta is uniquely derived in this labial horseshoe evolved in the last common taxon. The other taxa with a thickened labial ancestor of Lampronycteris, Macrotus, and border all possess a smooth rounded border Micronycteris, making this feature a syna- (e.g., Macrotus, Mimon bennettii, Lampron- pomorphy of this group. Lonchorhinini is ycteris, Phyllops, and Pygoderma). characterized by a derived condition, a thin, Finally, two genera, Chrotopterus and free edge. A thickened edge apparently Vampyrum, have a horseshoe in which the evolved subsequently in Mimon bennettii. thin, free edges are cupped around the nos- Similarly, the presence of an undifferenti- trils (see character 26; reconstruction not ated labial horseshoe border was primitive in shown). This feature is unique to these two Nullicauda, appearing in Carolliinae, Enchis- phyllostomids; in all other genera with free 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 161

Fig. 60. Evolution of the labial horseshoe inferred from optimization of character 25 on the strict consensus tree from our character congruence analysis. The morphology of the thickened labial horseshoe (character 27) is also indicated: asterisks indicate taxa with a V-shaped labial projection; a single cross indicates taxa in which all individuals have a V-shaped notch; a double cross indicates taxa in which some individuals have a V-shaped notch. The state for Centurio is not indicated because we scored this taxon ``?'' for all characters related to the noseleaf (see character 18). 162 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 edges on the horseshoe, the labial edge lies nes. Our analysis suggests that this short ¯at over the upper lip. structure may have originally evolved in the Noseleaf morphology appears to be useful last common ancestor of phyllostomids to in phylogenetic reconstruction at many tax- improve echolocating ability as a supplement onomic levels. Noseleaf characters diagnose to other senses (e.g., vision, olfaction, and numerous groups within Phyllostomidae thermoperception). The factors related to for- (e.g., Glossophaginae, Lonchorhinini, Vam- aging behavior that are involved in the evo- pyrini). In addition, congruence of these lution of a longer spear are not clear. Because characters with those of other systems indi- many phyllostomids rely so heavily on other cates that our hypotheses of homology for senses for prey detection, it seems unlikely the various structures seem to be valid. One that the longer spear seen in these bats par- notable exception is the sella of Chrotopte- allels an increasing reliance on echolocation rus, Lonchorhina, and Vampyrum (see for detection of food items. above). The noseleaf became a much more elab- TRACING THE DIVERSIFICATION orate and complex structure within phyllos- OF FEEDING HABITS tomids over evolutionary time. Primitively within the family, the spear was short, the Phyllostomids show a remarkable diversi- internarial region was ¯at, and the horseshoe ty of feeding specializations. However, data was undifferentiated from the upper lip. Sub- on the feeding ecology of phyllostomids are sequently, within the various subfamilies, the very incomplete. The diet of most species spear became more elongate, the central rib has been poorly studied, and seasonal and and other internarial structures evolved, and geographical variation in the diet has been the labial horseshoe became ¯aplike or inadequately documented. One example of cupped in some taxa. seasonal dietary change involves Carollia The evolutionary forces shaping noseleaf perspicillata, the phyllostomid whose diet morphology in phyllostomids remain un- has been perhaps most thoroughly investi- known. Earlier researchers proposed that gated. Several studies have suggested that noseleaf morphology was related to foraging this bat consumes more insects, nectar, and behavior and that echolocation played some pollen during the dry season than during the role in this relationship. However, many rainy season (Fleming et al., 1972; Heithaus phyllostomids rely on passive auditory cues, et al., 1975; Sazima, 1976; Fleming and olfaction, vision, or touch to detect and select Heithaus, 1986). Fleming et al. (1972) found food items (e.g., Tuttle and Ryan, 1981; Las- that insects make up 40% of the diet of Car- ka, 1990; Kalko and Condon, 1993). In ad- ollia perspicillata during the dry season, but dition, the link between morphology and only 10% during the wet season. echolocation is unclear. For example, in the Other serious problems with collecting di- group with the most variable noseleaves, the etary data involve over- or underestimation phyllostomines (Bogdanowicz et al., 1997), of various food types. Carollia perspicillata echolocation calls of the different species are reportedly consumes only the soft inner parts surprisingly similar (Belwood, 1988). Thus, of hard-bodied insects that it feeds upon, re- it is dif®cult to envision how foraging be- sulting in few recognisable insect remains in havior and diet are linked to phyllostomid fecal samples (Ayala and d'Allessandro, noseleaf morphology through echolocation 1973). Therefore, the proportion of insects in (Bogdanowicz et al., 1997). the diet may be underestimated. Similarly, Arita (1990) suggested that bats that rely many bats discard large seeds without swal- on senses other than echolocation to locate lowing them. Consequently, fecal sampling food (e.g., olfaction, vision, thermopercep- underestimates the proportion of these large- tion) may require less directionality in echo- seeded fruits in the diet (Thomas, 1988). Col- location calls than those that rely primarily lecting samples from night and day roosts on echolocation. He suggested that this may may provide a different picture of bat diets explain the truncated spears seen in desmo- (Thomas, 1988). In addition, geographic var- dontines, Brachyphylla, and phyllonycteri- iation in dietary habits may be important in 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 163 widespread taxa that occupy different habi- small vertebrates, but supplement this diet tats in different parts of their range. For ex- with insects and fruits (Ferrarezi and Gime- ample, Carollia perspicillata is found in nez, 1996). thorn scrublands, dry deciduous forests, and Coding dietary data for use as characters lowland rainforests (Pine, 1972; Fleming, is not always straightforward. For example, 1988; Cloutier and Thomas, 1992; Koopman, Ferrarezzi and Giminez (1996) coded Pter- 1993, 1994; Voss and Emmons, 1996), and onotus as a stict insectivore, despite evidence in arid regions this species may feed on cac- that more than 10% of the dietary samples tus fruits that are not available in moister in one study of P. quadridens included pol- habitats (Santos et al., 1996). len (Rodriguez-Duran and Lewis, 1987). We Despite such problems, Ferrarezi and Gi- follow Ferrarezzi and Giminez (1996), be- menez (1996) made the ®rst explicitly phy- cause other authors have suggested that P. logenetic attempt to consider the evolution of quadridens inadvertantly consumed pollen feeding specializations in the family as a while eating insects, its primary food source whole. These authors compiled a dietary data (Rodriguez-Duran and Kunz, 1992). Al- set based on previously published literature though similar problems are found in other accounts and mapped eight diet characters areas of the data set (e.g., reports of frugi- (one composite, seven multistate; see table 8 vory in the sanguivorous Desmodus), we for multistate) onto a phylogeny they assem- have only slightly modi®ed the original cod- bled from several studies (see ®g. 61; Ho- ings of Ferrarezi and Gimenez (1996; see our neycutt and Sarich, 1987a; Baker et al., table 8). 1989; Lim, 1993; Owen, 1991; Gimenez et In the context of the phylogeny used by al., 1996). The composite character com- Ferrarezzi and Gimenez (1996; ®g. 61), pre- bined all seven multistate characters in a dominant insectivory appeared to be primi- complex transformation series (not shown). tive for Phyllostomidae. In addition, it ap- The terminal taxa used in this analysis in- peared that most of the major shifts in diet cluded species, genera, tribes, and families of had occurred a single time each. Sanguivory noctilionoids (see table 8). evolved once in the last common ancestor of After assessing the relative importance of the three desmodontine genera. A predomi- food types in the diet, Ferrarezi and Gimenez nantly carnivorous way of life evolved in the (1996: table 1; see our table 8) de®ned one last common ancestor of Vampyrinae, and character for insectivory using the following strict insectivory may have evolved once or states (in addition to ``absent''): strict, pre- twice due to the unresolved relationships of dominant, and complementary. ``Strict insec- Lonchorhina and Macrophyllum. The last tivory'' indicates that no foods other than in- common ancestor of all nectarivores and fru- sects are consumed. ``Predominant insecti- givores primarily consumed fruit, indicating vory'' indicates that insects are a primary a single evolution of predominant frugivory. food source, but are not the only foods eaten. Predominant nectarivory and strict frugivory A ``complementary'' reliance on insects in- both evolved once. dicates that this food source is a secondary Our results are similar to those of Ferrarezi component of the diet. Ferrarezi and Gime- and Gimenez (1996). In the context of our nez (1996) de®ned two similar characters for phylogeny, strict insectivory appears to be frugivory (absent, complementary, predomi- primitive for Noctilionoidea (this cannot be nant, strict; other states in the original char- inferred unless noctilionoids are considered acter do not occur in phyllostomids), and in the context of Simmons and Geisler's nectarivory/palynivory (absent, complemen- [1998] microchiropteran phylogeny; see ®g. tary, predominant). Finally, Ferrarezi and Gi- 62). However, the primitive condition for menez (1996) included three presence/ab- Phyllostomidae is equivocal: either strict in- sence characters for sanguivory, carnivory, sectivory is primitive (DELTRAN) or the re- and folivory. They de®ned the derived state construction remains equivocal(ACCTRAN), of the carnivory character as ``predominant and either the absence of insectivory or the carnivory,'' indicating that taxa with the de- complementary state could be primitive for rived state of the character primarily feed on the family. Sanguivory (reconstruction not 164 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 165 166 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 61. Tree used by Ferrarezi and Gimenez (1996; redrawn from ®g. 4) with their feeding-habits character optimized on the topology. This character was ordered such that predominant insectivory evolved from strict insectivory; predominant carnivory, predominant frugivory, or sanguivory evolved from predominant insectivory; and predominant nectarivory or strict frugivory evolved from predominant frugivory. shown) is a synapomorphy of Desmodonti- genera, although this is not the only evolu- nae. At the node uniting Brachyphyllinae, tionary origin of this character, which also Hirsutaglossa, and the clade of Phyllostomi- occurs in Mimon (see ®g. 65). nae plus Nullicauda, insects appear to have Fruits appear to supplement the diet of become a less important part of the diet as most phyllostomids (see ®g. 63). However, the importance of plant parts (fruits, nectar, frugivory has become most important in Nul- and pollen) increased (regardless of the po- licauda; many members of this group rely al- sition of Brachyphylla; see ®gs. 63, 64). most exclusively on fruit and may not in- Despite the importance of plants to most clude any insects in their diet. Predominant phyllostomids, many phyllostomines rely frugivory evolved three times in phyllostom- principally on insects and small vertebrates ids: (1) in the last common ancestor of Nul- for food (see ®gs. 62, 65). Predominant in- licauda, (2) in the last common ancestor of sectivory evolved in the last common ances- all Ectophyllina genera except Artibeus and tor of Phyllostominae, with a strict habit Enchisthenes, and (3) in Artibeus (Artibeus). evolving in the last common ancestor of the In contrast, strict frugivory evolved only in sister taxa Lonchorhina and Macrophyllum. the last common ancestor of Stenodermatini. In Vampyrini, species of Tonatia may rely Although most phyllostomids apparently do more on insects for nutrients than do other not feed on plant products such as stems, members of this clade (this condition repre- leaves, and ¯owers, folivory has apparently sents either retention of the primitive condi- evolved ®ve times in the family (in Lam- tion [DELTRAN] or a reversal [ACCT- pronycteris, Glossophaga, Carollia, Arti- RAN]). Predominant carnivory appears as a beus, and Platyrrhinus; reconstruction not synapomorphy uniting all four Vampyrini shown; see table 8). 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 167

Fig. 62. Evolution of different types of insectivory (see text for description of character states) inferred from our optimization of the insectivory character of Ferrarezi and Gimenez (1996: table 1; see our table 8) on our strict consensus tree from the character congruence analysis. We inferred the state at the root with reference to a phylogeny of Microchiroptera (Simmons, 1998). The equivocal optimizations for Phyllostomidae and Vampyrini are due to differences in interpretation of the character under ACCTRAN or DELTRAN. The ``uncertain'' state for Noctilio is due to taxonomic polymorphism (see table 8).

A reliance on nectarivory/palynivory We found that sanguivory, strict insecti- evolved in the last common ancestor of the vory, predominant nectarivory, and strict fru- clade comprising Brachyphylla, Hirsutaglos- givory all have single evolutionary origins sa, Phyllostominae, and Nullicauda. The re- within Phyllostomidae. However, carnivory, construction is similar regardless of the po- predominant frugivory, and possibly pre- sition of Brachyphylla: predominant nectari- dominant insectivory all may have evolved vory is primitive for clade including all phyl- more than once (see ®gs. 63±65). These dif- lostomids except the desmodontines. ferences between our conclusions and those Complementary nectarivory evolved four of Ferrarezi and Gimenez (1996) are due to times (see ®g. 64). a change we made in the coding of Mimon 168 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 63. Evolution of different types of frugivory inferred from our optimization of the frugivory character of Ferrarezi and Gimenez (1996: table 1; see our table 8) on our strict consensus tree from the character congruence analysis. To prevent an equivocal reconstruction for the base of the clade that includes all phyllostomids except desmodontines, we examined trees in which the position of Brachy- phylla was resolved and ®xed the node at the base of the clade that includes all phyllostomids except desmodontines with the state that occurred under the two alternative placements for this genus. for the carnivory character and the different Brachyphylla, Hirsutaglossa, and Nullicauda tree topology we used (see table 8). (®g. 61). We found that plant products became an Despite the growing consensus concern- important food source for phyllostomids ear- ing the number of evolutionary origins of lier in their evolutionary history than Ferrar- feeding specializations in phyllostomids, ezi and Gimenez (1996) suggest. In our tree few hypotheses have been advanced to de- the switch to herbivory appears to character- scribe the evolutionary steps neccessary to ize the clade that includes all phyllostomids derive these specializations from their evo- except desmodontines (see ®gs. 63±64). lutionary precursors. Gillette (1975) put Concomitant with the increasing importance forth perhaps the only model to describe the of plants is a reduction in reliance on insects evolution of feeding strategies in bats. He (®g. 62). In contrast, the topology used by based his hypothesis on the following state- Ferrarezi and Gimenez (1996; ®g. 61) sug- ment by Jepsen (1970: 56), ``some forms gests that a reliance on plant products preferred fruit, after ®nding initially that it evolved only in the last common ancestor of was a good source for worms and bugs, and 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 169

Fig. 64. Evolution of different types of nectarivory inferred from our optimization of the nectarivory character of Ferrarezi and Gimenez (1996: table 1; see our table 8) on our strict consensus tree from the character congruence analysis. Note that the nectarivory character, as de®ned by Ferrarezi and Gimenez (1996) includes the consumption of pollen and petals. To prevent an equivocal reconstruction for the base of the clade that includes all phyllostomids except desmodontines, we examined trees in which the position of Brachyphylla was resolved and ®xed the node at the base of the clade that includes all phyllostomids except desmodontines with the state that occurred under the two alternative placements for this genus. in accord with such dietary changes, new of the alternative (plant) food source. As Fer- dental types appeared.'' rarezi and Gimenez (1996) found, this model Gillette's (1975) model relied on what he accurately describes their hypothesis of feed- called ``food source duality.'' From a con- ing specializations on their composite tree dition of generalized insectivory, he suggest- (see ®g. 61). ed that bats evolved to specialized insecti- Food source duality is also compatible vory, gleaning insects from fruit or ¯owers. with many parts of our hypothesis. For ex- At some point, these animals shifted to using ample, food source duality (predominant fru- the substrate (the plant material) as a food givory) is characteristic of basal nullicau- source in conjunction with the insects they dans, although some derived nullicaudans are had originally fed upon, thus relying on dual strictly frugivorous (®g. 63). However, the food sources. The culmination of this evo- two possible resolutions of predominant nec- lutionary sequence involved sole exploitation tarivory/palynivory on our tree indicate that 170 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 65. Evolution of carnivory inferred from our optimization of the carnivory character of Ferrarezi and Gimenez (1996: table 1; see our table 8) on our strict consensus tree from the character congruence analysis. The ``uncertain'' states for Phyllostomus and Tonatia are due to taxonomic polymorphism. this condition did not always arise from the nivores that began ingesting pieces of ¯esh less specialized condition involving some re- along with blood from these wound sites liance on both plants and insects (®g. 64). (Schutt, 1998). Only Slaughter's (1970) and Instead, our tree indicates that predominant Schutt's (1998) hypotheses were made in the nectarivory/palynivory may have arisen from context of a phylogeny. strict insectivory. Ferrarezi and Gimenez (1996) noted that For desmodontines, there is no direct evi- hypotheses proposing that blood feeding dence of food source duality. To explain the evolved from food source duality on mam- evolution of blood feeding in this group ®ve malian hosts (hypotheses 2 and 3) are not the hypotheses have been advanced: (1) desmo- most parsimonious interpretations. Given the dontines evolved from frugivorous species of relationships among the three desmodontine phyllostomids that were capable of piercing genera (Diphylla(Diaemus, Desmodus)), the thick fruit rinds (Slaughter, 1970); (2) desmodontines intially fed on the ectoparasites most parsimonious interpretation is that the of large mammals (Gillette, 1975; Turner, last common ancestor of extant desmodon- 1975); (3) desmodontines intially fed on in- tines was arboreal, as Sazima (1978) and sects attracted to the wounds of large mam- Schutt (1998) have suggested, and fed on the mals (Fenton, 1992); and (4) desmodontines blood of birds as do Diaemus and Diphylla initially fed on small vertebrates (Schmidt, today. Hypotheses that blood feeding 1978), and may have specialized on small evolved from frugivory and carnivory cur- arboreal prey items (e.g., Diaemus; Sazima, rently lack any corroborating phylogenetic 1978), (5) desmodontines were arboreal om- evidence.

SUMMARY AND CONCLUSIONS Phyllostomidae is a large (more than 140 sects, vertebrates, nectar, pollen, and fruits. species), diverse clade of Neotropical bats This group offers a number of problems for that includes species that feed on blood, in- systematists at many taxonomic levels. Pre- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 171 vious attempts to resolve phylogenetic prob- portions of the phyllostomid tree, and several lems in phyllostomids were hampered by traditionally recognized clades (e.g., Carol- limited taxonomic sampling, limited data liinae, Stenodermatinae) do not appear in any sets, and use of taxonomic congruence. The of the strict consensus trees of the separate result was a poorly resolved picture of phyl- data partitions. In addition to these problems, lostomid relationships. Our total evidence we ®nd no compelling evidence that parti- analysis of 150 morphological and molecular tioning morphological data re¯ects any un- characters resulted in a well-resolved hypoth- derlying biological reality. We therefore rec- esis of relationships within this family. Re- ommend character congruence or ``total ev- sults of parsimony analyses of our combined idence'' to make classi®catory recommen- data set (appendix 2) indicate that all tradi- dations and to provide a phylogenetic tionally recognized phyllostomid subfamilies framework. are monophyletic and that most taxa sharing Although both chromosomal and immu- feeding specializations form clades. These nological data provide additional support for results largely agree with studies that have several clades that we identi®ed (e.g., Des- used a taxonomic congruence approach to modontinae, Stenodermatinae, Nullicauda, evaluate karyological, immunological, and Hirsutaglossa, Phyllonycterinae), these data limited sets of morphological characters, al- sets are incongruent with many aspects of though our ®nding that Phyllostominae is our phylogenetic results, especially our ®nd- monophyletic is novel. Our results indicate ing of phyllostomine monophyly. These con- that several genera (Micronycteris, Artibeus, ¯icts may be due to methodological con- and Vampyressa) are not monophyletic. straints associated with the use of karyolog- Based on our phylogenetic results, we pro- ical and immunological data (e.g., problems pose a new classi®cation for Phyllostomidae with assessing homologies and distinguish- that better re¯ects hypothesized relation- ing primitive from derived traits). Among ships. Important features of this new classi- other observations, we ®nd that Macrotus ®cation include: (1) formal recognition of waterhousii, which has been thought to have Hirsutaglossa and Nullicauda to group nec- the primitive karyotype for Phyllostomidae tarivorous and frugivorous subfamilies, re- (Patton and Baker, 1978), nests well within spectively, (2) rede®nition of Glossophagi- the phyllostomine clade. This suggests that nae and inclusion of Glossophagini and Lon- results of previous analyses of chromosomal chophyllini within this subfamily, (3) rec- data may require reevaluation. ognition of Phyllostomini, Lonchorhinini, Mapping characters and behaviors on our Vampyrini, and Micronycterini as tribes phylogenetic tree provides a context for eval- within Phyllostominae, (4) formal recogni- uating hypotheses of evolution in Phyllo- tion of Stenodermatina (``short-faced'' sten- stomidae. Although previous studies of uter- odermatines) and Ectophyllina (``long- ine evolution in phyllostomids and other faced'' stenodermatines), (5) elevation of mammals have generally supported the uni- Glyphonycteris, Lampronycteris, Neonycter- directional progressive fusion hypothesis, our is, Trinycteris, and the nominate subgenus of results indicate that intermediate stages of Micronycteris to generic rank, (6) recogni- uterine fusion are often derived relative to tion of Mesophylla as a junior synonym of the fully simplex condition, and that rever- Ectophylla, (7) recognition of Enchisthenes sals occur with respect to internal uterine fu- as a distinct genus, and (8) retention of Der- sion. Uterine fusion therefore appears to be manura and Koopmania as subgenera of Ar- neither completely unidirectional nor pro- tibeus. gressive in Phyllostomidae. Our comparison of character congruence Evolution of the vibrissae and noseleaf are and taxonomic congruence indicates that also complex and homoplasy is commonly character congruence provides improved res- seen in the evolutionary history of these olution of relationships among phyllostom- structures; however, many of these features ids. Many of the data partitions we identi®ed diagnose clades of phyllostomids (see appen- and analyzed separately are informative only dix 4) and appear to be phylogenetically in- at limited hierarchical levels or in certain formative, indicating that they may be useful 172 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 in other mammalian and chiropteran groups. are represented by a single large clade within There is considerable reduction in numbers Phyllostomidae indicating that each feeding of vibrissae present in various clusters within specialization evolved once. However, rever- Phyllostomidae. These reductions occurred sals do occur (e.g., loss of nectar and pollen principally in Nullicauda, but also occurred feeding in many phyllostomines and sten- in Lonchorhinini and Hirsutaglossa, and may odermatines), and some specializations may be related to foraging behavior. have evolved more than once (e.g., carnivo- Within Phyllostomidae, the noseleaf seems ry). to have become a much more elaborate and Additional research effort needs to be ex- complex structure over evolutionary time. pended on resolving some remaining system- Primitively within the family, the spear was atic problems. The position of Brachyphylla short, the internarial region was ¯at, and the is unresolved in our most parsimonious to- horseshoe was undifferentiated from the up- pology, and it is not clear how immunolog- per lip. Subsequently, within the various sub- ical and karyological data that support place- families, the spear became more elongate, the ment of this genus with Hirsutaglossa should central rib and other internarial structures be interpreted. The status of the genus Vam- evolved, and the labial horseshoe became pyressa is also not clear. We agree with the ¯aplike or cupped in some taxa. results of some previous investigations that Presumably, noseleaf morphology is relat- indicated that this genus may not be mono- ed to foraging behavior in phyllostomids. We phyletic (e.g., Owen, 1987); however, our suspect, as have others (e.g., Neuweiler, analysis did not include all Vampyressa spe- 1984; Arita, 1990; Bogdanowicz et al., cies. Therefore, the genus should be re- 1997), that many characters of the noseleaf viewed before any nomenclatural change is in¯uence echolocation-emission patterns and made. Finally, several studies have suggested may be related to foraging strategies and that Mormoopidae, not a clade comprising habitat exploitation; however, most features Mormoopidae and Noctilionidae, may be the of the noseleaf have not been explored in this sister taxon of Phyllostomidae (Van Valen, context. 1979; Novacek, 1991; Kirsch et al., 1998; We suggest that the primitively short nose- Simmons and Conway, MS). In the future, leaf may have originally functioned to im- we hope to resolve this question by perform- prove echolocating ability as a supplement to ing a global parsimony analysis that inten- other senses (e.g., vision, olfaction, and ther- sively samples all noctilionoids. It is not moperception). Because many phyllostomids clear exactly how many of the weakly and rely so heavily on other senses for prey de- moderately supported clades we have iden- tection, it is unlikely that the larger spear ti®ed will be recovered in additional analyses seen in more derived taxa parallels an in- that add new data to our existing data matrix. creasing reliance on echolocation for detec- As our study built upon those of many pre- tion of food items. vious workers, we hope that the data and hy- Dietary evolution in phyllostomids ap- potheses presented here will provide a start- pears somewhat more complex than previ- ing point for new and productive investiga- ously thought. We ®nd that most of the major tions of phyllostomid relationships and evo- dietary guilds (e.g., frugivory, sanguivory) lution.

ACKNOWLEDGMENTS Special thanks to T. Grif®ths, C. Hood, J. ry), and P. Myers (University of Michigan C. Morales, R. Van Den Bussche, J. Wetterer, Museum of Zoology) for access to specimens and J. Wible for their careful reviews of an in their care. Special thanks to P. Brunauer earlier version of this manuscript. Thanks for locating numerous references cited in this also to M. Carleton, C. Handley, and L. Gor- contribution, T. Conway for her assistance don (United States National Museum), B. with many aspects of manuscript preparation, Patterson (Field Museum of Natural Histo- and P. Wynne for her excellent illustrations; 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 173 without their contributions this paper would and Conservation, and Columbia Universi- be considerably impoverished. A. Porzecan- ty's Center for Environmental Research and ski provided help with translations, and we Conservation. NSF Research Grants (DEB- are grateful for her assistance. ALW extends 9106868 and DEB-9873663) provided par- special thanks to J. Wetterer, whose assis- tial support for NBS. NSF Research Experi- tance and encouragement was essential to the ences for Undergraduates Grant BIO- completion of this manuscript. 9200351 to M. Stiassny provided funding to Fellowship support for ALW was provided ALW and MVR; we thank M. Stiassny and by the AMNH Of®ce of Grants and Fellow- D. Bynum for their help and support of the ships, the AMNH Center for Biodiversity AMNH REU program.

REFERENCES Allen, H. A. Baker, R. J. 1892a. On a new family of phyllostome bats. 1967. Karyotypes of bats of the family Phyl- Proc. U.S. Natl. Mus. 15: 437±439. lostomidae and their taxonomic impli- 1892b. Description of a new genus of phyllo- cations. Southwest. Nat. 12: 407±428. stome bats. Ibid. 15: 441±442. 1970. The role of karyotypes in phylogenetic 1898. On the Glossophaginae. Trans. Am. studies of bats. In B. H. Slaughter and Philos. Soc. 19: 237±266. W. D. Walton (eds.), About bats: 303± Altenbach, J.S. 312. Dallas: Southern Methodist Univ. 1979. Locomotor morphology of the vampire Press. bat, Desmodus rotundus. Spec. Publ. 1973. Comparative cytogenetics of the New Am. Soc. Mammal. 6: 1±137. World leaf±nosed bats (Phyllostomati- Anderson, K. dae). Period. Biol. 75: 37±45. 1906. On the bats of the genus Micronycteris 1979. Karyology. In R. J. Baker, J. K. Jones, and Glyphonycteris. Ann. Mag. Nat. Jr., and D. C. Carter (eds.), Biology of Hist. ser. 7, 18: 50±63. the bats of the New World family Phyl- 1912. Catalogue of the Chiroptera in the col- lostomatidae, p. III. Spec. Publ. Mus. lection of the British Museum, vol. I: Texas Tech Univ. 16: 1107±155. Baker, R. J., and R. A. Bass Megachiroptera. London: British Mu- 1979. Evolutionary relationship of the Bra- seum. chyphyllinae to the glossophagine gen- Anderson, S. A. era Glossophaga and Monophyllus.J. 1997. Mammals of Bolivia, taxonomy and Mammal. 60: 364±372. distribution. Bull. Am. Mus. Nat. Hist. Baker, R. J., and W. J. Bleier 231: 1±652. 1971. Karyotypes of bats of the subfamily Anderson, S. A., K. F. Koopman, and G. K. Carolliinae (Mammalia; Phyllostomati- Creighton dae) and their evolutionary implica- 1982. Bats of Bolivia: an annotated checklist. tions. Experientia 27: 220±222. Am. Mus. Novitates 2750: 24 pp. Baker, R. J., and T. C. Hsu Arita, H. T. 1970. Further studies on the sex±chromosome 1990. Noseleaf morphology and ecological systems of the American leaf-nosed correlates in phyllostomid bats. J. bats (Chiroptera, Phyllostomatidae). Mammal. 71: 36±47. Cytogenetics 9: 131±138. Arnold, M. L., R. L. Honeycutt, R. J. Baker, V. Baker, R. J., and G. Lopez M. Sarich, and J. K. Jones, Jr. 1970. Karyotypic studies of the insular pop- 1982. Resolving a phylogeny with multiple ulations of bats on Puerto Rico. Car- data sets: a systematic study of phyl- yologia 23: 465±472. lostomoid bats. Occas. Pap. Mus. Texas Baker, R. J., H. H. Genoways, W. J. Bleier, and J. Tech Univ. 77: 1±55. W. Warner Ayala, S. C., and A. d'Allessandro 1973. Cytotypes and morphometrics of two 1973. Insect feeding behavior of some Col- phyllostomatid bats, Micronycteris hir- ombian fruit-eating bats. J. Mammal. suta and Vampyressa pusilla. Occas. 54: 266±267. Pap. Mus. Texas Tech Univ. 17: 1±10. 174 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Baker, R. J., R. A. Bass, and M. A. Johnson somal homologies. M.S. thesis, Texas 1979. Evolutionary implications of chromo- Tech Univ., Lubbock. somal homology in four genera of sten- Belwood, J. J. odermatine bats (Phyllostomatidae: 1988. Foraging behavior, prey selection, and Chiroptera). Evolution 33: 220±226. echolocation in phyllostomine bats Baker, R. J., R. L. Honeycutt, M. L. Arnold, V. (Phyllostomidae). In P. E. Nachtigall M. Sarich, and H. H. Genoways and P.W.B. Moore (eds.), Animal sonar: 1981a. Electrophoretic and immunological 601±605. New York: Plenum. studies on the relationship of the Bra- Benedict, F. A. chyphyllinae and the Glossophaginae. 1957. Hair structure as a generic character in J. Mammal. 62: 665±672. bats. Univ. California Publ. Zool. 59: Baker, R. J., H. H. Genoways, and P. A. Seyfarth 285±548. 1981b. Results of the Alcoa Foundation±Suri- Bhatnager, K. P. name Expeditions. VI. Additional chro- 1980. The chiropteran vomeronasal organ: its mosomal data for bats (Mammalia: relevance to the phylogeny of bats. In Chiroptera) from Suriname. Ann. Car- D. E. Wilson and A. L. Gardner, (eds.), negie Mus. 50: 333±344. Proceedings of the ®fth international Baker, R. J., M. W. Haiduk, L. W. Robbins, A. bat research conference: 289±315. Lub- Cadena, and B. F. Koop bock: Texas Tech Univ. 1982. Chromosomal studies of South Ameri- 1985. Comparative morphology of the vom- can bats and their systematic implica- eronasal receptor-free epithelium of tions. In M. A. Mares and H. H. Gen- some phyllostomid bats: light micor- oways (eds.), Mammalian biology in scopic observations. Aust. J. Mammal. South America. Spec. Publ. Ser. Py- 8: 265±270. matuning Lab. Ecol. 6: 303±327. 1988. Anatomy. In A. M. Greenhall and U. Baker, R. J., M. B. Qumsiyeh, and C. S. Hood Scmidt (eds.), Natural history of vam- 1987. Role of chromosomal banding patterns in understanding mammalian evolution. pire bats: 41±70. Boca Raton, LA: In H. H. Genoways (ed.), Current CRC Press. mammalogy: 67±96. New York: Ple- Bhatnager, K. P., and F. C. Kallen num. 1974. Morphology of the nasal cavities and Baker, R. J., C. G. Dunn, and K. Nelson associated structures in Artibeus jamai- 1988a. Allozymic study of the relationships of censis and Myotis lucifugus. Am. J. Phylloderma and four species of Phyl- Anat. 139: 167±190. lostomus. Occas. Pap. Mus. Texas Tech Bhatnager, K. P., D. H. Matulionis, and W. Brei- Univ. 125: 1±14. pohl Baker, R. J., R. L. Honeycutt, and R. A. Bass 1982. Fine structure of the vomeronasal neu- 1988b. Genetics. In A. M. Greenhall and U. roepithelium of bats: a comparative Schmidt(eds.), Natural history of vam- study. Acta Anat. 112: 158±177. pire bats, pp. 31±40. Boca Raton, LA: Bogdanowicz, W., R. D. Csada, M. B. Fenton CRC Press. 1997. Structure of the noseleaf, echolocation, Baker, R. J., C. S. Hood, and R. L. Honeycutt and foraging behavior in the Phyllo- 1989. Phylogenetic relationships and classi®- stomidae (Chiroptera). J. Mammal. 78: cation of the higher categories of the 942±953. New World bat family Phyllostomidae. Bonaparte, C. L. Syst. Zool. 38: 228±238. 1831. Saggio di una distribuzione metodica Baron, G., H. Stephan, and H. D. Frahm degli animali vertebrati. Rome: Anto- 1996a. Comparative neurobiology in Chirop- nio Boulzaler. tera, vol. 2. Boston: Birkhauser. 1838. Synopsis vertebratorum systematis. 1996b. Comparative neurobiology in Chirop- Nuovi Ann. Sci. Nat., Anno I Tomo II: tera, vol. 3. Boston: Birkhauser. 105±133. Barrett, M., M. J. Donoghue, and E. Sober Brown, J. C. 1991. Against consensus. Syst. Zool. 40: 1971. The description of mammals. 1. The 486±493. external characters of the head. Mam- Bass, R. A. mal Rev. 1: 151±167. 1978. Systematics of the Desmodontinae and Bull, J. J, J. P. Huelsenbeck, C. W. Cunningham, Phyllonycterinae (Chiroptera: Phyllos- D. L. Swofford, and P. J. Waddell tomatidae) based on G±band chromo- 1993. Partitioning and combining data in phy- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 175

logenetic analysis. Syst. Biol. 42: 384± de Quieroz, K., and J. Gauthier 397. 1990. Phylogeny as a central principle in tax- Cabrera, A. onomy: Phylogenetic de®nitions of tax- 1958. CataÂlogo de los mamiferos de America on names. Syst. Zool. 39: 307±322. del Sur. Rev. Mus. Argent. Cienc. 1992. Phylogenetic taxonomy. Annu. Rev. Nat.``Bernardino Rivadavia'' 4: iv± Ecol. Syst. 23: 449±480. 308. 1994. Toward a phylogenetic system of bio- Cadena, A., and R. J. Baker logical nomenclature. Trends Ecol. 1976. Cariotipos de los murcielagos vampiros Evol. 9: 27±31. (Chiroptera: Desmodinae). Caldasia 11: de Queiroz, A., M. J. Donoghue, and J. Kim 159±163. 1995. Separate versus combined analysis of Chippendale, P. T., and J. J. Wiens phylogenetic evidence. Annu. Rev. 1994. Weighting, partitioning, and combining Ecol. Syst. 26: 657±681. characters in phylogenetic analysis. Desmarest, A. G. Syst. Biol. 43: 278±287. 1820. Mammalogie, ou description des espeÁ- Cloutier, D., and D. W. Thomas ces de mammifeÁres, partie 1. Paris: 1992. Carollia perspicillata. Mamm. Species Mme. Veuve Agasse. 417: 1±9. Dobson, G. E. Cooper, J. G., and K. P. Bhatnager 1875. Conspectus of the suborders, families, 1976. Comparative anatomy of the vomero- and genera of Chiroptera, arranged ac- nasal organ complex in bats. J. Anat. cording to their natural af®nities. Ann. 122: 571±601. Mag. Nat. Hist. ser. 4, 16: 345±357. 1878. Catalogue of the Chiroptera in the col- Corbet, G. B., and J. E. Hill lection of the British Museum. London: 1980. A world list of mammalian species, 1st British Museum. ed. London: British Museum. Dumeril, A. M. C. 1986. A world list of mammalian species, 2nd 1806. Zoologie analytique, ou meÂthode natu- ed. London: British Museum. relle de classi®cation des animaux, ren- Cracraft, J., and D. P. Mindell du plus facile aÁ l'aide de tableaux syn- 1989. The early history of modern birds: a optiques. Paris: Allais. comparison of molecular and morpho- Eernisse, D. J., and A. G. Kluge logical evidence. In B. Fernholm, K. 1993. Taxonomic congruence versus total ev- Bremer, and H. JoÈrnvall (eds.), The hi- idence, and amniote phylogeny inferred erarchy of life: molecules and mor- from fossils, molecules, and morphol- phology in phylogenetic analysis: 389± ogy. Mol. Biol. Evol. 10: 1170±1195. 403. Amsterdam: Elsevier. Emmons, L. H. Cuvier, G. 1997. Neotropical rainforest mammals: a ®eld 1817. Le reÁgne animal distribue d'apreÁs son guide, 2nd ed. Chicago: Univ. Chicago organisation, pour servir de base aÁ Press. l'histoire naturelle des animaux et Farris, J. S. d'introduction aÁ l'anatomie compareÂe, 1983. The logical basis of phylogenetic anal- tomus 1. Paris: Deterville. ysis. In N. I. Platnick and V. A. Funk Dalquest, W. W., and H. J. Werner (eds.), Advances in cladistics, vol. 2. 1954. Histological aspects of the faces of Proceedings of the second meeting of North American bats. J. Mammal. 35: the Willi Hennig Society: 1±36. New 147±160. York: Columbia Univ. Davis, D. E., D. C. Carter, and R. H. Pine Farris, J. S., M. KaÈllersjoÈ, A. G. Kluge, and C. 1964. Noteworthy records of Mexican and Bult Central American bats. J. Mammal. 45: 1995. Testing signi®cance of incongruence. 375±387. Cladistics 10: 315±319. de la Torre, L. Felsenstein, J. 1961. The evolution, variation, and system- 1978. Cases in which parsimony and compat- atics of the neotropical bats of the ge- ibility methods will be positively mis- nus Sturnira. Ph.D. diss., Univ. Illinois, leading. Syst. Zool. 27: 401±410. Urbana. Fenton, M. B. de Queiroz, A. 1984. Echolocation: implications for ecology 1993. For consensus (sometimes). Syst. Biol. and evolution of bats. Q. Rev. Biol. 59: 42: 368±372. 33±53. 176 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

1985. Communication in the Chiroptera. Frahm, H. D., and K. P. Bhatnager Bloomington: Indiana Univ. Press. 1980. Comparative morphology of the acces- 1992. Wounds and the origin of blood±feed- sory olfactory bulb in bats. J. Anat. ing in bats. Biol. J. Linn. Soc. 47: 161± 130: 349±365. 171. Gardner, A. L. Ferrarezzi, H., and E. d. A. Gimenez 1977a. Chromosomal variation in Vampyressa 1996. Systematic patterns and the evolution and a review of chromosomal evolution of feeding habits in Chiroptera (Ar- in the Phyllostomidae (Chiroptera). chonta: Mammalia). J. Comp. Biol. 1: Syst. Zool. 26: 300±318. 75±94. 1977b. Feeding habits. In R. J. Baker, J. K. Fischer von Waldheim, G. Jones, Jr., and D. C. Carter (eds.), Bi- 1813. Zoognosia tabulis synopticus illustrata, ology of the bats of the New World in usum praelectionum Academiae im- family Phyllostomatidae, p. II. Spec. perialis medico±chirugicae mosquensis Publ. Mus. Texas Tech Univ. 16: 293± edita. Moscow: Nicolai S. Vsevoloz- 350. sky. Gardner, A. L. and C. S. Ferrell Fleming, T. H. 1990. Comments on the nomenclature of 1988. The short±tailed fruit bat: a study in some Neotropical bats (Mammalia: plant±animal interactions. Chicago: Chiroptera). Proc. Biol. Soc. Washing- Univ. Chicago Press. ton 103: 501±508. Fleming, T. H., and E. R. Heithaus Gerber, J. D. 1968. Electrophoretic and immunologic com- 1986. Seasonal foraging behavior of Carollia parisons of the serum proteins of bats. perspicillata (Chiroptera: Phyllostomi- Ph.D. diss., Univ. Kansas, Lawrence. dae). J. Mammal. 67: 660±671. Gerber, J. D., and C. A. Leone Fleming, T. H., E. T. Hooper, and D. E. Wilson 1971. Immunologic comparisons of the sera 1972. Three Central American bat communi- of certain phyllostomatid bats. Syst. ties: structure, reproductive cycles, and Zool. 20: 160±166. movement patterns. Ecology 53: 555± Gervais, P. 569. 1854. Histoire naturelle des mammifeÁres, par- Flower, W. H., and R. Lydekker tie I. Paris: L. Curmer. 1891. An introduction to the study of mam- 1856. DeuxieÁme meÂmoire. Documents zool- mals living and extinct. London: Adam ogiques pour servir aÁ la monographie and Charles Black. des cheÂiropteÁres Sud±AmeÂricains. In P. Forman, G. L. Gervais (ed.), MammifeÁres: 25±88. In 1971. Gastric morphology in selected mor- F. Castelnau (ed.) Animaux nouveaux moopid and glossophagine bats as re- ou rares recueillis pendant l'expeÂdition lated to systematic problems. Trans. Il- dans les parties centrales de l'AmeÂrique linois Acad. Sci. 64: 273±282. du Sud, de Rio de Janeiro aÁ Lima, et 1973. Studies of gastric morphology in North de Lima au Para; executeÂe par ordre du American Chiroptera (Emballonuridae, gouvernement francËais pendant les an- Noctilionidae, and Phyllostomatidae). neÂes 1843 aÁ 1847, sous la direction du J. Mammal. 54: 909±923. Comte Francis Castelnau. 1(2): 1±116. Forman, G. L., R. J. Baker, and J. D. Gerber Paris: P. Bertran. [dated 1855, but pub- 1968. Comments on the systematic status of lished 1856; see Ann. Mag. Nat. Hist. vampire bats (family Desmodontidae). ser 7, 8: 164.]. Syst. Zool. 17: 417±425. Gill, T. Forman, G. L., C. J. Phillips, and C. S. Rouk 1872. Arrangement of the families of mam- 1979. Alimentary tract. In R. J. Baker, J. K. mals. Washington, D.C.: Smithsonian Jones, Jr., and D. C. Carter (eds.), Bi- Inst.. ology of the bats of the New World 1884. Animalivora. In J. S. Kingsley (ed.), family Phyllostomatidae, p. III. Spec. The standard natural history 5: 163± Publ. Mus. Texas Tech Univ. 16: 205± 177. Boston: S. E. Cassino. 227. Gillette, D. D. Frahm, H. D. 1975. Evolution of feeding strategies in bats. 1981. Volumetric comparison of the accesso- Tebiwa 18: 39±48. ry olfactory bulb in bats. Acta Anat. Gimenez, E. d. A. 109: 173±183. 1993. Morfologia lingual comparada, ®li- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 177

ogenia e evolucËaÄo dos haÂbitos alimen- Grif®n, D. R., and A. Novick tares na superfamilia Phyllostomoidea 1955. Acoustic orientation of neotropical (Mammalia: Chiroptera). M.S. thesis, bats. J. Exp. Zool. 130: 251±300. Univ. Estadual Paulista, SaÄo Paulo. Grif®ths, T. A. Gimenez, E. d. A., H. Ferrarezzi, and V. A. Taddei 1978. Muscular and vascular adaptations for 1996. Lingual morphology and cladistic anal- nectar-feeding in the glossophagine ysis of the New World nectar-feeding bats Monophyllus and Glossophaga.J. bats (Chiroptera: Phyllostomidae). J. Mammal. 59: 414±418. Comp. Biol. 1: 41±64. 1982. Systematics of the New World nectar- Goldfuss, G. A. feeding bats (Mammalia, Phyllostomi- 1820. Handbuch der zoologie, theil. 2. Nurn- dae), based on the morphology of the berg: Johann Leonhard Schrag. hyoid and lingual regions. Am. Mus. Goodwin, G. G., and A. M. Greenhall Novitates 2742: 45 pp. 1961. A review of the bats of Trinidad and 1983a. On the phylogeny of the Glossophagi- Tobago: descriptions, rabies infection, nae and the proper use of outgroup and ecology. Bull. Am. Mus. Nat. Hist. analysis. Syst. Zool. 32: 283±285. 122: 187±302. 1983b. Comparative laryngeal anatomy of the Gray, J. E. Big Brown bat, Eptesicus fuscus and 1821. On the natural arrangement of verte- the Mustached bat Pteronotus parnellii. brose animals. London Med. Repos. Mammalia 47: 377±394. 15: 296±310. 1985. Molar cusp patterns in the bat genus 1826. An attempt at a division of the family Brachyphylla: some functional and sys- Vespertilionidae into groups. Zool. J. 2: tematic observations. J. Mammal. 66: 242±243. 544±549. 1838. A revision of the genera of bats (Ves- Grif®ths, T. A., and B. B. Criley pertilionidae), and the description of 1989. Comparative lingual anatomy of the some new genera and species. Mag. bats Desmodus rotundus and Loncho- Zool. Bot. 2: 483±505. phylla robusta (Chiroptera: Phyllo- 1866a. Synopsis of the genera of Vespertilion- stomidae). J. Mammal. 70: 606±613. idae and Noctilionidae. Ann. Mag. Nat. Haiduk, M. W., and R. J. Baker Hist. ser. 3, 17: 89±93. 1982. Cladistical analysis of G±banded chro- 1866b. A revision of the genera of pteropine mosomes of nectar-feeding bats (Glos- bats (Pteropidae) and descriptions of sophaginae: Phyllostomidae). Syst. some apparently undescribed species. Zool. 31: 252±265. Proc. Zool. Soc. London 1866: 62±67 Hall, E. R. 1866c. A revision of the genera of Rhinolo- 1981. The mammals of North America, 2nd phidae, or horseshoe bats. Proc. Zool. ed. New York: Wiley. Soc. London 1866: 81±83. Hall, E. R., and K. R. Kelson 1866d. Revision of the genera of Phyllosto- 1959. The mammals of North America. New midae, or leaf-nosed bats. Proc. Zool. York: Ronald Press. Soc. London 1866: 111±118. Handley, C. O., Jr. Greenbaum, I. F., and C. J. Phillips 1959. A revision of American bats of the gen- 1974. Comparative anatomy and general his- era Euderma and Plecotus. Proc. U.S. tology of tongues of long-nosed bats Natl. Mus. 110: 95±246. (Leptonycteris sanborni and L. nivalis) 1960. Descriptions of new bats from Panama. with reference to infestation of oral Proc. U.S. Natl. Mus. 112: 459±479. mites. J. Mammal. 55: 489±504. 1980. Inconsistencies in formation of family- Greenbaum, I. F., R. J. Baker, and D. E. Wilson group and subfamily-group names in 1975. Evolutionary implications of the kar- Chiroptera. In D. E. Wilson and A. L. yotypes of the stenodermine genera Ar- Gardner (eds.), Proceedings of the ®fth dops, Ariteus, Phyllops, and Ectophyl- international bat research conference: la. Bull. S. California. Acad. Sci. 74: 9±13. Lubbock: Texas Tech Univ. 156±159. Hartley, D. J., and R. A. Suthers Greenhall, A. M. 1987. The sound emission pattern and the 1972. The biting and feeding habits of the acoustical role of the noseleaf in the vampire bat, Desmodus rotundus.J. echolocating bat Carollia perspicillata. Zool., (London) 168: 451±461. J. Acoust. Soc. Am. 82: 1892±1900. 178 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Heithaus, E. R., T. H. Flemming, and P. A. Opler International Commission on Zooloogical No- 1975. Foraging patterns and resource utiliza- menclature tion in seven species of bats in a sea- 1985. International code of zoological no- sonal tropical forest. Ecology 56: 841± menclature, 3rd ed. London: Int. Trust 854. Zool. Nomencl. Hernandez, J., and A. Cadena Jassal, M., and N. B. Simmons 1978. Notas para la revision del genero Lon- 1996. Phylogenetic relationships of ``phyllos- chorhina (Chiroptera: Phyllostomidae). tomine'' bats: results of an analysis us- Caldasia 12: 199±251. ing a dense-sampling approach. Bat Hill, J. E., and J. D. Smith Res. News 37: 136. 1984. Bats: a natural history. Austin: Univ. Jepsen, G. L. Texas Press. 1970. Bat origins and evolution. In W.A. Hillis, D. M. Wimsatt (ed.), Biology of bats, vol. 1: 1987. Molecular versus morphological ap- 1±64. New York: Academic. proaches to systematics. Ann. Rev. Johnson, M. A. Ecol. Syst. 18: 23±42. 1979. Evolutionary implications of G- and C- Honeycutt, R. L. banded chromosomes of 13 species of 1981. Molecular evolution in New World stenodermine bats. M.S. thesis, Texas leaf-nosed bats of the family Phyllo- Tech Univ., Lubbock. stomidae with comments on the super- Jones, J. K., Jr., and R. J. Baker family Noctilionoidea. Ph.D. diss., Tex- 1980. Chiroderma improvisum. Mamm. Spe- as Tech Univ., Lubbock. cies 134: 1±2. Honeycutt, R. L., and V. M. Sarich Jones, J. K., Jr., and D. C. Carter 1987a. Albumin evolution and subfamilial re- 1976. Annotated checklist, with keys to sub- lationships among New World leaf- families and genera. In R. J. Baker, J. nosed bats (family Phyllostomidae). J. K. Jones, Jr., and D. C. Carter (eds.), Mammal. 68: 508±517. Biology of the bats of the New World 1987b. Monophyly and molecular evolution family Phyllostomatidae, p. I. Spec. within three phyllostomid bat genera. J. Publ. Mus. Texas Tech Univ. 10: 7±38. Mammal. 68: 518±525. Kalko, E.K.V., and M. Condon Honeycutt, R. L., R. J. Baker, and H. H. Geno- 1993. Bat-plant interactions: how frugivorous ways leaf-nosed bats ®nd their food. Bat. 1980. Results of the Alcoa Foundation-Suri- Res. News 34: 115. name expeditions. III. Chromosomal Kirsch, J.A.W., J. M. Hutcheon, D.G.P. Byrnes, data for bats (Mammalia: Chiroptera) and B. D. Lloyd from Suriname. Ann. Carnegie Mus. 1998. Af®nities and historical zoogeography 49: 237±250. of the New Zealand short-tailed bat, Honeycutt, R. L., I. F. Greenbaum, R. J. Baker, Mystacina tuberculata Gray 1843, in- and V. M. Sarich ferred from DNA hybridization com- 1981. Molecular evolution of vampire bats. J. parisons. J. Mamm. Evol. 5: 33±64. Mammal. 62: 805±811. Kluge, A. G. Hood, C. S., and J. D. Smith 1989. A concern for evidence and a phylo- 1982. Cladistical analysis of female reproduc- genetic hypothesis of relationships tive histomorphology in phyllostoma- among Epicrates (Boidae, Serpentes). toid bats. Syst. Zool. 31: 241±251. Syst. Zool. 38: 7±25. 1983. Histomorphology of the female repro- Kluge, A. G., and A. J. Wolf ductive tract in phyllostomoid bats. Oc- 1993. Cladistics: what's in a word? Cladistics cas. Pap. Mus. Texas Tech Univ. 86: 1± 9: 183±199. 38. Koch, C. Hsu, T. C., R. J. Baker, and T. Utakoji 1862±63. Das Wesentliche der Chiropteren mit 1968. The multiple sex chromosome system besonder Beschreibung der in dem Her- of American leaf-nosed bats (Chirop- zogthum Nassau und den angranzenden tera, Phyllostomidae). Cytogenetics 7: Landestheilen vorkomenden Fleder- 27±38. maÈuse. Jahrg. Ver. Naturk. Herzogthum Huelsenbeck, J. P., J. J. Bull, and C. W. Cunning- Nassau 17±18: 261±593. ham Koop, B. F., and R. J. Baker. 1996. Combining data in phylogenetic anal- 1983. Electrophoretic studies of relationships ysis. Trends Ecol. Evol. 11: 152±157. of six species of Artibeus (Chiroptera: 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 179

Phyllostomidae). Occas. Pap. Mus. 1842. Nouveau table du reÁgne animal. Paris: Texas Tech Univ. 83: 1±12. Arthus Bertrand. Koopman, K. F. Lillegraven, J. A. 1978. Zoogeography of Peruvian bats with 1969. Latest Cretaceous mammals of upper special emphasis on the role of the An- part of Edmonton formation of Alberta, des. Am. Mus. Novitates 2651: 33 pp. Canada, and a review of the marsupial- 1984. Bats. In S. Anderson and J. K. Jones Jr. placental dichotomy in mammalian (eds.), Orders and families of Recent evolution. Univ. Kans. Paleontol. Con- mammals of the world: 145±186. New trib. Vertebr. 50(12): 1±222. York: Wiley. 1976. Biological considerations of the mar- 1993. Order Chiroptera. In D. E. Wilson and supial-placental dichotomy. Evolution D. M. Reeder (eds.), Mammal species 29: 707±722. of the world, a taxonomic and geo- Lim, B. K. graphic reference, 2nd ed.: 137±241. 1993. Cladistic reapprisal of Neotropical Washington, D.C.: Smithsonian Inst. stenodermatine bat phylogeny. Cladis- 1994. Chiroptera: systematics. Handb. der tics 9: 147±165. Zool., vol. 7, p. 60, Mammalia: 1±217. Linnaeus, C. Koopman, K. F., and E. L. Cockrum 1758. Systema naturae per regna tria naturae, 1967. Bats. In S. Anderson and J. K. Jones Jr. secundum classis, ordines, genera, spe- (eds.), Recent mammals of the world: cies cum characteribus, differentiis, 109±150. New York: Ronald Press. synonymis, locis, vol. 1, 10th ed. Koopman, K. F., and J. K. Jones, Jr. Stockholm: Laurentii Salvii. 1970. Classi®cation of bats. In B. H. Slaugh- Luckett, W. P. ter and W. D. Walton (eds.), About 1980. The use of fetal membrane data in as- bats: 22±28. Dallas: Southern Method- sessing chiropteran phylogeny. In D. E. ist Univ. Press. Wilson and A. L. Gardner (eds.), Pro- Koopman, K. F., and G. T. MacIntyre ceedings of the ®fth international bat 1980. Phylogenetic analysis of chiropteran research conference: 245±264. Lub- dentition. In D. E. Wilson and A. L. bock: Texas Tech Univ. Gardner (eds.), Proceedings of the ®fth Machado-Allison, C. E. international bat research conference: 1967. The systematic position of the bats 279±288. Lubbock: Texas Tech Univ. Desmodus and Chilonycteris, based on LaceÂpeÁde, B. G. host-parasite relationships (Mammalia; 1799. Tableau des divisions, sous-divisions, Chiroptera). Proc. Biol. Soc. Washing- ordres, et genres des mammifeÁres. Tabl. ton. 80: 223±226. Div. Subd. Order Genres MammifeÁres Maddison, W. P. Paris: Chez Plasson [1797]. 1993. Missing data versus missing characters Lanyon, S. M. in phylogenetic analysis. Syst. Biol. 42: 1993. Phylogenetic frameworks: towards a 576±581. ®rmer foundation for the comparative Maddison, W. P., and D. R. Maddison approach. Biol. J. Linn. Soc. 49: 45± 1992. MacClade: analysis of phylogeny and 61. character evolution, version 3.0. Com- Laska, M. puter program distributed by Sinauer, 1990. Olfactory sensitivity to food odor com- Sunderland, MA. ponents in the short-tailed fruit bat, Marques-Aguiar, S. A. Carollia perspicillata (Phyllostomati- 1994. A systematic review of the large spe- dae: Chiroptera). J. Comp. Physiol. cies of Artibeus Leach, 1821 (Mam- (166A): 395±399. malia: Chiroptera), with some phylo- Laurie, E.M.O. genetic inferences. Bol. Mus. Para. Em- 1955. Notes on some mammals from Ecua- ilio Goeldi ser. Zool. 10: 3±83. dor. Ann. Mag. Nat. Hist. ser. 12, 8: Maxson, L. R., and R. D. Maxson 268±276. 1990. Proteins II: immunological techniques. Le Gros Clark, W. E. In D. M. Hillis and C. Moritz (eds.), 1959. The antecedents of man. Edinburgh: Molecular systematics, 1st ed.: 127± Edinburgh Univ. Press. 155. Sunderland, MA: Sinaeuer. Lesson, R. P. McDaniel, V. R. 1827. Manuel de mammalogie. Paris: Librair- 1976. Brain Anatomy. In R. J. Baker, J. K. ie de Roret. Jones, Jr., and D. C. Carter (eds.), Bi- 180 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

ology of the bats of the New World Phyllostomatidae and Desmodontidae. family Phyllostomatidae, p. I. Spec. J. Mammal. 44: 44±56. Publ. Mus. Texas Tech Univ. 10: 147± 1977. Acoustic orientation. In W. A. Wimsatt 200. (ed.), Biology of bats, vol. 3: 73±289. McKenna, M. C., and S. K. Bell New York: Academic Press. 1997. Classi®cation of mammals above the Nowak, R. M. species level. New York: Columbia 1991. Walker's mammals of the world. Bal- Univ. timore: Johns Hopkins Univ. MedellõÂn,R.A. Oken, L. 1988. Prey of Chrotopterus auritus, with 1816. Lehrbuch der Naturgeschichte, theil 3, notes on feeding behavior. J. Mammal. abt., 2. Jena: August Schmidt. 69: 841±844. Owen, R. D. Miller, G. S. 1987. Phylogenetic analyses of the bat sub- 1907. The families and genera of bats. Bull. family Stenodermatinae (Mammalia: U.S. Natl. Mus. 57: 1±282. Chiroptera). Spec. Publ. Mus. Texas 1924. List of North American recent mam- Tech Univ. 26: 1±65. mals. Bull. U.S. Natl. Mus. 128: 1±673. 1991. The systematic status of Dermanura Miyamoto, M. M. concolor (Peters, 1865) (Chiroptera: 1985. Consensus cladograms and general Phyllostomidae), with description of a classi®cations. Cladistics 1: 186±189. new genus. Bull. Am. Mus. Nat. Hist. Miyamoto, M. M., and W. M. Fitch 206: 18±25. 1995. Testing species phylogenies and phy- Paradiso, J. L. logenetic methods with congruence. 1967. A review of the wrinkled-faced bats Syst. Biol. 44: 64±76. (Centurio senex Gray), with a descrip- MoÈhres, F. P., and G. Neuweiler tion of a new subspecies. Mammalia 1966. Die Ultraschallorientung der Gloss- 31: 595±604. blatt-FledermaÈuse (Chiroptera-Mega- Park, H., and E. R. Hall dermatidae). Z. Vergl. Physiol. 53: 1951. The gross anatomy of the tongues and 195±227. stomachs of eight New World bats. Mossman, H. W. Trans. Kansas Acad. Sci. 54: 64±72. 1953. The genital system and the fetal mem- Patton, J. C., and R. J. Baker branes as criteria for mammalian phy- 1978. Chromosomal homology and evolution logeny and taxonomy. J. Mammal. 34: of phyllostomatoid bats. Syst. Zool. 27: 289±298. 449±462. 1977. Comparative anatomy. In R. M. Wynn Patton, J. L, and A. L. Gardner (ed.), The biology of the uterus, 2nd 1971. Parallel evolution of multiple sex-chro- ed.: 19±34. New York: Plenum. mosome systems in the phyllostomatid 1987. Vertebrate fetal membranes. New bats, Carollia and Choeroniscus. Ex- Brunswick: Rutgers Univ. perientia 27: 105±106. Myers, P. Peters, W.C.H. 1981. Observations on Pygoderma bilabia- 1865. Vorlage von Abbildungen zu einer tum (Wagner). Z. SaÈugetierkd. 46: 146± Monographie der Chiropteren, und 151. UÈ bersicht der von ihm befolgten sys- Nagorsen, D.W., and R.L. Peterson. tematischen Ordnung der hieher gehoÈ- 1975. Karyotypes of six species of bats (Chi- rigen Gattungen. Monatsber. K. Preuss. roptera) from the Dominican Republic. Akad. Wiss. Berlin 1865: 265±258. R. Ont. Mus. Life Sci. Occas. Pap. 28: Peterson, R. L. 1±28. 1965. A review of the bats of the genus Ame- Neuweiler, G. trida, family Phyllostomidae. R. Ont. 1984. Foraging ecology and audition in bats. Mus. Life Sci. Contrib. 65: 1±13. Naturwissenschaften 71: 446±455. Phillips, C. J. Novacek, M. J. 1971. The dentition of glossophagine bats: 1991. Aspects of morphology of the cochlea development, morphological character- in microchiropteran bats: an investiga- istics, variation, pathology, and evolution of character transformation. Bull. tion. Univ. Kansas Mus. Nat. Hist. Am. Mus. Nat. Hist. 206: 84±100. Misc. Publ. 54: 1±138. Novick, A. Pierson, E. D. 1963. Orientation in Neotropical bats. II. 1986. Molecular systematics of the Microchi- 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 181

roptera: higher taxon relationships and tundus. Bol. Zool. Univ. SaÄo Paulo 3: biogeography. Ph.D. diss., Univ. of 97±119. California, Berkeley. Schmidt, U. Pine, R. H. 1978. Vampir¯edermaÈuse; familie Desmo- 1972. Bats of the genus Carollia. Texas dontidae (Chiroptera). Wittenberg Lu- Agric. Exp. Tech. Monogr. 8: 1±125. terhstadt: A. Zimsen. Platnick, N. I., C. E. Griswold, and J. A. Cod- Schnitzler, H. -U., and A. D. Grinnell dington 1977. Directional sensitivity of echolocation 1991. On missing entries in cladistic analysis. in the horseshoe bat Rhinolophus fer- Cladistics 7: 337±343. rumequinum. I. Directionality of sound Pleijel, F. emission. J. Comp. Physiol. 116: 51± 1995. On character coding for phylogeny re- 61. construction. Cladistics 11: 309±315. Schober, W., and K. Brauer Pocock, R. I. 1974. Makromorphologie des Zentralnerven- 1914. On the facial vibrissae of Mammalia. systems. In J. G. Helmcke, D. Starck, Proc. Zool. Soc. London 1914: 889± H. Wermuth (eds.), Handbuch der 912. Zoologie. Eine Naturgeschichte der Rehn, J. A. G. Stamme des Tierreiches. 8. Band/52. 1901. A study of the genus Centurio. Proc. Liefrung: 1±296. New York: Walter de Acad. Nat. Sci. Philadelphia. 53: 295± Gruyter. 302. Schutt, W. A., Jr. Robin, H. A. 1998. Chiropteran hindlimb morphology and the origin of blood-feeding in bats. In 1881. Recherches anatomiques sur les mam- T. H. Kunz, and P. A. Racey (eds.), Bat mifeÁres de l'ordre des cheÂiropteÁres. biology and conservation: 157±168. Ann. Sci. Nat., ser. 6 Zool., 12: 1±180. Washington D.C.: Smithsonian Inst. Rodrigo, A. G., M. Kelly-Borges, P. R. Bergquist, Silva-Taboada, G. S., and R. H. Pine and P. L. Bergquist. 1969. Morphological and behavioral evidence 1993. A randomisation test of the null hy- for the relationship between the bat ge- pothesis that two cladograms are sam- nus Brachyphylla and the Phyllonycter- ple estimates of a parametric phyloge- inae. Biotropica 1: 10±19. netic tree. New Zealand J. Bot. 31: Simmons, N. B. 257±268. 1993. The importance of methods: archontan Rodriguez-Duran, A., and T. H. Kunz phylogeny and cladistic analysis of 1992. Pteronotus quadridens. Mamm. Spe- morphological data. In R.D.E. Mac- cies 395: 1±4. Phee (ed.), Primates and their relatives Rodriguez-Duran, A., and A. R. Lewis in phylogenetic perspective: 1±61. New 1987. Patterns of population size, diet, and York: Plenum. activity time for a multispecies assem- 1996. A new species of Micronycteris (Chi- blage of bats at a cave in Puerto Rico. roptera: Phyllostomidae) from North- Caribb. J. Sci. 23: 352±360. eastern Brazil, with comments on phy- Santos, M., A. Ruiz, J. Cavalier, and P. Soriano logenetic relationships. Am. Mus. Nov- 1996. The phenology of cacti and its relation- itates 3158: 34 pp. ships with the bat community in a trop- 1998. A reappraisal of interfamilial relation- ical dry forest±thorn shrubland of Co- ships of bats. In T. H. Kunz and P. A. lombia. Bat Res. News 36: 105±106. Racey (eds.), Bats: biology and conser- [Dated 1995 but published in 1996]. vation: 3±26. Washington D.C.: Smith- Savage, D. E. sonian. 1951. A Miocene phyllostomatid bat from Simmons, N. B., and T. M. Conway Colombia, South America. Univ. Cali- MS. Phylogenetic relationships of mor- fornia Publ. Geol. Sci. 28: 357±366. moopid bats (Chiroptera: Mormoopi- Sazima, I. dae) based on morphological data. 1976. Observations on the feeding behavior Simmons, N. B., and J. H. Geisler of some phyllostomatid bats (Carollia, 1998. Phylogenetic relationships of Icaron- Anoura, and Vampyrops) in Southeast- ycteris, Archaeonycteris, Hassianycter- ern Brazil. J. Mammal. 57: 381±382. is, and Palaeochiropteryx to extant bat 1978. Aspectos do comportamento alimentar lineages, with comments on the evolu- do morceÃgo hematoÂfago, Desmodus ro- tion of echolocation and foraging strat- 182 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

egies in Microchiroptera. Bull. Am. mary, classi®cation, and phylogeny. Mus. Nat. Hist. 235: 1±182. Proc. Zool. Soc. London 1925: 701± Simmons, N. B., and C. O. Handley, Jr. 762. 1998. A revision of Centronycteris Gray Spix, J. B. (Chiroptera: Emballonuridae) with 1823. Simiarum et vespertilionum brasilien- notes on natural history. Am. Mus. sum: species novae ou histoire naturelle Novitates 3239: 28 pp. des espeÁces nouvelles des singes et de Simpson, G. G. chauve-souris, observeÂes et recueilles 1945. The principles of classi®cation and a pendant le voyage dans l'inteÂrieur du classi®cation of the mammals. Bull. BreÂsil. Monachii: F.S. Hubschmann. Am. Mus. Nat. Hist. 85: 1±350. Sprague, J. M. Sites, J. W., Jr., J. W. Bickham, and M. W. Haiduk 1943. The hyoid region of placental mammals 1981. Conservative chromosomal change in with especial reference to the bats. Am. the bat family Mormoopidae. Can. J. J. Anat. 72: 385±472. Genet. Cytol. 23: 459±467. Starett, A., and R. S. Casebeer Slaughter, B. H. 1968. Records of bats from Costa Rica. Los 1970. Evolutionary trends of chiropteran den- Angeles Cty. Mus. Contrib. Sci. 148: titions. In B. H. Slaughter and W. D. 1±21. Walton (eds.), About bats: 51±83. Dal- Stock, A. D. las: Southern Methodist Univ. Press. 1975. Chromosome banding pattern homolo- Smith J. D. gy and its phylogenetic implications in 1972. Systematics of the chiropteran family the bat genera Carollia and Choero- Mormoopidae. Univ. Kansas Mus. Nat. niscus. Cytogenet. Cell Genet. 14: 34± Hist. Misc. Publ. 56: 1±132. 41. 1976. Chiropteran evolution. In R. J. Baker, Straney, D. O. J. K. Jones, Jr., and D. C. Carter (eds.), 1980. Relationships of phyllostomatine bats: Biology of the bats of the New World evaluation of phylogenetic hypotheses. family Phyllostomatidae, p. I. Spec. Ph. D. diss., Univ. of California, Berke- Publ. Mus. Texas Tech Univ. 10: 49± ley. 69. Straney, D. O., M. H. Smith, I. F. Greenbaum, and Smith, J. D., and C. S. Hood R. J. Baker 1984. Geneology of New-World nectar-feed- 1979. Biochemical genetics. In R. J. Baker, J. ing bats reexamined: a reply to Grif- K. Jones, Jr., and D. C. Carter (eds.), ®ths. Syst. Zool. 33: 435±460. Biology of the bats of the New World Snow, J. L., J. K. Jones, and W. D. Webster family Phyllostomatidae, p. III. Spec. 1980. Centurio senex. Mamm. Species 138: Publ. Mus. Texas Tech Univ. 16: 157± 1±3. 176. Sokolov, B. V., and A. K. Makarov Strickler, T. L. 1971. Direction of the ultrasonic radiation 1978. Functional osteology and the myology and the role of the nasal leaf in Rhin- of the shoulder in the Chiroptera. Con- olophus ferrumequinum. J. Biol. Sci. trib. Vertebr. Evol. 4: 1±198. Moscow Univ. 7: 37±44. Strother, G. K., and M. Mogus Solmsen, E.-H. 1970. Acoustical beam patterns for bats some 1994. Vergleichende Untersuchungen zur theoretical considerations. J. Acoust. SchaÈdelkonstruktion der neuweltlichen Soc. Am. 48: 1430±1432. BluÈten¯edermaÈuse sowie zu ihrer sys- Swofford, D. L. tematischen Ordnung unter besonderer 1991. When are phylogeny estimates from BeruÈcksichtigung der Glossophaginae molecular and morphological data in- (Phyllostomatidae, Chiroptera, Mam- congruent? In M. M. Miyamoto and J. malia). Ph.D. Diss., Univ. Hamburg, Cracraft (eds.), Phylogenetic analysis Hamburg. of DNA sequences: 295±333. New Sonntag, C. F. York: Oxford Univ. Press. 1920. The comparative anatomy of the 1998. PAUP*: Phylogenetic analysis using tongues of the MammaliaÐI. general parsimony (*and other methods), ver- description of the tongue. Proc. Zool. sion 4.0b1. Computer program distrib- Soc. London 1920: 115±129. uted by Sinauer, Sunderland, MA. 1925. The comparative anatomy of the Swofford, D. L., and D. P. Begle tongues of the MammaliaÐXII. sum- 1993. PAUP: Phylogenetic analysis using 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 183

parsimony, version 3.1.1. User's man- from nuclear and mitochondrial DNA. ual distributed by the Smithsonian In- Mol. Biol. Evol. 10: 944±959. stitution, Washington, D.C. Van Den Bussche, R. A., J. L. Hudgeons, and R. Thomas, D. W. J. Baker 1988. Analysis of diets of plant-visiting bats 1998. Phylogenetic accuracy, stability, and and methods of estimating fruit avail- congruence: relationships within and ability to frugivorous bats. In T. H . among the New World bat genera Ar- Kunz (ed), Ecological and behavioral tibeus, Dermanura, and Koopmania. In methods for the study of bats: 211±231. T. H. Kunz and P. A. Racey (eds.), Bats: Washington, D.C.: Smithsonian Inst. biology and conservation: 59±71. Thomas, O. Washington D.C.: Smithsonian Inst. 1903. Two new glossophagine bats from Cen- Van Valen, L. tral America. Ann. Mag. Nat. Hist. ser. 1979. The evolution of bats. Evol. Theory 4: 7, 11: 286±289. 104±121. 1908. The missing premolar of the Chirop- Vaughan, T. A. tera. Ann. Mag. Nat. Hist. ser. 8: 346± 1959. Functional morphology of three bats: 348 Eumops, Myotis, and Macrotus. Univ. Trajano, E. Kansas Publ. Mus. Nat. Hist. 12: 1± 1985. Ecologia de populacËoÄes de morceÃgos 153. cavernõÂcolas em uma regiaÄo caÂrstica do 1986. Mammalogy, 3rd ed. Philadelphia: W. sudeste do Brasil. Rev. Bras. Zool. 2: B. Saunders. 255±820. Vaughan, T. A., and G. C. Bateman Trouessart, E. L. 1970. Functional morphology of the forelimb 1897. Catalogus mammalium tam viventum of mormoopid bats. J. Mammal. 51: quam fossilium. Nova edito (prima 217±235. completa). 1: 1±218. Berlin: R. Fried- laÈnder. Vierhaus, V. H. Turner, D. C. 1983. Wie Vampir¯edermaÈuse (Desmodus ro- 1975. The vampire bat, a ®eld study of be- tundus) ihre ZaÈhne schaÈrfen. Z. SaÈu- havior and ecology. Baltimore: Johns getierkd. 48: 269±277. Hopkins Univ. Press. Voss, R. S., and L. H. Emmons Tuttle, M. D., and M. J. Ryan 1996. Mammalian diversity in Neotropical 1981. Bat predation and the evolution of frog lowland rainforests: a preliminary as- vocalizations in the Neotropics. Sci- sessment. Bull. Am. Mus. Nat. Hist. ence 214: 677±678. 230: 1±115. Uchikawa, K. Wagner, J. A. 1987. Myobiid mites (Trombidiformes: Myo- 1840. Die SaÈugethiere in Abbildungen nach biidae) associated with the vampire der Natur mit Beschriebungen von Dr. bats (Chiroptera: Pyllostomatidae) and Johann Christian Daniel von Schreber, information on host taxonomy deduced Supplementband, Erst Abtheilung. from them. J. Parasitol. 73: 640±645. [Part one of Wagner's supplement to Van Den Bussche, R. A. Schreber's multivolume work]. Erlang- 1991. Phylogenetic analysis of restriction site en: in der Expedition des Schre- variation in the ribosomal DNA com- ber'schen SaÈughthier[e] und des Es- plex of New World leaf-nosed bat gen- per'schen Schmetterlingwerkes, etc. era. Syst. Zool. 40: 420±432. Waite, P.M.E. 1992. Restriction-site variation and molecular 1973a. Somatotropic organization of vibrissal systematics of New World leaf-nosed responses in the ventro-basal complex bats. J. Mammal. 73: 29±42. of the rat thalamus. J. Physiol. (Lon- Van Den Bussche, R. A., and R. J. Baker don) 228: 527±540. 1993. Molecular phylogenetics of the New 1973b. The responses of cells in the rat thala- World bat genus Phyllostomus based on mus to mechanical movements of whis- cytochrome b DNA sequence variation. kers. J. Physiol. (London) 228: 541± J. Mammal. 74: 793±802. 561. Van Den Bussche, R. A., R. J. Baker, H. A. Wich- Walton, D. W., and G. M. Walton man, and M. J. Hamilton 1968. Comparative osteology of the pelvic 1993. Molecular phylogenetics of Stenoder- and pectoral girdles of the Phyllosto- matini bat genera: congruence of data matidae (Mammalia: Chiroptera). J. 184 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Grad. Res. Cent. South. Methodist Wimsatt, W. A. Univ. 37: 1±35. 1975. Some comparative aspects of implan- Warner, R. M. tation. Biol. Reprod. 12: 1±40. 1983. Karyotypic megaevolution and phylo- 1979. Reproductive asymmetry and unilateral genetic analysis: New World nectar- pregnancy in Chiroptera. J. Reprod. feeding bats revisited. Syst. Zool. 32: Fertil. 56: 345±357. 279±282. Wineski, L. E. Weber, M. 1985. Facial morphology and vibrissal move- 1904. Die SaÈugetiere. Jena: Gustav Fischer. ment in the golden hamster. J. Morphol. Wenzel, R. L., V. J. Tipton, and A. Kiewlicz 183: 199±217. 1966. The streblid bat¯ies of Panama (Dip- Winge, H. tera Calypterae: Streblidae). In R. L. 1892. Jordfundne og nulevende Flagermus Wenzel and V. J. Tipton (eds.), Ecto- (Chiroptera) fra Lagoa Santa, Minas parasites of Panama: 405±675. Chica- Geraes, Brasilien. Museo Lundii 2: 1± go: Field Museum. 66. Wible, J. R., and K. P. Bhatnager 1941. The interrelationships of the mamma- 1996. Chiropteran vomeronasal complex and lian genera, trans. E. Deichmann and the interfamilial relationships of bats. J. G. M. Allen. Copenhagen: C. A. Re- Mamm. Evol. 3: 285±314. itzels. Wilder, B. D. Winkelmann, J. R. 1976. Comparative histology and ultrastruc- 1971. Adaptations for nectar-feeding in glos- ture of the epithelial elements of the pa- sophagine bats. Ph.D. diss., Univ. of rotid glands of selected glossophagine bats with reference to evolution and Michigan, Ann Arbor. systematics. Ph.D. diss., St. John's Wood Jones, F. Univ., New York. 1917. The genitalia of the Chiroptera. J. Anat. Wiley, E. O. Phys. London 51: 36±60. 1981. Phylogenetics: the theory and practice 1923. The mammals of South Australia. Pt. I: of phylogenetic systematics. New The monotremes and carnivorous mar- York: Wiley. supials. Adelaide: Gov. Printer. Wilkinson, M. Woolsey, T. A. 1995. A comparison of two methods of char- 1978. Some anatomical bases of cortical so- acter construction. Cladistics 11: 297± matotropic organization. Brain Behav. 308. Evol. 15: 325±371. Wille, A. Zucker E., and W. I. Welker 1954. Muscular adaptation of the nectar eat- 1969. Coding of somatic sensory input by vi- ing bats (subfamily Glossophaginae). brissae neurons in the rat trigeminal Trans. Kansas Acad. Sci. 57: 315±325. ganglion. Brain Res. 12: 128±156.

APPENDIX 1: SPECIMENS EXAMINED The following list includes all specimens ex- villii (AMNH 39559, 39561, 213894±213897, amined in this study; see individual character dis- 236656). Pteronotus davyi (AMNH 2698/2324, cussions for published data sources. All taxa are 26981, 27296, 29694, 29703±29704, 32083± arranged alphabetically. See Materials and Meth- 32085, 32089±32091, 32093±32094, 170201, ods for abbreviations. The dagger (²) indicates 170204, 175276, 178479±178481, 178487, that the noseleaf of a skin was rehydrated. 179954, 180437, 202316, 203563±203566, 204407, 204960, 204963, 204965, 207060, Mormoopidae: Mormoops megalophylla 214405±214407, 214410, 214412±214413, (AMNH 25090±25091, 25097±25100, 25121, 234961, 249065; USNM 6325, 362096, 508373, 27301, 27303, 27305, 144509±144511, 170194, 508376). Pteronotus parnellii (AMNH 23656, 170196±170197, 175564, 175575, 177423, 31570±31571, 31574±31576, 135245, 135249, 183315, 187842±187843, 187848, 187850± 238140, 267851). 187852, 204472±204473, 205340, 205342, Noctilionidae: Noctilio leporinus (AMNH 205344, 205347±205348, 206692±206693, 2757, 39574, 91936±91937, 91945, 92373, 214414, 239231±239232, 249066; USNM 37484/ 92378±92380, 92530, 93434, 93808, 173912, 8311, 314677, 407207, 417184). Mormoops blan- 175534, 175536, 175538, 213201±213202, 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 185

213956±213957, 230111-230112, 234277, 335401, 338091, 372482, 441630, 542757, 243719±243724, 243905±243906, 244904, 549502). Glossophaginae: Anoura geoffroyi 249993, 260049, 261372, 265974, 267398, (AMNH 60536, 60538, 78288, 203622±203624, 267408; USNM 391025, 508370, 522063, 203636±203637, 203678, 230270±230271, 523006). 233346, 235649, 237932±237933, 243758, Phyllostomidae: Brachyphylliinae: Brachy- 244620, 244947, 246191±246194, 246196, phylla cavernarum (AMNH 39287, 72278±72280, 246198±246199, 246473, 283347; USNM 49356, 72283, 72293±72295, 72297, 72299±72300, 385799, 523116±523117, 548093, 549371). An- 149367, 208181, 213706, 213731, 213735, oura caudifera (AMNH 176347, 212261, 213737, 213740, 213742, 213899, 213907± 246468±246471, 248013, 261393, 261395± 213909, 214203, 214251, 214258 235646, 261396). Anoura cultrata (AMNH 233257). Cho- 239064±239065, 246997±247005; USNM eroniscus godmani (AMNH 172779, 186162). 544829, 544830, 544832±544833, 544866). Choeroniscus intermedius (AMNH 95778, Brachyphylla nana (AMNH 175972±175973). 179956, 185314, 207065, 230285±230286, Carolliinae: Carollia perspicillata (AMNH 246156, 266120±266124, 266377, 267152± 130722, 202299, 202750±202752, 209407, 267153, 267946±267947). Choeroniscus minor 212944, 217534±217535, 217537±217538, (AMNH 67626, 140471, 207824, 248759; USNM 217540, 236043, 236047, 246206, 246210, 361575, 385925, 460101±460103). Choeroniscus 246212±246215, 248017, 260006±260007, periosus (AMNH 217038). Choeronycteris mexi- 266125, 266130, 266134, 266138±266139, cana (AMNH 27306, 27314±27316, 74465± 266141, 266143±266145, 266147, 266149, 74468, 173664±173665, 173871±173872, 266152±266153, 266155±266162, 266164, 173874, 180350, 212359, 212362, 212364, 266378, 267453±267454; USNM 536900± 213762, 237361±237362, 239233±239234, 536901, 536905, 537910, 537912±537914, 243862±243863; USNM 21903, 52178, 52181, 537916, 537918±537919). Carollia subrufa 52342, 52348, 52355±52358, 54354±54358, (AMNH 164023, 186375, 235717, 244091, 96492). Glossophaga commissarisi (AMNH 249087, 254622). Rhinophylla pumilio (AMNH 189605, 235715). Glossophaga leachii (AMNH 233521, 233523±233525, 236056, 237950, 97545, 135277, 189626). Glossophaga longiros- 239918, 239920, 262421±262423, 262468± tris (AMNH 182724±182725, 182904). Glosso- 262469, 262471, 266165, 266167, 266169, phaga soricina (AMNH 9755, 15157, 91238, 266171±266172, 266178, 266186±266187, 91926, 91929, 91932, 97553, 209354, 214137, 266189±266194, 266196, 267158±267160, 214139, 214415, 214417, 217530, 237911± 267162±267164, 267455, 267971; USNM 237912, 247984, 247989±247990, 254619, 361632, 393015, 574529). Rhinophylla ®scherae 260958, 260960, 260964±260965, 266090± (AMNH 76094, 94557, 230483, 230485±230496, 266091, 266099, 267133±267134, 267136± 230500, 233863, 239918, 262469, 266190). Rhin- 267143, 267448±267449, 267956; USNM ophylla alethina (AMNH 217032). Desmodonti- 312995, 313000±313003, 313119, 313121± nae: Desmodus rotundus (AMNH 172293, 313122, 545156, 545159). Hylonycteris under- 174303, 183304, 189215, 190149, 190152± woodi (AMNH 178904, 189687±189688, 238199; 190154, 202296, 203764, 205410, 205414, FMNH 69728; USNM 336355, 336455, 349922, 207903, 208894, 208897, 208899, 208904, 506578, 523119, 559562, 562787±562791, 209748, 210955±210956, 210962, 213654, 562793±562796, 562798±562799, 562802± 214423, 235326±235327, 235353±235354, 562805, 563276±563278, 566441). Leptonycteris 237373, 237375, 239943, 249100, 249102, curasoae (AMNH 7402±7403, 27321, 149385± 249193, 254670, 256983, 263625±263626, 149390, 169966, 173667, 173669, 180347, 267503±267505, 267211; USNM 114977, 180349, 189699, 189701, 189708±189709, 245140, 363077, 441621, 508828, 523438, 189711, 189714±189715, 189724±189725, 536939, 549500, 565049, 574562). Diaemus 204976±204979, 205353, 205355, 205357; youngi (AMNH 78285, 94559, 175654, 175997, USNM 314690, 389140, 389142±389143, 209100, 209742, 209744±209747, 215026, 389147±389150, 444719, 444722±444723, 237920, 246230±246232, 260249, 263182, 444732, 456798, 456800, 456804, 483373, 266347, 267297, 267299, 267301; USNM 511721). Leptonycteris nivalis (AMNH 172039, 335399±335400, 361772, 393029±393030, 249086). Lichonycteris obscura (AMNH 131769, 536938, 553710). Diphylla ecaudata (AMNH 244621, 267960; USNM 51546, 309402±309403, 63841, 66800±66801, 91283±91284, 126464, 315310, 319377, 319493, 331257±331258, 129041, 142201, 165640, 175000, 175004, 335187, 338727, 362595, 396476, 432194, 203837±203843, 203846, 206694, 213423, 483374, 506578, 514961, 519891±519892, 239944, 256847, 261777; USNM 315733, 537520, 537589, 575499±575500). Lionycteris 186 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248 spurrelli (AMNH 78431, 78433±78434, 97221, 31517±31519, 62940, 149217±149218, 149233, 97264, 97267±97269, 145504, 202295, 230207, 183850, 184700±184701, 230121±230122, 230209, 236006, 236010, 236014, 236016, 249947±249948, 256292±256293, 265950± 236019, 236022, 236024±236027, 239391, 265951; USNM 313101, 313105±313106, 260004, 265336; USNM 335186, 491672, 313726±313727, 313731, 313733, 491172, 491675±491676). Lonchophylla handleyi (AMNH 491174, 562890). Lonchorhina orinocensis 230214). Lonchophylla mordax (AMNH 235608). (USNM 491177, 491179). Macrophyllum macro- Lonchophylla robusta (AMNH 143761±143763, phyllum (AMNH 42744, 78747, 78750, 94551, 185380, 207820, 233761, 233177±233179, 96018, 97252, 98761, 119462±119463, 177662, 235760±235761, 243993, 267452; USNM 177664-177666, 178137, 178140, 178165, 306589, 311999, 313749, 362539, 362544, 178169, 178179, 209320±209321, 222040, 362546, 483361±483362, 498827, 562767± 235703±235706, 243730, 243908, 243974± 562773, 563275). Lonchophylla thomasi (AMNH 243976, 262424, 266039±266041; USNM 95493, 95772, 97271±97272, 209358, 210688, 312931, 312935, 312940, 312942±312943, 237930, 262429, 262434, 266100, 266104± 312946±312947, 312951, 312958, 312960). Mac- 266105, 266107±266108, 266117, 267452, rotus waterhousii (AMNH 2148, 2160/2146, 267942; USNM 339568, 446481, 554625). Mon- 2659/2145, 2661/2147, 4116, 22799, 41098, ophyllus plethodon (AMNH 72367). Monophyllus 41100±41101, 41103, 41130, 41133±41134, redmani (AMNH 19106, 23782±23783, 39431, 41140±41143, 41159, 41172, 41176, 45229± 39436, 39439±39440, 39444, 39447, 39505± 45231, 120970, 139569±139572, 139574, 39506, 39519, 39522, 39525, 41157, 176154± 166770±166771, 170844±170845, 170847, 176157, 219658±219660, 236659±236660, 176162±176163, 180523, 182518, 185074, 236662±236668, 236671; USNM 520514, 185076±185078, 204474±204478, 204483, 534895, 534899, 534900, 543893, 545161). Mu- 214818, 236658, 238145; USNM 511230± sonycteris harrisoni (AMNH 235179; USNM 511233, 534883, 538121±538122, 538124± 314639, 314689, 324971±324972; photo from 538125, 538127, 539736, 539738, 545147). Mi- Walker's Mammals of the World). Platalina gen- cronycteris hirsuta (AMNH 94534, 94560, ovensium (AMNH 257108; FMNH 54980, 69728; 139441±139442, 175604±175605, 175607, USNM 268765±268766). Scleronycteris ega 175610, 175612, 175615, 176287, 176623, (USNM 407889; photo from Walker's Mammals 179957, 179959, 182698, 217679±217680, of the World). Phyllonycterinae: Erophylla se- 230114±230116, 255705, 267093±267096, zekorni (AMNH 4824, 39341, 41056±41057, 267857; USNM 337373, 517305, 539797). Mi- 41059±41060, 41065±41066, 41091, 41096, cronycteris megalotis (AMNH 69154, 69164, 102059, 102061, 164280, 167115, 167117± 69194, 73497, 73500, 74476±74477, 91893, 167118, 167124±167125, 167127, 176100, 92386, 94517, 97206, 97219, 98912, 183159, 186976±186977, 194201±194203, 208979, 207775±207777, 207779, 212255±212256, 212998, 214945, 219702, 245691±245692, 216102, 230113, 230117, 243958, 246158, 247595±247597; USNM 538180±538183, 256349, 256820, 256845, 260003, 266020, 538185, 539744, 539746±539747). Phyllonycteris 267090±267092, 267411, 267862, 267864, poeyi (AMNH 10749±10750, 23758, 23762, 267090±267092). Micronycteris minuta (AMNH 103545, 176019±176022, 176024±176029, 71616, 71631, 92408±92411, 92689, 92693, 186987±186988, 214130, 236697±236698, 92695, 92846, 98916±98917, 175592±175593, 256331; USNM 49568, 103445, 103542, 103573± 175595, 175597, 183169, 183295, 230119± 103575, 535321, 535323, 535326). Phyllostomi- 230120, 233221, 237910, 244471, 267098, nae: Chrotopterus auritus (AMNH 32139, 267874). Mimon bennettii (AMNH 185862± 36987±36988, 175021±175022, 209353, 248306, 185869; 185872, 256823, 256294, 265107, 255714±255715, 256351, 256824, 260867, 265864, 267109±267110; USNM 316289). Mi- 260953, 261373, 267130±267131, 267443± mon crenulatum (AMNH 64538, 64541, 78651± 267444, 267852; USNM 113852, 335152, 78652, 78655±78656, 79526, 79528, 92225, 362400, 370090, 371568, 517315, 519868± 92387±92388, 130695, 130724, 131085±131086, 519869, 549357±549358, 554691). Glyphonycter- 175586, 207063, 233222, 236001, 256308, is sylvestris (AMNH 183297, 183299, 183846, 267111±267115, 267437±267438, 267883± 207061, 214316, 245656, 267879; USNM 267884, 267888; USNM 335120, 516015, 335101, 362394, 388734, 396399, 407260, 521580). Phylloderma stenops (AMNH 126869, 534240). Lampronycteris brachyotis (AMNH, 205371±205372, 256299, 266077±266078, 28336, 69969, 94601, 169020, 175628±175633, 267439±267441, 267890±267891; USNM 175637±175641, 182703±182704, 185856; MVZ 335142±335144, 336988±336989, 361532, 118762±118763). Lonchorhina aurita (AMNH 371547±371548, 373527, 388839±388841, 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 187

457937, 491318, 519694, 531030±531031, 113501, 361905, 544871±544872, 580675± 548468, 562334, 579583). Phyllostomus hastatus 580676). Ariteus ¯avescens (AMNH 4121, (AMNH 42340, 95409, 96954, 96956, 96958, 214944, 239109; USNM 53097, 96179±96181, 169946, 183870, 202308, 209334, 230170± 113926, 114042, 122658±122659, 252771, 230171, 233176, 239876, 239878±239879, 534909±534911, 545167-545175, 545177± 243860, 243983, 266070, 266072±266076, 545184, 545203±545204, 546358). Artibeus (Ar- 267128, 267433±267435, 267903, 267905; tibeus) amplus (AMNH 185366). Artibeus (Arti- USNM 102911, 306560±306564, 549353). Phyl- beus) fraterculus (AMNH 47238, 47249). Arti- lostomus discolor (AMNH 92398, 92400±92402, beus (Artibeus) hirsutus (AMNH 188750). Arti- 149219±149221, 182713, 207804, 207806, beus (Artibeus) jamaicensis (AMNH 39101, 243743±243744, 243981, 254612±254613, 72247, 74305, 131819, 140462, 202321, 207877, 267124). Phyllostomus elongatus (AMNH 92120, 217432±217433, 236678, 239933, 239937, 92417, 92419, 209331, 210676, 210679, 214317± 245331, 248326, 248510±248513, 263604± 214318, 230137±230138, 233226, 237927, 263605, 263610, 266321±266322, 266327± 266055, 266063). Tonatia brasiliense (AMNH 266328, 266331, 266334±266335, 266337, 267101±267104, 267426±267428; USNM 266340±266341, 266343±266344, 267202; 336990, 490096, 499293±499294, 562160, USNM 538135±538136, 544816, 544818± 566438). Tonatia carrikeri (AMNH 209322). 544819, 544822, 544824±544827). Artibeus (Ar- Tonatia evotis (AMNH 267635). Tonatia sauro- tibeus) lituratus (AMNH 180549, 210906, phila (AMNH 139438, 266042±266049, 267099± 266346, 267204, 267492, 268506, 268513). Arti- 267100, 267429±267431, 267910; USNM beus (Artibeus) obscurus (AMNH 246641, 554579, 581911). Tonatia schulzi (AMNH 266278±266279, 266288, 268523). Artibeus (Der- 267105±267106, 267420±267421). Tonatia silvi- manura) anderseni (AMNH 210841, 233759). Ar- cola (AMNH 61753, 69188, 78644, 78834± tibeus (Dermanura) aztecus (AMNH 175282, 78835, 95440±95441, 96965, 96976, 97009, 190008). Artibeus (Dermanura) cinereus (AMNH 97011±97013, 97340, 97350, 97353, 142906± 93500, 94179, 94181, 97075±97076, 212279, 142907, 182718, 237926, 262425, 267107± 212880±212881, 233782, 233784, 233777, 267108, 267924±267925; USNM 34509, 335107, 237964, 244084, 246627±246628, 246632, 335110, 335113, 370084, 503848, 518862, 248177, 256374, 266258±266263, 266265± 518864±518866, 539805, 549347, 574575, 266266, 266270, 266290±266291, 266302, 579580). Trachops cirrhosus (AMNH 209352, 266305, 266307, 267197±267198, 267499; 210683±210684, 233227, 235557, 236002, USNM 540677±540680, 549429, 549330, 255713, 256301, 266079±266089, 267129, 549437±549438). Artibeus (Dermanura) glaucus 267442, 267927, 267929, 267932, 267934± (AMNH 246631). Artibeus (Dermanura) gnomus 267936; USNM 244254, 244260, 361533± (AMNH 67650, 77531, 266293, 266301, 361534, 371559, 460090, 513435, 526241, 266310). Artibeus (Dermanura) phaeotis (AMNH 549354, 562763±562764). Trinycteris nicefori 21083, 209590, 210825, 217410). Artibeus (Der- (AMNH 78650, 149200, 175643±175645, manura) toltecus (AMNH 190015). Artibeus 179984±179985, 184557±184559, 256822, (Dermanura) watsoni (AMNH 18760, 265122). 266015±266017, 266019, 267409±267410, Artibeus (Koopmania) concolor (AMNH 80340, 267446, 267876). Vampyrum spectrum (AMNH 89945, 260014, 266267-266269, 267192±267195, 7435±7436, 18707, 28993, 42895, 66815, 267477±267478, 267480±267485, 267488, 175717±175718, 184694, 202292, 256825, 267502, 267981; USNM 405203, 405206, 261379, 267132, 267278, 267445; USNM 549439±549441). Centurio senex (AMNH 99645, 246253, 307233, 319373, 335159±335160, 175650±175651, 177419, 179989±179991, 335162±335163, 337312±337314, 346251, 180401, 183862, 208879, 214422, 243786± 346264±346265, 346267±346268, 393007, 243788, 256330, 256846, 265125; USNM 22039/ 562765±562766, 575471, 731786). Stenoderma- 6322, 310237±310239, 310242, 310245, 335392, tinae: Ametrida centurio (AMNH 142612± 346821±346822, 346827, 346832, 346840, 142613, 142909, 172127, 187224±187225, 503837, 511740±511741, 511855±511856, 207967, 247645, 267274±267276, 267278± 511859). Chiroderma salvini (AMNH 142484, 267279, 267375, 267973, 267975; USNM 261667, 262537). Chiroderma trinitatum (AMNH 370783±370786, 370795, 393758, 460135± 209521, 235607, 266255±266256, 267189). Chi- 460136, 494488, 494510, 494520, 531106± roderma villosum (AMNH 37041, 209542± 531108, 535108, 542285, 554827). Ardops ni- 209544, 209547, 209552, 209554, 209558, chollsi (AMNH 212557, 212896, 213925, 209560±209564, 209576, 210811, 233738, 213952±213955, 217104, 238150±238151; 243926, 243929, 245326±245327, 262526± USNM 110918, 113186, 113498±113499, 262532, 262534, 262539, 267190±267191, 188 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

267474±267476, 268533±268534, 268536; 208982±208983; USNM 239110, 271210, USNM 522434±522435, 534314±534316, 522706±522708, 549795). Sturnira bogotensis 574537). Ectophylla alba (FMNH 137219± (AMNH 207851, 207853, 246570, 260870, 137221, UMMZ 125633; USNM 315563, 261537). Sturnira erythromos (AMNH 233529, 319426, 324926, 335310, 335317±335325, 233597, 244948, 246572). Sturnira lilium 336259±336262, 563298, 579079). Ectophylla (AMNH 185320±185321, 189881, 189893± macconelli (AMNH 48270, 48274, 172126, 189894, 189946, 203705, 204713, 204724, 187223, 207068, 209577, 209579±209580, 209412, 210724, 214196, 233541±233542, 214368, 215025, 233745±233749, 246226, 248108, 249091, 249096±249097, 266199, 248886±248887, 262540±262541, 267281, 266203±266207, 266210, 266215, 266226, 267556±267558, 267560±267564, 268539; 266228±266234, 266236, 267167, 267169; USNM 315564, 361726, 460131±460132, USNM 502244±502250, 502257±502259). Stur- 548426±549427, 574540). Enchisthenes hartii nira ludovici (AMNH 189779, 237364, 254636, (AMNH 126239, 206872, 214367, 233598± 254638). Sturnira nana (AMNH 219172). Stur- 233599, 233753, 233791±233800; USNM nira tildae (AMNH 248863, 266240, 266241, 310230, 310213±310217, 323538±323539, 266243, 266247, 267167). Uroderma bilobatum 370726±370727, 370729±370731, 483909± (AMNH 171294, 179994±179995, 209472, 483910, 483919±483920, 483922, 494470, 209496, 217404, 217408, 217411, 233184± 507202, 539813, 562916±562917). Phyllops fal- 233185, 243770, 245322, 246350, 260193, catus (AMNH 41008, 143662, 176190, 213890± 260196±260197, 260210, 261041, 266257, 213891, 219701, 236690±236696, 238146± 267172±267178, 267183, 267465±267470, 238147; USNM 113250, 143844, 181300, 268564±268565; USNM 312881, 370401, 300504, 300590±300595, 520534, 535313, 522357±522358, 522361, 535093, 549406± 538317±538318, 538337±538345, 538374, 549407, 549410, 549412). Uroderma magniros- 538376, 542273). Platyrrhinus aurarius (AMNH trum (AMNH 209467, 210736±210737, 210740, 261225, 261227±261228). Platyrrhinus dorsalis 210749, 210768). Vampyressa bidens (AMNH (AMNH 214355, 233187, 233640, 236105, 67997±67998, 71656, 76090, 208072, 214351, 246610±246611, 246613). Platyrrhinus helleri 233734±233737, 233785, 261625±261627, (AMNH 207863, 207871±207872, 209511, 209517, 230635, 246222±246223, 261070± 261633, 261640; USNM 361713±361714, 261072, 261074±261075, 263616, 265119± 408573, 408578, 496590, 530997, 554778, 265120, 266254, 267179±267182, 267552, 562584±562585, 579598). Vampyressa nymphaea 267554±267555, 267562; USNM 315538± (AMNH 233189±233190, 235789±235790; 315540, 315542, 315544±315545, 544894± USNM 304903, 305382, 305384, 309062, 544895, 562896). Platyrrhinus infuscus (AMNH 309883±309889, 315555±315556, 318130, 230647, 248882). Platyrrhinus lineatus (AMNH 379076, 483686±483687, 483693±483696, 37007, 210805, 255932, 260231). Platyrrhinus 520005±520006, 520009±520011, 579076). Vam- ornatus (AMNH 235560). Pygoderma bilabiatum pyressa pusilla (AMNH 31501, 67986±67987, (AMNH 234286, 234289±234301, 246401, 71675, 71680, 141996±141997, 233191±233193, 246405, 246408, 248334, 248336±248339, 238219, 256827, 262559, 266311, 267471± 261758, 261761±261762, 261764; USNM 14816/ 267472, 267185; USNM 370525, 372143, 37502, 148161, 460507, 542744, 542745± 408571, 440660±440661, 440668, 483729± 542748). Sphaeronycteris toxophyllum (AMNH 483730, 483734±483736, 496566, 496584, 21344±21345, 24379±24380, 76251, 92248, 496586±496587, 503630, 503633±503635, 194213, 209741, 235561, 261765, 262637, 534302±534303, 535095, 574639). Vampyrodes 265354; FMNH 122939; USNM 35405, 35410, caraccioli (AMNH 29431, 175642, 186381, 122785, 143657, 168230, 241107, 370797± 209518, 230649±230651, 230653±230655, 370801, 370805, 370839±370840, 370843± 239254, 256828; USNM 6327, 309711, 393019± 370847, 372391, 372393, 405663, 455936, 393020, 393022, 447008, 447010, 447012± 494533, 494538±494539, 494544±494550, 447017 496565, 519716±519717, 519719, 496843, 522053). Stenoderma rufum (AMNH 519721, 539811, 567158, 574638, 579658).

APPENDIX 2: DATA MATRIX This data matrix includes all characters used in ranged alphabetically, but the familial and sub- all analyses discussed in the text. All taxa are ar- familial names are not shown. This matrix is 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 189 available in electronic format on the World Wide 11-00 0-001 0{02}00- 00010 --11- 003-0 10000 Web at ftp://ftp.amnh.org/pub/mammalogy 00000 110?? ????1 ???00 010{02}0 23101 111?1 20100 31-00 10111 00001 01-00 0010? Mormoops 21101 20000 00010 00000 00121 00000 00400 0---- -0--- --0-3 -0-00 --011 Choeroniscus 000{01}- 00002 --000 002-0 00000 00000 10201 00000 0{01}{12}02 10000 01012 --000 10000 11-00 00000 1010? ????? ????? ????? 11-00 0-0?? ?1-0- 10010 --11- 003-1 100-0 ?0000 200-- 00010 001-- --?00 00000 10000 00000 111?? ????? ???00 01000 23100 111?1 00000 00000 2011- 31-00 10111 00001 01-10 1010? ????0 Pteronotus 2{01}001 00011 0100? 000{024}0 00000 01200 0---- -0--- --1-3 -0-00 Choeronycteris 0-0{02}0 {01}000- 00002 --000 00100 00000 10201 00000 01202 10000 01012 --000 11-00 00000 00000 11001 00000 00100 00?00 ??-00 0-0?? ?1-0- 10010 --11- 003-1 100-0 00000 ?111- 00000 200-- 3-010 001-- --?00 0000{01} 11100 001?? ???00 01000 23100 111?1 2011- 10??? ????? ????? 31-00 10111 00001 01-10 101?? ????? 20001 Noctilio 00011 0100? 00000 02000 12400 0---- -0--- --0-3 -0-00 0-020 Glossophaga 1000- 00000 --010 100-0 00000 00000 00011 10201 00000 00402 10000 01022 --000 11-00 000?0 ???20 00000 00?00 ??-00 ?001- 0001- 0-001 0000- 00010 --000 003-0 10000 00000 100-- 2-000 000-- --?00 00000 {01}0000 00000 11010 00101 11100 01000 23101 11001 21100 00000 30000 10111 00001 01-11 12100 21101 20000 Brachyphylla 00010 0000? 00201 00000 01402 11011 00-02 --001 -0000 Hylonycteris 0-010 0000- 00000 --000 000-1 100-0 00200 10202 00000 01402 10000 01012 --000 11-00 000?0 0???? ???50 02021 12000 00-?0 00000 0-0?? ?1-0- 00010 --11- 003-0 10000 00000 1001- 100-- 00010 01-00 0001? 21100 20000 111?? ????? ???00 01000 23100 111?1 20100 00010 00010 31-00 10111 00011 01-10 101?? ????? 2???1 Carollia ???11 ?1?0? 10202 00000 0{01}412 10000 11-02 --001 Leptonycteris -0100 0-001 0100- 00000 --000 000-1 100-0 10221 00001 014021 00000 1022- -0001 1-000 00000 00000 01211 01150 11000 01000 00-10 -0010 000-0 0010- -0000 03-01 01-01 00001 01000 20000 100-- 00010 01-00 000?0 21111 10000 01??? ??000 10202 31011 11?12 21003 2{12}000 00010 00000 00001 01110 00010 1-111 21?? 21101 20000 Rhinophylla 00010 10000 10201 0000{01} 02412 20000 10-02 --000 Lichonycteris 00100 0-0?? ?100- 0000{04} --000 000-1 100-0 10202 00000 01402 10000 01012 --000 11-00 00??? 111?? ????? ???40 1122? ????? ????? 0-0?? ?1-0- 00010 --11- 003-1 101-0 10000 ????? ?0000 100-- 00010 000-- --??? 21110 111?? ????? ???00 01000 23100 110?1 21100 20000 00010 0000? 31-00 10111 00011 01-10 1010? ????? 20??? Desmodus ????? ????? 01011 00001 12402 10011 00-02 --100 00000 Lionycteris 0-001 0000- 001-1 --003 100-1 101-1 11??? 00220 00000 01602 10000 11022 --000 10000 -0-01 01101 01110 02020 12011 00-00 0001- 0-0?? ?100- 00000 --000 003-0 10000 00000 01-1- 120-- 00000 1------0?1 00001 20000 110?? ????? ???00 01000 23100 110?0 10000 00000 0001? 10000 11100 02010 01-02 020?? ????1 ?0000 Diaemus 00010 1010? 01011 00001 12402 10011 00-02 --100 00000 Lonchophylla 0-001 0000- 001-1 --003 100-1 101-1 11??? 10201 00000 0{02}602 10000 11022 --000 -00?? ????? ???10 0202? ????? ????? ????? ?1-1- 10000 0-0?? ?1-0- 00000 --000 003-0 10000 120-- 00000 1------??? ????1 20??? ????? ????? 00000 110?? ????1 ???00 01000 23100 110?0 Diphylla 10000 10000 11100 01010 01-02 020?? 21101 10001 00001 02402 10011 00-02 --100 00000 20000 00010 10100 0-001 0000- 00001 --003 100-1 101-0 11??? Monophyllus -0001 012?? ???10 0102? ????? ????? ????? 10200 00000 00402 10000 01022 --000 11-00 ?1-1- 100-- 00000 1------??? ????1 10000 0-0?? ?000- 00010 --000 003-0 20000 00000 00000 00010 100?? ????? ???00 01000 23101 110?1 21100 Anoura 30000 10111 00001 01-11 121?? 21101 20000 10221 00001 0{12}402 10000 01022 --000 00010 1000? 190 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Musonycteris ?0000 100-- 00010 001-- --??1 ????1 0?000 ????1 00000 00202 ??000 01012 --000 11-?0 00010 0001? 0-0?? ?1-0- 10010 --11- 003-1 100-0 00000 Macrotus 111?? ????? ???00 01?0? ????? ????? ????? 10201 00000 12402 10000 01001 -0000 00001 ????? ????? ????? ????? ????? ????? ??001 11001 1000- 00000 --000 003-0 00010 00000 00011 0100? 10000 01?11 00140 21012 01011 00-?0 00000 Platalina 20000 100-- 00010 01-00 000?1 1000? 00000 10201 00000 00602 10000 00-22 --000 10000 00000 00010 0-0?? ?1-0- 00000 --000 003-1 100-0 00000 Micronycteris hirsuta 111?? ????? ???00 01000 23100 110?0 10000 11201 00000 12402 10000 12001 -2000 00001 10000 11100 02000 01-02 020?? ????? ????? 100?? ?000- 00000 --000 002-0 00010 00000 ????? ????? 100?0 0???1 ???41 2020? ????? ????? ????? Scleronycteris ?0000 100-- 00010 01-00 00?0? 1000? 10??? ????1 00000 01?02 ??000 ????2 --000 11-?0 ????? ????? 0-0?? ?1-0- 00010 --11- 003-1 100-0 00000 Micronycteris megalotis 111?? ????? ???00 01?0? ????? ????? ????? 11201 00000 12402 10000 10-01 -1000 00001 ????? ????? ????? ????? ????? ????? ????? ????? 100?? ?000- 00000 --000 003-0 00010 00000 ????? 100?0 0???1 ???41 2000? ????? ????? ????? Erophylla ?0000 100-- 00010 01-00 00?01 10001 10??? 00201 00000 01402 11010 00-02 --000 11-00 ????? ????? 0-0?? ?000- 00000 --000 000-1 100-0 00211 Micronycteris minuta 110?? ???01 11151 21001 11000 00-?0 0111- ????1 00000 12402 10000 10-01 -1000 00001 10000 10112 00000 01-02 001?? ????? 10??? 110?? ?010- 00000 --000 002-0 00010 00000 ????? ????? 100?? ????? ???41 1100? ????? ????? ????? Phyllonycteris ?0000 100-- 00010 01-10 00??? ????1 10000 002{01}1 00000 02402 11010 00-02 --000 00000 1001? 11-00 0-0?? ?100- 00000 --000 000-1 100-0 Mimon bennettii 00211 110?? ????? ???51 02001 12000 00-10 10201 00000 00012 10000 20-01 -0000 00000 0111- 10000 10112 00000 01-02 011?? 2110? 0-0?? ?000- 00000 --010 000-0 00010 00000 10000 00010 0001? 000?0 0???? ???40 1000? ????? ????? ????? Chrotopterus ?0000 100-- 00010 01-00 00??? ????? ?0??? 10231 00000 12002 0-000 21100 1-000 00001 ????? ????? 0-0?? ?000- 00000 --012 00120 01010 00000 Mimon crenulatum 000?0 0???? ???40 1030? ????? ????? ????? 10201 01001 00012 10000 20-00 0-001 -0000 ?0000 000-- 00010 01-00 00?0? ????? ?0010 0-0?? ?000- 00000 --010 000-0 00010 00000 01000 10010 000?0 0???? ???50 0000? ????? ????? ????? Glyphonycteris ?0000 100-- 00010 01-10 00??? ????1 00010 ????2 00000 02402 10000 10-02 --000 00000 00000 1001? 0-0?? ?000- 00000 --000 003-0 00010 00000 Phylloderma 000?? ????? ???40 0100? ????? ????? ????? 00211 00000 02402 10000 20-02 --001 -0000 ?0000 100-- 02010 01-00 00??? ????? ????? 0-0?? ?000- 00000 --000 002-0 00010 00000 ????? ????? 000?? ????? ???40 0120? ????? ????? ????? Lampronycteris ?0000 100-- 00010 01-00 00??? 21001 20??? ????0 00000 12402 10000 10-01 -0000 00000 ????? ????? 0-0?? ?000- 00000 --000 002-0 00010 00000 Phyllostomus 000?? ????? ???10 0000? ????? ????? ????? 0021{01} 00000 02{01}12 10000 20-00 0-001 ?0000 100-- 00010 01-00 00??? ????1 ????? -0000 0-0?? ?000- 00000 --000 000-000010 ????? ????? 00000 00000 11210 00000 00202 01011 00-10 Lonchorhina 00000 20000 000-- 00010 01-00 00000 21001 11221 00000 {01}{02}{45}12 10000 21100 {02}0000 00000 10010 0-000 00000 0-0?? ?010- 00000 --000 00110 Tonatia 00010 00000 000?0 0???1 ???00 0011? ????? 11201 00000 {01}2{04}02 0-000 210{02}2 ????? ????? ?0000 100-- 00010 001-- --??? --001 -0001 0-0?? ?0{01}0- 00000 --010 002-0 ????? 00000 00000 10010 00010 00000 000?0 0???1 ???41 {12}030? Macrophyllum ????? ????? ????? ?0000 100-- 00010 01-00 10221 00001 00112 10000 21000 0-001 -0100 00?01 ????1 00000 00000 10010 0-0?? ?000- 00000 --000 00110 00010 00000 Trachops 000?0 0???1 ???50 0001? ????? ????? ????? 10201 00000 02102 0-000 21002 --001 -0001 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 191

0-0?? ?000- 00000 --000 00110 01010 00000 10222 1{01}000 02{01}12 10{02}00 20-00 000?0 0???1 ???40 1100? ????? ????? ????? 0-001 -0110 0-0?? ?000- 00000 --00{01} 000-1 ?0000 100-- 00010 01-00 00??0 10001 00000 10{01}-0 {01}0211 110?1 0???0 ???50 01000 10010 {01}122? ????? ????? ????? ?0002 100-- 1-100 Trinycteris 01-00 00?1? ????1 20010 00010 00000 ????2 02000 02402 10000 10-02 --000 00000 Ectophylla alba 0-0?? ?000- 00000 --000 002-0 00010 00000 10201 -0--0 02012 10100 20-00 0-001 -0110 100?0 0???1 ???00 01002 01001 00-?0 00000 0-0?? ?000- 00000 --000 000-1 101-0 10211 20000 100-- 00010 01-00 00001 ????? 10??? 111?? ????? ???50 1122? ????? ????? ????? ????? ????? ?0002 000-- 1-000 01-00 00??? 2221? ?0??? Vampyrum ????? ????? 10201 00000 12002 0-000 21100 1-000 00001 Ectophylla macconellii 0-0?? ?000- 00000 --002 002-0 01010 00000 10222 00000 02012 10100 20-00 0-011 -0110 100?? ????? ???40 1032? ????? ????? ????? 0-0?? ?000- 00000 --000 000-1 101-0 00211 ?0000 000-- 00010 000-- --??? ????1 00000 111?? ????? ???50 1122? ????? ????? ????? 00000 0001? ?0000 000-- 1-100 01-00 00??? ????1 10000 Ametrida 00010 00000 11202 00111 02011 20000 20-00 0-001 -0120 Enchisthenes 0-0?? ?00-- 01003 1-001 000-1 000-0 00200 10221 10001 02412 10000 20-02 --001 -0110 000?? ????? ???00 0122? ????? ????? ????? 0-0?? ?000- 00000 --001 000-1 000-0 00200 ?0002 200-- 1-000 01-10 10??? 2221? 11000 000?? ????? ???60 0122? ????? ????? ????? 10010 0000? ?0002 100-- 1-100 01-00 00??? 22211 21000 Ardops 00010 00000 00212 00101 02012 20000 20-01 -3001 -0120 Phyllops 0-0?? ?0030 00003 00001 010-1 000-0 00210 11222 00111 02012 20000 20-01 -0001 -0120 000?? ????? ???50 0122? ????? ????? ????? 0-0?? ?0031 00003 00001 010-1 000-0 00200 ?0002 200-- 1-000 01-10 10??? ????1 ?1000 000?? ????? ???40 21222 01000 00-?0 00001 10010 00000 20002 200-- 1-000 01-10 100?? ????? ?1??? Ariteus ????? ????? 11222 00101 02012 20000 20-01 -3001 -0120 0-0?? ?0030 00003 00001 010-1 001-0 00210 Platyrrhinus 000?? ????? ???50 0122? ????? ????? ????? 10221 11001 02{01}12 10{02}00 20-00 0-001 ?0002 200-- 1-000 01-10 10??? 2221? ?1000 -0110 0-0?? ?000- 00000 --001 010-1100-0 10110 00000 00{12}10 100?? ????? ???50 {01}1222 01001 Artibeus (Artibeus) 00-?0 00001 20001 000-- 1-000 01-10 000{01}? 10221 1000{01} 02412 10000 20-00 0-001 22211 {12}0100 00010 00000 -0110 0-001 0000- 00002 --001 010-1 00{01}-0 Pygoderma 00200 000?1 0??00 00150 01222 01000 00-10 10202 00111 01012 10000 20-01 -0001 -0120 00001 20002 100-- 1-100 01-00 000{01}1 0-0?? ?001- 00003 01001 010-1 001-0 10200 22211 {12}1010 00010 00001 000?? ????? ???40 0122? ????? ????? ????? Artibeus (Dermanura) ?0002 200-- 1-000 01-10 10??? ????? ?1000 10221 1000{01} 02{024}12 10{02}00 20-00 10110 00000 0-001 -0110 0-0?? ?000- 00002 --001 010-1 Sphaeronycteris 001-0 {01}0{12}00 000?? ????? ???50 0122? 10202 00111 02011 3-000 10-02 --022 -0-20 ????? ????? ????? ?0002 100-- 1-100 01-00 0-1?? ?0021 01003 1-001 000-1 000-0 00200 00?{01}? ????1 {12}{01}000 00010 00001 000?? ????? ???50 0122? ????? ????? ????? Artibeus (Koopmania) ?0102 200-- 1-000 01-10 10??? ????1 ?0??? ????2 00000 02312 10000 20-00 0-001 -0110 ????? ????? 0-0?? ?000- 00002 --001 000-1 000-0 00200 Stenoderma 000?? ????? ???50 0122? ????? ????? ????? 10222 00111 02012 20000 20-00 0-001 -0120 ?0002 100-- 1-100 01-00 00??? ????1 ?1??? 0-0?? ?0030 00003 01001 000-1 000-0 00210 ????? ????1 000?? ????? ???50 0122? ????? ????? ????? Centurio ?0002 200-- 1-000 01-10 10??? ????? 11000 11202 00111 12011 3------0-2 -0-20 0-1?? 10110 00000 ?0021 00003 00001 000-1 001-0 10200 00010 Sturnira 011?? ???00 2122? ????? ????? ????? ?011- 200-- 00202 00001 0{012}{04}12 10000 10-02 --001 1-000 01-10 10?1? 2221? 10000 10010 00000 -0110 0-001 0000- 00000 --0{01}0 000-1 100-0 Chiroderma {01}0{02}{01}{01} 000?0 0??10 00140 02020 192 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

00?00 ??-10 ?001- 20001 100-- 00010 01-00 0-0?? ?000- 00002 --000 000-1 101-0 10211 00?00 22211 {12}0010 00010 00010 111?? ????? ???50 0122? ????? ????? ????? Uroderma ?0001 100-- 1-100 01-00 00?1? ????? 20000 10221 11000 02012 10200 20-00 0-001 -0110 00010 00000 0-0?? ?000- 00002 --001 020-1 000-0 00210 Vampyressa pusilla 100?0 0???? ???50 01222 01000 00-?0 00001 11222 10000 02012 10200 20-00 0-011 -0110 20001 100-- 1-100 01-10 1001? 22211 20100 0-0?? ?000- 00002 --000 000-1 101-0 10211 00010 00000 111?? ????? ???40 11222 01001 00-?0 00001 Vampyressa bidens 20000 100-- 1-100 01-00 0001? 22211 10??? 10222 11000 02012 10200 20-00 0-001 -0110 ????? ????? 0-0?? ?000- 00002 --010 000-1 101-0 00211 Vampyrodes 111?? ????? ???50 0122? ????? ????? ????? 11221 11001 02412 10200 20-00 0-001 -0110 ?0001 100-- 1-100 01-00 00??? ????? ????? 0-0?? ?000- 00000 --001 020-1 101-0 00100 ????? ????? 000?? ????? ???50 0122? ????? ????? ????? Vampyressa nymphaea ?0001 000-- 1-100 01-10 00?1? 22211 10100 ????2 11000 02012 10200 20-00 0-001 -0110 00010 00000

APPENDIX 3: DISCRETE-STATE CHARACTERS NOT INCLUDED IN THIS STUDY In the course of this study we reviewed nu- 3.3.3.4, 3.3.3.6, 3.3.4.3, 3.3.4.4, 3.3.4.6, 3.3.4.12, merous characters that do not appear in any form 3.3.5.2, 3.3.5.4, 3.3.5.5, 3.3.5.7, 3.3.5.15; Lim in our data set. These characters have not been (1993): character 8; Marques-Aguiar (1994): used for a variety of reasons, which are noted be- characters 8, 12, 14, 15, 19, 20, 25, 29, 30, 31; low. Previously described characters that we omit- Gimenez et al. (1996): characters 1, 5, 7±9. ted include the following: (5) Discrete-state characters appropriate at lower taxonomic levels than those investigated here (1) Characters described by previous authors as (variable within most genera): Straney (1980): autapomorphies (or which would appear as auta- characters K4, K5, K17; Owen (1987): characters pomorphies in our analysis due to the level of tax- 12, 13, 16, 17; Marques-Aguiar (1994): characters onomic sampling): Gimenez (1993): character 1, 2, 3, 5, 6, 13, 16±18, 22, 23, 27, 28. Simmons 3.3.3.7.; Marques-Aguiar (1994): character 11; (1996): characters 2, 12, 13, 16, 17, 19, 21, 22. Owen (1991): character 1; Van Den Bussche (6) Characters that are variable within species: (1991, 1992): characters 29, 31, 34, 35, 39, 42, 44, Straney (1980): characters E15±16 (see Grif®ths, 48, 55, 56; Simmons (1996): characters 3, 15, 20. 1982: 26±27), E23 (see Grif®ths, 1982: 26±27), (2) Characters in which the derived condition H4, H5, J4, J5; Owen (1987): characters 8, 11, occurs in all members of the ingroup or all taxa 18. in our analysis: Straney (1980): character E1, (7) Characters for which our observations dif- K27; Hood and Smith (1982): characters 2, 4. fered from those of previous authors, and we were (3) Characters in which the derived condition unable to observe or con®rm the conditions re- only occurs in outgroup taxa: Van Den Bussche ported by others: Straney (1980): characters, G4, (1991, 1992): characters 40, 41, 45. G5, G21±23, H1, H2, H9, H11±14, J3, K28, K29; (4) Characters de®ned by previous authors that Gimenez (1993): characters 3.3.3.2, 3.3.4.1, vary continuously within Phyllostomidae, or 3.3.4.5, 3.3.4.13, 3.3.5.6, 3.3.5.10. based on descriptions that suggest continuous var- (8) Characters which we were unable to survey iation: Straney (1980): characters D3 ±5 (see due to limited specimen availability: Straney McDaniel, 1976), D6±12, H6, H7±8, H15, H16, (1980): characters K7, K9. J1, J2, J7, J8, K1, K2, K14±16, K18, K30; Owen (9) Characters for which the variation reported (1987): characters 5, 7, 15, 19, 20, 22; Giminez may be due to sampling juveniles: Straney (1980): (1993): characters 3.3.2.4, 3.3.2.6, 3.3.3.1, characters E2±8, H3.

APPENDIX 4: TAXONOMIC DIAGNOSES The following diagnoses apply to monophyletic genera of Mimon. Both ACCTRAN and DEL- groups described in the text and numbered in ®g- TRAN optimizations are given for each node (see ure 66 and the genera Lampronycteris, Glyphon- Materials and Methods for a description of opti- ycteris, Macrotus, Micronycteris, Phylloderma, mization procedures). Diagnoses are formatted as Phyllostomus, and Trinycteris and the two sub- follows: (character number; consistency index) 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 193 character description, state number → state num- atohyoideus insertion on stylohyoid, 0 → 1; (107; ber. For example, ``(30; 0.333) chin morphology, 0.500) Medial circumvallate papillae presence, 0 0 → 3'' indicates a change from state 0 to 3 for ⇒ 1; (109; 0.250) Lateral circumvallate papillae character 30. Unequivocal transformations (those presence, 0 → 1; (121; 1.000) Horny papillae ar- occurring in all reconstructions) are indicated by rangement, 0 ⇒ 1; (130; 0.250) Male accessory a double arrow ``⇒,'' equivocal transformations morphology, 0 → 1. DELTRAN: (10; 0.364) Uro- (occurring in only some reconstructions) by a sin- patagial fringe presence, 0 ⇒ 1; (20; 0.500) Trun- gle arrow ``→.'' Please refer to individual char- cated spear, 0 → 1; (28; 0.500) Dorsal skin ridge acter descriptions for character state information. on snout, 0 ⇒ 1; (50; 0.500) I1 occlusal margin Phyllostomidae: ACCTRAN: (5; 0.214) Dor- shape, 0 ⇒ 1; (55; 0.800) Incisor occlusion, 0 ⇒ sal fur banding pattern, 0 → 1; (15; 1.000) Num- 3; (56; 0.500) P3 presence, 0 → 1; (63; 0.273) ber of lateral vibrissae, 0 ⇒ 2; (16; 0.500) Vi- M3 presence, 0 ⇒ 1; (66; 0.400) m3 presence, 0 brissal papillae structure, 0 ⇒ 1; (30; 0.333) Chin ⇒ 1; (67; 1.000) m1 shearing ridge, 0 ⇒ 1; (75; morphology, 3 → 0; (40; 0.333) Vomeronasal car- 0.250) Number of m. occiptopollicalus muscle tilage shape, 0 → 1; (59; 1.000) p3 displacement, bellies, 0 → 1; (84; 0.278) Longest metacarpal, 0 0 → 1; (60; 0.167) W-shaped ectoloph presence → 1; (89; 0.333) Tail length, 0 → 2; (107; 0.500) on M1±M2, 0 ⇒ 1; (61; 0.333) M1 hypocone Medial circumvallate papillae presence, 0 ⇒ 1; presence, 0 ⇒ 1; (78; 0.500) Caput mediale of m. (109; 0.250) Lateral circumvallate papillae pres- triceps brachii insertion, 0 ⇒ 1; (80; 0.250) M. ence, 0 → 1; (121; 1.000) Horny papillae arrange- palmaris longus insertion on digit III, 0 → 1; (82; ment, 0 ⇒ 1. 0.333) M. palmaris longus insertion on digit V, 0 Node 1: ACCTRAN: (2; 0.100) Bulb on hair → 1; (83; 0.500) M. ¯exor digitorum profundus shaft base, 0 ⇒ 1; (4; 0.400) Hair scale margin insertion on digit IV, 0 → 1; (84; 0.278) Longest shape, 0 ⇒ 1; (11; 0.429) Superciliary vibrissae metacarpal, 0 → 1; (87; 0.200) Calcar length, 0 presence, 0 ⇒ 1; (48; 1.000) I2 presence, 0 ⇒ 1; ⇒ 1; (89; 0.333) Tail length, 0 → 2; (91; 0.667) (65; 1.000) m2 presence, 0 ⇒ 1; (87; 0.200) Cal- M. sternohyoideus medial ®bers origin, 0 ⇒ 1; car length, 1 ⇒ 2; (112; 1.000) Lingual sulcus (92; 0.600) M. sternohyoideus lateral ®bers ori- presence, 0 ⇒ 2. DELTRAN: (2; 0.100) Bulb on gin, 0 ⇒ 2; (122; 0.250) Horny papillae size, 0 hair shaft base, 0 ⇒ 1; (4; 0.400) Hair scale mar- → 1; (135; 0.400) Accessory olfactory bulb pres- gin shape, 0 ⇒ 1; (11; 0.429) Superciliary vibris- ence, 0 ⇒ 1; (136; 0.533) Inferior colliculi cov- sae presence, 0 ⇒ 1; (48; 1.000) I2 presence, 0 erage, 1 → 2; (149; 0.250) Restriction site 54, 0 ⇒ 1; (65; 1.000) m2 presence, 0 ⇒ 1; (87; 0.200) ⇒ 1. DELTRAN: (5; 0.214) Dorsal fur banding Calcar length, 1 ⇒ 2; (112; 1.000) Lingual sulcus pattern, 0 → 1; (15; 1.000) Number of lateral vi- presence, 0 ⇒ 2; (136; 0.533) Inferior colliculi brissae, 0 ⇒ 2; (16; 0.500) Vibrissal papillae coverage, 1 → 2. structure, 0 ⇒ 1; (40; 0.333) Vomeronasal carti- Node 2: ACCTRAN: (3; 1.000) Scale diver- lage shape, 0 → 1; (60; 0.167) W-shaped ectoloph gence from hair shaft, 0 ⇒ 2; (30; 0.333) Chin presence on M1±M2, 0 ⇒ 1; (61; 0.333) M1 hy- morphology, 0 → 1; (84; 0.278) Longest meta- pocone presence, 0 ⇒ 1; (78; 0.500) Caput me- carpal, 1 → 5; (99; 0.500) Right and left m. ge- diale of m. triceps brachii insertion, 0 ⇒ 1; (80; niohyoideus fusion, 0 → 1; (104; 0.250) M. colli 0.250) M. palmaris longus insertion on digit III, profundus lateral slip presence, 1 ⇒ 0; (106; 0 → 1; (83; 0.500) M. ¯exor digitorum profundus 1.000) M. cricopharyngeus slip number, 0 ⇒ 1; insertion on digit IV, 0 → 1; (87; 0.200) Calcar (119; 0.167) Basketlike papillae presence, 0 → 1; length, 0 ⇒ 1; (91; 0.667) M. sternohyoideus me- (131; 1.000) External uterine fusion, 0 ⇒ 2; (132; dial ®bers origin, 0 ⇒ 1; (92; 0.600) M. sterno- 0.667) Internal uterine fusion, 0 ⇒ 1; (133; 0.667) hyoideus lateral ®bers origin, 0 ⇒ 2; (135; 0.400) Attachment of oviduct to uterus, 0 ⇒ 1; (144; Accessory olfactory bulb presence, 0 ⇒ 1; (149; 0.333) Restriction site 47, 0 ⇒ 1. DELTRAN: (3; 0.250) Restriction site 54, 0 ⇒ 1. 1.000) Scale divergence from hair shaft, 0 ⇒ 2; Desmodontinae: ACCTRAN: (10; 0.364) Uro- (104; 0.250) M. colli profundus lateral slip pres- patagial fringe presence, 0 ⇒ 1; (20; 0.500) Trun- ence, 1 ⇒ 0; (106; 1.000) M. cricopharyngeus slip cated spear, 0 → 1; (28; 0.500) Dorsal skin ridge number, 0 ⇒ 1; (122; 0.250) Large central horny on snout, 0 ⇒ 1; (50; 0.500) I1 occlusal margin papillae presence, 0 → 1; (131; 1.000) External shape, 0 ⇒ 1; (55; 0.800) Incisor occlusion, 0 ⇒ uterine fusion, 0 ⇒ 2; (132; 0.667) Internal uter- 3; (56; 0.500) P3 presence, 0 → 1; (63; 0.273) ine fusion, 0 ⇒ 1; (133; 0.667) Attachment of M3 presence, 0 ⇒ 1; (66; 0.400) m3 presence, 0 oviduct to uterus, 0 ⇒ 1; (136; 0.533) Inferior ⇒ 1; (67; 1.000) m1 shearing ridge, 0 ⇒ 1; (75; colliculi coverage, 1 → 2; (144; 0.333) Restriction 0.250) Number of m. occiptopollicalus muscle site 47, 0 ⇒ 1. bellies, 0 → 1; (94; 0.333) M. ceratohyoideus in- Brachyphyllinae: ACCTRAN: (12; 0.500) Ge- sertion on ceratohyoid, 0 → 1; (95; 0.167) M. cer- nal vibrissae number, 2 ⇒ 1; (17; 0.500) Vibrissal 194 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Fig. 66. Strict consensus tree from our character congruence analysis with nodes numbered for reference to appendix 4, which presents apomorphies of the clades. 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 195 papillae in contact across dorsum of snout, 0 ⇒ (69; 0.400) m1 metaconid presence, 0 ⇒ 1; (70; 1; (20; 0.500) Truncated spear, 0 → 1; (39; 0.750) 0.667) m1 entoconid presence, 0 ⇒ 1; (84; 0.278) Vomeronasal tube development, 0 ⇒ 1; (40; Longest metacarpal, 0 → 5; (85; 0.333) Digit III 0.333) Vomeronasal cartilage shape, 1 ⇒ 0; (68; second phalanx length relative to ®rst, 0 ⇒ 1; (90; 0.714) m1 paraconid presence, 0 ⇒ 2; (87; 0.200) 0.500) M. mylohyoideus division, 0 ⇒ 1; (102; Calcar length, 1 ⇒ 2; (90; 0.500) M. mylohyoi- 0.333) M. stylohyoideus presence, 0 ⇒ 1; (103; deus division, 0 ⇒ 1; (109, 0.250) Lateral circum- 0.333) M. sphincter colli profundus anterolateral vallate papillae presence, 0 → 1; (129; 0.800) slip presence, 0 → 1; (104; 0.250) M. sphincter Brunner's glands presence, 0 ⇒ 1; (135; 0.400) colli profundus lateral slip presence, 0 ⇒ 1; (115; Accessory olfactory bulb presence, 1 ⇒ 0. DEL- 1.000) Hairlike papillae shape, 0 → 2; (128; TRAN: (12; 0.500) Genal vibrissae number, 2 ⇒ 0.500) Lingual artery number, 0 → 1; (136; 0.533) 1; (17; 0.500) Vibrissal papillae in contact across Inferior colliculi coverage, 2 ⇒ 1. dorsum of snout, 0 ⇒ 1; (20; 0.500) Truncated Glossophaginae: ACCTRAN: (1; 0.143) Pel- spear, 0 → 1; (39; 0.750) Vomeronasal tube de- age differentiation, 0 ⇒ 1; (12; 0.500) Genal vi- velopment, 0 ⇒ 1; (40; 0.333) Vomeronasal car- brissae number, 2 ⇒ 0; (19; 0.500) Spear length, tilage shape, 1 ⇒ 0; (68; 0.714) m1 paraconid 1 ⇒ 0; (22; 0.286) Internarial region morphology, presence, 0 ⇒ 2; (87; 0.200) Calcar length, 1 ⇒ 0 → 1; (24; 1.000) Lateral horseshoe morphology, 2; (89; 0.333) Tail length, 0 → 2; (90; 0.500) M. 0 ⇒ 2; (58; 0.273) p3 presence, 0 ⇒ 3; (60; mylohyoideus division, 0 ⇒ 1; (109, 0.250) Lat- 0.167) W-shaped ectoloph presence on M1-M2, 1 eral circumvallate papillae presence, 0 → 1; (119; → 0; (84; 0.278) Longest metacarpal, 5 → 0; (91; 0.167) Basketlike papillae presence, 0 → 1; (129; 0.667) M. sternohyoideus medial ®bers origin, 1 0.800) Brunner's glands presence, 0 ⇒ 1; (135; ⇒ 2; (92; 0.600) M. sternohyoideus lateral ®bers 0.400) Accessory olfactory bulb presence, 1 ⇒ 0. origin, 2 ⇒ 3; (93; 1.000) M. sternohyoideus in- Hirsutaglossa: ACCTRAN: (30; 0.333) Chin sertion, 0 ⇒ 1; (96; 1.000) M. hyoglossus origin, morphology, 1 → 0; (31; 1.000) Chin pad mor- 0 ⇒ 1; (97; 1.000) M. geniohyoideus insertion, 0 phology, 0 ⇒ 1; (32; 0.500) Chin cleft presence, 0 ⇒ 1; (99; 0.500) Right and left m. geniohyoideus → 1; (71; 0.125) P3±P4 diastema, 0 ⇒ 1; (72; fusion, 1 → 0; (101; 1.000) M. genioglossus in- 0.250) P4±M1 diastema, 0 ⇒ 1; (77; 0.500) M. sertion, 0 → 1; (127; 0.667) Flanking horny pa- spinodeltoideus origin, 1 → 0; (81; 1.000) M. pal- pillae presence, 0 ⇒ 2; (146; 0.143) Restriction maris longus insertion on digit IV, 0 ⇒ 1; (89; site 50, 0 → 1; (149; 0.250) Restriction site 54, 1 0.333) Tail length, 2 → 0; (103; 0.333) M. sphinc- ⇒ 0. DELTRAN: (1; 0.143) Pelage differentia- ter colli profundus anterolateral slip presence, 0 → tion, 0 ⇒ 1; (12; 0.500) Genal vibrissae number, 1; (113; 1.000) Hairlike papillae presence, 0 ⇒ 1; 2 ⇒ 0; (19; 0.500) Spear length, 1 ⇒ 0; (24; (119; 0.167) Basketlike papillae presence, 1 → 0; 1.000) Lateral horseshoe morphology, 0 ⇒ 2; (58; (125.0.667) Anterior horny papillae number, 0 ⇒ 0.273) p3 presence, 0 ⇒ 3; (91; 0.667) M. ster- 2; (128.0.500) Lingual artery number, 0 → 1. DEL- nohyoideus medial ®bers origin, 1 ⇒ 2; (92; TRAN: (31; 1.000) Chin pad morphology, 0 ⇒ 1; 0.600) M. sternohyoideus lateral ®bers origin, 2 (71; 0.125) P3±P4 diastema, 0 ⇒ 1; (72; 0.250) ⇒ 3; (93; 1.000) M. sternohyoideus insertion, 0 P4±M1 diastema, 0 ⇒ 1; (81; 1.000) M. palmaris ⇒ 1; (96; 1.000) M. hyoglossus origin, 0 ⇒ 1; longus insertion on digit IV, 0 ⇒ 1; (82; 0.333) M. (97; 1.000) M. geniohyoideus insertion, 0 ⇒ 1; palmaris longus insertion on digit V, 0 → 1; (113; (127; 0.667) Flanking horny papillae presence, 0 1.000) Hairlike papillae presence, 0 ⇒ 1; (125; ⇒ 2; (149; 0.250) Restriction site 54, 1 ⇒ 0. 0.667) Anterior horny papillae number, 0 ⇒ 2. Lonchophyllini: ACCTRAN: (13; 0.667) In- Phyllonycterinae: ACCTRAN: (17; 0.500) Vi- terramal vibrissae number, 4 ⇒ 6; (32; 0.500) brissal papillae in contact across dorsum of snout, Chin cleft presence, 1 → 0; (42; 0.400) Zygomatic 0 ⇒ 1; (68; 0.714) m1 paraconid presence, 0 ⇒ arch completeness, 0 ⇒ 1; (103; 0.333) M. 2; (69; 0.400) m1 metaconid presence, 0 ⇒ 1; (70; sphincter colli profundus anterolateral slip pres- 0.667) m1 entoconid presence, 0 ⇒ 1; (85; 0.333) ence, 1 → 0; (112; 1.000) Lingual sulcus pres- Digit III second phalanx length relative to ®rst, 0 ence, 0 ⇒ 1; (114; 1.000) Hairlike papillae dis- ⇒ 1; (90; 0.500) M. mylohyoideus division, 0 ⇒ tribution, 1 ⇒ 0; (117; 0.667) Anteriorly directed 1; (102; 0.333) M. stylohyoideus presence, 0 ⇒ patch of medial-posterior papillae presence, 0 ⇒ 1; (104; 0.250) M. sphincter colli profundus lat- 2; (128; 0.500) Lingual artery number, 1 → 0; eral slip presence, 0 ⇒ 1; (115; 1.000) Hairlike (148; 1.000) Restriction site 53, 0 → 1. DEL- papillae shape, 0 → 2; (136; 0.533) Inferior col- TRAN: (13; 0.667) Interramal vibrissae number, liculi coverage, 2 ⇒ 1. DELTRAN: (17; 0.500) 4 ⇒ 6; (42; 0.400) Zygomatic arch completeness, Vibrissal papillae in contact across dorsum of 0 ⇒ 1; (101; 1.000) M. genioglossus insertion, 0 snout, 0 ⇒ 1; (32; 0.500) Chin cleft presence, 0 → 1; (112; 1.000) Lingual sulcus presence, 0 ⇒ → 1; (68; 0.714) m1 paraconid presence, 0 ⇒ 2; 1; (114; 1.000) Hairlike papillae distribution, 1 ⇒ 196 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

0; (117; 0.667) Anteriorly directed patch of me- circumvallate papillae presence, 0 ⇒ 1; (125; dial-posterior papillae presence, 0 ⇒ 2. 0.667) Anterior horny papillae number, 1 ⇒ 0; Node 3: ACCTRAN: (21; 0.286) Central rib (127; 0.667) Flanking horny papillae presence, 2 length, 0 ⇒ 1; (119; 0.167) Basketlike papillae ⇒ 0. presence, 0 ⇒ 1. DELTRAN: (21; 0.286) Central Node 7: ACCTRAN: (4; 0.400) Hair scale mar- rib length, 0 ⇒ 1; (22; 0.286) Internarial region gin shape, 2 → 0; (10; 0.364) Uropatagial fringe morphology, 0 → 1; (60; 0.167) W-shaped ecto- presence, 1 → 0; (24; 1.000) Lateral horseshoe loph presence on M1-M2, 1 → 0; (119; 0.167) morphology, 2 ⇒ 1; (42; 0.400) Zygomatic arch Basketlike papillae presence, 0 ⇒ 1; (146; 0.143) completeness, 0 ⇒ 1; (60; 0.167) W-shaped ec- Restriction site 50, 0 → 1; (148; 1.000) Restric- toloph presence on M1-M2, 0 ⇒ 1; (73; 0.250) tion site 53, 0 → 1. M1-M2 diastema, 0 ⇒ 1; (95; 0.167) M. ceratoh- Glossophagini: ACCTRAN: (49; 1.000) Size yoideus insertion on stylohyoid, 1 ⇒ 0; (135; of I1 verus I2, 0 ⇒ 1; (95; 0.167) M. ceratohyoi- 0.400) Accessory olfactory bulb presence, 1 → 0; deus insertion, 0 ⇒ 1; (100; 1.000) M. styloglos- (140; 1.000) Restriction site 36, 0 ⇒ 1; (145; sus insertion, 0 ⇒ 1; (101; 1.000) M. genioglossus 1.000) Restriction site 49, 0 ⇒ 1; (147; 1.000) insertion, 1 → 2; (106; 1.000) M. cricopharyngeus Restriction site 52, 0 ⇒ 1. DELTRAN: (24; 1.000) slip number, 1 ⇒ 3; (115; 1.000) Hairlike papillae Lateral horseshoe morphology, 2 ⇒ 1; (42; 0.400) shape, 0 → 1; (120; 1.000) Horny papillae loca- Zygomatic arch completeness, 0 ⇒ 1; (60; 0.167) tion, 0 ⇒ 1; (124; 0.167) Central horny papillae W-shaped ectoloph presence on M1-M2, 0 ⇒ 1; number, 0 ⇒ 1; (125; 0.667) Anterior horny pa- (73; 0.250) M1-M2 diastema, 0 ⇒ 1; (95; 0.167) pillae number, 2 ⇒ 1;(126; 0.250) Posterior horny M. ceratohyoideus insertion on stylohyoid, 1 ⇒ papillae number, 0 ⇒ 1. DELTRAN: (22; 0.286) 0; (140; 1.000) Restriction site 36, 0 ⇒ 1; (145; Internarial region morphology, 0 → 1; (32; 0.500) 1.000) Restriction site 49, 0 ⇒ 1; (147; 1.000) Chin cleft presence, 0 → 1; (49; 1.000) Size of I1 Restriction site 52, 0 ⇒ 1. verus I2, 0 ⇒ 1; (60; 0.167) W-shaped ectoloph Node 8: ACCTRAN: (5; 0.214) Dorsal fur presence on M1-M2, 1 → 0; (77; 0.500) M. spi- banding pattern, 1 ⇒ 2; (119; 0.167) Basketlike nodeltoideus origin, 1 → 0; (95; 0.167) M. cera- papillae presence, 0 ⇒ 1. DELTRAN: (5; 0.214) tohyoideus insertion, 0 ⇒ 1; (100; 1.000) M. sty- Dorsal fur banding pattern, 1 ⇒ 2; (119; 0.167) loglossus insertion, 0 ⇒ 1; (101; 1.000) M. ge- Basketlike papillae presence, 0 ⇒ 1. nioglossus insertion, 0 → 2; (103; 0.333) M. Node 9: ACCTRAN: (13; 0.667) Interramal vi- sphincter colli profundus anterolateral slip pres- brissae number, 4 ⇒ 2; (46; 1.000) Pterygoid lam- ence, 0 → 1; (106; 1.000) M. cricopharyngeus slip inae in¯ation, 0 ⇒ 1; (104; 0.250) M. sphincter number, 1 ⇒ 3; (115; 1.000) Hairlike papillae colli profundus lateral slip presence, 0 ⇒ 1. DEL- shape, 0 → 1; (120; 1.000) Horny papillae loca- TRAN: (13; 0.667) Interramal vibrissae number, tion, 0 ⇒ 1; (124; 0.167) Central horny papillae 4 ⇒ 2; (46; 1.000) Pterygoid laminae in¯ation, 0 number, 0 ⇒ 1; (125; 0.667) Anterior horny pa- ⇒ 1; (104; 0.250) M. sphincter colli profundus pillae number, 2 ⇒ 1; (126; 0.250) Posterior lateral slip presence, 0 ⇒ 1. horny papillae number, 0 ⇒ 1; (128; 0.500) Lin- Node 10: ACCTRAN: (1; 0.143) Pelage dif- gual artery number, 0 → 1. ferentiation, 0 → 1; (14; 0.333) Lateral vibrissal Node 4: ACCTRAN: (74; 0.333) Distal tip of column presence, 0 → 1; (19; 0.500) Spear length, clavicle attachment, 0 → 1; (102; 0.333) M. sty- 1 ⇒ 0; (21; 0.286) Central rib length, 0 → 1; (64; lohyoideus presence, 0 ⇒ 1. DELTRAN: (102; 1.000) M3 ectoloph shape, 0 → 1; (78; 0.500) M. 0.333) M. stylohyoideus presence, 0 ⇒ 1. triceps brachii caput mediale insertion, 1 → 2; Node 5: ACCTRAN: (4; 0.400) Hair scale mar- (79; 0.500) M. palmaris longus insertion on digit gin shape, 0 → 2; (10; 0.364) Uropatagial fringe II, 0 ⇒ 1; (82; 0.333) M. palmaris longus inser- presence, 0 → 1; (12; 0.500) Genal vibrissae num- tion on digit V, 1 → 0; (84; 0.278) Longest meta- ber, 0 ⇒ 1; (98; 0.500) M. geniohyoideus split carpal, 5 ⇒ 4; (91; 0.667) M. sternohyoideus me- insertion type, 0 ⇒ 1. DELTRAN: (12; 0.500) Ge- dial ®bers origin, 1 ⇒ 0; (92; 0.600) M. sterno- nal vibrissae number, 0 ⇒ 1; (98; 0.500) M. ge- hyoideus lateral ®bers origin, 2 ⇒ 1; (106; 1.000) niohyoideus split insertion type, 0 ⇒ 1. M. cricopharyngeus slip number, 1 ⇒ 2. DEL- Node 6: ACCTRAN: (53; 1.000) i1 presence, 0 TRAN: (19; 0.500) Spear length, 1 ⇒ 0; (30; ⇒ 1; (54; 0.286) i2 presence, 0 ⇒ 1; (107; 0.500) 0.333) Chin morphology, 0 → 1; (79; 0.500) M. Medial circumvallate papillae presence, 0 ⇒ 1; palmaris longus insertion on digit II, 0 ⇒ 1; (84; (125; 0.667) Anterior horny papillae number, 1 ⇒ 0.278) Longest metacarpal, 0 ⇒ 4; (91; 0.667) M. 0; (127; 0.667) Flanking horny papillae presence, sternohyoideus medial ®bers origin, 1 ⇒ 0; (92; 2 ⇒ 0; (146; 0.143) Restriction site 50, 1 → 0. 0.600) M. sternohyoideus lateral ®bers origin, 2 DELTRAN: (53; 1.000) I1 presence, 0 ⇒ 1; (54; ⇒ 1; (99; 0.500) Right and left m. geniohyoideus 0.286) I2 presence, 0 ⇒ 1; (107; 0.500) Medial fusion, 0 → 1; (106; 1.000) M. cricopharyngeus 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 197 slip number, 1 ⇒ 2; (119; 0.167) Basketlike pa- Mimon (Anthorhina): ACCTRAN: (7; 0.500) pillae presence, 0 → 1. Dorsal stripe presence, 0 ⇒ 1; (10; 0.364) Uro- Phyllostominae: ACCTRAN: (21; 0.286) Cen- patagial fringe presence, 0 ⇒ 1; (124; 0.167) Cen- tral rib length, 1 → 2; (41; 0.667) Nasopalatine tral horny papillae number, 0 ⇒ 1. DELTRAN: duct presence, 0 → 1; (60; 0.167) W-shaped ec- (7; 0.500) Dorsal stripe presence, 0 ⇒ 1; (10; toloph presence on M1-M2, 1 ⇒ 0; (61; 0.333) 0.364) Uropatagial fringe presence, 0 ⇒ 1; (84; M1 hypocone presence, 1 ⇒ 0; (89; 0.333) Tail 0.278) Longest metacarpal, 4 → 5; (124; 0.167) length, 2 → 0; (90; 0.500) M. mylohyoideus di- Central horny papillae number, 0 ⇒ 1; (139; vision, 0 ⇒ 2; (94; 0.333) M. ceratohyoideus in- 0.200) Restriction site 32, 0 → 1. sertion on ceratohyoid, 0 → 1; (95; 0.167) M. cer- Mimon (Mimon): ACCTRAN: (25; 0.222) La- atohyoideus insertion on stylohyoid, 0 ⇒ 1; (133; bial horseshoe morphology, 0 ⇒ 1; (30; 0.333) 0.667) Location of oviductal entrance to uterus, 1 Chin morphology, 1 ⇒ 0; (84; 0.278) Longest ⇒ 0; (144; 0.333) Restriction site 47, 1 ⇒ 0; (146; metacarpal, 5 → 4; (86; 0.385), Digit IV second ⇒ 0.143) Restriction site 50, 0 1. DELTRAN: (21; phalanx length relative to ®rst, 0 ⇒ 1. DEL- → 0.286) Central rib length, 0 2; (60; 0.167) W- TRAN: (25; 0.222) Labial horseshoe morphology, ⇒ shaped ectoloph presence on M1-M2, 1 0; (61; 0 ⇒ 1; (30; 0.333) Chin morphology, 1 ⇒ 0; (86; ⇒ 0.333) M1 hypocone presence, 1 0; (64; 1.000) 0.385), Digit IV second phalanx length relative to → M3 ectoloph shape, 0 1; (90; 0.500) M. my- ®rst, 0 ⇒ 1. lohyoideus division, 0 ⇒ 2; (95; 0.167) M. cera- ⇒ Node 14: ACCTRAN: (14; 0.333) Lateral vi- tohyoideus insertion on stylohyoid, 0 1; (133; brissal column presence, 1 → 0; (26; 1.000) Mor- 0.667) Location of oviductal entrance to uterus, 1 phology of thin free labial edge of horseshoe, 0 ⇒ 0; (144; 0.333) Restriction site 47, 1 ⇒ 0; (146; → ⇒ ⇒ 1; (58; 0.273) p3 presence, 0 2. DELTRAN: 0.143) Restriction site 50, 0 1. (58; 0.273) p3 presence, 0 ⇒ 2; (131; 1.000) Ex- Node 11: ACCTRAN: (130; 0.250) Male ac- ternal uterine fusion, 2 → 1; (132; 0.667) Internal cessory gland morphology, 0 ⇒ 1; (131; 1.000) uterine fusion, 1 → 0. External uterine fusion, 2 → 1; (132; 0.667) In- Vampyrini: ACCTRAN: (13; 0.667) Interra- ternal uterine fusion, 1 → 0; (136; 0.533) Inferior mal vibrissae number, 4 → 0; (16; 0.500) Vibrissal colliculi coverage, 2 ⇒ 0. DELTRAN: (1; 0.143) papillae morphology, 1 ⇒ 0; (22; 0.286) Inter- Pelage differentiation, 0 → 1; (59; 1.000) p3 dis- narial region morphology, 0 ⇒ 1; (35; 0.500) Pin- placement, 0 → 1; (130; 0.250) Male accessory na morphology, 0 ⇒ gland morphology, 0 ⇒ 1; (136; 0.533) Inferior 1; (62; 0.500) Metastyle ⇒ length on M1-M2, 0 → 1; (86; 0.385) Digit IV colliculi coverage, 2 0. ⇒ Lonchorhinini: ACCTRAN: (12; 0.500) Genal second phalanx length relative to ®rst, 0 1. vibrissae number, 2 ⇒ 0; (25; 0.222) Labial horse- DELTRAN: (16; 0.500) Vibrissal papillae mor- ⇒ phology, 1 ⇒ 0; (22; 0.286) Internarial region shoe morphology, 2 0; (84; 0.278) Longest ⇒ metacarpal, 4 → 5; (87; 0.200) Calcar length, 1 morphology, 0 1; (35; 0.500) Pinna morphol- ⇒ ogy, 0 ⇒ 1; (86; 0.385) Digit IV second phalanx 0; (123; 0.500) Size of horny papillae in clus- ⇒ ter, 0 → 1. DELTRAN: (12; 0.500) Genal vibris- length relative to ®rst, 0 1. sae number, 2 ⇒ 0; (14; 0.333) Lateral vibrissal Node 15: ACCTRAN: (11; 0.429) Superciliary → column presence, 0 → 1; (25; 0.222) Labial horse- vibrissae presence, 0 1; (54; 0.286) i2 presence, → → shoe morphology, 2 ⇒ 0; (87; 0.200) Calcar 0 1; (59; 1.000) p3 displacement, 1 2; (87; ⇒ length, 1 ⇒ 0. 0.200) Calcar length, 1 0; (88; 0.429) Plagio- ⇒ Node 12: ACCTRAN: (4; 0.400) Hair scale patagium attachment location, 0 3. DELTRAN: ⇒ margin shape, 0 ⇒ 2; (22; 0.286) Internarial re- (87; 0.200) Calcar length, 1 0; (88; 0.429) Pla- ⇒ gion morphology, 0 ⇒ 1; (58; 0.273) p3 presence, giopatagium attachment location, 0 3. 0 ⇒ 1; (89; 0.333) Tail length, 0 ⇒ 1; (122; Node 16: ACCTRAN: (23; 0.500) Sella pres- 0.250) Large central horny papillae presence, 1 ⇒ ence, 0 ⇒ 1; (25; 0.222) Labial horseshoe mor- 0. DELTRAN: (4; 0.400) Hair scale margin shape, phology, 2 ⇒ 0; (30; 0.333) Chin morphology, 1 0 ⇒ 2; (22; 0.286) Internarial region morphology, ⇒ 0; (55; 0.800) Incisor occlusion, 0 ⇒ 2; (111; 0 ⇒ 1; (58; 0.273) p3 presence, 0 ⇒ 1; (89; 0.333) Papillae distribution on pharyngeal tongue, 0.333) Tail length, 0 ⇒ 1; (122; 0.250) Large cen- 1 ⇒ 0. DELTRAN: (11; 0.429) Superciliary vi- tral horny papillae presence, 1 ⇒ 0; (123; 0.500) brissae presence, 0 → 1; (13; 0.667) Interramal Size of horny papillae in cluster, 0 → 1. vibrissae number, 4 → 0; (23; 0.500) Sella pres- Mimon: ACCTRAN: (13; 0.667) Interramal vi- ence, 0 ⇒ 1; (25; 0.222) Labial horseshoe mor- brissae number, 4 ⇒ 0; (54; 0.286) i2 presence, 0 phology, 2 ⇒ 0; (26; 1.000) Morphology of thin ⇒ 1; (139; 0.200) Restriction site 32, 0 → 1. free labial edge of horseshoe, 0 → 1; (30; 0.333) DELTRAN: (13; 0.667) Interramal vibrissae num- Chin morphology, 1 ⇒ 0; (55; 0.800) Incisor oc- ber, 4 ⇒ 0; (54; 0.286) i2 presence, 0 ⇒ 1. clusion, 0 ⇒ 2; (62; 0.500) Metastyle length on 198 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

M1-M2, 0 → 1; (111; 0.333) Papillae distribution hair shaft base, 0 → 1; (27; 1.000) Thickened la- on pharyngeal tongue, 1 ⇒ 0. bial horseshoe morphology, 0 ⇒ 1; (85; 0.333) Micronycterini: ACCTRAN: (21; 0.286) Cen- Digit III second phalanx length relative to ®rst, 0 tral rib length, 2 ⇒ 1; (30; 0.333) Chin morphol- ⇒ 1. DELTRAN: (27; 1.000) Thickened labial ogy, 1 ⇒ 0; (136; 0.533) Inferior colliculi cov- horseshoe morphology, 0 ⇒ 1; (85; 0.333) Digit erage, 0 → 1. DELTRAN: (21; 0.286) Central rib III second phalanx length relative to ®rst, 0 ⇒ 1; length, 2 ⇒ 1; (30; 0.333) Chin morphology, 1 ⇒ 0. (136; 0.533) Inferior colliculi coverage, 0 → 1. Node 17: ACCTRAN: (5; 0.214) Dorsal fur Node 20: ACCTRAN: (37; 1.000) Notch depth banding pattern, 1 ⇒ 2; (94; 0.333) M. cerato- in interauricular band, 1 ⇒ 0; (87; 0.200) Calcar hyoideus insertion on ceratohyoid, 1 → 0. DEL- length, 1 ⇒ 0. DELTRAN: (2; 0.100) Bulb on hair TRAN: (5; 0.214) Dorsal fur banding pattern, 1 shaft base, 0 → 1; (37; 1.000) Notch depth in ⇒ 2. interauricular band, 1 ⇒ 0; (87; 0.200) Calcar Glyphonycteris: ACCTRAN: (58; 0.273) p3 length, 1 ⇒ 0. presence, 2 ⇒ 3; (117; 0.667) Anteriorly directed Phyllostomini: ACCTRAN: (1; 0.143) Pelage patch of medial-posterior papillae presence, 0 ⇒ differentiation, 1 → 0; (4; 0.400) Hair scale mar- 2. DELTRAN: (58; 0.273) p3 presence, 2 ⇒ 3; gin shape, 0 ⇒ 1; (76; 0.500) M. occipitopolli- (117; 0.667) Anteriorly directed patch of medial- calus ventral attachment number, 0 → 1; (80; posterior papillae presence, 0 ⇒ 2. 0.250) M. palmaris longus insertion on digit III, Trinycteris: ACCTRAN: (7; 0.500) Dorsal 1 → 0; (83; 0.500) M. ¯exor digitorum profundus stripe presence, 0 ⇒ 2; (71; 0.125) P3±P4 diaste- insertion, 1 → 0; (88; 0.429) Plagiopatagium at- ma, 0 ⇒ 1; (84; 0.278) Longest metacarpal, 4 ⇒ tachment location, 0 ⇒ 2. DELTRAN: (4; 0.400) 0. DELTRAN: (7; 0.500) Dorsal stripe presence, Hair scale margin shape, 0 ⇒ 1; (88; 0.429) Pla- 0 ⇒ 2; (71; 0.125) P3±P4 diastema, 0 ⇒ 1; (84; giopatagium attachment location, 0 ⇒ 2. 0.278) Longest metacarpal, 4 ⇒ 0; (136; 0.533) Phylloderma: ACCTRAN: (14; 0.333) Lateral Inferior colliculi coverage, 0 → 1. vibrissal column presence, 1 → 0; (58; 0.273) p3 Node 18: ACCTRAN: (11; 0.429) Superciliary presence, 0 ⇒ 2. DELTRAN: (58; 0.273) p3 pres- vibrissae presence, 0 ⇒ 1; (25; 0.222) Labial ence, 0 ⇒ 2. horseshoe morphology, 2 ⇒ 1. DELTRAN: (11; Phyllostomus: ACCTRAN: (13; 0.667) Inter- 0.429) Superciliary vibrissae presence, 0 ⇒ 1; ramal vibrissae number, 4 ⇒ (01); (25; 0.222) La- (25; 0.222) Labial horseshoe morphology, 2 ⇒ 1. bial horseshoe morphology, 2 ⇒ 0; (84; 0.278) Lampronycteris: ACCTRAN: (5; 0.214) Dorsal Longest metacarpal, 4 ⇒ 0; (87; 0.200) Calcar fur banding pattern, 1 ⇒ 0; (84; 0.278) Longest length, 1 ⇒ 0; (111; 0.333) Papillae distribution metacarpal, 4 ⇒ 1; (87; 0.200) Calcar length, 1 on pharyngeal tongue, 1 ⇒ 0. DELTRAN: (13; ⇒ 0. DELTRAN: (5; 0.214) Dorsal fur banding 0.667) Interramal vibrissae number, 4 ⇒ (01); (14; pattern, 1 ⇒ 0; (84; 0.278) Longest metacarpal, 4 0.333) Lateral vibrissal column presence, 0 → 1; ⇒ 1; (87; 0.200) Calcar length, 1 ⇒ 0. (25; 0.222) Labial horseshoe morphology, 2 ⇒ 0; Node 19: ACCTRAN: (35; 0.500) Pinna mor- (76; 0.500) M. occipitopollicalus ventral attach- phology, 0 ⇒ 1; (36; 1.000) Interauricular band ment number, 0 → 1; (78; 0.500) M. triceps bra- presence, 0 ⇒ 1; (71; 0.125) P3±P4 diastema, 0 chii caput mediale insertion, 1 → 2; (80; 0.250) ⇒ 1; (86; 0.385) Digit IV second phalanx length M. palmaris longus insertion on digit III, 1 → 0; relative to ®rst, 0 ⇒ 2. DELTRAN: (35; 0.500) (83; 0.500) M. ¯exor digitorum profundus inser- Pinna morphology, 0 ⇒ 1; (36; 1.000) Interauric- tion, 1 → 0; (84; 0.278) Longest metacarpal, 4 ⇒ ular band presence, 0 ⇒ 1; (71; 0.125) P3±P4 0; (87; 0.200) Calcar length, 1 ⇒ 0; (94; 0.333) diastema, 0 ⇒ 1; (86; 0.385) Digit IV second pha- M. ceratohyoideus insertion on ceratohyoid, 0 → lanx length relative to ®rst, 0 ⇒ 2. 1; (111; 0.333) Papillae distribution on pharyn- Macrotus: ACCTRAN: (21; 0.286) Central rib geal tongue, 1 ⇒ 0. length, 1 ⇒ 0; (22; 0.286) Internarial region mor- Nullicauda: ACCTRAN: (5; 0.214) Dorsal fur phology, 0 ⇒ 1; (58; 0.273) p3 presence, 2 ⇒ 3; banding pattern, 1 → 2; (33; 0.500) Central pa- (89; 0.333) Tail length, 0 ⇒ 1; (136; 0.533) In- pilla presence on chin, 0 ⇒ 1; (134; 1.000) Lo- ferior colliculi coverage, 1 → 0; (146; 0.143) Re- cation of oviductal entrance to uterus, 0 ⇒ 1; striction site 50, 1 ⇒ 0. DELTRAN: (21; 0.286) (149; 0.250) Restriction site 54, 1 → 0. DEL- Central rib length, 1 ⇒ 0; (22; 0.286) Internarial TRAN: (14; 0.333) Lateral vibrissal column pres- region morphology, 0 ⇒ 1; (41; 0.667) Nasopal- ence, 0 → 1; (21; 0.286) Central rib length, 0 → atine duct presence, 0 → 1; (58; 0.273) p3 pres- 1; (33; 0.500) Central papilla presence on chin, 0 ence, 2 ⇒ 3; (89; 0.333) Tail length, 0 ⇒ 1; (94; ⇒ 1; (134; 1.000) Location of oviductal entrance 0.333) M. ceratohyoideus insertion on ceratohy- to uterus, 0 ⇒ 1. oid, 0 → 1; (146; 0.143) Restriction site 50, 1 ⇒ 0. Carolliinae: ACCTRAN: (42; 0.400) Zygo- Micronycteris: ACCTRAN: (2; 0.100) Bulb on matic arch completeness, 0 ⇒ 1; (82; 0.333) M. 2000 WETTERER ET AL.: PHYLOGENY OF PHYLLOSTOMID BATS 199 palmaris longus insertion on digit V, 0 → 1; (86; brissal papillae morphology, 1 → 2; (34; 1.000) 0.385) Digit IV second phalanx length relative to Internal labial papillae distribution, 1 ⇒ 2; (44; ®rst, 0 ⇒ 1; (102; 0.333) M. stylohyoideus pres- 1.000) Posterior hard palate length, 0 → 1; (50; ence, 0 → 1. DELTRAN: (1; 0.143) Pelage dif- 0.500) I1 occlusal margin morphology, 0 ⇒ 3; ferentiation, 0 → 1; (42; 0.400) Zygomatic arch (111; 0.333) Papillae distribution on pharyngeal completeness, 0 ⇒ 1; (86; 0.385) Digit IV second tongue, 1 ⇒ 2; (124; 0.167) Central horny papil- phalanx length relative to ®rst, 0 ⇒ 1; (149; lae number, 0 ⇒ 1; (126; 0.250) Posterior horny 0.250) Restriction site 54, 1 → 0. papillae presence, 0 ⇒ 1; (136; 0.533) Inferior Stenodermatinae: ACCTRAN: (10; 0.364) colliculi coverage, 2 ⇒ 1; (141; 1.000) Restriction Uropatagial fringe presence, 0 ⇒ 1; (34; 1.000) site 38, 0 ⇒ 1. DELTRAN: (5; 0.214) Dorsal fur Internal labial papillae distribution, 0 ⇒ 1; (68; banding pattern, 1 → 2; (8; 1.000) White shoulder 0.714) m1 paraconid presence, 0 → 2; (74; 0.333) spot presence, 0 ⇒ 1; (9; 0.500) White neck spot Distal tip of clavicle attachment, 0 → 1; (78; presence, 0 ⇒ 1; (13; 0.667) Interramal vibrissae 0.500) M. triceps brachii caput mediale insertion, number, 4 ⇒ 0; (34; 1.000) Internal labial papillae 2 → 1; (80; 0.250) M. palmaris longus insertion distribution, 1 ⇒ 2; (50; 0.500) I1 occlusal margin on digit III, 1 ⇒ 0; (105; 1.000) Passage direction morphology, 0 ⇒ 3; (111; 0.333) Papillae distri- of m. sphincter colli profundus lateral slip, 0 → bution on pharyngeal tongue, 1 ⇒ 2; (124; 0.167) 1; (110; 0.333) Lateral circumvallate papillae lo- Central horny papillae number, 0 ⇒ 1; (126; cation, 0 → 1; (132; 0.667) Internal uterine fusion, 0.250) Posterior horny papillae presence, 0 ⇒ 1; 1 ⇒ 2; (133; 0.667) Location of oviductal en- (136; 0.533) Inferior colliculi coverage, 2 ⇒ 1; trance to uterus, 1 ⇒ 2. DELTRAN: (10; 0.364) (141; 1.000) Restriction site 38, 0 ⇒ 1. Uropatagial fringe presence, 0 ⇒ 1; (34; 1.000) Node 21: ACCTRAN: (2; 0.100) Bulb on hair Internal labial papillae presence, 0 ⇒ 1; (80; shaft, 0 → 1; (15; 1.000) Number lateral vibrissae, 0.250) M. palmaris longus insertion on digit III, 2 ⇒ 1; (44; 1.000) Posterior hard palate length, 1 1 ⇒ 0; (89; 0.333) Tail length, 0 → 2; (132; → 2; (47; 0.500) Rostral morphology, 0 → 1; (51; 0.667) Internal uterine fusion, 1 ⇒ 2; (133; 0.667) 0.500) Morphology of main cusp of pointed upper Location of oviductal entrance to uterus, 1 ⇒ 2. incisor, 0 → 1; (84; 0.278) Longest metacarpal, 4 Stenodermatini: ACCTRAN: (21; 0.286) Cen- ⇒ 0. DELTRAN: (15; 1.000) Number lateral vi- tral rib length, 1 → 2; (55; 0.800) Incisor occlu- brissae, 2 ⇒ 1; (84; 0.278) Longest metacarpal, 4 sion, 0 ⇒ 1; (61; 0.333) M1 hypocone presence, ⇒ 0. 1 ⇒ 0; (79; 0.500) M. palmaris longus insertion Node 22: ACCTRAN: (16; 0.500) Vibrissal pa- on digit II, 1 → 0; (88; 0.429) Plagiopatagium pillae morphology, 2 → 3; (21; 0.286) Central rib attachment location, 0 ⇒ 2; (90; 0.500) M. my- length, 2 → 1; (29; 0.667) Presence of outgrowth lohyoideus division, 0 ⇒ 2; (110; 0.333) Lateral posterior to spear, 0 → 2; (30; 0.333) Chin mor- circumvallate papillae location, 1 → 2; (116; phology, 1 ⇒ 2; (38; 1.000) Facial hood presence, 1.000) Medial-posterior mechanical papillae in- 0 ⇒ 1; (108; 1.000) Medial circumvallate papillae clination, 0 ⇒ 1; (119; 0.167) Basketlike papillae fusion, 0 ⇒ 1; (137; 0.571) Sex chromosomes, 1 presence, 1 ⇒ 0; (129; 0.800) Brunner's glands ⇒ 0. DELTRAN: (16; 0.500) Vibrissal papillae presence, 0 ⇒ 1; (130; 0.250) Male accessory morphology, 1 → 3; (30; 0.333) Chin morpholo- gland morphology, 0 → 1; (137; 0.571) Sex chro- gy, 1 ⇒ 2; (38; 1.000) Facial hood presence, 0 ⇒ mosomes, 0 ⇒ 1. DELTRAN: (1; 0.143) Pelage 1; (44; 1.000) Posterior hard palate length, 0 → differentiation, 0 → 1; (21; 0.286) Central rib 2; (108; 1.000) Medial circumvallate papillae fu- length, 1 → 2; (55; 0.800) Incisor occlusion, 0 ⇒ sion, 0 ⇒ 1; (137; 0.571) Sex chromosomes, 1 ⇒ 0. 1; (61; 0.333) M1 hypocone presence, 1 ⇒ 0; (68; Node 23: ACCTRAN: (25; 0.222) Labial 0.714) m1 paraconid presence, 0 → 2; (88; 0.429) horseshoe morphology, 2 ⇒ 1; (57; 0.400) P4 ac- Plagiopatagium attachment location, 0 ⇒ 2; (90; cessory cusp presence, 0 ⇒ 1; (143; 0.500) Re- 0.500) M. mylohyoideus division, 0 ⇒ 2; (105; striction site 46, 0 ⇒ 1. DELTRAN: (25; 0.222) 1.000) 0 → 1; (110; 0.333) Lateral circumvallate Labial horseshoe morphology, 2 ⇒ 1; (57; 0.400) papillae location, 0 → 2; (116; 1.000) Medial-pos- P4 accessory cusp presence, 0 ⇒ 1; (143; 0.500) terior mechanical papillae inclination, 0 ⇒ 1; Restriction site 46, 0 ⇒ 1. (119; 0.167) Basketlike papillae presence, 1 ⇒ 0; Node 24: ACCTRAN: (4; 0.400) Hair scale (129; 0.800) Brunner's glands presence, 0 ⇒ 1; margin shape, 0 ⇒ 2; (44; 1.000) Posterior hard (137; 0.571) Sex chromosomes, 0 ⇒ 1; (149; palate length, 1 → 3. DELTRAN: (4; 0.400) Hair 0.250) Restriction site 54, 1 → 0. scale margin shape, 0 ⇒ 2; (16; 0.500) Vibrissal Stenodermatina: ACCTRAN: (8; 1.000) papillae morphology, 1 → 2; (44; 1.000) Posterior White shoulder spot presence, 0 ⇒ 1; (9; 0.500) hard palate length, 0 → 3. White neck spot presence, 0 ⇒ 1; (13; 0.667) In- Node 25: ACCTRAN: (27; 1.000) Thickened terramal vibrissae number, 4 ⇒ 0; (16; 0.500) Vi- labial horseshoe morphology, 0 → 3; (45; 1.000) 200 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 248

Emarginate hard palate shape, 1 ⇒ 0; (69; 0.400) (124; 0.167) Large central horny papillae number, m1 metaconid presence, 0 ⇒ 1; (84; 0.278) Lon- 0 ⇒ 1; (138; 1.000) Restriction site 28, 0 ⇒ 1. gest metacarpal, 4 ⇒ 5. DELTRAN: (45; 1.000) DELTRAN: (57; 0.400) P4 accessory cusp pres- Emarginate hard palate shape, 1 ⇒ 0; (69; 0.400) ence, 0 ⇒ 2; (110; 0.333) Lateral circumvallate m1 metaconid presence, 0 ⇒ 1; (84; 0.278) Lon- papillae location, 2 → 1; (124; 0.167) Large cen- gest metacarpal, 4 ⇒ 5. tral horny papillae number, 0 ⇒ 1; (138; 1.000) Node 26: ACCTRAN: (9; 0.500) White neck Restriction site 28, 0 ⇒ 1. spot presence, 1 ⇒ 0. DELTRAN: (9; 0.500) Node 31: ACCTRAN: (10; 0.364) Uropatagial White neck spot presence, 1 ⇒ 0; (27; 1.000) fringe presence, 0 ⇒ 1; (50; 0.500) I1 occlusal Thickened labial horseshoe morphology, 0 → 3. margin morphology, 2 → 0; (68; 0.714) m1 para- Ectophyllina: ACCTRAN: (4; 0.400) Hair conid presence, 2 → 1; (111; 0.333) Papillae dis- scale margin shape, 0 ⇒ 2; (5; 0.214) Dorsal fur tribution on pharyngeal tongue, 1 ⇒ 0; (136; banding pattern, 2 → 1; (6; 0.333) Facial stripes 0.533) Inferior colliculi coverage, 2 → 1. DEL- presence, 0 ⇒ 1; (75; 0.250) Number of m. oc- TRAN: (10; 0.364) Uropatagial fringe presence, 0 ciptopollicalus muscle bellies, 0 → 1; (84; 0.278) ⇒ 1; (61; 0.333) M1 hypocone presence, 0 → 1; Longest metacarpal, 4 → 5; (118; 0.333) Long- (111; 0.333) Papillae distribution on pharyngeal tipped bi®d papillae presence, 0 ⇒ 1. DELTRAN: tongue, 1 ⇒ 0. (4; 0.400) Hair scale margin shape, 0 ⇒ 2; (6; Node 32: ACCTRAN: (5; 0.214) Dorsal fur 0.333) Facial stripes presence, 0 ⇒ 1; (118; 0.333) banding pattern, 1 → 2; (55; 0.800) Incisor occlu- Long-tipped bi®d papillae presence, 0 ⇒ 1. sion, 1 → 0; (63; 0.273) Presence of M3, 0 → 1; Node 27: ACCTRAN: (10; 0.364) Uropatagial (70; 0.667) m1 entoconid presence, 0 ⇒ 1; (72; fringe presence, 1 ⇒ 0; (25; 0.222) Labial horse- 0.250) P4±M1 diastema, 0 ⇒ 1. DELTRAN: (5; shoe morphology, 2 ⇒ 0; (50; 0.500) I1 occlusal 0.214) Dorsal fur banding pattern, 1 → 2; (61; margin morphology, 0 → 2. DELTRAN: (10; 0.333) M1 hypocone presence, 0 → 1; (70; 0.667) 0.364) Uropatagial fringe presence,1 ⇒ 0; (25; m1 entoconid presence, 0 ⇒ 1; (72; 0.250) P4± 0.222) Labial horseshoe morphology, 2 ⇒ 0; (84; M1 diastema, 0 ⇒ 1. 0.278) Longest metacarpal, 4 → 5. Node 33: ACCTRAN: (73; 0.250) M1-M2 dia- Artibeus: ACCTRAN: (150; 1.000) EcoRI de- stema, 0 ⇒ 1. DELTRAN: (50; 0.500) I1 occlusal ®ned satellite DNA repeat, 0 ⇒ 1. DELTRAN: margin morphology, 0 → 2; (55; 0.800) Incisor oc- (50; 0.500) I1 occlusal margin morphology, 0 → clusion, 1 → 0; (63; 0.273) M3 presence, 0 → 1; 2, (150. 1.000) EcoRI de®ned satellite DNA re- (73, 0.250) M1-M2 diastema, 0 ⇒ 1; (110; 0.333) peat, 0 ⇒ 1. Lateral circumvallate papillae location, 2 → 1. Node 28: ACCTRAN: (57; 0.400) P4 acces- Node 34: ACCTRAN: (66; 0.400) m3 pres- sory cusp presence, 0 ⇒ 1; (63; 0.273) M3 presence, 0 ⇒ 1. DELTRAN: (66; 0.400) m3 presence, 0 → 1. DELTRAN: (57; 0.400) P4 acces- ence, 0 ⇒ 1. sory cusp presence, 0 ⇒ 1. Node 35: ACCTRAN: (7; 0.500) Dorsal stripe Node 29: ACCTRAN: (7; 0.500) Dorsal stripe presence, 1 ⇒ 0; (29; 0.667) Presence of out- presence, 0 ⇒ 1; (13; 0.667) Interramal vibrissae growth posterior to spear, 0 → 1; (86; 0.385) Digit number, 4 ⇒ 0; (18; 1.000) Noseleaf color, 0 ⇒ IV second phalanx length relative to ®rst, 0 ⇒ 1; 2; (61; 0.333) M1 hypocone presence, 0 → 1; (69; (110; 0.333) Lateral circumvallate papillae loca- 0.400) m1 metaconid presence, 0 ⇒ 1; (71; 0.125) tion, 1 ⇒ 0; (136; 0.533) Inferior colliculi cov- P3±P4 diastema, 0 ⇒ 1; (95; 0.167) M. cerato- erage, 2 ⇒ 1. DELTRAN: (7; 0.500) Dorsal stripe hyoideus insertion on stylohyoid, 0 → 1; (110; presence, 1 ⇒ 0; (86; 0.385) Digit IV second pha- 0.333) Lateral circumvallate papillae location, 2 lanx length relative to ®rst, 0 ⇒ 1; (110; 0.333) → 1; (137; 0.571) Sex chromosomes, 1 ⇒ 0. Lateral circumvallate papillae location, 1 ⇒ 0; DELTRAN: (7; 0.500) Dorsal stripe presence, 0 (136; 0.533) Inferior colliculi coverage, 2 ⇒ 1. ⇒ 1; (13; 0.667) Interramal vibrissae number, 4 Ectophylla: ACCTRAN: (6; 0.333) Facial ⇒ 0; (18; 1.000) Noseleaf color, 0 ⇒ 2; (69; stripes presence, 1 → 0; (18; 1.000) Noseleaf col- 0.400) m1 metaconid presence, 0 ⇒ 1; (71; 0.125) or, 2 ⇒ 1; (50; 0.500) I1 occlusal margin mor- P3±P4 diastema, 0 ⇒ 1; (137; 0.571) Sex chro- phology, 2 ⇒ 0; (111; 0.333) Papillae distribution mosomes, 1 ⇒ 0. on pharyngeal tongue, 1 ⇒ 0. DELTRAN: (18; Node 30: ACCTRAN: (57; 0.400) P4 acces- 1.000) Noseleaf color, 2 ⇒ 1; (50; 0.500) I1 oc- sory cusp presence, 0 ⇒ 2; (75; 0.250) Number clusal margin morphology, 2 ⇒ 0; (111; 0.333) of m. occipitopollicalus muscle bellies, 1 → 0; Papillae distribution on pharyngeal tongue, 1 ⇒ 0.