Icaronycteris index (holotype; PU 18150) from the Green River Formation, Wyoming.

2 CONTENTS Abstract ...... 4 Introduction ...... 5 Relationships and Classi®cation of Eocene Bats: A Historical Overview ...... 12 Relationships Among Extant Lineages of Bats ...... 31 Goals of the Current Study ...... 40 Materials and Methods ...... 40 Taxonomic Sampling and Monophyly ...... 40 Outgroups ...... 42 The Data Set ...... 46 Characters Examined in Fossil Bats ...... 48 Skull and Dentition ...... 48 Anterior Axial Skeleton ...... 64 Pectoral Girdle ...... 70 Forelimb ...... 77 Posterior Axial Skeleton and Pelvis ...... 81 Hindlimb ...... 84 Completeness ...... 85 Methods of Phylogenetic Analysis ...... 86 Analysis of Character Transformations ...... 87 Phylogenetic Results ...... 88 Results of Analyses ...... 88 Analysis 1: All Characters, Fossil Taxa Excluded ...... 88 Analysis 2: All Characters, All Taxa ...... 89 Analysis 3: All Taxa, Characters Limited to Those Scored in Fossil Forms ...... 92 Summary ...... 94 Character Evolution in Early Chiropterans ...... 94 Character Transformations Associated with Basal Nodes ...... 94 Features Diagnosing the Microchiropteran Crown Group ...... 99 Character Transformations at Basal Nodes: A Functional Perspective ...... 101 Evolution of Echolocation and Foraging Strategies ...... 107 Timing of the Origin of Flight and Echolocation ...... 108 Vision and the Evolution Of Echolocation ...... 111 Foraging Ecology of Eocene Bats ...... 119 Evolution of Foraging Strategies: A Phylogenetic Perspective ...... 129 Classi®cation of Eocene Bats ...... 133 Conclusions and Summary ...... 139 Acknowledgments ...... 143 References ...... 143 Appendix 1: Specimens Examined ...... 170 Appendix 2; Character Descriptions ...... 172 Appendix 3: Data Matrix ...... 180 4 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

ABSTRACT The Eocene fossil record of bats (Chiroptera) and obstacle detection but not detection or track- includes four genera known from relatively com- ing of airborne prey. Owing to the mechanical plete skeletons: Icaronycteris, Archaeonycteris, coupling of ventilation and ¯ight, the energy costs Hassianycteris, and Palaeochiropteryx. Phyloge- of echolocation to ¯ying bats were relatively low. netic relationships of these taxa to each other and In contrast, the bene®ts of aerial insectivory were to extant lineages of bats were investigated in a substantial, and a more sophisticated low-duty-cy- parsimony analysis of 195 morphological char- cle echolocation system capable of detecting, acters, 12 rDNA restriction site characters, and tracking, and assessing airborne prey subsequent- one character based on the number of R-1 tandem ly evolved rapidly. The need for an increasingly repeats in the mtDNA d-loop region. Results in- derived auditory system, together with limits on dicate that Icaronycteris, Archaeonycteris, Has- body size imposed by the mechanics of ¯ight, sianycteris, and Palaeochiropteryx represent a se- echolocation, and prey capture, may have resulted ries of consecutive sister-taxa to extant microchi- in reduction and simpli®cation of the visual system ropteran bats. This conclusion stands in contrast as echolocation became increasingly important. to previous suggestions that these fossil forms Our analysis con®rms previous suggestions that represent either a primitive grade ancestral to both Icaronycteris, Archaeonycteris, Hassianycteris, Megachiroptera and Microchiroptera (e.g., Eochi- and Palaeochiropteryx used echolocation. Forag- roptera) or a separate clade within Microchirop- ing strategies of these forms were reconstructed tera (e.g., Palaeochiropterygoidea). A new higher- based on postcranial osteology and wing form, level classi®cation is proposed to better re¯ect hy- cochlear size, and stomach contents. In the con- pothesized relationships among Eocene fossil bats text of our phylogeny, we suggest that foraging and extant taxa. Critical features of this classi®- behavior in the microchiropteran lineage evolved cation include restriction of Microchiroptera to in a series of steps: (1) gleaning food objects dur- the smallest clade that includes all extant bats that ing short ¯ights from a perch using vision for ori- use sophisticated echolocation (Emballonuridae ϩ entation and obstacle detection; prey detection by Yinochiroptera ϩ Yangochiroptera), and formal passive means, including vision and/or listening recognition of two more inclusive clades that en- for prey-generated sounds (no known examples in compass Microchiroptera plus the four fossil gen- fossil record); (2) gleaning stationary prey from a era. perch using echolocation and vision for orienta- Comparisons of results of separate phylogenet- tion and obstacle detection; prey detection by pas- ic analyses including and subsequently excluding sive means (Icaronycteris, Archaeonycteris); (3) the fossil taxa indicate that inclusion of the fossils perch hunting for both stationary and ¯ying prey changes the results in two ways: (1) altering per- using echolocation and vision for orientation and ceived relationships among extant forms at a few obstacle detection; prey detection and tracking us- poorly supported nodes; and (2) reducing per- ing echolocation for ¯ying prey and passive ceived support for some nodes near the base of means for stationary prey (no known example, al- the tree. Inclusion of the fossils affects some char- though Icaronycteris and/or Archaeonycteris may acter polarities (hence slightly changing tree to- have done this at times); (4) combined perch hunt- pology), and also changes the levels at which ing and continuous aerial hawking using echolo- transformations appear to apply (hence altering cation and vision for orientation and obstacle de- perceived support for some clades). Results of an tection; prey detection and tracking using echo- additional phylogenetic analysis in which soft-tis- location for ¯ying prey and passive means for sta- sue and molecular characters were excluded from tionary prey; calcar-supported uropatagium used consideration indicate that these characters are for prey capture (common ancestor of Hassianyc- critical for determination of relationships among teris and Palaeochiropteryx; retained in Palaeo- extant lineages. chiropteryx); (5) exclusive reliance on continuous Our phylogeny provides a basis for evaluating aerial hawking using echolocation and vision for previous hypotheses on the evolution of ¯ight, orientation and obstacle detection; prey detection echolocation, and foraging strategies. We propose and tracking using echolocation (Hassianycteris; that ¯ight evolved before echolocation, and that common ancestor of Microchiroptera). The tran- the ®rst bats used vision for orientation in their sition to using echolocation to detect and track arboreal/aerial environment. The evolution of prey would have been dif®cult in cluttered en- ¯ight was followed by the origin of low-duty-cy- vionments owing to interference produced by cle laryngeal echolocation in early members of multiple returning echoes. We therefore propose the microchiropteran lineage. This system was that this transition occurred in bats that foraged in most likely simple at ®rst, permitting orientation forest gaps and along the edges of lakes and rivers 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 5 in situations where potential perch sites were ad- The evolution of continuous aerial hawking jacent to relatively clutter-free open spaces. Aerial may have been the ``key innovation'' responsible hawking using echolocation to detect, track, and for the burst of diversi®cation in microchiropteran evalute prey was apparently the primitive foraging bats that occurred during the Eocene. Fossils re- strategy for Microchiroptera. This implies that ferable to six major extant lineages are known gleaning, passive prey detection, and perch hunt- from Middle±Late Eocene deposits, and recon- ing among extant microchiropterans are second- struction of ghost lineages leads to the conclusion arily derived specializations rather than retentions that at least seven more extant lineages were min- of primitive habits. Each of these habits has ap- imally present by the end of the Eocene. parently evolved multiple times.

Only extensive phylogenetic analysis, based on as many suites of characters as pos- sible, and carried out in conjunction with adaptational and aerodynamic studies, can form the basis for reconstruction of evolutionary change. Padian (1987: 19) The behavior of fossil animals and the evolution of ¯ight will probably always be a subject of contention, and although we may never know for certain if we have found the right answers, we can always distinguish the possible from the impossible, the probable from the improbable. Norberg (1990: 268)

INTRODUCTION Bats ®rst appear in the fossil record in Ear- ne and Swisher, 1995). At least three addi- ly Eocene deposits of North America, Eu- tional skeletons of Icaronycteris were sub- rope, Africa, and Australia (table 1). The di- sequently discovered in the same deposits versity of Eocene bats is remarkableÐ24 (Novacek, 1985a, 1987; Habersetzer and genera are currently recognized, and new Storch, 1987; see appendix 1). Isolated teeth species are described almost every year. Of from Clarkforkian deposits in North America the eight bat genera that make their ®rst ap- were referred to cf. Icaronycteris sp. by Gin- pearance in the Early Eocene, almost half are gerich (1987), thus potentially extending the known from spectacular, nearly complete range of the genus to the Late Paleocene (®g. skeletons: Icaronycteris, Archaeonycteris, 1). Other fragmentary material suggests that Hassianycteris, and Palaeochiropteryx. Icaronycteris may have persisted in North Much of what is known (or hypothesized) America until the Gardenerbuttian (McKen- about the early evolution of bats is based on na and Bell, 1997), an early Bridgerian in- studies of these taxa (e.g., Jepsen, 1966, terval that ended approximately 50 Ma 1970; Russell and SigeÂ, 1970; Richter and (Woodburne, 1987; Woodburne and Swisher, Storch, 1980; Novacek 1985a, 1987; Haber- 1995). Only one species, Icaronycteris index setzer and Storch, 1987, 1988, 1989; Haber- Jepsen, 1966, is currently recognized in setzer et al., 1989, 1992, 1994; Norberg, North America. However, the geographic 1989; Storch, 1989). range of Icaronycteris may have extended Icaronycteris was described by Jepsen beyond North America to Europe. Russell et (1966) on the basis of a beautifully preserved al. (1973) recognized a new species (Icaron- skeleton from the Early Eocene Green River ycteris? menui) based on a large collection Formation of Fossil Basin, Wyoming (see of jaw fragments and isolated teeth from Ear- Frontispiece). A small bat with a long tail, ly Eocene (®g. 1; MP1 8±9) deposits in Icaronycteris had a wingspan of approxi- mately 30 cm and probably weighed 10±16 1 Mammal Paleogene (MP) reference levels are bio- g (Habersetzer and Storch, 1987; Norberg, stratigraphic intervals used to subdivide the Paleogene 1989). The holotype was collected from beds of Europe. These units, which are of unspeci®ed dura- that are Middle Wasatchian in age, approxi- tion, are based on terrestrial mammalian faunas. The MP mately 53 Ma (Woodburne, 1987; Woodbur- system is based on the assumption that the fossil record 6 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

France. However, the af®nities of this mate- recognized: Palaeochiropteryx tupaiodon rial remain questionable because of the frag- Revilliod, 1917 (®g. 2) and Palaeochiropte- mentary nature of available specimens and ryx spiegeli Revilliod, 1917 (Smith and lack of any diagnostic apomorphies shared Storch, 1981; Habersetzer and Storch, 1987). with the North American form. P. tupaiodon was the smaller of the two spe- Palaeochiropteryx, Archaeonycteris, and cies, probably weighing 7±10 g and having Hassianycteris are known principally from an estimated wingspan of 24±28 cm (Haber- the famous Early/Middle Eocene ``Grube setzer and Storch, 1987; Norberg, 1989). In Messel'' deposits (MP 11) near Darmstadt, contrast, P. spiegeli had an estimated wing- Germany. Hundreds of complete and partial span of 26±30 cm and probably weighed 10± skeletons of bats have been found at Messel, 18 g (Habersetzer and Storch, 1987; Nor- many preserved with stomach contents indi- berg, 1989). In addition to the Messel ma- cating that they were insectivores that had terial, isolated fragments referable to Palaeo- been foraging successfully just prior to death chiropteryx have also been reported from (Smith et al., 1979; Richter and Storch, 1980; Sparnacian, Cuisian, Lutetian, and Bartonian Richter, 1987; Habersetzer et al., 1992, (MP 10, 11±13, 16) deposits from elsewhere 1994). This unusual concentration of fossil in Europe (Russell et al., 1973, 1982; Savage bats that show no evidence for cause of death and Russell, 1983). may have resulted from release of poisonous Archaeonycteris is known from at least six gas that overcame bats foraging over the sur- skeletons from Messel, and two species are face of Lake Messel (Habersetzer and Storch, currently recognized based on this material: 1988; Habersetzer et al., 1992). Another pos- Archaeonycteris trigonodon Revilliod, 1917b sibility is that the bats died as a result of (®g. 3) and A. pollex Storch and Habersetzer, poisoning by toxic alkaloids from a blue- 1988 (Russell and SigeÂ, 1970; Smith and green algal bloom (see Pybus et al. [1986] Storch, 1981; Habersetzer and Storch, 1987; for a modern example). Storch and Habersetzer, 1988). A. trigonodon Palaeochiropteryx is by far the most com- probably had a wingspan of 32±37 cm and mon bat at Messel, accounting for almost may have weighed 17±27 g (Habersetzer and 75% of all bat ®nds (Habersetzer and Storch, Storch, 1987; Norberg, 1989). A. pollex was 1989; Habersetzer et al., 1992). More than slightly larger, with an estimated body 50 skeletons of Palaeochiropteryx are known weight of 30±35 g (Storch and Habsersetzer, from Messel, and two species are currently 1988). In addition to the Messel material, fragmentary specimens referable to Archaeo- nycteris have been reported from Sparnacian, consists of continuously evolving lineages; therefore, Cuisian, and possibly Bartonian deposits in separation between ancestor and descendant species (and France (MP 8±13, 16; Russell et al, 1973, the temporal range of each species) is completely arbi- 1982; Godinot, 1981; Savage and Russell, trary. In an effort to prevent this subjective element from 1983; Schmidt-Kittler, 1987). This material confounding biostratigraphy, the boundaries of the MP units were not de®ned by the ®rst and last appearance includes at least one additional species, Ar- of taxa; accordingly, they are not true biostratigraphic chaeonycteris brailloni Russell, Louis, and zones (Schmidt-Kittler, 1987). Each MP level has a ref- Savage, 1973 (MP 8±9; Schmidt-Kittler, erence locality whose fauna functions as the type for the 1987). Another species, Archaeonycteris re- reference level (Schmidt-Kittler, 1987). The method villiodi Russell and SigeÂ, 1970, is based on used to order MP levels not correlated to marine stages a partial dentition from Messel. Always con- is not clear. It appears to be based on faunal similarity sidered somewhat problematic, this form was and the evolutionary stage of the lineages within the transferred to Hassianycterididae as Hassi- respective faunas (Russel et al., 1982; Schmidt-Kittler, anycteris? revilliodi by Habersetzer and 1987). Ordering of the MP levels by radiometric ages, Storch (1987). Smith and Russell (1992) sub- magnetostratigraphy, and the superposition of strata has not been possible (Schmidt-Kittler, 1987). Although the sequently reported discovery of a skull of re- use of faunal evolutionary stage in developing the se- villiodi and con®rmed that this species quence of MP levels is subjective and of questionable should be placed in Hassianycteris. validity, it is currently the only option for placing Eu- Hassianycteris is known from at least 20 ropean fossil mammals in geologic time. skeletons from Messel. Three species are 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 7 currently recognized based on this material: cene chiropteran taxa, not just the four gen- Hassianycteris messelensis Smith and era examined in the present study. The tax- Storch, 1981; Hassianycteris magna Smith onomy and spelling of names are those of and Storch, 1981; and Hassianycteris revil- the original author(s) of each study. Question liodi Russell and SigeÂ, 1970 (Smith and marks associated with taxonomic names to Storch, 1981; Habersetzer and Storch, 1988; indicate uncertainty (e.g., Icaronycteris? Smith and Russell, 1992). Wingspans and menui) also re¯ect the usage of the original weights have been estimated for only those author(s). species known from relatively complete skel- Twenty-four genera of bats are now rec- etons (i.e., H. messelensis and H. magna). H. ognized from the Eocene, although some messelensis probably had a wingspan of 35± were originally described based on material 40 cm and weighed 25±45 g (Habersetzer from younger deposits (table 1). Several ear- and Storch, 1987; Norberg, 1989). H. magna ly Tertiary bat fossils were reported in the had a wingspan of 45±50 cm and may have literature prior to 1875, but most were iso- weighed 65 g, making it the largest known lated teeth or fragments of jaws referred to Eocene bat (Habersetzer and Storch, 1987; either Chiroptera, Rhinolophus,orVesperti- Norberg, 1989). A fourth species, Hassi- lio, the latter being a waste-basket taxon that anycteris joeli Smith and Russell, 1992, is once included most bats. Summaries of the known from a partial dentary with teeth from history of fossil bat discoveries prior to 1875 upper Ypresian (Early Eocene) deposits in were provided by Revilliod (1922) and Le- Belgium (Smith and Russell, 1992). gendre and Sige (1983). Most modern bat classi®cations are based RELATIONSHIPS AND CLASSIFICATION largely upon that of Dobson (1875: 345), OF EOCENE BATS: who was the ®rst to provide a comprehensive A HISTORICAL OVERVIEW classi®cation of extant bats ``arranged ac- cording to their natural af®nities.'' Dobson Phylogenetic relationships among Icaron- (1875) brie¯y considered the origins of Chi- ycteris, Palaeochiropteryx, Archaeonycteris, roptera, although he did not explicitly discuss and Hassianycteris have been the subject of any of the fossil material known at the time. considerable debate, as have their relation- Accompanying Dobsons (1875) account was ships to extant lineages. It has long been a ®gure ``illustrating the af®nities of the fam- thought that many Eocene fossils represent ilies and genera of Chiroptera, and probable early members of extant microchiropteran lines of descent from ancestral forms . . .'' families, but Icaronycteris, Palaeochiropte- (®g. 5). In this diagram, Dobson (1875) used ryx, Archaeonycteris, and Hassianycteris (of- Palaeochiroptera as a name for the group of ten referred to as the ``archaic'' Eocene bats) largely unknown fossil bats that he presumed have remained enigmatic. Most workers have were ancestral to all extant bat lineages. regarded some or all of these forms as rep- Schlosser (1887) named Pseudorhinolo- resentatives of an early grade of chiropteran phus and Vespertiliavus based on material evolution, but opinions have differed con- from the Late Eocene to Early Oligocene cerning their relationships to each other, Quercy phosphorite deposits in France (cur- other Eocene taxa, Megachiroptera, Micro- rently referred to MP 16±23; Crochet et al., chiroptera, and to extant microchiropteran 1981; Sige and Legendre, 1983; Schmidt- superfamilies. This uncertainty has been Kittler, 1987). Drawing on extensive com- re¯ected in classi®cations, which have var- parisons with extant genera, Schlosser (1887) ied considerably over the last few decades. suggested a close relationship between Pseu- Because ideas concerning relationships of dorhinolophus and Rhinolophus and noted Icaronycteris, Palaeochiropteryx, Archaeo- several points of similarity between Vesper- nycteris, and Hassianycteris developed in the tiliavus, species of Vespertilio (many now context of a rich body of literature on early placed in other genera and even other fami- Tertiary bats, we review here (in chronolog- lies; see below), and Taphozous. He subse- ical order) the history of classi®cation and quently concluded that Pseudorhinolophus phylogenetic hypotheses regarding all Eo- should be grouped with rhinolophids and 8 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Fig. 1. Chart showing geologic time scale, standard geologic ages based on marine strata, and the correlation of land mammal ages and levels. Data complied from Fahlbusch (1976), Russell et al. (1982), Savage and Russell (1983), Berggren et al. (1995), Woodburne and Swisher (1995), and McKenna and Bell (1997). Correlations of MP levels to marine strata of known age are based on the following: MP 7 (Hooker, 1991; Woodburne and Swisher, 1995); MP 10, 13, 14, and 15 (Russell et al., 1982; Aubry, 1983); and MP 21 (Hooker, 1992). Abbreviations: EU., European; Ma., milleannus, millions of years before present; N.A., North American. For a de®nition of MP levels, see footnote 1 on page 5. aAges according to Fahlbusch (1976). bAges according to Savage and Russell (1983). cThese MP reference levels are not correlated to marine strata; instead, they are placed in an approx- imate chronological order (based on Russell et al. [1982] and Schmidt-Kittler [1987]) between MP levels that can be correlated to marine strata.

Vespertiliavus with vespertilionids. Schlosser (1887) concluded that Pseudorhinolophus (1887) also described fragmentary material was more closely related to Phyllorhina (ϭ that he referred to Rhinolophus sp. and to a Hipposideros) than to Rhinolophus.Here- later Tertiary species of Rhinolophus de- ferred the isolated humerus described by scribed by Filhol (1872). Schlosser (1887) Schlosser (1887) to ``Taphozous(?),'' noting also discussed an isolated humerus that he that it was clearly different from the typical suggested was similar to that of rhinolophids. rhinolophid form. Weithofer (1887) also de- Weithofer (1887) discussed new material scribed several new taxa based on cranioden- of Pseudorhinolophus and suggested modi- tal material from the Quercy deposits, in- ®cations of Schlosser's (1887) description cluding Alastor (which he associated with and allocation of this taxon. Based on com- Pseudorhinolophus and Phyllorhina), Nec- parisons with extant material, Weithofer romantis (which he concluded was related to 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 9

Fig. 2. Palaeochiropteryx tupaiodon (SMF ME 10) from Messel, Germany. From Habersetzer and Storch (1987: ®g. 2). Photo by E. Pantak (Senkenbergmuseum).

Phyllostomidae), and a fossil species referred phora is somewhat more specialized. Winge to ``Rhinolophus(?).'' Recent collections (1892) also discussed an undescribed, almost have suggested that Necromantis is restricted complete skull of Vespertiliavus from the to the MP 17 fauna (Sige and Legendre, collections at Copenhagen. Several features 1983; Schmidt-Kittler, 1987). of this specimen led him to conclude that Winge (1892 [translated into English in Vespertiliavus is a relatively specialized Winge, 1941]) discussed the status of Pseu- member of Emballonuridae, closely related dorhinolophus and Alastor, noting that al- to Taphozous. He noted that this conclusion though some authors had treated these as was not surprising given the taxonomic his- genera separate from Phyllorhina (ϭ Hip- tory of species previously compared with posideros), few had noted any signi®cant dif- Vespertiliavus. Citing Schlossers (1887) ferences. He concluded that Palaeophyllo- original description of Vespertiliavus, Winge phora clearly belonged to the same group as (1892) observed that the ``vespertilionid'' Pseudorhinolophus, although Palaeophyllo- that Schlosser thought was most similar to 10 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Fig. 3. Archaeonycteris trigonodon (SMF 80/1379) from Messel, Germany. From Habersetzer and Storch (1987: ®g. 3). Photo by E. Pantak (Senkenbergmuseum).

VespertiliavusÐVespertilio alectoÐwas (1887) undoubtedly also represents Vesper- considered by Dobson (1878) to be conspe- tiliavus. ci®c with Emballonura monticola. Winge Meschinelli (1903) described Archaeop- further noted that the isolated humerus re- teropus based on a skeleton with a poorly ferred to ``Taphozous(?)'' by Weithofer preserved skull from late Oligocene deposits 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 11

Fig. 4. Hassianycteris messelensis (SMF ME 1414a) from Messel, Germany. From Habersetzer and Storch (1987: ®g. 5). Photo by E. Pantak (Senkenbergmuseum). 12 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

TABLE 1 Temporal and Geographic Distribution of Genera of Fossil Bats Known from Eocene Depositsa

First occurrence in Early Eoceneb: Ageina Russell, Louis, and Savage, 1973 Early Eocene, Europe and North America Archaeonycteris Revilliod, 1917b Early Eocene±Middle Eocene, Europe Australonycteris Hand, Novacek, Godthelp, and Archer, 1994 Early Eocene, Australia Dizzya SigeÂ, 1991a Early Eocene, Africa Eppsinycteris Hooker, 1996 Early Eocene, Europe Hassianycteris Smith and Storch, 1981 ?Early Eocene, Middle Eocene, Europe Honrovits Beard, SigeÂ, and Krishtalka, 1992 Early Eocene, North America Icaronycteris Jepsen, 1966 ?Late Paleocene, Early Eocene, ?Middle Eocene, North America; ?Early Eocene, Europe Palaeochiropteryx Revilliod, 1917b Early Eocene±Middle Eocene, Europe First occurrence in Middle Eocene: Cecilionycteris Heller, 1935 Middle Eocene, Europe Hipposideros Gray, 1831c Middle Eocene±Early Pliocene, Europe; Early Miocene±Recent, Africa; ?Pleistocene±Recent, Madagascar; Mid- dle Pleistocene±Recent, Asia; Late Pleistocene±Recent, East Indies; Recent, Australia and New Guinea Matthesia Sige and Russell, 1980 Middle Eocene, Europe Palaeophyllophora Revilliod, 1917b Middle Eocene±Early Miocene, Europe Rhinolophus LaceÂpeÁde, 1799 Middle Eocene±Recent, Europe; Middle Miocene±Recent, Africa; Early Pleistocene±Recent, Asia; Late Pleisto- cene±Recent, East Indies; Late Pleistocene±Recent, Australia and New Guinea Stehlinia Revilliod, 1919d Middle Eocene±Late Oligocene, Europe Vespertiliavus Schlosser, 1887 Middle Eocene±Middle Oligocene, Europe Wallia Storer, 1984e Middle Eocene, North America First occurrence in Late Eocene: Chadronycteris Ostrander, 1983 Late Eocene, North America Cuvierimops Legendre and SigeÂ, 1983 Late Eocene±Early Oligocene, Europe Necromantis Weithofer, 1887f Late Eocene, Europe Paraphyllophora Revilliod, 1922 Late Eocene and/or Early Oligocene, Middle Miocene, Europe Philisis SigeÂ, 1985 Late Eocene, Africa Vampyravus Schlosser, 1910g Late Eocene, Africa Vaylatsia SigeÂ, 1990h Late Eocene±?Early Miocene, Europe 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 13 in Italy. Although not explicit with regards Based on comparisons with extant pteropod- to classi®cation, Meschinelli (1903) clearly ids and phyllostomids of similar size, Schlos- regarded Archaeopteropus as an early me- ser (1911) suggested that Provampyrus gachiropteran. Although Archaeopteropus should be referred to Phyllostomidae. Later does not appear in Eocene or even early Oli- authors noted that Provampyrus and Vam- gocene deposits, this taxon (still known only pyravus appear to have been based on the from a single specimen that was destroyed in same specimen (e.g., SigeÂ, 1985; see below). World War II) has frequently been classi®ed Leche (1911) compared a relatively well- with Eocene bats and has had an important preserved skull of Pseudorhinolophus with impact on discussions of the early diversi®- extant specimens of Phyllorhina (ϭ Hippos- cation of Chiroptera (see below). ideros) and Rhinolophus, and concluded that In the course of a preliminary description Pseudorhinolophus should be considered of an early Tertiary fauna from the Jebel El congeneric with Phyllorhina. He also con- Qatrani Formation in Egypt, Schlosser curred with Winge's (1892) conclusion that (1910) recognized the existence of a large bat Vespertiliavus is an early member of Embal- based on a single humerus. He noted simi- lonuridae, probably a close relative of Ta- larities between this specimen and Vampyrus phozous. and Stenoderma, and he named the new form Andersen (1912) considered the status of Vampyravus. The next year, Schlosser (1911) Archaeopteropus in his monograph on Me- published a more detailed account of the gachiroptera. He noted that Meschinelli mammals from Fayum (including the Jebel (1903) had confused the third and ®fth digits El Qatrani Formation), and described Pro- of the wing; however, upon making this cor- vampyrus based on an isolated humerus. rection, Andersen (1912: xxxix) found that

←

a Unless otherwise noted, age ranges are from McKenna and Bell (1997); see Russell et al. (1982) and Savage and Russell (1983) for summaries of species associated with many different formations, ages, and faunas in the Paleogene. Isolated teeth of uncertain af®nities have been referred to Chiroptera based on material from Early and Middle Eocene deposits in France (``Eochiroptera incertae sedis, Fam. gen. et sp. indeÂt.,'' Crochet et al., 1988: 422; ``Eochiroptera, fam., gen., et sp. indet. 1 and 2,'' Marandat, 1991: 90; cf. Vesperiliavus, cf. Necromantis, Marandat et al., 1993), latest Early Eocene deposits of the Kuldana Formation in Pakistan (``Chiroptera indet. A and B''; Russell and Gingerich, 1981: 284, 286), and Middle Eocene Bridger Formation deposits in North America (``Chiroptera, Family uncertain, Undescribed genus and species''; McKenna et al., 1962: 27). An isolated tooth from Late Eocene deposits in Thailand was referred to Megachiroptera by Ducrocq et al. (1993). b Wyonycteris Gingerich, 1987, which was described as an Early Eocene palaeochiropterygid bat, is omitted from this list based on arguments developed by Hand et al. (1994). Wyonycteris is based on fragmentary dental remains that exhibit several features common to but not uniquely diagnostic of bats (Gingerich, 1987; Hand et al., 1994). Hand et al. (1994) noted the Wyonycteris seems to lack some putative dental synapomophies of early bats, and concluded that no convincing evidence supports assignment of this form to chiroptera. Similarly, Paradoxonycteris Revilliod, 1919 from the Late Eocene of Europe is omitted based on arguments presented by Russell and Sige (1970) and Sige (1976). As with Wyonycteris, in the absence of more complete material there seems little reason to refer Paradoxonycteris to Chiroptera. Zanycteris Matthew, 1917 from the Wasatchian of Colorado is omitted because it is now thought to be a picrodontid primate (Simpson, 1945; Romer, 1966; Savage and Russell, 1983; Carroll, 1988). c Includes Pseudorhinolophus Schlosser, 1887 and Alastor Weithofer, 1887. The range for Hipposideros (subgenus Pseudorhinolophus) is Middle Eocene±Middle Miocene, Europe (Hand and Kirsch, 1988). d Includes Nycterobius Revilliod, 1922, Paleunycteris Revilliod, 1922, and Revilliodia Simpson, 1945. See Handley (1955) for a discussion of the taxonomic history of Stehlinia and other names associated with Revilliod's 1919 and 1922 publications. Sige (1974) provided a thorough revision of the genus and con®rmed the synonymies. e Originally described as a proscalopid insectivoran, but transferred to Molossidae by Legendre (1985). f Includes Necromanter Lydekker, 1888 and Necronycteris Palmer, 1903. g Includes Provampyrus Schlosser, 1911, which is apparently an objective junior synonym of Vampyravus (SigeÂ, 1985). Vampyravus may be a senior synonym of Philisis, but this cannot be determined based on available material; see discussion in Sige (1985). h Age range following Hand and Kirsch (1998). 14 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Fig. 5. Dobson's (1875) view of bat relationships (redrawn from Dobson, 1875: unumbered foldout facing p. 350). The original caption for this ®gure read ``Diagram illustrating the af®nities of the families and genera of Chiroptera, and probable lines of descent from ancestral forms (Palaeochiroptera). The families are indicated by circles, the subfamilies by semicircles, and the relative position of both indi- cates their af®nity. In the same manner, the af®nity of the generic groups to each other, and to groups of other families, is indicated (as far as possible) by the relative position of the names of these groups in each circle . . . The position of the circle representing the Pteropidae is not intended (as in other cases) to indicate their descent from the Phyllostomidae, but to show their position with regard to the whole suborder Microchiroptera.'' We have retained Dobson's (1875) spellings of group names, many of which are no longer in common usage.

``the hand of Archaeopteropus was a genuine lostomid, its af®nities were questioned by Megachiropteran hand.'' subsequent authors (e.g. Winge, 1941), and Matthew (1917) described Zanycteris it was later recognized as a picrodontid pri- based on a partial maxilla from the Wasatch mate (Simpson, 1945; Romer, 1966; Savage Formation in Colorado. Although Matthew and Russell, 1983; Carroll, 1988). (1917) considered Zanycteris to be a phyl- Revilliod (1917b) described Palaeochi- 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 15 ropteryx and Archaeonycteris based on cra- cies in this genus. Based on comparisons niodental and skeletal material from Messel with extant forms, he concluded that Vesper- (MP 11; Schmidt-Kittler, 1987). He com- tiliavus is a relatively primitive emballonurid pared these forms with other Eocene and Re- that is either closely related to Taphozous or cent taxa, and proposed that Palaeochirop- ancestral to the family as a whole. teryx and Archaeonycteris be placed in their Finally, in part III of his monograph, Re- own families, Palaeochiropterygidae and Ar- villiod (1922) described Paleunycteris based chaeonycteridae. These family-group names on several dentary fragments with teeth from have formed the basis for most subsequent deposits at Quercy (MP 17±19; Crochet et classi®cations of archaic Eocene fossil bats. al., 1981; Sige and Legendre, 1983; Schmidt- Revilliod (1917a, 1920, 1922) published a Kittler, 1987) and Egerkingen (MP 14; three-part monograph on fossil bats of the Schmidt-Kittler, 1987), and described Para- Tertiary. A source of some confusion has doxonycteris based on a maxillary fragment been the fact that he published a preliminary with teeth from Mormont (MP 17±19; Sige note summarizing the entire monograph in and Legendre, 1983; Schmidt-Kittler, 1987). 1919, one year before part II was published Revilliod (1922) concluded that Paleunycter- and three years before part III was published. is was related either to vespertilionids This summary paper (Revilliod, 1919) in- (broadly de®ned to include natalids) or phyl- cluded the ®rst references to many named lostomids (de®ned as including mormoop- taxa described in later parts of the mono- ids), and that Paradoxonycteris resembled graph, but did not include the descriptions emballonurids and primitive phyllostomids. required to make all of these names avail- Apparently because of the fragmentary na- able. Handley (1955) resolved the taxonomic ture of the available material, he declined to problems introduced by this work by noting refer either genus to a particular family, al- that many names published in Revilliod though he placed both in Microchiroptera. (1919) were nomina nuda, but that descrip- Revilliod (1922) also described several taxa tions provided by Revilliod (1920, 1922) that he had mentioned but not described in subsequently made these names available his 1919 publication (see above and Handley, from the later dates. Most of the conclusions 1955). Initial descriptions were provided for described in Revilliod's (1919) ``Note preÂli- Palaeophyllophora and Paraphyllophora minaire'' were discussed in greater depth in (MP 18±19; Crochet et al., 1981; Schmidt- Revilliod (1922; see below). However, there Kittler, 1987), and a revised description was are some minor differences between the 1919 presented for Pseudorhinolophus (including and 1922 accounts. Revilliod (1919) sug- Alastor). Revilliod (1922) referred all of gested that Paleunycteris might represent a these genera (as well as the Eocene repre- distinct family ``des . . . PaleunycteÂrides,'' sentatives of Rhinolophus) to Rhinolophidae but by 1922 he had apparently decided sensu lato. He placed another new form, Nyc- against formally naming a separate family terobius Revilliod, 1922 from Quercy Sainte for this genus. Considering the origin of bats, Neboule (MP 18; Schmidt-Kittler, 1987), in Revilliod (1919) noted that Archaeonycteris ``Vespertilionides'' and noted a close resem- exhibited the most primitive dentition, one blance to Myotis. from which all other chiropteran dentitions Revilliod (1922) additionally reviewed all could be derived. of the other Tertiary bats considered in his In part II of his monograph, Revilliod earlier works (i.e., parts I and II of his mono- (1920) discussed fossil specimens of Rhino- graph; Revilliod, 1917a, 1920) and presented lophus, described two new species in this ge- ®nal conclusions regarding each taxon. He nus from Eocene deposits in France, and referred Archaeopteropus to Megachiroptera, similarly discussed and named two new spe- placed Necromantis in Megadermatidae, re- cies of Necromantis. Based on extensive ferred Zanycteris and Provampyrus to Phyl- comparisons with extant forms, he placed lostomidae, and retained Palaeochiropteryx Necromantis in ``MeÂgadermideÂs'' (ϭ Mega- and Archaeonycteris in their own families dermatidae). Revilliod (1920) also redescribed within Microchiroptera. With respect to Pa- Vespertiliavus and described three new spe- laeochiropteryx, he noted that ``Ce genre re- 16 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 preÂsente une famille eÂteinte isoleÂe que l'on own subfamily (Palaeophyllophorinae) with- ne peut rattacher aÁ aucune famille reÂcente.'' in Hipposideridae, Pseudorhinolophus in Archaeonycteris was noted to ``repreÂsentant Hipposiderinae, and Stehlinia and Revilliodia d'une famille eÂtient treÁs primitive'' (Revil- (a new name for Nycterobius) in Vesperti- liod, 1922: 177). lioninae. Simpson (1945) proposed a new Heller (1935) described Cecilionycteris subfamily (Archaeopteropodinae) for Ar- based on a series of skeletons and cranioden- chaeopteropus, which he placed in Pteropod- tal fragments from the Middle Eocene ``Gru- idae. Paleunycteris and Paradoxonycteris be Cecilie'' deposits (MP 13; Franzen and were referred to Chiroptera incertae sedis. Haubold, 1987) in Germany. He compared Savage (1951) described Notonycteris,a Cecilionycteris with specimens and descrip- phyllostomid from the Miocene La Venta tions of Palaeochiropteryx, Palaeophyllo- fauna of Colombia. In the course of his dis- phora, Paleunycteris, Paraphyllophora, Nec- cussion, Savage (1951: 362) noted that Pro- romantis, Vespertiliavus, and several extant vampyrus ``is clearly separable'' from phyl- taxa. Noting similarities with representatives lostomids based on location of the capitulum from several different family-level groups, of the humerus, which is in line with the Heller (1935) declined to refer Cecilionyc- shaft in Provampyrus but not in phyllostom- teris to any particular family, instead leaving ids. He did not comment further on the af- it incertae sedis within Chiroptera. ®nities of Provampyrus, but observed that Winge (1941) discussed Revilliod's Notonycteris was the only Tertiary represen- (1917a, 1917b, 1920, 1922) papers on early tative of Phyllostomidae. Tertiary bats of Europe and reevaluated the Handley (1955) addressed the nomencla- morphological data reported for Palaeochi- tural confusion resulting from Revilliod's ropteryx and Archaeonycteris. Winge (1941: (1919) publication, and discussed the status 304) concluded that of Nycterobius Revilliod, 1922 and Revil- liodia Simpson, 1945. He concluded that . . . there is hardly anything to prevent us from re- both are junior synonyms of Stehlinia Revil- garding Palaeochiropteryx as belonging to the ves- liod, 1919. pertilionids, as one of the primitive forms, most likely of the nataline group. Friant (1963) revised the Tertiary rhino- lophids, referring Eocene Rhinolophus to Winge (1941) found that Archaeonycteris Rhinolophinae, and Palaeophyllophora, Par- was too poorly known to assess its af®nities aphyllophora, and Pseudorhinolophus to with other chiropteran taxa, but attributed Ar- Hipposiderinae. She also reviewed the high- chaeopteropus to Pteropodidae. He also er-level classi®cation of Microchiroptera, commented on Provampyrus, noting that recognizing two families of Eocene bats, Ar- while the holotype humerus resembles that of chaeonycteridae and Palaeochiropterygidae phyllostomids, a similar form might be ex- (Friant, 1963). Paleunycteris and Paradox- pected in primitive genera of other families onycteris were placed incertae sedis in Mi- such as Vespertilionidae and Rhinolophidae. crochiroptera. Simpson (1945) considered the af®nities Romer (1966) summarized the classi®ca- of Eocene bats in his in¯uential classi®cation tion of fossil bats and placed many Eocene of mammals. He (1945: 60) placed Archaeo- and Oligocene fossils in extant families follow- nycteris in Archaeonycteridae and Palaeo- ing the recommendations of previous authors. chiropteryx in Palaeochiropterygidae, and re- He placed Vespertiliavus in Emballonuridae, ferred both families to Microchiroptera, ``Su- Necromantis in Megadermatidae, Eocene perfam. uncertain.'' Cecilionycteris was re- Rhinolophus in Rhinolophidae, Palaeophyl- ferred to ?Palaeochiropterygidae incertae lophora, Paraphyllophora, Hipposideros, sedis, and several Eocene fossil forms were and Pseudorhinolophus in Hipposideridae, placed in extant microchiropteran families Provampyrus and ?Vampyravus in Phyllo- and subfamilies: Vespertiliavus in Emballon- stomidae, Stehlinia in Vespertilionidae, and uridae, Necromantis in Megadermatidae, Eo- Archaeopteropus in Pteropodidae. He rec- cene Rhinolophus in Rhinolophidae, Palaeo- ognized two archaic families, Archaeonyc- phyllophora and Paraphyllophora in their teridae (including only Archaeonycteris) and 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 17

Palaeochiropterygidae (including Palaeochi- under the heading ``Fossil Chiroptera of un- ropteryx and ?Cecilionycteris). Both of these certain status'' (Koopman and Jones, 1970: families were left in ``Superfamily uncer- 28). Paleunycteris and Paradoxonycteris tain'' within Microchiroptera (Romer, 1966: were referred to Chiroptera incertae sedis, 382). Paradoxonycteris was referred to Chi- and several Eocene taxa were placed in mod- roptera incertae sedis (Romer, 1966). ern families: Vespertiliavus in Emballonuri- Jepsen (1966) described Icaronycteris dae, Necromantis in Megadermatidae, and from a single, beautifully preserved skeleton Stehlinia in Vespertilionidae (Koopman and from Fossil Basin in the Green River For- Jones, 1970). Palaeophyllophora and Para- mation of Wyoming. He placed this form in phyllophora were placed in their own tribe the new family Icaronycteridae within Mi- Palaeophyllophorini in Hipposiderinae crochiroptera. Jepsen discussed other puta- (Koopman and Jones, 1970). Archaeoptero- tive bats from early Tertiary deposits in pus was placed in Archaeopteropodinae North America, but did not make any de- within Megachiroptera. tailed comparisons with Eocene bats from Slaughter (1970) discussed evolution of Europe. the chiropteran dentition, drawing conclu- Sige (1968) described the bats of an Early sions about the af®nities of many fossil taxa Miocene fauna from Bouzigue, France, in- based on their dental morphology. Slaughter cluding a new species of Pseudorhinolophus. (1970: 59) observed that the dentition of Ica- He reviewed the contents and stratigraphic ronycteris is ``more like the ancestral con- range of Pseudorhinolophus and presented a dition than any other known chiropteran.'' revised diagnosis of this taxon, which he He went on to note that Cecilionycteris, Pa- considered to represent a subgenus of Hip- laeochiropteryx, and Archaeonycteris have posideros. He suggested several possibilities slightly more derived dentitions, but did not for relationships among Pseudorhinolophus, associate them with any extant lineages. He Brachyhipposideros (a subgenus known from observed that Necromantis is clearly a me- Oligocene and Miocene deposits), and mod- gadermatid, although somewhat less derived ern species of Hipposideros, but he reached than extant members of the family. Slaughter no de®nitive conclusions other than that placed Pseudorhinolophus on the lineage these forms are very closely related. leading to modern Hipposideros, and recog- Jepsen (1970) published detailed stereo- nized Palaeophyllophorinae as close relatives photographs of Icaronycteris in an essay of hipposiderines. The ``Eocene Rhinolo- concerning the evolution of bats and pow- phus'' were placed at the base of the lineage ered ¯ight. However, the text of his publi- leading to modern Rhinolophus, palaeophyl- cation contained little information on Icaron- lophorines, and hipposiderines (Slaughter, ycteris that was not available in the original 1970: ®g. 5), with Nycteris shown as the sis- description (Jepsen, 1966), and no compari- ter-group to this rhinolophid clade. He also sons were drawn between Icaronycteris and recognized Vespertiliavus as the oldest mem- Eocene bats from Europe. Concerning the re- ber of Emballonuridae, noting that the pro- lationships of Icaronycteris to other bats, totypic dentition for the superfamily Embal- Jepsen (1970: 40) noted only that ``Icaron- lonuroidea (which he de®ned as including ycteris index as a species may have been di- noctilionids) would be like that of Vesperti- rectly ancestral to all or to some living mi- liavus except for the form of the hypocone. crobats or megabats or to none of our con- Finally, Slaughter noted that Stehlinia ap- temporary chiropts.'' pears to be intermediate between ancestral In a short paper on the classi®cation of vespertilionids and the lineage leading to all bats, Koopman and Jones (1970) recognized vespertilionids except miniopterines and mu- three families of Eocene bats: Icaronycteri- rinines. dae (including only Icaronycteris), Archaeo- Russell and Sige (1970) were among the nycteridae (Archaeonycteris), and Palaeochi- ®rst to present detailed comparisons of the ropterygidae (Palaeochiropteryx and Ceci- Messel bats with extant taxa. Their study lionycteris, the latter of which was placed in- concluded with an updated classi®cation of certae sedis). These families were listed the archaic Eocene bats and emended diag- 18 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Fig. 6. Russell, Louis, and Savage's hypothesis of phylogeny and divergence times of Eocene bats (redrawn from Russell et al., 1973: ®g. 11). The original caption for this ®gure read ``Hypothetical relationships between the earliest European bats.'' noses for each taxon. They placed the archaic prevented a con®rmed identi®cation. Hand Eocene genera in a new superfamily Palaeo- (1990) referred this specimen to Tadaridinae chiropterygoidea within Microchiroptera. indet. (Tadaridinae was named by Legendre Two families were recognized within this [1984] in a review of extant molossids). group: Palaeochiropterygidae and Icaronyc- Russell et al. (1973) described fragmen- terididae. Icaronycterididae was de®ned as tary dental remains from the Early Eocene of including Icaronycteris and possibly Ar- France (MP 8±9; Schmidt-Kittler, 1987) that chaeopteropus. In contrast, Palaeochiropter- they referred to new species of Icaronycter- ygidae was divided into two subfamilies, is? and Archaeonycteris as well as a new ge- Palaeochiropteryginae (including Palaeochi- nus, Ageina. Based on dental morphology, ropteryx and Cecilionycteris) and Archaeo- they hypothesized that Icaronycteris? menui nycteridinae (containing only Archaeonyc- was closely related to Palaeochiropteryx. Ar- teris). chaeonycteris was recognized as a distinct Sige (1971) described an isolated forelimb yet related lineage, while Ageina was placed from Stampien deposits in France (MP 21± on a distant branch of uncertain af®nities 24; Russell et al., 1982; Schmidt-Kittler, (®g. 6; Russell et al., 1973: ®g. 11). Al- 1987) and referred it to Tadarida sp. Al- though classi®cation was not explicitly dis- though this specimen may represent Cuvier- cussed, they placed Icaronycteris? menui in imops, lack of associated dental material has Palaeochiropterygoidea: Icaronycteridae?, 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 19

Fig. 7. Smith's hypotheses of bat relationships (redrawn from Smith, 1976: ®g. 1). The original caption for this ®gure read ``A, cladogram of the generally accepted view of chiropteran phylogeny with the Microchiroptera dervied from a common emballonuridlike ancestry. B, an alternative proposal for chiropteran evolution with several microchiropteran lineages being derivied, independently, from a world-wide paleochiropteran grade and the Megachiroptera (Pteropodidae) derived either separately from an insectivorous stock or from the paleochiropteran grade. a, Emballonuroidea; b, Rhinolophoidea; c, Phyllostomatoidea; d, Vespertilionoidea.'' Note that Smith (1976) used ``Myzapodidae'' instead of ``My- zopodidae,'' and ``Phyllostomatoidea'' instead of ``Phyllostomoidea'' (ϭ Noctilionoidea).

Archaeonycteris brailloni in Archaeonycter- rial of Stehlinia and compared it to other fos- inae (family not speci®ed), and Ageina in sil and extant taxa. Based on his analysis, ``Family uncertain'' (Russell et al., 1973: 35). Sige recognized Nycterobius, Paleunycteris, Sige (1974) reviewed the available mate- and Revilliodia as junior synonyms of Steh- 20 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 linia, and concluded that Stehlinia was best malian faunas of Europe. He referred to the placed in the family Kerivoulidae within Messel bats as ``chiropteÁres paleÂochiropteÂr- Vespertilionoidea. He compared Stehlinia ygoõÈdeÂs,'' including within this group Icaron- with ``les paleÂochiropteÂrygoõÈdeÂs'' but did not ycterididae and Palaeochiropterygidae. The comment on contents or classi®cation of this latter family contained two subfamilies: Ar- group. chaeonycteridinae and Palaeochiropterygi- Smith (1976) discussed Eocene fossil bats nae. Vespertiliavus was referred to Embal- in a review of phylogenetic relationships and lonuridae, Stehlinia to Vespertilionoidea patterns of evolutionary radiation of extant (Kerivoulidae), Hipposideros (Pseudorhino- bats (®g. 7). Apparently drawing on Dob- lophus) and Palaeophyllophora to Rhinolo- son's (1875) concept, Smith (1976: 52±53) phoidea (Hipposideridae), and Necromantis used the informal name ``Paleochiroptera'' to to Megadermatidae. In the absence of con- collectively refer to Icaronycteris, Palaeo- tradictory evidence, Sige suggested that pa- chiropteryx, Archaeonycteris, Cecilionycter- laeochiropterygines, emballonuroids, vesper- is, and Ageina. He perceived this group as tilionoids, and rhinolophoids might have representing ``a world-wide paleochiropteran originated from unspeci®ed endemic Euro- grade in early to middle Eocene times . . . pean groups. Based on its supposed ``denture [that] was primitive and generalized in most apparemment insectivore'' (insectivorous respects.'' He suggested that extant Micro- dentition), Sige (1977: 185) interpreted Ar- chiroptera could be easily derived from the chaeopteropus as a form intermediate be- paleochiropteran grade, and that this diver- tween megachiropterans and icaronycterids. gence probably occurred in the Paleocene. In a review paper concerning the evolution He clearly did not think that any of the Eo- of bat ¯ight, Smith (1977) noted that one of cene taxa listed above were ancestral to ex- the ``®ve monophyletic superfamilies'' of tant microchiropterans. Smith further sug- Microchiroptera was Palaeochiropterygo- gested that Archaeopteropus was probably idea, which included Palaeochiropterygidae also derived from the paleochiropteran grade, and Icaronycteridae. While Russell and Sige although he questioned the megachiropteran (1970) had tentatively referred Archaeopter- af®nities of this form. Smith (1976: 54) ob- opus to Icaronycteridae, Smith (1977) noted served that ``The distinctness and marked de- that he had examined the only known spec- parture of megachiropteran dentition from imen of Archaeopteropus and had found no that of the Microchiroptera, as well as from known Tertiary fossils, suggests to me that justi®cation for this arrangement. According- this group of bats had their origin much ear- ly, he restricted Icaronycteridae to only Ica- lier in the paleochiropteran grade or perhaps ronycteris. To our knowledge, Smith was the . . . separately from an insectivorous ances- ®rst to suggest that Icaronycteris, Palaeo- tral stock.'' chiropteryx, Archaeonycteris, Cecilionycter- Smith (1976) also discussed other Eocene is, and Ageina formed a clade (rather than a fossils that he thought could be referred to grade) distinct from other microchiropteran extant families. These included Vespertilia- bats. vus (referred to Emballonuridae); Palaeo- Barghoorn (1977) described two new phyllophora, Paraphyllophora, Rhinolophus, skulls of Vespertiliavus from the Quercy Pseudorhinolophus, and Hipposideros (Rhin- Phosphorites in southern France. Based on a olophidae); Necromantis and Provampyrus cladistic analysis of cranial characters in ex- (Megadermatidae); and Stehlinia and Nycter- tant and fossil emballonurids and several out- obius (Vespertilionidae). The presence of groups, Barghoorn concluded that Vesperti- modern families (and even genera) in Middle liavus is the sister-group of a clade com- and Late Eocene deposits was interpreted as prised of Taphozous and Saccolaimus. additional evidence for an early divergence Butler (1978) reviewed the evolutionary of microchiropterans from the paleochirop- history of bats in Africa and provided a clas- teran stock (Smith, 1976). si®cation of fossil forms known from that Sige (1977) discussed the fossil record of continent. In this context, he placed Vampy- Eocene bats in a review of Paleogene mam- ravus in Microchiroptera incertae sedis, ex- 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 21

Fig. 8. Van Valen's (1979) hypothesis of bat relationships (redrawn from Van Valen, 1979: ®g. 1). The original caption for this ®gure read ``Phylogeny of the known families of bats . . . Infraorders, superfamilies, and families of the Microchiroptera are separated by solid, dashed, and dotted lines respectively.'' The numbers refer to taxa discussed in an appendix; we reproduce these here because they are informative with respect to Van Valen's unconventional concepts of group membership: 1, Megachiroptera, Pteropodidae; 2, Microchiroptera, undiscovered family; 3, Vespertilionia, Vespertilion- oidea, Natalidae; 4, Recent Natalidae; 5, Vespertilionidae; 6, Recent Vespertilionidae; 7, Mystacinidae; 8, Recent Mystacinidae; 9, Molossidae; 10, Phyllostomatia, Rhinopomatoidea, undiscovered Family; 11, Emballonurid±rhinolophoid stem; 12, Rhinolophoidea; 13, Megadermatidae; 14, Nycteridae; 15, Rhin- olophidae; 16, Rhinopomatidae; 17, Emballonurid±noctilionoid stem; 18, Emballonuridae (family should perhaps extand as far back as 11); 19, Recent Emballonuridae; 20, Craseonycterididae; 21, Noctilion- oidea, perhaps early Mormoopidae; 22, Noctilionidae; 23, Mormoopidae; 24, Recent Mormoopidae; 25, Phyllostomatidae; 26, Recent Phyllostomatidae; 27, Desmodontidae. Spelling of group names follows those of the original author. plicitly rejecting the conclusions of Schlosser suborder Eochiroptera for the archaic Eocene (1911) and Smith (1976). bats, within which he recognized one family Van Valen (1979) published the ®rst cla- (Palaeochiropterygidae). The latter group distic assessment of higher-level bat relation- was described as ``including Archaeonyc- ships (®g. 8). This study was based on di- ter(id)idae and probably Icaronycter(id)idae verse morphological characters, although and Archaeopteropodinae as subfamilies'' speci®c methods of analysis were not de- (Van Valen, 1979: 109). Van Valen indicated scribed and a taxon±character matrix was not that Eochiroptera was ancestral to all extant included. The focus of Van Valen's study was bats including Megachiroptera and Micro- extant bats, but he also discussed some fea- chiroptera. He noted that extant megachirop- tures of the better known fossil forms as re- terans were probably derived from persistent ported in the literature. He proposed the new eochiropterans such as Archaeopteropus. The 22 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 criteria used to lump forms ancestral to dif- Legendre and Sige (1983) went back to ferent clades into one group (Eochiroptera) one of the ®rst fossil bats discussed in the appears to have been perceived adaptive sim- scienti®c literatureÐCuvier's (1822) ``Ves- ilarity and their shared retention of numerous pertilion de Montmartre''Ðand found that primitive traits. Van Valen (1979: 109) ac- additional preparation of the original speci- knowledged that many of the taxa in his clas- men (a partial skeleton) revealed features di- si®cation were paraphyletic, which he noted agnostic of Molossidae. This form, which ``re¯ects the progressive evolution of grades was variously referred to Vespertilio pari- of adaptation.'' siensis, Serotinoides antiquus,``Vespertilio'' Sige and Russell (1980) provided detailed serotinoides, and cf. Tadarida sp. by previ- descriptions and comparisons of Palaeochi- ous 19th-century authors, was named Cu- ropteryx tupaiodon and Cecilionycteris, and vierimops by Legendre and Sige (1983). It is described a new genus Matthesia based on currently thought to be Late Eocene (Pria- dental material from Geiseltal, ``Grube Ce- bonian) in age (MP 19; Russell et al., 1982; cilie'' (MP 13; Franzen and Haubold, 1987). Schmidt-Kittler, 1987). Legendre and Sige All of these forms were placed in Eochirop- (1983) suggested that Cuvierimops was pos- tera: Palaeochiropterygidae: Palaeochiropter- sibly ancestral to a large complex including yginae by Sige and Russell (1980). all extant molossids with the exception of Smith and Storch (1981) described the Tadarida and Mormopterus. new genus Hassianycteris from Messel (MP Sige and Legendre (1983) reviewed the 12; Schmidt-Kittler, 1987). Hassianycteris fossil record of bats in different depositional was described as differing signi®cantly from environments and regions in the Mediterra- palaeochiropterygoids in possessing an ad- nean basin. They referred Icaronycteris, Pa- vanced degree of dental reduction and other laeochiropteryx, Cecilionycteris, Matthesia, derived dental and osteological features. Archaeonycteris, and Ageina to Eochirop- tera. Hassianycteris was described as an ear- Smith and Storch (1981: 164) concluded that ly microchiropteran not clearly related to any Hassianycteris ``is more closely associated of the extant superfamilies. Several addition- with the emballonuroid/rhinolophoid section al Eocene forms were referred to extant fam- of the Microchiroptera than to any other ilies, including Vespertiliavus, Hipposideros group.'' As a result, they referred Hassianyc- (Pseudorhinolophus), Palaeophyllophora, teris to Microchiroptera incertae sedis rather Necromantis, Rhinolophus, Stehlinia, Cu- than placing it in Palaeochiropterygoidea. vierimops, and cf. Tadarida. Although fam- Following many earlier authors, Smith and ily af®nities were not speci®ed, these refer- Storch (1981) recommended removal of Ar- ences were clearly intended to follow past chaeopteropus from Palaeochiropterygoidea classi®cations of these taxa. and subsequent placement in Pteropodidae Ostrander (1983) described Chadronycter- based on morphological evidence that Ar- is based on a maxillary fragment with teeth chaeopteropus is an early megachiropteran from what are now considered to be Late Eo- (for details of this argument, see Habersetzer cene (Chadronian) deposits in Nebraska. and Storch, 1987). In doing so, Smith and Based on comparisons with Stehlinia (re- Storch explicitly restricted Palaeochiropter- ferred to Kerivoulidae by SigeÂ, 1974) and ygoidea to Palaeochiropteryx, Archaeonyc- ``Palaeochiropterygidae,'' Ostrander referred teris, Icaronycteris, Cecilionycteris, Ageina, Chadronycteris to Vespertilionoidea: Keri- and Matthesia. They (1981: 163) noted the voulidae. following concerning possible relationships Gupta (1984) discussed bat phylogeny and of Palaeochiropterygoidea as thus de®ned: relationships of Icaronycteris. Apparently failing to draw any distinction between sym- While we would not now go so far as including these plesiomorphy and synapomorphy, Gupta bats in the family Vespertilionidae Gray 1821, they suggested that Icaronycteris may be a me- may well share a close sister-group relationship with the superfamily Vespertilionoidea or they may be gachiropteran based on several shared prim- placed as the sister-group of a larger one including itive features (e.g., presence of a claw on the the Phyllostomatoidea and Vespertilionoidea. index ®nger). Gupta (1984: 42) also pro- 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 23

Fig. 9. Cladogram of interfamilial relationships inferred from auditory characters (redrawn from Novacek, 1980a: ®g. 1). The original caption for this ®gure read ``Wagner tree resulting from analysis of 18 auditory characters for Recent chiropteran families. Craseonycteridae and Myzopodidae are omit- ted. Horizontal bars bridging two or more branches are synapomorphies; short black bars are indepen- dently derived characters under this arrangement; striped bars represent character state reversals.'' posed an unusual hypothesis concerning the ®cation of all chiropteran genera in their origin of bats: book on the natural history of bats. They rec- ognized three families in Palaeochiroptery- The close similarity in the structure of the patagium of Pterosaurs and of bats, and the presence of hair in goidea: Palaeochiropterygidae (including both, can compel us to think that the bats might have Palaeochiropteryx, Cecilionycteris, and evolved from the Pterosaurs with modi®cations of Matthesia), Archaeonycteridae (including mammalian characters during 12 million years [of Archaeonycteris and Ageina), and Icaronyc- the] Paleocene, the period during which no fossils [of teridae (including only Icaronycteris). Par- pterosaurs or bats] have so far been reported. adoxonycteris and Hassianycteris were left None of these ideas have been accepted by as ``Family incertae sedis'' within Microchi- subsequent authors because all published roptera (Hill and Smith, 1984: 221). Follow- data sets indicate that bats are therian mam- ing Smith and Storch (1981), Hill and Smith mals (e.g., Novacek, 1980a, 1986, 1990; No- (1984) suggested that Palaeochiropterygo- vacek and Wyss, 1986; Simmons, 1993a, idea appeared to have af®nities with Vesper- 1994, 1995). Pterosaurs are almost univer- tilionoidea, while Hassianycteris might be sally regarded as archosaurian diapsid rep- associated with Emballonuroidea and Rhin- tiles that are more closely related to croco- olophoidea. Many Eocene and Oligocene diles and birds than to mammals (e.g., Pa- fossils were referred to extant families by dian, 1984, 1987; Gauthier, 1986). Hill and Smith (1984). They placed Ar- Hill and Smith (1984) provided a classi- chaeopteropus in its own subfamily Ar- 24 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 chaeopteropodinae within Pteropodidae, Ves- ments in the text suggest that Sige included pertiliavus in Emballonurinae within Embal- Eochiropterato represent the basal stock of lonuridae, Necromantis and Provampyrus in Chiroptera. Megadermatidae, Pseudorhinolophus in Hip- HoraÂcek (1986) discussed relationships of posiderinae within Hipposideridae, Palaeo- Stehlinia and Kerivoula and concluded (con- phyllophora and Paraphyllophora in Palaeo- tra SigeÂ, 1974) that these taxa were not close- phyllophorinae within Hipposideridae, and ly related. Instead, HoraÂcek suggested that Stehlinia in Vespertilioninae within Vesper- Stehlinia might be related to Miniopterinae. tilionidae. Mein and Tupinier (1986) brie¯y reviewed Legendre (1985) reviewed the fossil rec- the early Tertiary record of bats in a short ord of molossids and reevaluated the af®ni- paper on the evolution of echolocation sys- ties of Wallia, a taxon originally described tems. They followed Van Valen (1979) in by Storer (1984) as a proscalopid insectivore considering Icaronycteris, Archaeonycteris, from the Uintan (Middle Eocene) Swift and Palaeochiropteryx as representatives of Creek Local Fauna of Saskatchewan, Cana- Eochiroptera, while referring other fossil da. Legendre concluded that Wallia probably taxa to extant superfamilies or families. Mein represents the oldest known molossid, and he and Tupinier listed Vespertiliavus in Embal- placed both Wallia and Cuvierimops in Tad- lonuridae, Stehlinia in Vespertilionoidea, and aridinae, the subfamily to which all Tertiary referred Necromantis and possibly Provam- fossil molossids have been referred (Hand, pyrus to Megadermatidae. 1990). Gingerich (1987) described Wyonycteris Sige (1985) focused on fossil bats from based on a partial dentary and isolated teeth Late Eocene deposits of the Fayum area, from Late Paleocene (Clarkforkian) deposits Egypt, and redescribed and discussed the sta- of the Willwood Formation of Wyoming. He tus of Vampyravus Schlosser, 1910 and Pro- referred Wyonycteris to Palaeochiropterygi- vampyrus Schlosser, 1911. He concluded that dae on the basis of dental similarities, but these two taxa are objective synonyms based subsequent authors (e.g., Habersetzer et al., on a single specimen, a well-preserved hu- 1994; Hand et al., 1994) have questioned the merus. Sige (1985) concluded that Vampy- chiropteran af®nities of this form (see foot- ravus was too poorly known to be assigned notes to table 1). Gingerich also described to any family, but suggested that morphology teeth referred to cf. Icaronycteris from the of the holotype indicated af®nities with ei- Willwood Formation. ther hipposiderids, natalids, or phyllostom- Habersetzer and Storch (1987) discussed ids. Sige also described a new taxon based in detail the classi®cation and functional on craniodental material, Philisis, which he morphology of Paleogene bats, concentrating placed in its own family, Philisidae. Based on the fossil bats from Messel. They argued on comparisons with fossil and extant forms, against usage of Palaeochiropterygoidea and Sige placed Philisidae within Vespertiliono- Eochiroptera, recognizing that each was idea and suggested that it was more closely probably paraphyletic even if Archaeoptero- related to Vespertilionidae sensu lato than to pus was removed. Instead, they recognized Molossidae (®g. 12). Speculating that Vam- and provided revised diagnoses for three pyravus and Philisis might eventually be families placed in Microchiroptera incertae shown to be conspeci®c if more complete sedis: Archaeonycterididae (including Ar- material were to be discovered, Sige (1985) chaeonycteris and Icaronycteris), Palaeochi- hypothesized that Vampyravidae (which ropterygidae (Palaeochiropteryx), and a new would replace Philisidae as the correct family family, Hassianycterididae (Hassianycteris, name in this case) would prove to be more including ``Archaeonycteris'' revilliodi Rus- closely related to Natalidae than to Vesper- sell and SigeÂ, 1970). The relationships of Ce- tilionidae or Molossidae. In each of SigeÂ's cilionycteris, Ageina, and Matthesia were not phylogenetic trees (®g. 12), Eochiroptera was addressed. shown as occupying the basal branch. It is Novacek (1987) focused on phylogenetic not clear, however, whether this was meant relationships of the best known Eocene taxa to imply monophyly of Eochiroptera; com- (Icaronycteris and Palaeochiropteryx) rather 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 25

than on issues of classi®cation. Following a detailed review of a wide array of morpho- logical features, Novacek (1987) concluded that both Icaronycteris and Palaeochiropte- ryx are more closely related to extant micro- chiropterans than to extant megachiropterans (®g. 14). He concluded that Icaronycteris and probably Palaeochiropteryx are out- groups of all Recent families of Microchi- roptera, but could not rule out the possibility that one or both of these Eocene taxa might have special af®nities within Microchiroptera (i.e., might be more closely related to one or more extant families than to others). Nova- cek (1987: 15±16) further concluded that . . . the Palaeochiropterygoidea seems merely an ar- ti®cial convention to group several early microchi- ropterans whose relationships with modern families of this suborder remain poorly known . . . Icaronyc- teris and Palaeochiropteryx are more accurately des- ignated as Microchiroptera incertae sedis. Moreover, there seems no justi®cation for a formal designation of an ``ancestral'' group (e.g., Eochiroptera sensu Van Valen, 1979) to distinguish Icaronycteris, Palaeochi- ropteryx, and other Eocene bats from the Recent chi- ropteran suborders. In a classi®cation of fossil vertebrates, Carroll (1988) named a new superfamily for the archaic Eocene bats, Icaronycteroidea, which he placed within Microchiroptera. He Fig. 10. Relationships of selected taxa in- included two families in Icaronycteroidea: ferred from features of fetal membrane structure and development (redrawn from Luckett, 1980a: Icaronycteridae (Icaronycteris) and Palaeo- ®g. 5). The original caption for this ®gure read chiropterygidae (Archaeonycteris, Palaeo- ``Character state distribution of fetal membrane chiropteryx, Cecilionycteris, and Matthesia). and reproductive features of Chiroptera . . .'' In concept and contents, Icaronycteroidea is identical to Palaeochiropterygoidea as rec- ognized by most previous authors. Carroll (1977), and included a brief discussion of the gave no justi®cation for this apparently un- af®nities of Vespertiliavus. Robbins and Sar- necessary name change, and no subsequent ich concluded by assigning Vespertiliavus to authors have used Icaronycteroidea. Carroll a new tribe, Vespertiliavini, within the sub- additionally provided a classi®cation for oth- family Taphozoinae. er Eocene bats, placing Vespertiliavus in Em- Sige (1988) described the fossil bats from ballonuridae, Necromantis in Megadermati- the Marinesian (ϭ Robiacian; MP 16) Le dae, Rhinolophus in Rhinolophidae, Hippo- Bretou fauna from the Quercy Phosphorites sideros, Palaeophyllophora, and Paraphyl- in France. He referred Vespertiliavus to Em- lophora in Hipposideridae, ?Vampyravus in ballonuridae, and Palaeophyllophora and Phyllostomidae, and Stehlinia in Vesperti- Hipposideros (Pseudohipposideros) to Hip- lionidae. He also placed Archaeopteropus in posideridae. Sige (1988: 93) also reported the Pteropodidae. presence of a new form that he referred to Robbins and Sarich (1988) used protein ``Rhinolophoidea, sp. indet.'' electrophoresis and immunological distance Storch and Habersetzer (1988) described a data to address relationships among extant new species of Archaeonycteris (A. pollex) Emballonuridae. They also considered mor- based on two skeletons from Messel, and phological data presented by Barghoorn compared this taxon with other Eocene bats. 26 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

They also reported the discovery of a new ``probably represents the stem group of the skull that con®rmed Habersetzer and Storch's genus Rhinolophus.'' (1987) suggestion that ``Archaeonycteris'' Sige (1991a) later described the genus revilliodi should be referred to Hassianycter- Dizzya from the Early Eocene of Chambi in is rather than to Archaeonycteris or Archaeo- northern Africa. Following broad compari- nycterididae. sons with extant and extinct forms, he re- Habersetzer and Storch (1989) discussed ferred Dizzya to Vespertilionoidea: Philisi- ecology and echolocation abilities of Eocene dae. A specimen referred to Rhinolophoidea bats from Messel as inferred from wing and gen. and sp. indet. was also described, and cochlear morphology. Based on comparisons Sige went on to extensively discuss the bio- with various extinct and extant bats, Haber- stratigraphic, biogeographic, and paleoeco- setzer and Storch (1989: 214) noted that logical implications of the bat fauna from the African Eocene. In a ®nal note, Sige dis- Archaeonycteris trigonodon Revilliod 1917 (family cussed usage of the name Eochiroptera. Con- Archaeonycterididae) is an archaic species.... Pa- tra Habersetzer and Storch (1987, 1989) and laeochiropteryx tupaiodon Revilliod 1917, and P. spiegeli Revilliod 1917, (family Palaeochiropterygi- Novacek (1987), he (1991a: 372) argued that dae) are small specialized species....Hassianycteris this taxon ``est preÂfeÂre par logique, commo- messelensis Smith and Storch 1981, H. magna Smith diteÂ, et ef®caciteÂaÁ celui super-familial des and Storch 1981, and H. revilliodi (Russell and Sige Palaeochiropterygoidea inclus dans les Mi- 1970) (Family Hassianycterididae) are the most ad- vanced species in dental and skeletal features. crochiroptera.'' Sige (1991a: 373) argued that recognizing the ``taxon-grade'' Eochi- Habersetzer and Storch (1989) argued roptera is a simple, operational approach to against Van Valen's (1979) concept of Eochi- classifying organisms of uncertain af®nities roptera, which was originally de®ned as in- that share a large number of primitive char- cluding all of the taxa mentioned in the ex- acters. Expressing the unusual opinion that cerpt above. They noted that Eochiroptera as Microchiroptera must itself be considered conceived by Van Valen would be a group paraphyletic (``l'unite sub-ordinale Microchi- of primitive and unspecialized species that roptera, qui . . . doit eÃtre clairment percËue ultimately gave rise to both Megachiroptera comme un rassemblement paraphyleÂtique, and Microchiroptera. Noting that the Messel mais se valeur empirique. . ..''; SigeÂ, 1991a: bat fauna includes ecologically diverse forms 373), he emphasized the empirical value of that are all ``completely developed'' micro- formally grouping taxa based on their per- chiropterans, Habersetzer and Storch (1989: ceived grade of evolution. He argued that 231) suggested that Eochiroptera was a mis- such a classi®cation system should perhaps leading concept, at least when applied to the be maintained for the chiropteran suborders Messel bats. (to indicate their adaptive level in the contin- Sige (1990) and Sudre et al. (1990) de- uum of bat evolution), even if it becomes scribed new species of Vespertiliavus from possible to diagnose strongly supported, Stampian (MP 25) and early Bartonian (MP monophyletic superfamilies including some 14) deposits of France, respectively. Both re- of the archaic Eocene taxa. tained Vespertiliavus in Emballonuridae. Sige (1991b) discussed morphology of the Sige additionally described a new species of deciduous dentition of Eocene bats and pre- Hipposideros (subgenus Pseudorhinolo- sented a ®gure summarizing relationships phus), a new species of Stehlinia (which he and stratigraphic ranges (®g. 15). He indi- referred to Natalidae sensu Van Valen, 1979), cated that ®ve extant lineages (Megachirop- and a new genus Vaylatsia. All four of the tera, Phyllostomatoidea [ϭ Noctilionoidea], new taxa described by Sige (1990) were Rhinolophoidea, Emballonurioidea, and Ves- based on fragmentary dentitions and isolated pertilionoidea) and three extinct lineages postcranial elements from the Quercy Phos- (Hassianycterididae, Archaeonycterididae, phorites. Vaylatsia was referred to Hipposi- and Palaeochiropterygidae) were in existence deridae based on dental and humeral char- in the Eocene. He placed Archaeonycteridi- acters, but Sige (1990: 1132) noted that it dae and Hassianycterididae in Eochiroptera, 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 27

Fig. 11. Eisenberg's hypothesis of bat relationships (redrawn from Eisenberg, 1981: ®g. 27). The original caption read ``Diagram of the classi®cation of the Chiroptera. Based in part on Smith 1976.''

but excluded Hassianycterididae from this comparisons with extant taxa, they referred group. Honrovits to Natalidae sensu Van Valen, Smith and Russell (1992) described a new 1979 along with Stehlinia, Chadronycteris, species of Hassianycteris from Upper Ypre- Chamtwaria (an Early Miocene form from sian deposits in Belgium. This discovery ex- Africa described by Butler, 1984), and prob- tended the age range of Hassianycteris (pre- ably Ageina. In the course of comparisons viously known only from Messel) into the with other Eocene forms, Beard et al. (1992) Early Eocene. Because the basal Middle Eo- recognized Archaeonycterididae as including cene age generally supposed for Messel is an Archaeonycteris and Icaronycteris; Palaeo- estimate, this temporal range extension may chiropterygidae as including Palaeochirop- be more apparent than real. teryx, Cecilionycteris, and Matthesia; and Beard et al. (1992) described the fossil ge- Hassianycterididae as including Hassianyc- nus Honrovits from Early Eocene (Late Was- teris. This usage essentially follows that of atchian) beds in North America. Based on Habersetzer and Storch (1987, 1988), differ- 28 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 ing only in the explicit referral of Cecilion- Eppsinycteris based on a partial dentary from ycteris and Matthesia to Palaeochiropterygi- early Ypresian beds (MP 8/9) near London, dae. England. This specimen had been previously Hand et al. (1994) described the ®rst Eo- identi®ed as a geolabid insectivoran, but cene bat from Australia, which they named Hooker noted the presence of a buccal cin- Australonycteris and placed in Microchirop- gulum on the lower molars, which was iden- tera, family indeterminate. Drawing compar- ti®ed by Hand et al.(1994) as a possible syn- isons with other Eocene bats, Hand et al. in- apomorphy of Chiroptera. The presence of a dicated that Australonycteris is most similar buccal cingulum, combined with other char- to Archaeonycteridae (de®ned as including acters thought to have a very limited distri- Archaeonycteris and ?Icaronycteris) in terms bution among insectivorous mammals, led of overall dental morphology and lack of de- Hooker to identify Eppsinycteris as a bat. rived features characteristic of extant fami- Based on three other derived features com- lies. They suggested, however, that morphol- mon in emballonurids, he referred Eppsinyc- ogy of an isolated periotic indicated that Aus- teris to Emballonuridae. This allocation ex- tralonycteris might instead represent ``a very tended the temporal range of Emballonuridae early vespertilionoid with a plesiomorphic by 10 million years, back into the earliest dentitionÐor that more than one species is Eocene. represented'' (i.e, that the periotic and dental McKenna and Bell (1997) summarized an fragments represent different species; Hand enormous amount of information from the et al., 1994: 378). They went on to discuss literature and provided a classi®cation for all the af®nities of Wyonycteris, concluding that mammals at the genus level and above. there is no convincing evidence that this tax- Among bats, they referred Aegina, Austra- on is closely related to bats. lonycteris, Provampyrus (ϭ Vampyravus), Habersetzer et al. (1994) discussed paleo- and Chadronycteris to Microchiroptera in- ecology of the Messel bats (Palaeochiropte- certae sedis. Archaeonycteris and Icaronyc- ryx, Archaeonycteris, and Hassianycteris) teris were placed in Archaeonycteridae, based on morphology of the wing and coch- while Palaeochiropteryx, Matthesia, and Ce- lea as well as fossilized stomach contents. In cilionycteris were referred to Palaeochirop- their introduction, they reviewed the Eocene terygidae. Hassianycteris was referred to record of bats and noted that Wyonycteris Hassianycterididae. All three of the archaic should be removed from Palaeochiroptery- familiesÐArchaeonycteridae, Palaeochirop- gidae and possibly from Chiroptera. Haber- terygidae, and HassianycterididaeÐwere re- setzer et al. (1994) followed Habersetzer and ferred to Microchiroptera incertae sedis. Storch (1987) in abandoning both Palaeochi- Among the extant lineages, Vespertiliavus ropterygoidea and Eochiroptera, instead rec- was referred to Emballonuridae: Taphozo- ognizing three families of uncertain af®nities inae: Vespertiliavini, Necromantis was re- (Palaeochiropterygidae, Archaeonycteridi- ferred to Megadermatidae, and Vaylatsia was dae, and Hassianycterididae). They noted placed incertae sedis in Rhinolophidae. In a that these families share some symplesiom- major nomenclatural change based upon pri- orphic features, but differ markedly from one ority of stem generic names, McKenna and another in morphology and apparent grade of Bell recognized Hipposiderinae as a junior evolution. Like Habersetzer and Storch synonym of Rhinonycterinae. Hipposideros (1989), Habersetzer et al. (1994: 236) sug- (including Pseudorhinolophus) was subse- gested that Archaeonycteridae retains a large quently placed in Rhinolophidae: Rhinonyc- number of primitive features, that Palaeochi- terinae: Rhinonycterini: Hipposiderina, and ropterygidae has a primitive dentition but Palaeophyllophora and Paraphyllophora more specialized wing skeleton, and that were placed in Rhinolophidae: Rhinonycter- members of Hassianycterididae ``are clearly inae: Palaeophyllophorini. Philisidae, includ- more advanced than the preceding families ing Dizzya and Philisis, was placed in Ves- and are of rather more modern appearance pertilionoidea by McKenna and Bell (1997), dentally and osteologically.'' and Honrovits was referred to Natalidae. Hooker (1996) described a new genus Stehlinia was placed in Vespertilionidae: 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 29

Fig. 12. SigeÂ's (1985) hypotheses of relationships of Philisidae and Vespertilionoidea (redrawn from SigeÂ, 1985: ®gs. 8, 9). The original caption for A read ``PhylogeÂnie preÂsumeÂe des Philisidae''; the caption for B read ``Branchements majeurs preÂsumeÂs Vespertilionoidea et leur chronologie.'' See text for explanation of ``Vampyravidae.''

Vespertilioninae: Myotini. Wallia and Cu- teropus to Megachiroptera: Pteropodidae: vierimops were referred to Molossidae: Mo- Archaeopteropodinae. lossinae, incertae sedis. Because Tomopeati- Hand and Kirsch (1998) examined rela- nae was included in Molossidae, Molossinae tionships among extant and extinct Hipposi- sensu McKenna and Bell is equivalent to deridae in a cladistic analysis of craniodental Molossidae sensu Legendre and others. Fi- and postcranial osteological characters. Sev- nally, McKenna and Bell referred Archaeop- eral species of Rhinolophus were included as 30 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Fig. 13. Pierson's hypotheses of relationships and temporal patterns of divergence based on trans- ferrin immunological distance data (redrawn from Pierson, 1986: ®gs. 31, 37). The original caption for A read ``A proposed phylogeny for microchiropteran families, showing hypothesized times of divergence for major lineages . . . Calibration of immunological distances assumed Paleocene (55 MYA) origins for the current radiation.'' The caption for B read ``vespertilionid relationships [were derivied] using anti-Antrozous transferrin''; and the caption for C read `` molossid relationships [were derived] using anti-Tadarida tranferrin.'' 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 31

nent of cladistic methods, drew upon this rich history and presented a pair of clado- grams (®g. 7) intended to summarize the then ``generally accepted view'' of bat phylogeny (Smith, 1976: 56). These trees differed only in the perceived origins of lineages from var- ious ancestral stock(s); relationships among the extant families were the same in both trees. Although Smith (1976, 1980) stressed the importance of cladistic methods (i.e., dis- tinguishing primitive from derived similari- ty), his phylogeny was apparently not based Fig. 14. Novacek's (1987) hypothesis of re- on an explicit character analysis, but rather lationships of Icaronycteris to extant bats (re- on his perceptions of character polarities as drawn from Novacek, 1987: ®g. 10). The original informed by his own research on mormoop- caption read ``Cladogram favoring a close rela- ids (Smith, 1972) and the ideas of earlier tionship between Icaronycteris and recent Micro- workers (e.g., those listed above, as well as chiroptera ...,''andwent on to list morphologi- Hill, 1974). cal characters associated with each branch Smith's (1976) trees (®g. 7) indicated (omitted here). Novacek (1987) also concluded monophyly of several higher-level taxa in- that Palaeochiropterx was more closely related to cluding four extant superfamilies: (1) Em- Microchiroptera than to Megachiroptera, but he did not include it in the ®gured phylogenetic tree. ballonuroidea, including Rhinopomatidae, Craseonycteridae, and Emballonuridae; (2) Rhinolophoidea, including Megadermatidae, outgroup taxa, and a variety of other micro- Nycteridae, and Rhinolophidae (including chiropterans were used to assess character Hipposiderinae); (3) Phyllostomatoidea (ϭ polarity. Hand and Kirsch concluded that Noctilionoidea), including Phyllostomatidae Pseudorhinolophus and Palaeophyllophora (ϭ Phyllostomidae), Mormoopidae, and represent relatively derived hipposiderids, Noctilionidae; and (4) Vespertilionoidea, in- while Vaylatsia may be the plesiomorphic cluding Molossidae, Mystacinidae, Natali- sister-group of extant rhinolophids (ϭ rhin- dae, Thyropteridae, Furipteridae, Vesperti- olophines). lionidae, and Myzopodidae. Monophyly of Vespertilionidae was apparently assumed. RELATIONSHIPS AMONG EXTANT Two clades subsequently named as infraor- LINEAGES OF BATS ders by Koopman (1985) also appeared in Smith's trees: Yinochiroptera (Emballonuro- Relationships among extant bats have been idea ϩ Rhinolophoidea) and Yangochiroptera a source of debate for as long as scientists (Noctilionoidea ϩ Vespertilionoidea). have been classifying mammals. Excellent Mein and Tupinier (1977) discussed the summaries of the early history of chiropteran dental formula of Miniopterus, a genus usu- classi®cation can be found in Winge (1941) ally placed in its own subfamily (Miniopter- and Smith (1976, 1980). As noted by both inae) in Vespertilionidae. Based on their authors, most recent bat classi®cations are ul- analysis, they suggested that this group be timately based on that of Dobson (1875: removed from Vespertilionidae to its own 345), who purported to arrange the subor- family, Miniopteridae. ders, families, and genera of bats ``according Van Valen (1979) presented a signi®cantly to their natural af®nities'' (see ®g. 5). In the different hypothesis of bat relationships (®g. hundred years that followed, many major and 8) based on a cladistic analysis of morpho- minor changes to this classi®cation scheme logical traits, including many features de- were suggested by such workers as Winge scribed for the ®rst time in studies published (1892), Miller (1907), Simpson (1945), Da- in the 1970s (e.g., Henson, 1970; Strickler, vis (1970), and Koopman and Jones (1970). 1978). Although Van Valen published lists of In 1976, James Dale Smith, an early propo- derived characters supporting each clade in 32 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Fig. 15. SigeÂ's (1991b) summary of biostratigraphic data for Eocene bats (redrawn from SigeÂ, 1991b: ®g. 5). The original caption read ``Biostratigraphic position of the main Early Paleogene bat localities and temporal distribution of major bat groups. The geological frame is that of Berggren et al. 1985 and the mammalian biostratigraphy MP that of Mainz 1987 (Schmidt-Kittler, 1987). Bold lines: fossil evi- dence; double dotted lines: unevidenced existance; thin lines: presumed existence.''

his tree, he did not discuss the methods used uroidea, Yinochiroptera, and Yangochirop- to identify these traits as derived. Van Val- tera did not appear as monophyletic groups. en's tree indicated monophyly of three of the Although Miniopterinae was retained in Ves- four superfamilies (Rhinolophoidea, Nocti- pertilionidae, Kerivoulinae was removed lionoidea, and Vespertilionoidea); Emballon- from the latter family and placed in Natalidae 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 33

Fig. 16. Two alternative hypotheses of interfamilial relationships suggested by Novacek (redrawn from Novacek, 1991: ®gs. 6, 7). The original caption for A read `` This cladogram . . . is based on discussions in Koopman (1984). The position of Palaeochiropteryx . . . follows Novacek (1987). Note the remote branch position of Rhinopomatidae.'' the caption for B read ``as in [the previous ®gure] modi®ed by relocation of Rhinopomatidae as a sister taxon of Rhinolophoidea (following Pierson, 1986).'' together with Myzopodidae, Thyropteridae, Novacek focused on morphology of the au- and Furipteridae, which Van Valen (1979) re- ditory region; Luckett concentrated on mor- duced in rank to subfamily level. phology and development of fetal mem- Novacek (1980a) and Luckett (1980a) branes. Primarily because the tree derived published cladograms of bat relationships from auditory characters (®g. 9) differed sig- based on analyses of single organ systems. ni®cantly from all previous hypotheses of 34 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

assumed by both Novacek (1980a) and Luck- ett (1980a). Eisenberg (1981) published a phylogeny of bats in his book on mammalian radiations. Eisenberg's (1981: 147) tree (®g. 11) differed only slightly from that presented by Smith (1976; see ®g. 7). Like Smith (1976), Eisen- berg depicted each of the microchiropteran superfamilies as monophyletic. However, Ei- senberg placed Emballonuroidea, Rhinolo- phoidea, and Phyllostomatoidea (ϭ Nocti- lionoidea) together in an unresolved clade with Vespertilionoidea as the sister-group, rather than identifying Emballonuroidea and Rhinolophoidea as sister-taxa as had Smith (1976). Gopalakrishna and Chari (1983) described fetal membrane development in Miniopterus, a taxon generally placed in its own subfamily in Vespertilionidae. Based on their ®ndings, which indicated major differences between miniopterines and other vespertilionids, Go- palakrishna and Chari recommended removal of Miniopterus to its own family, Miniopter- idae. Pierson (1986) proposed a series of phy- logenetic hypotheses based on an analysis of transferrin immunological distance data (®g. 13). Her results differed signi®cantly from those of previous studies in many areas, in- cluding (1) placing Rhinopomatidae within Fig. 17. Two alternative hypotheses of yino- Rhinolophoidea, (2) associating Mystacini- chiropteran relationships based on hyoid muscu- dae with Noctilionoidea rather than Vesper- loskeletal morphology (redrawn from Grif®ths et tilionoidea, (3) placing Furipteridae and Na- al., 1992: ®gs. 10, 11). Both trees were equally talidae at the base of the microchiropteran parsimonious in the context of the hyoid data pre- tree, and (4) grouping Tomopeas and Min- sented by Grif®ths and Smith (1991) and Grif®ths iopterus with Molossidae rather than with et al. (1992), but these authors preferred tree A Vespertilioninae. In the context of Pierson's on the basis of hypothesized patterns of transfor- trees, Yinochiroptera is monophyletic, but mation in two characters. Yangochiroptera and the four superfamilies are not. One of the more striking results of higher-level relationships, Novacek (1980a) Pierson's (1986) immunological studyÐthe warned against using that cladogram (or any placement of MystacinidaeÐwas explored in other derived from a single organ system) as greater depth in Pierson et al. (1986), which a basis for a new phylogenetic reconstruction also included a discussion of some morpho- or classi®cation. Luckett's (1980a) study was logical features. unable to resolve many relationships (®g. Hill and Harrison (1987) reviewed cranio- 10), but provided some support for mono- dental morphology and structure of the bac- phyly of Noctilionoidea and a close relation- ulum in selected vespertilionids, and pro- ship between Megadermatidae and a clade posed a new generic classi®cation of those containing Vespertilionidae and Thyropteri- forms traditionally included in Vespertilion- dae. Monophyly of Vespertilionidae sensu inae (e.g., by Miller, 1907). Hill and Harri- Koopman and Jones (1970) was apparently son (1987) included Myotini and Antrozoini 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 35

Fig. 18. Fenton's (1992) phylogeny of bat families with Altringham's (1996) addition of subfamilies (redrawn from Fenton [1992: p. 10] and Altringham [1996: ®g. 1.12]). Fenton's (1992) caption read ``A family tree or phylogeny shows the presumed evolutionary relationships between the living families of bats. The families are grouped according to superfamilies and suborders.'' Altringham's (1996) caption read ``An evolutionary tree for the bats to subfamily level. Considerable uncertainty surrounds some parts of this tree, in particular the relationships between megabats and microbats, and the detailed classi®cation of the Phyllostomidae and Vespertilionidae.'' as tribes within Vespertilioninae, and rec- curred with Gopalakrishna and Chari's ognized Nyctophininae as a distinct subfam- (1983) recommendation that Miniopteridae ily. should be recognized as a family distinct Tiunov (1989) examined variation in mor- from Vespertilionidae. In contrast, Tiunov phology of the tongue and male accessory found no differences between rhinolophines glands (e.g., prostate, seminal vessicles, and hipposiderines, and therefore recom- Cowper's gland, ampullary glands) in a num- mended that they be placed in a single fam- ber of Old World species, and discussed the ily, Rhinolophidae. phylogenetic implications of these data. Baker et al. (1991a) analyzed variation in Based on observed differences, Tiunov con- rDNA restriction sites in a study designed to 36 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Fig. 19. Volleth and Heller's hypothesis of relationships among vespertilionids (redrawn from Vol- leth and Heller, 1984: ®g. 7). The original caption read ``Cladogram of Vespertilionidae based on kary- ological features . . . Where the names of species are lacking, only one species was studied (12 cases). From karyologically heterogeneous genera all species studied are shown. The dotted line between Ep- tesicus and Scotophilus indicates a second possibility for the relations of Eptescini . . .[that tribe] could be grouped together with Scotophilus .... [Abbreviations:] H. ϭ Hesperopternus, Hyps. ϭ Hypsugo, Ny. ϭ Nyctalus, P. ϭ Pipistrellus, T. ϭ Tylonycteris.'' 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 37 test bat monophyly (see below). Although unresolved, and suggested monophyly of the their data set could not resolve subordinal or remaining three superfamilies. Monophyly of interfamilial relationships of bats, support Vespertilionidae (including kerivoulines, to- was found for several groups, including Yin- mopeatines, and miniopterines) was not dis- ochiroptera, Noctilionoidea, and a clade con- cussed. taining Mormoopidae and Noctilionidae. Grif®ths and his colleagues used osteo- In the 1980s and early 1990s, a signi®cant myological characters of the hyoid region to controversy arose in chiropteran systematics explore relationships among various groups regarding bat monophyly. The history of this of yinochiropteran bats (Grif®ths and Smith, controversy was reviewed by Simmons 1991; Grif®ths et al., 1992). Grif®ths et al. (1994) and will not be repeated here. Al- (1992) presented two alternative phylogenies though bat diphyly has been proposed by (®g. 17), neither of which supported mono- several authors (e.g., Smith and Madkour, phyly of Emballonuroidea or Rhinolophoi- 1980; Hill and Smith, 1984; Pettigrew, 1986, dea. The only clade that appeared in both 1991a, 1991b, 1994, 1995; Pettigrew and Ja- trees was Rhinolophidae ϩ Hipposideridae. mieson, 1987; Pettigrew et al., 1989; Rayner, Gopalakrishna and Badwaik (1992) dis- 1991b; Pettigrew and Kirsch, 1995), a grow- cussed fetal membrane structure and used ing body of data provides very strong sup- phenetic similarity to evaluate phylogenetic port for bat monophyly. Data supporting chi- relationships among bat families. They con- ropteran monophyly include morphological cluded that features from many organ systems (Luckett, 1980a, 1993; Wible and Novacek, 1988; . . . similarities between Molossidae and Pteropodidae and differences between Molossidae and Vesperti- Kovtun, 1989; Thewissen and Babcock, lionidae suggest a closer relationship between Ptero- 1991, 1993; Kay et al., 1992; Novacek, 1992, podidae and Molossidae than between Molossidae 1994; Beard, 1993; Simmons, 1993a, 1994, and Vespertilionidae. It is, therefore, suggested on 1995; Wible and Martin, 1993; Simmons and purely embryological grounds that Molossidae be Quinn, 1994; Miyamoto, 1996), DNA±DNA separated from the Super-family Vespertilionidae and be placed somewhere between Pteropodidae and Em- hybridization data (Kirsch et al., 1995; ballonuridae (Gopalakrishna and Badwaik, 1992: 7). Hutcheon and Kirsch, 1996; Kirsch, 1996), and sequence data from numerous mitochon- Fenton (1992) provided a phylogeny of bat drial and nuclear genes (Adkins and Honey- families in his semipopular book on bats, but cutt, 1991, 1993, 1994; Mindell et al., 1991; did not discuss the source of this hypothesis. Ammerman and Hillis, 1992; Bailey et al., The same topology was reproduced by Fen- 1992; Stanhope et al., 1992, 1993, 1996; Ho- ton (1995). This tree (®g. 18) was developed neycutt and Adkins, 1993; Knight and Min- to summarize possible relationships of bats dell, 1993; Novacek, 1994; Allard et al., as re¯ected in numerous systematic studies 1996; Miyamoto, 1996; Porter et al., 1996). and classi®cations (e.g., Hill, 1974; Pierson Bat monophyly now represents one of the et al., 1986); it was not based on any new most strongly supported phylogenetic hy- data or data analyses (Fenton, personal com- potheses within Mammalia (Simmons, 1994; mun.). Altringham (1996) reproduced Fen- Miyamoto, 1996). ton's tree (again with no mention of source), Novacek (1991) used two novel phyloge- modifying it only by adding branches for nies as a framework for discussing evolution each subfamily (®g. 18). of cochlear features in bats (®g. 16). These Volleth and Heller (1994) provided the trees were not derived from an explicit char- ®rst cladistic hypothesis of relationships acter analysis, but were instead based on among genera of Vespertilionidae sensu lato consideration of the morphological charac- (®g. 19). Using data from G-banded chro- ters discussed in Koopman (1984) and No- mosomes, they identi®ed homologous arms vacek (1987) as well as Pierson's (1986) im- and presented a cladogram based on an anal- munological results (see caption to ®g. 16). ysis of derived chromosomal features (e.g., Both of Novacek's (1991) trees indicated translocations, Robertsonian fusions, ®s- paraphyly of Yinochiroptera and Emballon- sions). They concluded that (1) Miniopteri- uroidea, left monophyly of Yangochiroptera nae is the sister-group to a clade containing 38 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Fig. 20. Results of Simmons' (1998) analysis of higher-level relationships based on morphology and rDNA restriction sites. The original caption of this ®gure read ``Strict consensus of two equally most-parsimonious trees (608 steps each; CI ϭ 0.410, RI ϭ 0.592) found in a heuristic analysis . . . The numbers below internal branches indicate the percentage of bootstrap replicates in which each clade appeared; numbers above the branches are decay values (the minimum number of additional steps required to collapse each clade).''

Kerivoulinae ϩ Murininae ϩ Vespertilioni- cluded that these data support a close rela- nae; (2) Myotini falls outside a clade con- tionship between Tomopeas and molossids taining the remaining vespertilionines; (3) rather than vespertilionids, and recommend- Nyctophilus is a member of Vespertilionini; ed that Tomopeatinae be transferred from (4) Vespertilionini and Pipistrellini are sister- Vespertilionidae to Molossidae. taxa; and (5) Eptesicini (including Hesper- Kirsch and his colleagues (Pettigrew and optenus), ``Nycticeiini,'' and Plecotini fall Kirsch, 1995; Hutcheon and Kirsch, 1996; outside the Vespertilionini ϩ Pipistrellini Kirsch, 1996; Kirsch and Hutcheon, 1997; clade. Antrozous, Tomopeas, Lasiurus, and Hutcheon et al., in press; Kirsch and Petti- other New World taxa were omitted from grew, in press; Pettigrew and Kirsch, in Volleth and Heller's study. press) reported results of a series of DNA- Sudman et al. (1994), drawing in part DNA hybridization studies of relationships upon the unpublished work of Barkley among a set of taxa including both megachi- (1984), investigated relationships of Tomo- ropterans and microchiropterans. Surprising- peas (the only member of Tomopeatinae) to ly, these experiments suggested that Rhino- vespertilionids and molossids using protein lophoidea and Pteropodidae may be sister- electrophoresis, cytochrome b gene sequenc- taxa, implying microchiropteran paraphyly. es, and morphological characters. They con- However, this result has been questioned by 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 39 the authors themselves, who have suggested TABLE 2 that base compositional bias might be re- Higher-level Classi®cation of Recent Bats sponsible for producing a false phylogenetic Proposed by Simmons (1998) signal linking two A-T rich clades (Ptero- podidae and Rhinolophoidea; Hutcheon and Order Chiroptera Suborder Megachiroptera Kirsch, 1996; Kirsch, 1996; Kirsch and Family Pteropodidae Hutcheon, 1997; Hutcheon et al., in review; Suborder Microchiroptera Kirsch and Pettigrew, in review; Pettigrew Infraorder incertae sedis and Kirsch, in review). Although topology of Superfamily Emballonuroidea other parts of the tree remains suspect, many Family Emballonuridae recovered groupings are congruent with Infraorder Yinochiroptera those identi®ed in previous studies. As noted Superfamily Rhinopomatoidea by Hutcheon et al. (in press), ``a tree as star- Family Craseonycteridae tling as ours obviously must be veri®ed by Family Rhinopomatidae additional studies . . .'' Superfamily Rhinolophoidea Family Nycteridae The most recent comprehensive attempt to Family Megadermatidae resolve higher-level relationships among ex- Family Rhinolophidae tant family-level lineages of bats was that of Subfamily Rhinolophinae Simmons (1998). She assembled a data set Subfamily Hipposiderinae of 192 discrete characters including many Infraorder Yangochiroptera new morphological characters, all of the rel- Superfamily incertae sedis evant morphological data discussed by most Family Mystacinidae previous workers (e.g., Van Valen, 1979; Superfamily Noctilionoidea Luckett, 1980a; Novacek, 1980a, 1991; Family Noctilionidae Barkley, 1984; Grif®ths and Smith, 1991; Family Mormoopidae Family Phyllostomidae Grif®ths et al., 1992), and the rDNA restric- Superfamily Molossoidea tion site data presented by Baker et al. Family Antrozoidae (1991a). To test vespertilionid monophyly Family Molossidae and evaluate relationships of its subfamilies, Subfamily Tomopeatinae Vespertilionidae sensu lato was split into sev- Subfamily Molossinae en subgroups for analysis: Vespertilioninae, Superfamily Vespertilionoidea Myotinae, Miniopterinae, Murininae, Keri- Family Vespertilionidae voulinae, Antrozoinae, and Tomopeatinae. Subfamily Vespertilioninae Parsimony analyses of the resulting data set Subfamily Miniopterinae produced a well-resolved tree (®g. 20) in Subfamily Myotinae Subfamily Murininae which many nodes were strongly supported. Subfamily Kerivoulinae Major results included the following: (1) Superfamily Nataloidea Emballonuridae appears to be the sister- Family Myzopodidae group of all other microchiropterans, there- Family Thyropteridae fore Emballonuroidea and Yinochiroptera (as Family Furipteridae traditionally recognized) are not monophy- Family Natalidae letic; (2) Rhinopomatidae and Craseonycter- idae are sister-taxa; (3) Rhinolophoidea, Noctilionoidea, Vespertilionoidea, and Yan- phylogenetic results, restricting Emballonu- gochiroptera each appear to be monophylet- roidea to Emballonuridae, placing Rhino- ic; (4) Vespertilionidae sensu lato is not pomatidae and Craseonycteridae in Rhino- monophyletic; Antrozoinae and Tomopeati- pomatoidea, restricting Yinochiroptera to nae are more closely related to Molossidae Rhinopomatoidea ϩ Rhinolophoidea, raising than to other vespertilionids; and (5) Myzo- Antrozoinae to family rank as Antrozoidae, podidae, Thyropteridae, Natalidae, and Fu- referring Tomopeatinae to Molossidae, rec- ripteridae form a clade. ognizing Antrozoidae ϩ Molossidae as a Simmons (1998) proposed a number of new superfamily Molossoidea, recognizing nomenclatural changes (table 2) based on her Myzopodidae ϩ Thyropteridae ϩ Natalidae 40 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

ϩ Furipteridae as a new superfamily Nata- logeny of extant lineages, and to investigate loidea, and restricting Vespertilionidae to methodological issues involving fossil taxa Vespertilioninae ϩ Miniopterinae ϩ Myot- and missing data in phylogenetic analyses. inae ϩ Murininae ϩ Kerivoulinae. Noting The goals of the current study are as follows: that several of the latter subfamilies might (1) to determine the relationships of Icaron- eventually be raised to family rank, Simmons ycteris, Archaeonycteris, Hassianycteris, and (1998) restricted Vespertilionoidea to Ves- Palaeochiropteryx to each other and to fam- pertilionidae as de®ned above, pending fur- ily-level lineages of extant bats; (2) to eval- ther study. uate the effects of including fossil genera on perceived relationships among extant forms; GOALS OF THE CURRENT STUDY (3) to evaluate the effects of including soft- The data set developed by Simmons tissue and molecular characters in a study in- (1998)Ðwhich of course includes many cluding both fossil and extant groups; (4) to characters originally described by others (see document patterns of character transforma- appendix 2)Ðincludes more than 80 cranio- tion in the basal part of the chiropteran tree; dental and postcranial osteological charac- and (5) to consider the implications of these ters. This provides a unique opportunity to data for theories concerning the early evo- evaluate the evolutionary relationships of the lution of echolocation and foraging strategies archaic Eocene taxa in the context of a phy- in Microchiroptera.

MATERIALS AND METHODS TAXONOMIC SAMPLING AND (1966, 1970) noted 14 characters of Icaron- MONOPHYLY ycteris that he considered as ``primitive'' or ``generalized'' relative to the conditions seen The present study consists of a phyloge- in extant bats, close examination of these netic analysis of 24 family-level lineages of features by Novacek (1987) reduced this list extant bats, two ordinal-level extant outgroup considerably. Novacek identi®ed only four taxa (Scandentia, Dermoptera), and four Eo- primitive features apparently found in Ica- cene fossil bat genera (Icaronycteris, Ar- ronycteris that have been lost or modi®ed in chaeonycteris, Palaeochiropteryx, Hassi- all other known fossil and extant bats: (1) anycteris; see appendix 1 for specimens ex- unfused sternal elements, (2) relatively short amined). The extant lineages are the same as radius, (3) complete phalangeal formula (2- those analyzed by Simmons (1998); taxo- 3-3-3) on the digits of the wings, and (4) nomic names used for bat clades follow the head and neck of femur set at an angle to the classi®cation proposed by Simmons (1998; shaft. Our examination of additional speci- table 2). Monophyly of Vespertilioninae (in- mens of extant bats plus material of Icaron- cluding Nyctophilinae) was assumed for ycteris that was not available to Jepsen or practical reasons following Volleth and Hell- Novacek indicate that two more of these fea- er (1994) and Simmons (1998). Monophyly tures should be removed from the list. of each of the other extant bat lineages used Our observations indicate that sternal ele- as OTUs (operational taxonomic units) is ments are typically unfused in juvenile, sub- well established (see Simmons [1998] and adult, and some young adult bats. Novacek references cited therein). Each of the fossil (1987) noted that the holotype of Icaronyc- genera also appears to be monophyletic (ta- teris index (PU 18150) might have unfused ble 3), with the possible exception of Ica- sternal elements simply because it was a ronycteris. young individual. This was con®rmed by our In the absence of apomorphic traits, most examination of UW 21481a±b (®g. 21), a workers have diagnosed Icaronycteris on the previously undescribed specimen of Icaron- basis of primitive features that have been ycteris index from Wyoming.2 The sternal el- modi®ed in all other bats (Jepsen, 1966, 1970; Novacek, 1987). Although Jepsen 2 We identi®ed UW 21481a±b as Icaronycteris index 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 41

ements in this individual are fused with the Another feature cited by Jepsen (1966, exception of the joints between the manubri- 1970) and Novacek (1987) as a primitive um and mesosternum and between the me- character of Icaronycteris is the orientation sosternum and xiphisternum. Sutures are vis- of the head and neck of femur, which they ible between three mesosternal elements, but described as set at an angle to the long axis the remaining mesosternal elements are fully of the shaft (although not to the extent seen fused (®g. 22). This degree of fusion is sim- in terrestrial mammals). Novacek (1987: 13) ilar to that found in young adult bats of many noted that in this trait Icaronycteris is ``clear- extant families. Because UW 21481a±b and ly more conservative than living mega- and PU 18150 are similar in all other respects, microchiropterans, where the head is nearly we conclude that PU 18150 was a young aligned with the shaft and the neck is very adult Icaronycteris, whereas UW 21481a±b short or absent.'' However, our comparisons represents a somewhat older individual. A of the femur of Icaronycteris with those of similar pattern of variation is also seen in extant bats indicate that Novacek was misled some of the Messel bats. For example, all by the angle of presentation of the femur in mesosternal elements are fused in some spec- PU 18150, which provides an oblique view imens of Archaeonycteris trigonodon (e.g., of the head and neck (®g. 23). We found that SMF Me 80/1379), while all sutures are still this view accentuates a relatively small de- visible in others (e.g., SMF Me 963a). Ica- gree of offset of the femoral head. Compar- ronycteris therefore appears to have had an isons of the femur of PU 18150 and UW ontogenetic pattern of sternal fusion similar 21481a±b with those of extant megachirop- to that seen in Archaeonycteris and many ex- terans and noctilionoids of similar body size tant bats. indicate that the head and neck of the femur of Icaronycteris are set at the same angle (relative to the shaft of the femur) as seen in based on morphology and collection locality. Based on many extant bats. Indeed, the proximal femur comparisons with Jepsens original stereophotographs of Rousettus aegyptiacus is virtually identical (many of which remain unpublished), we found that UW to that of Icaronycteris when viewed from 21481a±b appears identical to the holotype of Icaron- ycteris index PU 18150 in vitually all respects (although the same angle. Icaronycteris thus cannot be see text for a discussion of sternal fusion). Forearm considered more primitive than extant bats in length in UW 21481a±b is 47.5 mm; forearm length in terms of femur morphology. PU 18150 is 48.0 mm. Measurements of skull length, Our survey thus limits the list of primitive tibia length, and length of various wing elements in UW features found in Icaronycteris (but modi®ed 21481a±b are within Ϯ2 mm of those reported by Jepsen in all other known bats) to the following: (1) (1966) for PU 18150. This range of variation is well relatively short radius, and (2) complete pha- within that seen in extant species of similar body size langeal formula (2-3-3-3) on the digits of the (e.g., Swanepoel and Genoways, 1979). UW 21481a±b wings. Novacek (1987) noted that relative was collected from early Eocene sediments of the Green length of the radius had never been quanti- River Formation of Wyoming, the same deposits from which PU 18150 was collected (Jepsen, 1966). Because ®ed, but this was subsequently done by Ha- detailed dental comparisons between PU 18150 and UW bersetzer and Storch (1987). They found that 21481a±b were precluded by preservation of the latter the ratio of humerus length to radius length specimen, it remains possible that UW 21481a±b rep- was somewhat larger (and the radius there- resents a species of Icaronycteris distinct from index; fore relatively shorter) in Icaronycteris com- however, we found no data to support such an interpre- pared to other Eocene bats, although the dif- tation. Accordingly, we refer UW 21481a±b to Icaron- ference between the range of values for Ica- ycteris index Jepsen, 1966. ronycteris (0.715±0.732) and Archaeonycter- Grande (1980: ®g. III.26) ®gured a badly preserved is (0.680±0.715) is not great. bat fossil from Middle Eocene deposits of the Green The diagnosis of Icaronycteris therefore River Formation of Colorado that he noted ``probably represents an undescribed species.'' Unfortunately, this rests on just two plesiomorphic features. specimen is in a private collection and has never been However, it seems unlikely that multiple lin- described. This record suggests, however, that bat spe- eages are represented in this group (at least cies other than Icaronycteris index may have been pres- among the four specimens that we examined) ent in western North America in the Eocene. because the range of variation in size and 42 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

TABLE 3 Apomorphiesa Diagnosing Genera of Eocene Bats

Icaronycteris (one species, I. index) ● No apomorphies Archaeonycteris (three species, A. trigonodon, A. pollex, A. brailloni) ● Ecto¯exus deeply retracted between parastyle and metastyle on M1 and M2 ● Ventral process of manubrium of sternum oriented at approximately 90Њ to axis of body of manubium ● Posterior xiphisternum narrow, without lateral ¯are ● Dorsal ischial tuberosity present Palaeochiropteryx (two species, P. tupaiodon and P. spiegeli) ● Metaconule present on M1 and M2 ● Length of postparacrista equals length of premetacrista on M3 ● Ventral accessory process present on cervical vertebra 5 ● Infraspinous fossa of scapula relatively broad ● Ventral projection present on anteromedial ¯ange of scapula Hassianycteris (four species, H. messelensis, H. magna, H. revilliodi, and H. joeli) ● Protoconid and hypoconid on molar teeth exceptionally tall and robust ● Last lower premolar not molariform, metaconid lacking and talonid short ● First upper premolar reduced to tiny peg or absent ● Mandible deep, thickened dorsoventrally ● Radius unusually long and strongly curved ● Trochiter extends well beyond level of head humerus ● Metacarpal of digit V relatively short compared to metacarpals III and IV a Not found in any other Eocene taxon. Membership in extant families is precluded by documented or inferred absence of derived characters diagnostic of those taxa (for a list of characters diagnosing extant lineages see Simmons, 1998: tables 1, 2). Sources: Russell and SigeÂ, 1970; Smith and Storch, 1981; Habersetzer and Storch, 1987, 1989; Storch and Habersetzer, 1988; personal obs.). shape of skeletal elements among referred in- congruence studies have failed to resolve ex- dividuals appears to fall within limits char- isting con¯icts between morphological data acteristic of extant species (see footnote 2). and several molecular data sets (e.g., Allard We have therefore assumed monophyly of et al., 1996; Miyamoto, 1996; Stanhope et Icaronycteris for the purposes of the current al., 1996). Nevertheless, practical considera- study, although we recognize that the Euro- tions require that a choice of outgroups be pean material referred to ?Icaronycteris men- made to facilitate analyses of relationships ui and ?Icaronycteris sp. may represent one and character evolution within Chiroptera. A or more different clades. close relationship between Chiroptera and Dermoptera (colugos or ¯ying lemurs) is OUTGROUPS strongly supported by morphological data Only one outgroup is necessary to root a (Wible and Novacek, 1988; Novacek, 1992, phylogenetic tree (Nixon and Carpenter, 1994; Simmons, 1993a, 1994, 1995; Szalay 1993), but at least two outgroups are usually and Lucas, 1993; Wible, 1993; Simmons and included in cladistic analyses to establish Quinn, 1994; Miyamoto, 1996; Grif®ths, character polarities and to permit testing of MS), 12S rDNA sequences (sampled from ingroup monophyly (Maddison et al., 1984). more than 150 species representing 20 or- Ideally, outgroups should comprise the near- ders; Vrana, 1994), and combined morpho- est sister-taxa to the ingroup because the logical data and cytochrome oxidase II probability of homoplasy increases with time (COII) gene sequences (Novacek, 1994). Ac- since divergence from a common ancestor. cordingly, Simmons (1998) used Dermoptera The identity of the sister-group of bats is still as an outgroup for investigating interrelation- the subject of considerable debate, and recent ships of bats. Morphological synapomor- 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 43

Fig. 21. New specimen of Icaronycteris index (UW 21481a±b) from the Green River Formation, Wyoming. Photograph by C. Tarka. phies supporting monophyly of Volitantia was included as an additional outgroup by (Dermoptera ϩ Chiroptera) are listed in table Simmons (1998). Monophyly of extant Der- 4. Scandentia (tree shrews), which together moptera and monophyly of Scandentia are with Primates form the sister-group of the each strongly supported (Zeller, 1986; Volitantia in optimal morphological trees MacPhee et al., 1989; Beard, 1993; Wible (Novacek, 1992, 1994; Miyamoto, 1996), and Zeller, 1994). 44 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Fig. 22. Icaronycteris index (UW 21481a±b). Close-up ventral view of the ribcage, sternum, and right shoulder region. AB, axillary border of scapula; CC, costal cartilages; CL, clavicle; HU, humerus; LP, lateral process of manubrium of sternum; MN, manubrium of sternum; MS, mesosternum; PL, posterior laminae of ribs; SC, scapula; SN, suprascapular notch; VP, ventral process of manubrium of sternum; XI, xiphisternum. Photograph by C. Tarka.

Several DNA nucleotide sequence data dinal relationships when analyzed alone and sets (epsilon ␤-globin, interphotoreceptor re- in combined analyses (Bailey et al., 1992; tinoid binding protein [IRBP], COII, von Stanhope et al., 1992, 1993, 1996; Adkins Willebrand factor [vWF]) have produced al- and Honeycutt, 1993, 1994; Novacek, 1994; ternative hypotheses of mammalian interor- Allard et al., 1996; Miyamoto, 1996; Porter 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 45

Fig. 23. Icaronycteris index (holotype; PU 18150). Stereophotographs of close-up ventral view of pelvic region. From Jepsen (1970: ®g. 16); reprinted and labels revised by Novacek (1987: ®g. 8). A, acetabulum; Ap, anterolateral process of seventh lumbar vertebra; Cf, fourth caudal vertebra; F, ®bula; Fl, lesser trochanter of femur; G, greater trochanter of femur; Hf, head of femur; Ls, seventh lumbar vertebra; O, obturator foramen; Ps, pubic symphisis; Sa, sacrum; Sp, pubic spine.

et al., 1996). Rather than placing Dermoptera tree was rooted using Pteropodidae). These and Scandentia as close relatives of Chirop- results indicate that morphological character tera (i.e., in a monophyletic Archonta), these data for the ingroup taxa are highly struc- studies have alternatively suggested that In- tured, and lead us to suspect that the same sectivora, Carnivora, Artiodactyla, a clade tree topology would be obtained no matter comprising Artiodactyla ϩ Cetacea, an Ar- what eutherian group was used to root the tiodactyla ϩ Cetacea ϩ Perissodactyla clade, tree. Accordingly, we chose to retain the out- or a Carnivora ϩ Perissodactyla clade might groups used by Simmons (1998) rather than be the sister-taxon of bats. We considered in- expanding the study to include additional cluding some or all of these orders in our outgroup taxa. study, but preliminary examinations of char- Although choice of outgroups may not be acter variation indicated that all of these taxa critical for determining ingroup relation- exhibited high degrees of taxonomic poly- ships within Chiroptera, it can signi®cantly morphism with respect to the characters in affect assessments of character polarity. We our data set, thus limiting their value as out- therefore note that the polarity assessments groups for the current study. discussed below are based on the assump- Simmons (1998) demonstrated that out- tion that Scandentia and Dermoptera are ap- group choice does not appear to be critical propriate outgroups for investigating pat- for determining ingroup relationships of bats, terns of morphological character evolution at least when morphological data are em- in bats. This assumption was accepted in the ployed. She obtained identical tree topolo- current study because (1) as noted above, gies for Chiroptera regardless of whether considerable morphological and some mo- both outgroups, one outgroup, or no out- lecular data support a close relationship be- groups were included (in the latter case, the tween bats, dermopterans, and tree shrews; 46 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

TABLE 4 (Morphological Synapomorphiesa of Volitantia (Chiroptera ؉ Dermoptera Identi®ed by Simmons (1995)

1) Tooth enamel with horseshoe-shaped prisms with associated minor boundary planes (seams)b 2) Fenestra cochleae (round window) faces directly posteriorlyc 3) Subarcuate fossa greatly expanded and dorsal semicircular canal clearly separated from endocranial wall of squamosalc 4) Tegmen tympani reduced, tapered to a round process, does not form roof over mallear±incudal articulation or entire ossicle chaind 5) Ramus infraorbitalis of the stapedial artery passes through the cranial cavity dorsal to the alisphenoide,f 6) Neural spines on cervical vertebrae 3±7 weak or absentc 7) Ribs ¯attened, especially near vertebral endsc,g 8) Forelimbs markedly elongatedc 9) Proximal displacement of the areas of insertion for the pectoral and deltoid muscles; coalesced single proximal humeral torusg 10) Presence of humeropatagialis musclec 11) Reduction of proximal ulnac 12) Modi®cation of distal radius and ulna: fusion of distal ulna to distal radius; distal radius transversely widened, manus effectively rotated 90Њ; deep grooves for carpal extensors on dorsal surface of distal radius; disengagement and reduction of the ulna from anterior humeral contactf,g 13) Fusion of scaphoid, centrale, and lunate into scaphocentralunateg 14) Patagium continuously attached between digits of manusc,g 15) Elongation of the fourth and ®fth pedal raysg 16) Ungual phalanges both proximally and distally deep, compressed mediolaterallyg 17) Presence of a tendon-locking mechanism on digits of feeth a Features listed have been discussed by other authors as noted. Dermoptera is de®ned here to include extant gliding lemurs (Galeopithecidae ϭ Cynocephalus) ϩ extinct Paromomyidae. This grouping is equivalent to Eudermoptera sensu Beard (1993). Several fossil taxa included in Dermoptera by Beard (micromomyids, plesiadapids, carpolestids, and saxonellids) are excluded here due to ambiguity concerning their relationships (Simmons, 1993a). b Lester et al. 1988. c Wible and Novacek 1988. d Wible and Martin 1993. e Wible 1993. f Simmons 1994. g Szalay and Lucas 1993. h Simmons and Quinn 1994.

(2) dermopterans and tree shrews exhibit shrews similarly constitute a plausible mor- many character states that are widely re- photype for a scansorial arboreal ancestral garded as plesiomorphic for Eutheria form (Novacek, 1980b; Szalay and Dra- (Slaughter, 1970; Cartmill and MacPhee, whorn, 1980). Nevertheless, It must be not- 1980; Novacek, 1980b, 1986; 1990; Nova- ed that our polarity assessments may require cek and Wyss, 1986), thus providing an ap- further modi®cation as better resolution of propriate guide to polarity of many charac- interordinal relationships is achieved. ters used in this study (see character de- Monophyly of Chiroptera was assumed in scriptions for further comments); (3) at least the current study based on the studies pre- with respect to nondental skeletal charac- viously cited (see above). Morphological fea- ters, dermopterans constitute a plausible tures identi®ed by Simmons (1994) as syn- morphotype for a gliding chiropteran ances- apomorphies of Chiroptera are listed in table tor with specializations for underbranch 5. Because bat monophyly was assumed a hanging behavior (Winge, 1941; Smith, priori, characters relevant only to bat mono- 1977; Hill and Smith, 1984; Rayner, 1986; phyly (e.g., most of the features in table 5) Szalay and Lucas, 1993; Simmons and were not included in the data set used in the Quinn, 1994; Simmons, 1995); and (4) tree current study. 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 47

TABLE 5 Morphological Synapomorphies of Chiroptera Identi®ed by Simmons (1994)

1) Deciduous dentition does not resemble adult dentition; deciduous teeth with long, sharp, recurved cusps 2) Palatal process of premaxilla reduced; left and right incisive foramina fused in midsaggital plane 3) Postpalatine torus absent 4) Jugal reduced and jugolacrimal contact lost 5) Two entotympanic elements in the ¯oor of the middle-ear cavity: a large caudal element and a small rostral element associated with the internal carotid artery 6) Tegmen tympani tapers to an elongate process that projects into the middle-ear cavity medial to the epitympanic recess 7) Proximal stapedial artery enters cranial cavity medial to the tegmen tympani; ramus inferior passes anteriorly dorsal to the tegmen tympani 8) Modi®cation of scapula: reorientation of scapular spine and modi®cation of shape of scapular fossae; reduction in height of spines presence; of a well-developed transverse scapular ligament 9) Modi®cation of elbow: reduction of olecranon process and humeral articular surface on ulna; presence of ulnar patella; absence of olecranon fossa on humerus 10) Absence of supinator ridge on humerus 11) Absence of entepicondylar foramen in humerus 12) Occipitopollicalis muscle and cephalic vein present in leading edge of propatagium 13) Digits II±V of forelimb elongated with complex carpometacarpal and intermetacarpal joints, support enlarged interdigital ¯ight membranes (patagia); digits III±V lack claws 14) Modi®cation of hip joint: 90Њ rotation of hindlimbs effected by reorientation of acetabulum and shaft of femur; neck of femur reduced; ischium tilted dorsolaterally; anterior pubes widely ¯ared and pubic spine present; absence of m. obturator internus 15) Absence of m. gluteus minimus 16) Absence of m. sartorius 17) Vastus muscle complex not differentiated 18) Modi®cation of ankle joint: reorientation of upper ankle joint facets on calcaneum and astragalus; trochlea of astragalus convex, lacks medial and lateral guiding ridges; tuber of calcaneum projects in plantolateral direction away from ankle and foot; peroneal process absent; sustentacular process of calcaneum reduced, calcaneoastra- galar and sustentacular facets on calcaneum and astragalus coalesced; absence of groove on astragalus for tendon of m. ¯exor digitorum ®bularis 19) Presence of calcar and m. depressor ossis styliformis 20) Entocuneiform proximodistally shortened, with ¯at, triangular distal facet 21) Elongation of proximal phalanx of digit I of foot 22) Embryonic disc oriented toward tubo±uterine junction at time of implantation 23) Differentiation of a free, glandlike yolk sac 24) Preplacenta and early chorioallantoic placenta diffuse or horseshoe-shaped, with de®nitive placenta reduced to a more localized discoidal structure 25) De®nitive chorioallantoic placenta endotheliochorial 26) Cortical somatosensory representation of forelimb reverse of that in other mammals

THE DATA SET restricted to fewer than half of the extant families, and no published data set includes The basis of our study was the morphol- representatives of more than 7 of the 24 ex- ogy ϩ rDNA restriction site data set origi- nally developed by Simmons (1998) to ad- tant taxa included in the present study (e.g., dress family-level relationships among extant Pettigrew et al., 1989; Adkins and Honey- bats. Although nucleotide sequence data hold cutt, 1993; Stanhope et al., 1993, 1996; Por- great promise for resolving bat phylogeny, ter et al., 1996). Accordingly, nucleotide se- sampling has been poor in previous studies quence data are not included in our data set. with respect to the taxonomic levels under The data set used in our study (see appen- consideration here. As noted by Simmons dices 2, 3) comprises 195 morphological (1998), taxonomic sampling in amino acid characters, 12 rDNA restriction site charac- and nucleotide sequence data sets has been ters, and one character based on the number 48 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 of R-1 tandem repeats in the mtDNA d-loop tures and rDNA restriction sites) are sum- region. Although fundamentally similar to marized in appendix 2. The character order that presented by Simmons (1998), our data established by Simmons (1998) is retained to set includes corrections, some new observa- facilitate comparison between that data set tional data (e.g., hyoid and shoulder myology and the modi®ed version used in this study; data for Antrozoidae), several modi®ed char- gaps in the numerical sequence given below acter de®nitions (e.g., an additional state de- re¯ect characters that could not be scored in ®ned for character 21; characters 75±77 any of the fossil taxa. Higher-level taxonom- jointly replace a single character de®ned by ic names employed below are sensu Sim- Simmons, 1998), and a number of entirely mons (1998; table 2). new characters (i.e., characters 19, 125±127, 138, 143, 154±157, 192, 208). SKULL AND DENTITION Multistate characters were coded hierar- chically following methods described by Character 9: Premaxilla articulates with Simmons (1993a), except in cases where so maxilla via sutures (0); or premaxilla fused doing would result in uninformative charac- to maxilla (1); or premaxilla articulates with ters. For example, if ``absent'' represented maxilla via ligaments, premaxilla freely mov- one condition in a multistate transformation, able (2). The articulation between the max- we de®ned two characters: (A) a present/ab- illa and the premaxilla consists of a simple sent character (e.g., character 86); and (B) a sutured joint in Icaronycteris, Archaeonyc- character describing variations in form of the teris, Hassianycteris, and Palaeochiropteryx. structure when it is present (e.g., character Although Smith and Storch (1981: 158) not- 87). This method perserves homology infor- ed in the original description of Hassianyc- mation without requiring ordered character teris messelensis that ``the premaxillary ap- transformations (Simmons, 1993b). Howev- pears to have a strong nasal branch which er, if the ``absent'' condition occurs in only appears to have been fused to the maxillary,'' one taxon in the analysis, a present/absent examination of new material (e.g., SMF ME character would not be informative. In these 1469b) led us to conclude that the premaxilla cases, we de®ned a multistate character in- was not fused to the maxilla in Hassianyc- cluding ``absent'' as one of the character teris. Instead, the premaxilla±maxilla articu- states (e.g., character 91). lation appears to have consisted of a simple The complete data set used in this study, suture in this taxon, as in the other Eocene including character state descriptions and ref- forms. Among extant bats, a similar sutured erences for data sources, is available elec- joint is seen only among megachiropterans tronically via the World Wide Web at ftp:// (e.g., Pteropus, Rousettus). In contrast, the ftp.amnh.org/pub/mammalogy. premaxilla is fully fused to the maxilla and the sutures are obliterated in adults of some CHARACTERS EXAMINED IN FOSSIL megachiropteran species (e.g., Hypsignathus) BATS and all yangochiropteran bats. Emballonurids and yinochiropterans exhibit yet another de- As a result of imperfect preservation, the rived condition, one in which the premaxilla Eocene specimens available for study (see articulates with the maxilla via ligaments that appendix 1 for list of specimens examined) allow the premaxilla to be freely movable. permitted us to score Icaronycteris, Archaeo- The articulation between the maxilla and the nycteris, Hassianycteris, and Palaeochirop- premaxilla consists of a simple suture in both teryx for most but not all of the hard-tissue outgroups and most other mammals. In this characters in the data set. Below, we present context, the state seen in the Eocene fossil our observations concerning the characters bats apparently represents the primitive con- that we could score in these taxa, as well as dition. the range of variation seen among extant lin- Character 10: Nasal branches of premax- eages with respect to these features. Char- illae well developed (0); or reduced or ab- acter descriptions and complete references sent (1). The nasal branches of the premax- for all characters (including soft-tissue fea- illae are de®ned as those portions of the pre- 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 49

maxillae that lie on the face adjacent to the radiographs (Smith and Storch, 1981: ®gs. 1, external narial opening. The nasal branches 2), an image of the anterior dentition is su- of the premaxillae are relatively well devel- perimposed on the anterior palate. Consid- oped in Icaronycteris, Archaeonycteris, Has- ering the complex nature of this region and sianycteris, and Palaeochiropteryx, as they the high density of the dental images, any are in most extant bats. In contrast, the nasal observations of palatal morphology in these branches of the premaxillae in rhinolophoids specimens must be viewed with caution. We are either entirely absent or are reduced to considered scoring Hassianycteris as having tiny splints of bone. The nasal branches of a reduced palatal branch of the premaxilla the premaxillae are well developed in both based on the report by Smith and Storch outgroups, suggesting that the state seen in (1981), but subsequently noted that reduction the Eocene fossil bats is primitive. and lack of fusion of the palatal branches Character 11: Palatal branches of pre- were omitted from Habersetzer and Storch's maxillae well developed (0); or reduced or (1987) diagnosis of Hassianycterididae absent (1). The palatal branches of the pre- (which contains only Hassianycteris). Given maxillae are de®ned as those portions of the the prominence of these features in Smith premaxillae that contribute to the anterior and Storch's original diagnosis of Hassianyc- palate. The palatal branches of the premax- teris, we can only conclude that there is con- illae are relatively well developed in Icaron- siderable uncertainty about the structure of ycteris and many extant bats (Nycteridae, the premaxilla in these bats. Accordingly, we Rhinolophidae, Phyllostomidae, Noctilioni- score Hassianycteris as ``?'' for this charac- dae, Mormoopidae, Mystacinidae, Myzopod- ter. idae, Thyropteridae, Natalidae, some Molos- Character 15: Hard palate extends pos- sinae). In contrast, the palatal branches of the teriorly into interorbital region (0); or ter- premaxillae are either reduced to tiny splints minates either at or anterior to level of zy- of bone or are entirely absent in other extant gomatic roots (1). The hard palate, which lineages. This reduction is apparently a de- forms a bony separation between the oral and rived condition since well developed palatal nasal passages, terminates posteriorly at the branches are present in both outgroups and mesopterygoid fossae. Position of the pos- most other mammals. terior edge of the hard palate varies indepen- The condition of the anterior palate in Ar- dently of the posterior extent of the molar chaeonycteris and Palaeochiropteryx could toothrow, and is apparently linked to the not be determined from available specimens. structure of the nasal passages and soft tis- We found that the palatal branches of the pre- sues of the pharyngeal region. The hard pal- maxillaries are not visible in specimens pre- ate extends posteriorly into the interorbital pared in dorsal view, and that the mandibular region in Archaeonycteris, Palaeochiropte- rami obscure the palate in all specimens pre- ryx, one of the outgroups (Scandentia), and pared in ventral view. Accordingly, these many extant bats (Pteropodidae, some Em- forms are scored ``?'' for this character. ballonuridae, some Hipposiderinae, some Smith and Storch (1981: 154, 164) noted Phyllostomidae, Mormoopidae, Noctilioni- that in Hassianycteris the ``palatal branch [of dae, Mystacinidae, Myzopodidae, Thyropter- the premaxilla] not well developed, premax- idae, some Furipteridae, Natalidae, Antrozo- illaries not fused [at midline],'' and that ``the idae, Tomopeatinae, some Molossidae, and apparent shape and reduction of the premax- Vespertilionidae). In contrast, the hard palate illary is quite reminiscent of the derived con- terminates either at or anterior to the level of dition of this element in emballonurids.'' the zygomatic roots in the other outgroup However, the source(s) of these observations (Dermoptera) and all remaining extant bats. were not discussed, and we were unable to Lack of agreement between the two out- con®rm them during our microscopic exam- groups precludes a priori determination of ination of specimens. Smith and Storch ap- the primitive condition for this character. The parently described these features based on ra- postition of the posterior end of the hard pal- diographs, although the published versions ate in Icaronycteris and Hassianycteris could are too fuzzy to be of use. In both published not be determined due to the position of the 50 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 mandibular rami in all avialable specimens, lioninae. The upper incisors are entirely ab- so these forms are scored ``?'' for this char- sent in some Pteropodidae and Megaderma- acter. tidae. Both outgroups have two upper inci- Character 16: Two upper incisors in each sors on each side of the jaw. In this context, side of jaw (0); or one incisor (1); or upper the state seen in Archaeonycteris, Hassianyc- incisors absent (2). Because of dif®culties teris, and Palaeochiropteryx apparently rep- associated with determining homologies of resents the primitive condition. We were un- the anterior dental loci among different bat able to unambiguously determine the number lineages (reviewed by Slaughter, 1970), we of upper incisors in Icaronycteris, so this tax- chose to score the number of teeth in various on is scored ``?'' for this character. dental regions (i.e., upper incisors, lower in- Character 17: Three lower incisors in cisors, upper premolars, lower premolars) each side of jaw (0); or two incisors (1); or rather than attempting to score presence/ab- one incisor (2); or incisors absent (3). Three sence of teeth at speci®c loci. Although some lower incisors are present on each side of the potentially informative patterns may be over- jaw in Icaronycteris, Archaeonycteris, Has- looked by this method, this approach pre- sianycteris, and Palaeochiropteryx. This is serves that homology information of which the same lower incisor formula seen in many we are most con®dent and permits scoring of extant bats (some Emballonuridae, Nycteri- all taxa (including the outgroups) for each dae, Nataloidea, some Antrozoidae, some character. Molossinae, and Vespertilionidae). Two low- The size of one or more teeth in a number er incisors are present on each side of the of species has been reduced to the point lower jaw in some Pteropodidae, some Em- where these teeth are considered vestigial ballonuridae, Rhinopomatoidea, Megader- (Slaughter, 1970). This is particularly com- matidae, Rhinolophidae, some Phyllostomi- mon in the premolar dentition. In some in- dae, Mormoopidae, some Antrozoidae, To- stances there is within-species (and even mopeatinae, and some Molossinae. A single within-individual) polymorphism with re- lower incisor is present on each side in some spect to presence/absence of vestigial teeth Pteropodidae, some Phyllostomidae, Nocti- (e.g., the anterior upper premolar in Rhino- lionidae, Mystacinidae, and some Molossi- lophus clivosus, the middle lower premolar nae. Finally, the lower incisors are entirely in Chrotopterus auritus). In our experience, absent in some Pteropodidae and some Phyl- older individuals frequently lack such teeth, lostomidae. Both outgroups have three lower while they are often present in younger ani- incisors on each side. In this context, the mals. This pattern suggests that vestigial state seen in the Eocene bat genera appar- teeth are often lost during the lifetime of the ently represents the primitive condition. individual. Consequently, we have scored the Character 18: Three upper premolars in taxa in question as having the higher dental each side of jaw (0); or two premolars (1); formula in cases of within-species polymor- or one premolar (2). Three upper premolars phism in presence/absence of vestigial teeth. are present on each side of the jaw in Ica- Two upper incisors are present on each ronycteris, Archaeonycteris, Hassianycteris, side of the jaw in Archaeonycteris, Hassi- and Palaeochiropteryx magna. Hassianycter- anycteris, and Palaeochiropteryx. This is the is messelensis has only two upper premolars same upper incisor formula as seen in many (®g. 24), apparently having lost the tiny peg- extant bats (some Pteropodidae, some Em- like anterior premolar seen in H. magna. ballonuridae, Nycteridae, some Phyllostomi- Among extant bats, three upper premolars dae, Mormoopidae, Noctilionidae, Natalo- are present in some Pteropodidae, some idea, some Vespertilioninae, Miniopterinae, Phyllostomidae, Myzopodidae, Thyropteri- Myotinae, Murininae, and Kerivoulinae). In dae, Natalidae, some Myotinae, and Kerivou- contrast, only one incisor is present on each linae. In contrast, two upper premolars are side of the upper jaw in some Pteropodidae, present on each side of the jaw in some Pter- some Emballonuridae, Rhinopomatoidea, opodidae, Emballonuridae, some Megader- Rhinolophidae, some Phyllostomidae, Mys- matidae, Rhinolophinae, some Hipposideri- tacinidae, Molossoidea, and some Vesperti- nae, some Phyllostomidae, Mormoopidae, 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 51

nae. Three roots are present on P3 in both outgroups, suggesting that the triple-rooted condition seen in Icaronycteris is primitive. The double-rooted condition seen in the oth- er Eocene taxa is apparently derived. Be- cause it is thought that P3 is often the ®rst tooth lost from the premolar dentition in spe- cies with a reduced number of teeth (Slaugh- ter, 1970), we scored this character only in forms that retain three upper premolars (state ``0'' of character 18 above). Accordingly, Emballonuridae, Yinochiroptera, Mormoopi- dae, Noctilionidae, Mystacinidae, Furipteri- dae, Molossoidea, Vespertilioninae, Miniop- terinae, and Murininae are scored ``-'' for Fig. 24. Hassianycteris messelensis (SMF this character. ME 1414a). Closeup lateral view of skull. Note Character 20: Three lower premolars in the robust dentary and dentition, and the absence each side of jaw (0); or two premolars (1). of the ®rst upper premolar. A vestigal ®rst upper Three premolars are present on each side of premolar is present in H. magna. AN, angular the lower jaw in Icaronycteris, Archaeonyc- process of dentary; VAP, ventral accessory pro- teris, Hassianycteris and Palaeochiropteryx. cess of vertebra (C3?). Scale bar ϭ 10 mm. From Among extant bats, three lower premolars Habersetzer and Storch (1987: ®g. 5). Photo by are present in Rhinolophinae, some Phyllos- E. Pantak (Senkenbergmuseum). tomidae, Mormoopidae, Nataloidea, some Vespertilioninae, Miniopterinae, some My- Mystacinidae, Furipteridae, some Molossi- otinae, and Kerivoulinae. Only two lower nae, some Vespertilioninae, Miniopterinae, premolars are present in all other lineages. some Myotinae, and Murininae. A single up- Among the outgroups, Scandentia has three per premolar is present on each side in Rhin- lower premolars, while Dermoptera has two. opomatoidea, Nycteridae, some Megader- Lack of agreement between the two out- matidae, some Hipposiderinae, some Phyl- groups precludes a priori determination of lostomidae, Noctilionidae, Antrozoidae, To- the primitive condition for this character. mopeatinae, some Molossinae, and some Character 21: Lower ®rst and second mo- Vespertilioninae. Among the outgroups, lars with primitive tribosphenic arrangement Scandentia has three upper premolars, while of cusps and cristids (0); or nyctalodont (1); Dermoptera has two. Lack of agreement be- or myotodont (2); or teeth modi®ed for fruit tween the two outgroups precludes a priori and/or nectar or blood feeding, cusps and determination of the primitive condition for cristids not distinct (3). Menu and Sige this character. (1971) discussed molar morphology in bats, Character 19: Middle upper premolar and distinguished two forms of the talonid of with three roots (0); or with two roots (1); m1 and m2 that they termed ``nyctalodonty'' or with one root (2). Among bats with three and ``myotodonty.'' In both morphotypes, upper premolars, the middle premolar (usu- the hypoconulid lies adjacent to the entocon- ally designated P3) may have either one, two, id on the lingual edge of the tooth, having or three roots. P3 has three roots in Icaron- shifted lingually from the midline position ycteris. In contrast, this tooth has only two (where it is more or less equidistant from the roots in Archaeonycteris, Hassianycteris, and labial and lingual borders of the tooth) char- Palaeochiropteryx. Among extant bats, P3 is acteristic of primitive therian tribosphenic triple-rooted only in Natalidae. P3 has two molars. Nyctalodonty is de®ned by the po- roots in Pteropodidae, some Phyllostomidae, sition of the postcristid, which connects the Thyropteridae, and some Kerivoulinae. P3 is hypoconid with the hypoconulid in nyctalo- single-rooted in some Phyllostomidae, My- dont forms. Myotodonty is distinguished by zopodidae, Myotinae, and some Kerivouli- an alternative arrangement in which the post- 52 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 cristid bypasses the hypoconulid to connect rhinolophids that feed on soft-bodied prey instead with the entoconid. The postcristid is (e.g., Rhinolophus blasii) and those that feed relatively high and forms a sharp shearing on hard-bodied prey (e.g., Hipposideros edge in both conditions, but no cristid runs commersoni) both have nyctalodont teeth between the entoconid and hypoconulid. (Strait, 1993). It therefore seems premature Although it is easy to classify molar mor- to draw any functional conclusions from phology of most extant bats, Menu and Sige transformations in talonid structure in bats. (1971) noted that problems arise when Eo- The ®rst lower molars exhibit the primi- cene bats are considered because the hypo- tive tribosphenic arrangement of the talonid conulid is located in the midline position in cusps and cristids in Archaeonycteris. In con- some taxa, always connected to the hypocon- trast, the lower molars are nyctalodont in id and sometimes connected to the entoconid Hassianycteris. Both nyctalodont and myoto- by low cristids. The posterior edge of the tal- dont conditions appear in Palaeochiropteryx, onid is thus dominated by a large conical although never in the same individual. Both cusp (the hypoconulid) rather than by a nyctalodont and myotodont specimens have shearing crest (the postcristid). This mor- been referred to Palaeochiropteryx tupaio- phology, termed ``archaic'' by Menu and don (e.g., by Russel and SigeÂ, 1970), raising Sige (1971), appears in numerous Late Cre- the possiblity of within-species polymor- taceous and Paleocene mammals, including phism. However, we are unaware of any forms thought to represent the earliest mem- cases of within-species polymorphism in this bers of several eutherian radiations (e.g., pri- character in extant bats. This suggests that P. mates, insectivores; Menu and SigeÂ, 1971). tupaiodon may be a composite species, a hy- This apparently represents the primitive tri- pothesis that will require testing with addi- bosphenic condition. tional data. Meanwhile, we score Palaeochi- Possible differences in function of the ropteryx as exhibiting both nyctalodonty and three talonid types discussed above have not myotodonty. yet been examined. Nyctalodont and myoto- Van Valen (1979) claimed that the lower dont teeth are at least super®cially similar in molars of Icaronycteris are nyctalodont, an terms of the length and height of the post- observation that was subsequently cited by cristid, which may indicate an increased re- other authors (e.g., Gingerich, 1987). We dis- liance on shearing as compared to the prim- agree with Van Valen's assessment. Our ob- itive tribosphenic condition. Increased shear- servations indicate that Icaronycteris index ing has been associated with specializations had primitive tribosphenic lower molars. The for feeding on hard-shelled beetles (Freeman, European specimens referred to ?Icaronyc- 1979) and on soft-bodied prey (Strait, 1993). teris by Russell et al. (1973) include both Freeman (1979) found that molossid bats that nyctalodont and primitive tribosphenic mor- feed on beetles have relatively larger teeth photypes. However, this material consists of with longer shearing cusps than do molossids isolated teeth only, and we consider the af- that feed on soft-bodied prey. Alternatively, ®nites of these specimens to each other and Strait (1993) found that mammalian insecti- to Icaronycteris index to be uncertain at best. vores that feed predominantly on soft-bodied Accordingly, we score Icaronycteris as ex- prey (e.g., moths and caterpillars) have rel- hibiting only the primitive tribosphenic con- atively longer shearing crests (summed dition. across the entire molar) than do close rela- Among the extant families, the primitive tives that have more generalized feeding hab- tribosphenic condition of the talonid is seen its and consume large quantities of hard-bod- in fossil members of Hipposiderinae (i.e., ied beetles. Whatever the relationship be- Palaeophyllophora) and Megadermatidae tween shearing and prey type, these studies (i.e., Necromantis; Beard et al., 1992). Nyc- suggest that variation in talonid form may be talodonty occurs in Emballonuridae, Rhino- indicative of differences in dietary habits. pomatoidea, Nycteridae, extant Megaderma- However, there does not seem to be any ob- tidae, Rhinolophinae, extant Hipposiderinae, vious correlation between feeding habits and some Phyllostomidae, Mormoopidae, Furip- talonid type in extant bats. For example, teridae, Natalidae, some Molossinae, some 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 53

Fig. 25. Icaronycteris index (UW 2244). Ventral view of the basicranium and posterior lower jaws. From Novacek (1987: ®g. 2). An, angular process of dentary; Co, cochlea; Ect, ectotympanic; Gl, glenoid fossa; Oa, orbicular apophysis; Sl, stylohyal; Z, zygomatic arch.

Vespertilioninae, Miniopterinae, some My- and on a common raphe dorsal, posterior, and otinae, and Murininae. In contrast, myoto- ventral to the angular process (Turnbull, donty is found in some Phyllostomidae, Noc- 1970). This pattern of muscle insertion is tilionidae, Mystacinidae, Myzopodidae, Thy- seen in bats that have an elongate angular ropteridae, Antrozoidae, Tomopeatinae, some process (Storch, 1968; Kallen and Gans, Molossinae, some Vespertilioninae, some 1972). Both m. masseter and m. pterygoideus Myotinae, and Kerivoulinae. In Pteropodidae internus act to close and protrude the lower and some Phyllostomidae the teeth are mod- jaw. M. masseter also pulls the ipsilateral i®ed for fruit and/or nectar or blood feeding dentary (the dentary on the same side of the and the cusps and cristids are not distinct. skull as the muscle) laterally, while m. pter- Among the outgroups, Scandentia exhibits ygoideus internus pulls the ipsilateral dentary the primitive tribosphenic condition, but we lingually (Storch, 1968; Kallen and Gans, could not score Dermoptera owing to its 1972). unique dental morphology. Although polarity The lower jaw has an elongate angular cannot be ascertained unambiguously, the process in Icaronycteris (®gs. 25, 26), Ar- data suggest that the primitive tribosphenic chaeonycteris, Hassianycteris (®g. 24), and condition is primitive for bats. Palaeochiropteryx (®g. 27). This is similar to Character 22: Lower jaw with elongate the condition seen in most bats. In contrast, angular process (0); or without elongate an- the angular process is either very short (rel- gular process (1). The angular region of the ative to its dorsoventral width) or is effec- lower jaw in mammals is the site of insertion tively absent in Pteropodidae, Craseonycter- of jaw adductor muscles including m. mas- idae, and some Nycteridae. An elongate an- seter pars super®cialis, m. masseter pars pro- gular process is present in Scandentia but not fundus, m. zygomaticomandibularis, m. man- in Dermoptera. Lack of agreement between dibuloauricularis, and m. pterygoideus inter- the two outgroups precludes a priori deter- nus (ϭ medial pterygoid; Turnbull, 1970). In mination of the primitive condition for this many ``generalized'' eutherian mammals and character. However, occurrence of an elon- some specialized carnivores, a bony exten- gate angular process in many presumably sion of the angleÐtermed the angular pro- primitive eutherians (e.g., leptictids, Asio- cessÐlies between the insertions of m. pter- ryctes, lipotyphlans) suggests that presence ygoideus internus and m. masseter. These of an elongate angular process may be prim- muscles insert both on the angular process itive for Eutheria. 54 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Fig. 26. Icaronycteris index (UW 21481a±b). Stereophotographs of a ventral view of the basicranial region (top) with labelled key photograph (bottom center). AN, angular process of dentary, BS, basi- sphenoid; CP, coronoid process of dentary; CTP, caudal process of tympanic; ER, epitympanic recess; PGF, postglenoid foramen; PR, periotic; SL, tylohyal; SQ, squamosal; Z, zygomatic arch. Photographs by C. Tarka.

Character 23: Angular process projects at chiropteryx (®g. 27). This is similar to the or below level of occlusal plane of toothrow, condition seen in most extant bats and both well below coronoid process (0); or angular outgroups. In contrast, the angular process process projects above level of occlusal plane projects above the level of the occlusal plane of toothrow, at same level as the coronoid of the toothrow (roughly at the same level as process (1). The angular process, when pres- the coronoid process) in Myzopodidae, Thy- ent, extends from the posteroventral ``corner'' ropteridae, and Furipteridae. In this context, of the lower jaw. The angular process projects the condition in the Eocene fossil bats appears at or below the level of the occlusal plane of to be primitive. This character cannot be eval- the toothrow (well below the coronoid pro- uated in taxa that lack an angular process cess) in Icaronycteris (®gs. 25, 26), Archaeo- (e.g., taxa scored ``1'' for 22); these forms are nycteris, Hassianycteris (®g. 24), and Palaeo- therefore scored ``-'' for this character). 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 55

Fig. 27. Palaeochiropteryx tupaiodon, positive prints of radiographs; from Novacek (1987: ®g. 4). A. Dorsoventral view of the skull of SMF ME 788b. B. Lateral view of SMF ME 1127. An, angular process of dentary; Co, cochlea; Cp, coronoid process of dentary; Ect, ectotympanic; Sl, stylohyal; Z, zygomatic arch.

Character 24: Postorbital process present extant microchiropterans (with the exception (0); or absent (1). The postorbital process is of some Emballonuridae) is not fused with a laterally projecting process of the frontal or sutured to the basisphenoid. Instead, the that forms part of the posterodorsal rim of pars cochlearis is loosely attached to the ba- the orbit. A postorbital process is present in sisphenoid via ligaments and/or thin splints Archaeonycteris, Hassianycteris, and Pa- of bone (®g. 28). This condition (sometimes laeochiropteryx. Among extant forms, a referred to as ``cochlear isolation'') is postorbital process is present only in Ptero- thought to function in reducing bone con- podidae, Emballonuridae, Nycteridae, and duction of laryngeal vibrations (Henson, Megadermatidae. All other extant bats lack a 1970; Van Valen, 1979). Although there is postorbital process. A postorbital process is some variation in the degree of isolation of present in both outgroups, suggesting that the cochlea as judged by the relative sizes of presence of this structure represents the prim- the openings that surround the periotic (the itive condition. We were not able to ascertain anterolateral pyriform fenestra, anteromedial whether this process is present in Icaronyc- to medial parts of the basicochlear ®ssure, teris; this taxon is therefore scored ``?'' for and posteromedial jugular foramen; Nova- this feature. cek, 1991), this ``loosely attached'' condition Character 25: Pars cochlearis of petrosal clearly differs from the typical mammalian sutured to basisphenoid (0); or loosely at- pattern in which the periotic is ®rmly sutured tached to basisphenoid via ligaments and/or to the surrounding bones of the basicranium thin splints of bone (1). The petrosal of all including the basisphenoid. 56 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Fig. 28. Nyctalus noctula (Vespertilioninae; redrawn from Henson, 1970: ®g. 9). Ventral view of the basicranium and audtory region showing major osteological features, blood vessels, and muscles. The ectotympanic and malleus have been re¯ected laterally on the right side of the skull. APP, anterior process of petrosal; BF, basicochlear ®ssure; BO, basioccipital; BOP, basioccipital pit; CO, cochlea; ECT, ectotympanic; ETR, epitympanic recess; FC, fenestra cochleae (ϭ round window); FM, foramen mag- num; FO, foramen ovale; FV, fenestra vestibuli (ϭ oval window); GF, glenoid fossa; I, incus; ICA, internal carotid artery; M, malleus; MS, m. stapedius; MTT, m. tensor tympani; OAM, orbicular apoph- ysis of malleus; OC, occipital condyle; PF, pyriform fenestra; PGF, postglenoid foramen; PP, paroccipital process; PT, pterygoid hamulus; SF, stapedial fossa; ST, stapes; STA, stapedial artery; T, tendon of m. tensor tympani; TM, tympanic membrane; Z, zygomatic arch.

Based on observed gaps and patterns of turn) is a specialization for perceiving the breakage in the basicranium of preserved echoes of high-frequency echolocation sig- specimens of Archaeonycteris and Palaeo- nals (Henson, 1970; Segall, 1971; Bruns et chiropteryx, it appears that the periotic was al., 1983±1984; Novacek, 1985a, 1987, loosely attached to the basisphenoid (and the 1991; Habersetzer and Storch, 1992). By cochlea relatively isolated) in these taxa. In plotting maximum external cochlear width contrast, the periotic is sutured to the basi- against skull length, Novacek (1985a, 1987, sphenoid in Pteropodidae and some Embal- 1991) found that extant Microchiroptera and lonuridae. A similar condition is seen in both Megachiroptera have nonoverlapping distri- outgroups and most other mammals, sug- butions, with microchiropterans having con- gesting that the ``sutured'' condition is prim- sistently larger cochleae. The megachiropter- itive, and that the ``loosely attached'' con- an condition (i.e., a relatively small cochlea) dition seen in Archaeonycteris and Palaeo- was interpreted as the primitive condition be- chiropteryx is derived. We were unable to de- cause it also occurs in many other small to termine the type of periotic connection in medium-sized mammals (Novacek, 1985a, Icaronycteris and Hassianycteris; these taxa 1987, 1991). are therfore scored ``?'' for this character. Novacek (1985a, 1987, 1991) also esti- Character 26: Cochlea not enlarged (0); mated cochlear width and skull length of Ica- or moderately enlarged (1); or greatly en- ronycteris and Palaeochiropteryx, and found larged (2). Much has been written about that these fossil taxa fell within the range of cochlear size and echolocation in bats, with variation observed in extant Microchiroptera. most authors agreeing that cochlear enlarge- Because an enlarged cochlea is derived and ment (speci®cally enlargement of the basal is associated with use of sophisticated echo- 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 57 location,3 these observations led Novacek methods were biased by his use of skull (1985a, 1987, 1991) to conclude that Icaron- length as a measure of body size, because ycteris and Palaeochiropteryx were echolo- skull length also re¯ects dietary habits and cating bats closely related to extant Micro- structure of the dentition. Habersetzer and chiroptera. Storch alleviated this problem by using ba- In another study of relative cochlear size sicranial width (measured between the out- in bats, Habersetzer and Storch (1992) sug- ermost bony margins of the semicircular ca- gested that Novacek's (1985a, 1987, 1991) nals) as an indicator of size, and also adopted a more precise method of measuring cochlear 3 We recognize two categories of echolocation sys- width. Using radiographs, they measured tems based on the information that is obtained from re- cochlear width from the end of the ®rst half sulting echoes. We use the term ``sophisticated echolo- turn of the cochlea to the end of the second cation'' to describe systems that can be used for orien- half turn, thus guaranteeing homologous tation in cluttered environments (e.g., within vegetation measurements that excluded the promonto- or under the forest canopy) and for detection, tracking, rium and vestibular system. Using these and evaluation of moving objects including prey. Calls methods, Habersetzer and Storch obtained associated with sophisticated echolocation are produced results different from those of Novacek in the larynx and are often (but not always) highly struc- (1985a, 1987, 1991). Rather than having tured in terms of call length, frequency, and type of modulation. Call structure, duration, and pulse interval completely nonoverlapping distributions, may be intentionally varied by the animal depending on they found that Microchiroptera (represented the circumstances (e.g., searching for prey, approaching by more than 200 species) and Megachirop- prey, attacking prey), and call structure is often species tera (more than 70 species) exhibited a nar- speci®c. There is a strong correlation between call type row zone of overlap in relative cochlear size and ¯ight/foraging strategies, in part because the infor- (®g. 29). Some microchiropterans with un- mation content of returning echoes varies depending on usually small cochleae for their body sizes call structure and pulse interval. However, all sophisti- (e.g., Phyllostomus hastatus, Leptonycteris cated echolocators apparently record and process infor- nivalis, Carollia perspicillata, Megaderma mation of several types, including relative timing of lyra, Megaderma spasma) fell within or on pulse and echo (including different arrival times at the right and left ears), differences in intensity between the border of the smallest polygon containing pulse and echo, and (in some cases) frequency shifts nonecholocating megachiropterans (Haber- between pulse and echo. Sophisticated echolocation ap- setzer and Storch, 1992). The observed pat- parently provides the animal with a detailed acoustic tern of distribution (®g. 29) was interpreted map of its environment, a map that changes moment by as evidence that cochlear size in bats varies moment depending on movement of both the echoloca- essentially continuously from Megachirop- tor and the objects in its environment. This form of tera (with the smallest cochleae, lacking so- echolocation occurs in all extant microchiropterans but phisticated echolocation) through Rhinolo- not in megachiropterans. phidae (with the largest cochlea and highly In contrast, we use the term ``primitive echolocation'' sophisticated echolocation), with cochlear to refer to less complex systems in which echoes of broadband clicks produced by the animal (by tongue- size mirroring the functional signi®cance of clicks, voice, or some other means) are used to obtain the acoustic sense. general information about the animals surroundings. Habersetzer and Storch (1992) also mea- Pulses of sound used in primitive echolocation systems sured cochlear and basicranial width in Ar- are always broadband, low in intensity, short in duration, chaeonycteris trigonodon, A. pollex, Palaeo- and never exhibit structured changes in frequency over chiropteryx tupaiodon, P. spiegeli, Hassi- time. The data used by the animal appears to consist anycteris messelensis, and H. revilliodi, and only of differences in timing of pulse and returning compared the resulting measurements with echo. The information obtained using primitive echolo- data from extant forms. They found that both cation is probably limited to detection of relatively large, species of Archaeonycteris had a relatively stationary obstacles (e.g., cave walls); this form of echo- location is not useful for detecting and tracking prey or small cochlea, falling in the zone of overlap orientation in cluttered environments. See discussion un- between Megachiroptera and Microchirop- der ``Evolution of Echolocation and Foraging Strate- tera, close to the point representing Megad- gies'' for references and more information about the tax- erma spasma. In contrast, both species of Pa- onomic distribution of different echolocation systems. laeochiropteryx and both species of Hassi- 58 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Fig. 29. Plot of the oblique diameter (width) of the cochlea versus basicranial width; see text for discussion. This graph was redrawn from Habersetzer and Storch (1992: ®g. 4) and several new data points were added. The zone of overlap between Megachiroptera and Microchiroptera is enclosed with a dashed line. Numbered data points refer to the following taxa: 1, Icaronycteris index;2,Archaeonyc- teris trigonodon;3,Archaeonycteris pollex;4,Palaeochiropteryx tupaiodon;5,Palaeochiropteryx spie- geli;6,Hassianycteris revilliodi;7,Hassianycteris messelensis;8,Rousettus aegypticus;9,Rousettus leschenaulti; 10, Rhinopoma microphyllum; 11, Rhinopoma hardwickei; 12, Craseonycteris thonglon- gyai; 13, Megaderma lyra; 14, Megaderma spasma; 15, Nycteris grandis; 16, Nycteris macrotis; 17, Nycteris thebaica; 18, Nycteris hispida; 19, Phyllostomus hastatus; 20, Trachops cirrhosus; 21, Lepto- nycteris nivalis; 22, Carollia perspicillata; 23, Desmodus rotundus; 24, Noctilio albiventris; 25, Mor- moops megalophylla; 26, Mystacina robusta; 27, Myzopoda aurita; 28, Thyroptera tricolor; 29, Antro- zous pallidus; 30, Tomopeas ravus; 31, Kerivoula pelucida; 32, Tupaia glis; 33, Tupaia javanica. Values for Cynocephalus variegatus (not shown on graph): basicranial width ϭ 28.1; cochlear width ϭ 3.85. 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 59 anycteris had a relatively larger cochlea, fall- plied to specimens from which the bullae and ing slightly above the zone of overlap in the ear ossicles had been removed. This method lower edge of the vespertilionid range. allowed us to plot cochlea width/basicranial In our attempts to score cochlear size as a width for several taxa not mentioned in their phylogenetic character, we evaluated the study, including outgroup taxa (Tupaia, Cy- studies of both Novacek (1985a, 1987, 1991) nocephalus) and several extant bats not ex- and Habersetzer and Storch (1992) and found plicitly mentioned by them (e.g., Nycteris evidence that led us to prefer the approach [additional species], Noctilio, Mystacina, adopted by the latter authors. As previously Myzopoda, Thyroptera, Furipterus, Antro- noted, skull length (as used by Novacek) is zous, Tomopeas, Kerivoula). correlated with diet and is not a good mea- We estimated cochlear width and basicra- sure of body size. For example, long-nosed nial width for Icaronycteris based on UW nectar-feeding phyllostomids such as Lepto- 2244 (®g. 25). Basicranial width was esti- nycteris have skull lengths equivalent to mated by doubling the observed width of the those seen in bats more than twice their body left half of the basicranium, which is less dis- mass. Basicranial width (as measured by Ha- torted than the right side. To obtain cochlear bersetzer and Storch, 1992) is much less sub- width, we estimated the width of the basi- ject to this bias. We also found Novacek's occipital, width of both epitympanic reces- method of measuring cochlear size somewhat ses, and width of portions of the squamosal problematic. He measured the maximum exposed lateral to the ear region. These mea- width of the cochlea in a plane perpendicular surements were summed and the total sub- to the long axis of the skull (Novacek, per- tracted from basicranial width. The resulting sonal commun.). We examined many of the value was divided by 2 to provide an esti- same specimens used by Novacek, and found mate of width of the cochlea. This method that the region of maximum width of the assumes that the basicapsular fenestra, if cochlea varied in location depending on present, was not of signi®cant width. cochlear shape, orientation, size of the fenes- To re¯ect the variation seen in cochlear tra cochleae, and thickness of the bone form- size in bats (and the correlation between ing the cochlear wall lateral to the fenestra width of the basal turn of the cochlea and cochleae. Accordingly, Novacek's measure- use of sophisticated echolocation), we adopt- ments may not have been taken from ho- ed a tripartite scoring scheme for this feature mologous points in all taxa, and may have based on ®gure 29. Three states are recog- regularly overestimated size of the cochlea in nized: cochlea not enlarged, cochlea moder- taxa with a large fenestra cochleae with a ately enlarged, and cochlea greatly enlarged. thickened lateral wall. In contrast, Habersetz- Taxa that fall below the zone of overlap er and Storch's (1992) measure of cochlear between Megachiroptera and Microchirop- width (taken across the second half of the tera (as plotted in ®g. 29) were scored as basal turn of the cochlea) has several advan- having a cochlea that is not enlarged. This tages, including (1) it is based on homolo- condition is seen in most Pteropodidae, and gous points and is not affected by differences our estimates of values for Scandentia and in size of the fenestra cochleae or thickness Dermoptera (also Habersetzer and Storch's of the lateral wall of the cochlea in that re- [1992] calculations for small primates, insec- gion; and (2) it is biologically meaningful be- tivores, and rodents) indicate that a relatively cause it measures the width of the basal turn small cochlea is the primitive condition. of the cochlea, which is the region where None of the extant members of these groups echolocation calls are perceived (Henson, use sophisticated echolocation comparable to 1970; Burns et al., 1983±1984; Habersetzer that seen in extant microchiropterans (see and Storch, 1992). discussion under ``Evolution of Echolocation Using the ®gures reported by Habersetzer and Foraging Strategies'' below). and Storch (1992) for several species (e.g., Taxa were scored as having a moderately Trachops cirrhosus, Megaderma spasma), enlarged cochlea when they fell within the we found that we could reproduce their mea- zone of overlap between Megachiroptera and surements quite closely using calipers ap- Microchiroptera as outlined in ®gure 29. Two 60 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 fossil bats fall in this zone: Icaronycteris and labyrinth from external view in adults (No- Archaeonycteris. Among extant taxa, this vacek, 1985b, 1991). condition is seen in some Pteropodidae, some Novacek (1991) hypothesized that the dif- Megadermatidae, some Phyllostomidae, and ferences between the cryptocochlear and at least one member of Mystacinidae (Mys- phanerocochlear conditions resulted from tacina robusta, now extinct). Some of these differing rates of ossi®cation of the cochlea, forms (e.g., the megachiropterans) do not use with phanerocochlear forms having a slower echolocation, but others do (e.g., the micro- rate of bone deposition (paedomorphosis). chiropterans). However, none of the latter are An alternative explanation is that these dif- typical aerial insectivores. Instead, most ap- ferent cochear conditions may result from parently rely on passive acoustic cues and/or differential growth rates of the cochlear lab- vision to detect large invertebrate or small yrinth (whose growth requires bone resorp- vertebrate prey that is captured by gleaning tion from the inner side of the cochlear wall) or landing on the prey, or they feed on fruit and the outer wall of the periotic (whose or nectar detected by vision or olfaction growth requires bone deposition in the area (Norberg and Rayner, 1987; Habersetzer and of the promontorium). The cochlea of most Storch, 1992). Mystacina tuberculata appar- embryonic mammals has a very thin, trans- ently combines all of these foraging strate- parent promontorial wall (Novacek, 1991). In gies with some aerial hawking for ¯ying in- the cryptocochlear condition, growth of the sects (Daniel, 1976, 1979, 1990; B. Lloyd cochlear labyrinth apparently does not keep and S. Parsons, personal commun.). How- pace with growth of the outer wall of the ever, nothing is known of the foraging be- promontorium during ontogeny. Resorption havior of Mystacina robusta, which was the on the inner surface does not compensate for only mystacinid available for our study. deposition on the outer surface, and thus a Taxa were scored as having a greatly en- thick cochlear wall forms as development progresses. This stands in contrast to the pha- larged cochlea when they fell above the zone nerocochlear condition, in which both pro- of overlap between Megachiroptera and Mi- cesses (resorption and deposition) appear to crochiroptera as plotted in ®gure 29. This keep pace throughout development, resulting condition is seen in two fossil taxa (Hassi- in a thin, transparent promontorium that re- anycteris and Palaeochiropteryx) and in all sembles that seen in the embryo. Although a extant microchiropteran groups except some reduction in rate of external bone deposition Megadermatidae, some Nycteridae, some (paedomorphosis) is possible, it seems much Phyllostomidae, and Mystacinidae. All ex- more likely that an increase in growth rate of tant forms with a greatly enlarged cochlea (as the cochlear labyrinth (peramorphosis) is re- de®ned here) use sophisticated echolocation, sponsible. This developmental hypothesis is and most are expert aerial hawkers (Norberg supported by the observation that many and Rayner, 1987; Habersetzer and Storch, cryptocochlear microchiropterans (e.g., Me- 1992). gadermatidae, Nycteridae, some Phyllostom- Character 27: Cochlea cryptocochlear idae) exhibit cochlear widths that fall in the (0); or phanerocochlear (1). Novacek zone of overlap with megachiropterans (i.e., (1985b, 1991) described variation in cochlear a ``moderately enlarged'' cochlea), while structure among bats and recognized two dis- most phanerocochlear microchiropterans tinct patterns of petrosal ossi®cation and have what we term a ``greatly enlarged'' adult cochlear morphology. The ``phanero- cochlea (see character 26 above). However, cochlear'' state occurs when the petrosal wall some exceptions exist. For example, hippos- is thin and poorly ossi®ed, resulting in a con- iderines have a very large cochlea relative to dition where the cochlear labyrinth is clearly most other microchiropterans (Habersetzer visible externally in the adult (Novacek, and Storch, 1992), but many have a crypto- 1985b, 1991). In contrast, a ``cryptococh- cochlear cochlea. Accordingly, we treat lear'' condition occurs when strong petrosal cochlear enlargement and form of the coch- ossi®cation produces a thicker encasement of lear wall (cryptocochlear or phanerococh- bone around the cochlea, hiding the cochlear lear) as separate characters. 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 61

As judged from comparisons of radio- Pteropodidae, Nycteridae, Megadermatidae, graphs of fossil and extant forms published and some Phyllostomidae. In both outgroups, by Habersetzer and Storch (1989, 1992) and the lateral process of the tympanic is poorly Habersetzer et al.(1994), Palaeochiropteryx developed or absent. Accordingly, the con- and Hassianycteris apparently exhibit the dition seen in the Eocene fossil bats appears phanerocochlear condition. This conclusion to be primitive. was supported by our microscopic examina- Character 30: Epitympanic recess shal- tions of several specimens (SMF Me 1205, low and broad (0); or deep, often constricted Me 1494) in which the cochlear wall easily in area (1). The epitympanic recess (ϭ re- transmits light. Among extant bats, the pha- cessus epitympanicus, atticus tympanicus, or nerocochlear condition occurs in Emballon- aditus ad antrum) is that part of the primary uridae, Rhinopomatoidea, some Nycteridae, tympanic cavity that contains the middle ear some Rhinolophinae, some Hipposiderinae, ossicles (Klaauw, 1931). In bats, the epitym- some Phyllostomidae, some Mormoopidae, panic recess lies in the dorsolateral roof of Mystacinidae, Nataloidea, Antrozoidae, To- the middle ear space, and it effectively hous- mopeatinae, some Molossinae, Vespertilion- es only the incus and the joint between the inae, Miniopterinae, Myotinae, and Murini- incus and malleus (®g. 28). The stapes lies nae. In contrast, the cryptocochlear condition just medial to the recess in articulation with is seen in Pteropodidae, some Nycteridae, the fenestra ovalis of the periotic, and most Megadermatidae, some Rhinolophinae, some of the malleus is located in a more ventro- Hipposiderinae, some Phyllostomidae, some lateral position in the tympanic cavity. Mormoopidae, Noctilionidae, some Molos- The epitympanic recess is shallow and sidae, and Kerivoulinae. Both outgroups are broad in Icaronycteris (®g. 26) and Palaeo- also cryptocochlear, suggesting that this con- chiropteryx. Among extant bats, this condi- dition is primitive. Accordingly, the phaner- tion is seen in Pteropodidae, Emballonuridae, ocochlear condition seen in Palaeochiropte- Rhinopomatidae, Megadermatidae, Phyllos- ryx and Hassianycteris is derived. We were tomidae, Myzopodidae, Nataloidea, Molos- unable to ascertain which condition is pres- soidea, and Vespertilionidae. In contrast, the ent in Icaronycteris and Archaeonycteris; epitympanic recess is deep and often con- these taxa are therefore scored ``?'' for this stricted in area in Craseonycteridae, Nycter- character. idae, Rhinolophidae, Mormoopidae, Nocti- Character 28: Lateral process of ectotym- lionidae, and Mystacinidae. The epitympanic panic weak or absent (0); or well developed, recess is shallow and broad in both out- forms tubular external auditory meatus (1). groups, suggesting that the similar condition The ectotympanic in bats is a ring-shaped or seen in Icaronycteris and Palaeochiropteryx horseshoe-shaped bone that supports the is relatively primitive. We were not able to tympanic membrane (®g. 28). In most bats, observe the epitympanic recess in Archaeo- the medial edge of the ectotympanic is ex- nycteris and Hassianycteris; these taxa are panded and encloses part of the middle ear, therefore scored ``?'' for this character. thus forming a partial bulla that encloses the Character 31: Fossa for m. stapedius in- primary tympanic cavity as de®ned by distinct (0); or shallow and broad (1); or Klaauw (1931). In some bats, an additional deep, constricted in area, often a crescent- lateral outgrowth of the ectotympanic forms shaped ®ssure (2). The fossa for the origin the ¯oor of a bony enclosure of what Klaauw of m. stapedius is located in the posterior (1931) termed the ``recessus meatus acousti roof of the middle ear space (®g. 28). When externi'' (external auditory meatus). The lat- well-de®ned, it occupies a cavity dorsal to eral process of the ectotympanic is absent in the crista parotica. The fossa for m. stapedius Icaronycteris, Archaeonycteris, Hassianyc- is apparently deep and constricted in Palaeo- teris, and Palaeochiropteryx. This is similar chiropteryx. Among extant bats, a similar to the condition seen in most extant bats. In condition is seen in Emballonuridae, Rhino- contrast, the lateral process of the tympanic pomatoidea, Phyllostomidae, Noctilionidae, is well developed and forms the ¯oor of a Molossoidea, some Vespertilioninae, Min- tubular external auditory meatus in some iopterinae, Murininae, and Kerivoulinae. In 62 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 these forms, the fossa for m. stapedius is of- or absent (0); or large (1). The orbicular ten a deep, crescent-shaped ®ssure that lies apophysis is a dense osseous protrusion that lateral to the posterolateral surface of the lies at the end of the body of the malleus, cochlea. In contrast, the fossa for m. stape- distal to the base of the manubrium of the dius is comparatively shallow and broad in malleus (®g. 28). The function of the orbic- Megadermatidae, Myzopodidae, Thyropteri- ular apophysis remains uncertain, although dae, Furipteridae, some Vespertilioninae, and several hypotheses have been proposed. The Myotinae. The fossa for m. stapedius is poor- principal axis of rotation of the malleus runs ly developed (not distinguishable from the through the long axis of the gonial process, remainder of the roof of the middle ear) in perpendicular to the transverse body of the Nycteridae, Rhinolophidae, Mormoopidae, malleus (Fleisher, 1978, 1980; Wilson and Mystacinidae, and Natalidae. The fossa for Bruns, 1983). Fleischer (1978, 1980) sug- m. stapedius is similarly indistinct in both gested that the orbicular apophysis might outgroups, suggesting that the presence of a function as the center for a second mode of deep fossa in Palaeochiropteryx is a rela- vibration at high frequencies, in which the tively derived condition. We were unable to malleus vibrates around an axis running determine the condition of the fossa for m. through the transverse body of the malleus, stapedius in Icaronycteris, Archaeonycteris, perpendicular to the gonial process. This hy- and Hassianycteris; these taxa are therefore pothesis impies that complex motions of the scored ``?'' for this character. malleus are possible, perhaps including ¯ex- Character 32: Fenestra cochleae small or ion at the malleus-incus joint if the short pro- of moderate size, maximum diameter Ͻ20% cess of the incus were constrained (Wilson of the external width of the ®rst half-turn of and Bruns, 1983). However, Wilson and the cochlea (0); or enlarged, maximum di- Bruns (1983) found no evidence of a change ameter Ͼ25% of the external width of the in the rotational axis with in increasing fre- ®rst half turn of the cochlea (1). The fenestra quency in their experimental study of mid- cochleae (ϭ fenestra rotundum) is a mem- dle-ear mechanics in Rhinolophus, suggest- brane-covered opening in the tympanic wall ing that rotational movement of the malleus of the petrosal. It faces posteriorly or poster- occurs only around the primary (gonial) axis olaterally, and separates the scala tympani (at of rotation or that rotation around both axes the base of the cochlear labyrinth) from the occurs at all frequencies. middle ear cavity (®g. 28). The fenestra A second hypothesis proposed by Fleisch- cochleae appears relatively large in Hassi- er (1978, 1980) is that presence of an orbic- anycteris and Palaeochiropteryx, where it ular apophysis serves to shift the center of has a maximum diameter Ͼ25% of the ex- mass of the malleus±incus complex away ternal width of the ®rst half turn of the coch- from the axis of rotation of the ossicle sys- lea. Among extant bats, a similarly large fe- tem, thus lowering its natural resonant fre- nestra cochleae is seen in Pteropodidae, quency. Fleischer (1978: 34) suggested that Rhinopomatoidea, Noctilionoidea, and Mys- presence of an orbicular apophysis, which is tacinidae. In contrast, the fenestra cochleae is correlated with minute size of the ossicle of small or moderate size (Ͻ20% of the ex- chain and rigid fusion of the gonial process ternal width of the ®rst half turn of the coch- to the ectotympanic, may be necessary to lea) in Emballonuridae, Rhinolophoidea, Na- tune the middle ear because ``without the or- taloidea, Molossoidea, and Vespertilionidae. bicular apophysis, the natural frequency of The fenestra cochleae is similarly small in the malleus±incus complex might simply be both outgroups, suggesting that the relatively too high, even for bats.'' However, Wilson enlarged fenestra cochleae seen in Hassi- and Bruns (1983) argued that this explana- anycteris and Palaeochiropteryx is relatively tion was unlikely since the extra mass would derived. We were unable to determine the have been better employed for this purpose condition of the fenestra cochleae in Icaron- if it were applied to stiffening the manubrium ycteris and Archaeonycteris; these taxa are of the malleus. They also noted that the res- therefore scored ``?'' for this character. onating frequency of the membrane ossicle Character 35: Orbicular apophysis small system was not sharply tuned to the CF fre- 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 63 quency used by the bat in their study, Rhin- and both outgroups. This suggests that the olophus ferrumequinum. large orbicular apophysis seen in the Eocene Wilson and Bruns (1983: 12) suggested al- bats is a derived condition. ternative hypotheses for function of the or- Character 74: Stylohyal occurs as gently bicular apophysis in bats: curved bar with no enlargement or other modi®cation to the lateral edge or cranial tip . . . the middle-ear may be acting as a transmission (0); or with cranial tip slightly expanded (1); line with distributed mass and stiffness. In an ideal transmission line . . . a signal is transmitted without or with bifurcated tip (2); or with large, ¯at loss even at high frequencies, but with a delay (phase expansion or ``foot'' on lateral cranial tip shift proportional to frequency) depending on length (3); or with very large, ¯at, axe-shaped en- ....Considering the dimensions involved, the major source of delay [of transmission of sound through the largement at tip (4); or lateral half of entire middle ear] would appear to be transverse motion of stylohyal swollen (5). The stylohyal is the the manubrium ....Onthis hypothesis the orbicular distalmost element of the anterior cornu of apophysis might help to match transverse vibration the hyoid apparatus. It typically consists of along the manubrium into a torsional vibration along the transversal segment of the malleus. Alternatively, an elongate bar that extends between the hy- it might play a similar role to a telephone line loading oid apparatus and the ear region of the skull. coil which, contrary to intuition, improved high-fre- In Icaronycteris (®gs. 25, 26), Archaeonyc- quency transmission. teris, and Palaeochiropteryx (®g. 27) the dis- In other words, the orbicular apophysis may tal stylohyal (the end nearest the skull) has improve the ability of the middle ear to trans- as a slightly expanded, paddle-shaped tip. mit high-frequency sounds with a minimum Among extant bats, a similar condition is time delay. Unfortunately, Wilson and Bruns seen in some Emballonuridae, Craseonycter- (1983) did not explain how this ``matching'' idae, and all Yangochiroptera with the ex- of transverse to tortional vibrations occurs. ception of Myzopodidae. In contrast, the dis- Another possible function for the orbicular tal stylohyal has a bifurcated tip in some Em- aphophysis may be in the system employed ballonuridae and Rhinopomatidae. The sty- by echolocating bats to avoid self-deafening lohyal has a large, ¯at, lateral expansion or by ``freezing'' the middle-ear ossicle during ``foot'' in Rhinolophoidea, and a very large, the emission of echolocation calls (for de- axe-shaped enlargement (which extends both scriptions of this mechanism, see Henson, laterally and medially) at its tip in Myzopod- 1964, 1965, 1966, 1967a, 1970; Jen and idae. Only three extant bat groupsÐPtero- Suga, 1976; Fenton et al., 1995). One com- podidae, Nycteridae, and MegadermatidaeÐ ponent of this system is m. tensor tympani, lack any enlargement or modi®cation of the which inserts onto the transverse body of the distal tip of the stylohyal. In these forms, the malleus just proximal to the orbicular apoph- stylohyal consists of a simple, curved bar. ysis. In bats that use short FM (frequency Among the outgroups, the stylohyal is a sim- modulated) calls for echolocation, contrac- ple, curved bar in Scandentia, but Dermop- tion of m. tensor tympani begins 6±8 msec tera has a unique stylohyal that is laterally before the onset of vocalization, and relaxa- swollen throughout much of its length. Al- tion occurs very rapidly, within 2±8 msec af- though there is disagreement between the ter vocalization (Jen and Suga, 1976). Pos- sible roles that the orbicular apophysis may two outgroups used in the current study, play in this system (e.g., in changing the comparisons with the stylohyal of other speed with which the ear ossicles freeze and mammalian orders (Klaauw, 1931; Sprague, return to full sensitivity) have not been ex- 1943) indicate that the primitive form of the plored. stylohyal is a simple, curved bar. This sug- The orbicular apophysis is relatively large gests that the expanded distal tip seen in Ica- in Icaronycteris (®g. 25), Archaeonycteris, ronycteris, Archaeonycteris, and Palaeochi- Hassianycteris, and Palaeochiropteryx.A ropteryx represents a derived condition. We similar condition is seen in all known extant were unable to determine the condition of the Microchiroptera. In contrast, the orbicular stylohyal in Hassianycteris, which is there- apophysis is small or absent in Pteropodidae fore scored ``?'' for this character. 64 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

ANTERIOR AXIAL SKELETON accessory processes present on C5 (1). Pos- teriorly directed ventral accessory processes Character 75: Posteriorly directed ven- are absent from C5 in Icaronycteris, Ar- tral accessory processes not present on cen- chaeonycteris, and Hassianycteris, but are tra of cervical vertebrae 2 and 3 (0); or ven- present on C5 in Palaeochiropteryx. Among tral accessory processes present on C2 and extant bats, ventral accessory processes are C3 (1). Ventral accessory processes are present on C5 in all Microchiroptera, but are paired projections from the posteroventral absent from this vertebra in Pteropodidae and edges of the vertebral centra (®g. 24). These both outgroups. In this context, absence of processes, which may be fused across the ventral accessory processes (state ``0'') in midline to form a single large projection, ar- Icaronycteris, Archaeonycteris, and Hassi- ticulate with corresponding facets on succes- anycteris appears to represent the primitive sive vertebrae. Many mammals have ventral condition, while presence of accessory pro- accessory processes on the axis (C2), but cesses on C5 in Palaeochiropterx is a de- these processes rarely occur on other verte- rived condition. brae except in bats. Presence of ventral ac- Character 78: Seventh cervical vertebra cessory processes on multiple vertebrae ap- not fused to ®rst thoracic vertebra (0); or C7 parently facilitates extensive dorsi¯exion of and T1 at least partially fused (1). There is the neck in many bats, allowing greater free- no evidence of fusion between the seventh dom of movement of the head while roosting cervical and ®rst thoracic vertebrae in Ica- (Crerar and Fenton, 1984). ronycteris, Archaeonycteris, and Palaeochi- Posteriorly directed ventral accessory pro- ropteryx. This condition is similar to that cesses are present on the centra of cervical seen in extant Pteropodidae, Emballonuridae, vertebrae 2 and 3 in Icaronycteris, Archaeo- Rhinopomatidae, some Nycteridae, Phyllos- nycteris, Hassianycteris, and Palaeochirop- tomidae, some Mormoopidae, Noctilionidae, teryx. Among extant bats, similar accessory Mystacinidae, Myzopodidae, Antrozoidae, processes are present on C3 in some Ptero- and Vespertilionidae. In contrast, the centra, podidae and all Microchiroptera. Accessory zygopophyses, and transverse processes of processes are absent from these vertebrae in C7 and T1 are partially to fully fused in adult some Pteropodidae and both outgroups. In Craseonycteridae, some Nycteridae, Megad- this context, presence of accessory processes ermatidae, Rhinolophidae, some Mormoopi- on C2 and C3 in the Eocene bats apparently dae, Thyropteridae, Furipteridae, Natalidae, represents a derived condition. and Molossidae. Sutures may be visible be- Character 76: Posteriorly directed ven- tween C7 and T1 in some young adults, but tral accessory processes not present on cen- synovial joints are lacking. These vertebrae trum of cervical vertebra 4 (0); or ventral are not fused in either outgroup, nor are they accessory processes present on C4 (1). Pos- fused in most other mammals. In this con- teriorly directed ventral accessory processes text, lack of C7±T1 fusion in the Eocene bats are absent on C4 in Icaronycteris, but are apparently represents the primitive condition. present on C4 in Archaeonycteris, Hassi- We were unable to determine the condition anycteris, and Palaeochiropteryx. Among of C7 and T1 in Hassianycteris; this taxon is extant bats, ventral accessory processes are therefore scored ``?'' for this character. present on C4 in all Microchiroptera, but are Character 79: First and second thoracic absent from this vertebra in Pteropodidae and vertebrae not fused (0); or T1 and T2 fused both outgroups. In this context, absence of (1). There is no evidence of fusion between ventral accessory processes in Icaronycteris the ®rst and second thoracic vertebrae in Ica- appears to represent the primitive condition, ronycteris and Palaeochiropteryx. This con- while presence of accessory processes on C4 dition is similar to that seen in most extant in the other Eocene bats is a derived condi- bats. In contrast, the centra, zygopophyses, tion. and transverse processes of T1 and T2 are Character 77: Posteriorly directed ven- fused in adult Hipposiderinae and Thyrop- tral accessory processes not present on cen- teridae. Sutures may be visible between T1 trum of cervical vertebra 5 (0); or ventral and T2 in some young adults, but synovial 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 65

joints are lacking. T1 and T2 are not fused mal clavicles (with which it articulates an- in either outgroup, nor are they fused in most terolaterally), the costal cartilages of rib 1 other mammals. In this context, lack of T1± (with which it articulates posterolaterally), T2 fusion in Icaronycteris and Palaeochirop- and the mesosternum (with which it articu- teryx apparently represents the primitive con- lates posteriorly). The manubrium extends dition. We were unable to determine the con- laterally only to the level of the clavicular dition of T1 and T2 in Archaeonycteris and joint in Icaronycteris (®g. 22), Archaeonyc- Hassianycteris; these taxa are therefore teris (®g. 3), Hassianycteris, and Palaeochi- scored ``?'' for this character. ropteryx, and there is no evidence of fusion Character 80: Anterior ribs not fused to of the ®rst costal cartilage to the manubrium vertebrae (0); or ®rst rib fused to vertebrae or ®rst rib. This is similar to the condition (1); or at least ®rst ®ve ribs fused to verte- seen in most extant bats. In contrast, the ma- brae (2). The anterior ribs are not fused to nubrium has a winglike projection on each vertebrae in Icaronycteris, Hassianycteris, side that extends laterally well beyond the and Palaeochiropteryx. Among extant bats, clavicular joint in Rhinopomatoidea, Megad- a similar condition is seen in Pteropodidae, ermatidae, and Rhinolophidae. In these Emballonuridae, Rhinopomatidae, Nycteri- forms, this winglike lateral projection is dae, Megadermatidae, and Yangochiroptera. fused directly to the ®rst rib. It seems likely In contrast, the ®rst rib is fused to the ver- that these manubrial ``wings'' are formed by tebral column in Craseonycteridae, and at ossi®cation and fusion of the ®rst costal car- least the ®rst ®ve ribs are fused to vertebrae tilage (which is typically broad and has a in Rhinolophidae. No fusion is seen between similar shape) with the body of the manu- ribs and vertebrae in the outgroups and most brium. Manubrial wings are lacking and the other mammalian groups. This suggests that ®rst rib is not fused to the manubrium or ribs the lack of fusion seen in the Eocene bats is in both outgroups and most mammals, sug- primitive. We were unable to determine if rib gesting that the condition seen in the Eocene fusion was present or absent in Archaeonyc- bats is primitive. teris; this taxon is therefore scored ``?'' for Character 83: Second costal cartilage ar- this character. ticulates with sternum at manubrium±mesos- Character 81: Width of ®rst rib similar to ternum joint (0); or second rib articulates other ribs (0); or ®rst rib at least twice the with manubrium, no contact between rib (or width of other ribs (1). The width of the ®rst costal cartilage) and mesosternum (1). The rib is similar to that of the other ribs in Ica- second costal cartilage articulates with the ronycteris and Palaeochiropteryx. This con- sternum at the manubrium±mesosternum dition is also seen in most extant bats. In joint in Icaronycteris (®gs. 21, 22), Archaeo- contrast, the ®rst rib is at least twice the nycteris (®g. 3) Hassianycteris, and Palaeo- width of the other ribs in Rhinolophoidea. chiropteryx. This condition is similar to that The width of the ®rst rib is similar to that of seen in most extant bats. In contrast, in Rhin- the other ribs in both outgroups and most olophoidea the second rib articulates with the other mammals, suggesting that this condi- manubrium and there is no contact between tion, which is seen in the Eocene bats, is rib (or costal cartilage) and mesosternum. primitive. We were unable to determine the The second costal cartilage articulates with width of the ®rst rib in Archaeonycteris and the sternum at the manubrium±mesosternum Hassianycteris; these taxa are therefore joint in both outgroups. This suggests that scored ``?'' for this character. the condition seen in the Eocene bats is rel- Character 82: First costal cartilage not atively primitive. ossi®ed or fused with manubrium or ®rst rib Character 84: Second rib articulates with (0); or ®rst costal cartilage ossi®ed and sternum via costal cartilage (0); or second fused to manubrium (where it appears to rib fused to sternum, costal cartilage absent form a winglike lateral process of the ma- or ossi®ed (1). The second rib articulates nubrium) and fused to ®rst rib (1). The ma- with the sternum via a costal cartilage in Ica- nubrium is the anteriormost element of the ronycteris (®gs. 21, 22), Archaeonycteris sternum. It lies at the junction of the proxi- (®g. 3), Hassianycteris, and Palaeochirop- 66 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Fig. 30. Icaronycteris index (holotype; PU 18150). Stereophotographs of close-up dorsal view of the shoulder region. From Jepsen (1970: ®g. 9); reprinted from original negatives. teryx. This condition is similar to that seen Molossidae, some Vespertilioninae, and in most extant bats. In contrast, in Rhinolo- some Myotinae. In contrast, the mesosternum phidae the second rib is fused to the sternum articulates with only four costal cartilages and the costal cartilage is absent or ossi®ed. posterior to the second rib in some Pteropod- In these forms, an extensive sheet of thin idae, Craseonycteridae, Rhinolophoidea, and bone runs between the second rib and the some Phyllostomidae. This number is further lateral process of the manubrium. The second reduced to three costal cartilages in some rib articulates with the sternum via a costal Vespertilioninae, Murininae, and Kerivouli- cartilage in both outgroups and most other nae. The mesosternum articulates with ®ve mammals, suggesting that the condition seen or more costal cartilages in both outgroups, in the Eocene bats is relatively primitive. suggesting that the condition seen in the Eo- Character 85: Mesosternum articulates cene bats is relatively primitive. with at least ®ve costal cartilages posterior Character 86: Ribs with no anterior lam- to second rib (0); or articulates with four inae (0); or anterior laminae present (1). costal cartilages posterior to second rib (1); Anterior laminae are thin plates of bone that or articulates with only three costal carti- run along the leading edges of ribs anterior lages posterior to second rib (2). The me- to the main body of the rib in some bats. sosternum articulates with at least ®ve costal These structures, which are often nearly cartilages posterior to the second rib in Ica- transparent, appear to provide an increased ronycteris (®gs. 21, 22), Archaeonycteris area for muscle attachment. Anterior laminae (®g. 3), Hassianycteris, and Palaeochirop- are absent in Icaronycteris (®gs. 22, 30), but teryx. Among extant lineages, a similar con- are present in Palaeochiropteryx. Among ex- dition is seen in some Pteropodidae, some tant forms, anterior laminae are absent in Phyllostomidae, Mormoopidae, Noctilioni- some Emballonuridae, Rhinolophidae, some dae, Mystacinidae, Myzopodidae, Natalidae, Phyllostomidae, Mystacinidae, Antrozoidae, 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 67

be either relatively narrow (lamina width less than that of the main body of the rib) or wide (equal to or wider than the main body). The anterior laminae are narrow in Palaeochirop- teryx. Among extant bats, narrow anterior laminae are found in Pteropodidae, Embal- lonuridae, Craseonycteridae, Phyllostomidae, Myzopodidae, Molossidae, Vespertilioninae, Myotinae, Murininae, and Kerivoulinae. Wide anterior laminae are seen in Rhinopom- atidae, Nycteridae, Megadermatidae, Mor- moopidae, Noctilionidae, Thyropteridae, Fu- ripteridae, Natalidae, and Miniopterinae. This character cannot be evaluated for taxa that lack anterior laminae (state ``0'' of char- acter 86 above); these forms (including Ica- ronycteris) are scored ``-'' for this character. Because both outgroups lack anterior lami- nae, the polarity of this character cannot be determined a priori. As noted above, we were unable to adequately determine presence/ab- sence of anterior laminae in Archaeonycteris and Hassianycteris; these taxa are therefore scored ``?'' for this character. Character 88: Ribs with no posterior lam- Fig. 30 (continued). Key to sterophotographs inae (0); or posterior laminae present (1). on preceding page. AB, axillary border of scap- Posterior laminae are thin plates of bone that ula; AC, acromion process of scapula; CL, clav- run along the trailing edges of ribs posterior icle; HE, head of humerus; HU, humerus; IF, in- to the main body of the rib. Like the anterior fraspinous fossa of scapula; SF, supraspinous fos- sa of scapula; SN, suprascapular notch of scapula; laminae, these structures may be nearly T, trochiter. transparent, and they appear to provide an increased area for muscle attachment. Pos- terior laminae are present in Icaronycteris Tomopeatinae, and some Molossinae. Ante- (®gs. 22, 30), Hassianycteris (®g. 4), and Pa- rior laminae are present in some Pteropodi- laeochiropteryx (®g. 2), and they are also dae, some Emballonuridae, Rhinopomato- present in most extant bats. Posterior laminae idea, Nycteridae, Megadermatidae, some are absent in Mystacinidae and some Molos- Phyllostomidae, Mormoopidae, Noctilioni- sinae. Posterior laminae are also absent in dae, Nataloidea, some Molossinae, and Ves- both outgroups, suggesting that their pres- pertilionidae. Both outgroups lack anterior ence in Icaronycteris and Palaeochiropteryx laminae, suggesting that absence of anterior is derived. We were unable to adequately de- laminae in Icaronycteris is relatively primi- termine presence/absence of these structures tive, while presence in Palaeochiropteryx is in Archaeonycteris; this taxon is therefore relatively derived. We were unable to ade- scored ``?'' for this character. quately determine presence/absence of these Character 89: Posterior laminae on ribs structures in Archaeonycteris and Hassianyc- narrow, lamina width less than that of main teris; these taxa are therefore scored ``?'' for body of rib (0); or posterior laminae wide, this character. equal to or wider than main body of rib (1). Character 87: Anterior laminae on ribs Like the anterior laminae, posterior laminae narrow, lamina width less than that of main may be either relatively narrow (lamina body of rib (0); or anterior laminae wide, width less than that of the main body of the equal to or wider than main body of rib (1). rib) or wide (equal to or wider than the main The anterior laminae on the ribs of bats may body). The posterior laminae are narrow in 68 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Icaronycteris (®gs. 22, 30) but are wide in the manubrium is small in Scandentia but is Hassianycteris and Palaeochiropteryx. broad and triangular in Dermoptera. Lack of Among extant bats, narrow posterior laminae agreement between the outgroups precludes are found in Pteropodidae, some Emballon- a priori determination of the primitive con- uridae, Craseonycteridae, Phyllostomidae, dition for this character. We were unable to Noctilionidae, Natalidae, Molossoidea, and determine the form of the anterior face of the some Vespertilioninae. Wide posterior lami- manubrium in Icaronycteris; this taxon is nae are seen in some Emballonuridae, Rhin- therefore scored ``?'' for this character. opomatidae, Rhinolophoidea, Mormoopidae, Character 91: Ventral process of manu- Myzopodidae, Thyropteridae, Furipteridae, brium absent (0); or ventral process present, some Vespertilioninae, Miniopterinae, My- distal tip blunt or rounded (1); or ventral otinae, Murininae, and Kerivoulinae. This process present, distal tip laterally com- character cannot be evaluated for taxa that pressed (2). The ventral process of the ma- lack posterior laminae (state ``0'' of character nubrium provides the anterior attachment 88 above); these forms are scored ``-'' for point for a series of ligamentous sheets that this character. Because both outgroups lack run down the midline of the sternum posterior laminae, polarity of this character (Vaughan, 1959, 1970b; Norberg, 1970, cannot be determined a priori. As noted 1972a; Strickler, 1978; Hermanson and Al- above, we were unable to adequately deter- tenbach, 1983, 1985). The ventral process, mine presence/absence of anterior laminae in together with these ligamentous sheets, Archaeonycteris and Hassianycteris; these forms the origin for the m. pectoralis com- taxa are scored ``?'' for this character. plex (Vaughan, 1959, 1970b; Norberg, 1970; Character 90: Anterior face of manubri- Strickler, 1978; Hermanson and Altenbach, um small (0); or broad, de®ned by elevated 1983, 1985). The distal tip of the ventral pro- ridges (1). In addition to its role in connect- cess of the manubrium is blunt and some- ing the body of the sternum to the clavicles what rounded in Icaronycteris (®gs. 21, 22) and anterior ribs, the manubrium is also an and Archaeonycteris (®g. 3). The ventral pro- important site of muscle origin for parts of cess in these forms is triangular in cross sec- the m. pectoralis complex, which provides tion and points somewhat posteriorly. In con- the majority of the power for the downstroke trast, the tip of the ventral process is laterally of the wings (Vaughan, 1959, 1970b; Nor- compressed (keel-like) in Hassianycteris and berg, 1970; Strickler, 1978; Hermanson and Palaeochiropteryx. In cross section, the ven- Altenbach, 1983, 1985). The anterior face of tral process in these forms is lens-shaped the manubrium provides the point of origin with the long axis running anteroposteriorly. for part of the anterior division of m. pec- Among extant bats, a ventral process with a toralis, which rotates the humerus and helps blunt or rounded tip is found in Rhinopom- to pull it downward and forward and during atidae, Megadermatidae, Rhinolophinae, the downstroke when the wing is protracted some Hipposiderinae, some Phyllostominae, against the force of the airstream (Vaughan, Mystacinidae, Antrozoidae, Tomopeatinae, 1959, 1970b; Norberg, 1970; Strickler, 1978; some Molossinae, and some Vespertilioni- Hermanson and Altenbach, 1983, 1985). The nae. A laterally compressed ventral process anterior face of the manubrium of the ster- is seen in Pteropodidae, Emballonuridae, num is relatively small and poorly de®ned in Craseonycteridae, Nycteridae, some Hippos- Archaeonycteris, Hassianycteris, and Pa- iderinae, some Phyllostominae, Mormoopi- laeochiropteryx. This condition is similar to dae, Noctilionidae, Nataloidea, some Molos- that seen in most extant bats. In contrast, the sinae, some Vespertilioninae, Miniopterinae, anterior face of the manubrium is a broad, Myotinae, Murininae, and Kerivoulinae. triangular surface that extends onto the lat- Among the outgroups, a blunt ventral pro- eral processes and is de®ned by three ele- cess occurs in Dermoptera, but the ventral vated ridges in some Pteropodidae, Megad- process is absent in Scandentia. Although the ermatidae, Rhinolophidae, some Phyllostom- outgroup evidence is ambiguous, this distri- idae, Natalidae, and some Molossinae. bution suggests that a ventral process with a Among the outgroups, the anterior face of blunt, rounded tip (as in Icaronycteris and 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 69

Archaeonycteris) may be primitive, while the outgroup evidence is ambiguous, presence of laterally compressed, keel-like condition an obtuse ventral process in Dermoptera sug- seen in Hassianycteris and Palaeochiropte- gests that this condition, which is also seen ryx appears to be derived. in Icaronycteris and Palaeochiropteryx, may Character 92: Angle between axis of ven- be primitive. The approximate 90Њ angle of tral process and body of manubrium acute the ventral process in Archaeonycteris would (0); or approximately 90Њ (1); or obtuse (2); thus be interpreted as a derived condition. or ventral process bilobed with one acute We were unable to determine the angle of the and one obtuse process (3). The axis of the ventral process in Hassianycteris; this taxon ventral process is de®ned as the long axis of is therefore scored ``?'' for this character. the thickened base and central body of the Character 93: Length of manubrium pos- process. The orientation of the axis of the terior to lateral processes Ͼ2.5 times trans- ventral process (as seen in lateral view) var- verse width (0); or length Ͻ2 times trans- ies among bats. In some taxa, the angle be- verse width (1). The length of the manubrium tween the ventral process and the body of the posterior to the lateral processes is less than manubrium is acute, and the ventral process twice the transverse width of this portion of appears to project posteroventrally. In other the manubrium in Icaronycteris (®gs. 21, forms, the angle is approximately 90Њ (so that 22), Archaeonycteris (®g. 3), Hassianycteris, the ventral process projects ventrally) or ob- and Palaeochiropteryx. This condition is tuse (ventral processes projects anteroven- similar to that seen in most extant bats. In trally). Although a full range of variation be- contrast, the manubrium is relatively elon- tween these conditions is theoretically pos- gated (posterior portion Ͼ2.5 times trans- sible, we found that most species could be verse width) in Nycteridae, some Phyllos- easily placed in one of the categories de®ned tomidae, and Molossidae. Both conditions above. The only exceptions were a few taxa occur among the outgroupsÐthe manubrium that have a bilobed ventral process with one is relatively short in Dermoptera but is elon- acute and one obtuse lobe. gate in Scandentia. Accordingly, the primi- Our examinations of the Eocene fossils in- tive condition of this feature cannot be de- dicated that the ventral process is obtuse in termined a priori. Icaronycteris (®gs. 21, 22) and Palaeochi- Character 94: Mesosternum narrow, ropteryx, and is oriented at approximately mean width less than half the distance be- 90Њ in Archaeonycteris (®g. 3). Among ex- tween clavicles at sternoclavicular joint (0); tant forms, an obtuse ventral process is seen or mesosternum broad, mean width greater in some Pteropodidae, some Emballonuridae, than three-fourths the distance between clav- Noctilionidae, some Mormoopidae, some icles (1). The mesosternum (ϭ body of ster- Mystacinidae, Myzopodidae, Molossoidea, num) articulates with the manubrium anteri- and Vespertilioninae. The ventral process is orly and with the xiphisternum posteriorly. oriented at approximately 90Њ in some Pter- The relative width of the mesosternum in a opodidae, some Emballonuridae, Rhinopom- bat can be estimated by comparing mean me- atoidea, some Phyllostomidae, some Mor- sosternal width with the transverse distance moopidae, Thyropteridae, Myotinae, Muri- between the right and left clavicles at their ninae, and Kerivoulinae. An acute ventral joints with the manubrium.4 The mesoster- process occurs in Megadermatidae, Rhino- num is relatively narrow (mesosternal width lophidae, some Phyllostomidae, and Natali- less than half the interclavicular distance) in dae. The ventral process is bilobed in Nyc- Icaronycteris (®gs. 21, 22), Archaeonycteris, teridae, Furipteridae, and Miniopterinae. Among the outgroups, an obtuse ventral pro- 4 cess occurs in Dermoptera. This feature can- Modi®cations of the manubrium and ribs in a num- ber of taxa (e.g., rhinolophoids) preclude meaningful not be evaluated in taxa that lack a ventral size comparisons of the mesosternum with these ele- process on the manubrium (state ``0'' of ments. Accordingly, we chose to compare width of the character 91 above). Scandentia, whose mesosternum with the interclavicular distance because members lack a ventral process, is therefore our observations suggest that the distance between the scored ``-'' for this character. Although the clavicles is correlated principally with body size. 70 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Hassianycteris, and Palaeochiropteryx. This Emballonuridae, Rhinopomatoidea, Nycteri- is similar to the condition seen in Pteropod- dae, Megadermatidae, Hipposiderinae, Noc- idae, Emballonuridae, Rhinopomatoidea, tilionoidea, Mystacinidae, Thyropteridae, Megadermatidae, Rhinolophidae, Noctilion- Molossoidea, and Vespertilionidae. There is oidea, Mystacinidae, Myzopodidae, Molos- no lateral ¯are in some Pteropodidae, some soidea, and some Vespertilionidae. In con- Emballonuridae, Rhinolophinae, Myzopodi- trast, the mesosternum is broad (width great- dae, and Furipteridae. Both outgroups have er than three-fourths the interclavicular dis- a xiphisternum with a wide lateral ¯are. This tance) in Nycteridae, Thyropteridae, pattern suggests that the ¯ared morphology Furipteridae, Natalidae, some Vespertilioni- seen in Icaronycteris, Hassianycteris, and nae, Miniopterinae, Myotinae, Murininae, Palaeochiropteryx may be relatively primi- and Kerivoulinae. The mesosternum is rela- tive, while absence of a xiphisternal ¯are in tively narrow in both outgroups, suggesting Archaeonycteris may represent a relatively that the narrow condition seen in the Eocene derived condition. bats is relatively primitive. Character 95: Xiphisternum without keel PECTORAL GIRDLE (0); or with prominent median keel (1). The xiphisternum lacks a median longitudinal Character 97: Acromion process without keel on its ventral surface in Icaronycteris medial shelf (0); or with shelf that projects (®g. 22), Archaeonycteris, and Palaeochi- medially over supraspinous fossa or medial ropteryx. This condition is seen among ex- base of acromion process (1). The distal half tant bats in Pteropodidae, Rhinopomatidae, of the medial surface of the acromion process Hipposiderinae, some Phyllostomidae, Noc- serves as the attachment point for one end of tilionidae, Myzopodidae, Molossoidea, and the transverse scapular ligament (ϭ dorsal some Vespertilioninae. In contrast, a promi- scapular ligament), which stretches between nent, ventrally projecting median keel is the acromion and the anteromedial rim of the present on the xiphisternum in Emballonur- scapula dorsal to m. supraspinatus (Vaughan, idae, Craseonycteridae, Nycteridae, Megad- 1959, 1970b; Norberg, 1970; Strickler, 1978; ermatidae, Rhinolophidae, some Phyllostom- Hermanson and Altenbach, 1983, 1985). In idae, Mormoopidae, Mystacinidae, Thyrop- various bats the transverse scapular ligament teridae, Furipteridae, Natalidae, some Ves- provides an increased attachment area for m. pertilioninae, Miniopterinae, Myotinae, supraspinatus (which elevates, extends, and Murininae, and Kerivoulinae. A xiphisternal rotates the humerus), m. acromiodeltoideus keel is absent in both outgroups, suggesting (which elevates the humerus), and m. spi- that absence of a keel is relatively primitive. nodeltoideus (which elevates and ¯exes the We were unable to determine if a xiphisternal humerus; Vaughan, 1959, 1970b; Norberg, keel is present in Hassianycteris due to dis- 1970; Strickler, 1978; Hermanson and Alten- tortion and ¯attening of the sternal region in bach, 1983, 1985). Presence of a medial shelf all available specimens; this taxon is there- on the distal acromion is probably correlated fore scored ``?'' for this character. with changes in the extent and orientation of Character 96: Posterior xiphisternum the transverse scapular ligament and in the with wide lateral ¯are (0); or not laterally origins of those parts of m. supraspinatus and ¯ared (1). The posterior xiphisternum has a m. acromiodeltoideus that originate directly wide lateral ¯are in Icaronycteris (®g. 22), from the acromion process. Hassianycteris, and Palaeochiropteryx. This The acromion process of the scapula lacks ¯are is produced by a steady increase in a medial shelf in Icaronycteris, Archaeonyc- width of the xiphisternum from the anterior teris, Hassianycteris, and Palaeochiropteryx. to the posterior end of the element. In con- This is similar to the condition seen in most trast, the xiphisternum is not ¯ared (and extant bats. In contrast, the acromion process width of the posterior end is approximately has a shel¯ike projection that extends medi- equal to the width of the anterior end) in Ar- ally over the supraspinous fossa or medial chaeonycteris. Among extant forms, a lateral base of the acromion process in Thyropteri- ¯are is present in some Pteropodidae, some dae, Furipteridae, Natalidae, some Vesperti- 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 71 lioninae, Miniopterinae, and Kerivoulinae. anycteris are similar in size; accordingly, this No medial shelf is seen in either outgroup, variation may represent either within-species suggesting that absence of a medial shelf on polymorphism (within H. messelensis)or the acromion process is the primitive condi- taxonomic polymorphism (it may differenti- tion. ate H. messelensis from a previously uniden- Character 98: Tip of acromion process ti®ed specimen of H. revilliodi). We are un- without anterior projection (0); or with tri- aware of any within-species polymorphism angular anterior projection (1). The tip of in this character in extant bat species, so we the acromion process lacks an anterior pro- favor the latter interpretation. jection in Icaronycteris, Hassianycteris, and No posterolateral projection is present on Palaeochiropteryx. The acromion similarly the acromion process in the outgroups, sug- lacks such a projection in Pteropodidae, Em- gesting that absence of a posterolateral pro- ballonuridae, Yinochiroptera, Phyllostomi- jection in the Eocene bats is primitive. Pres- dae, Natalidae, and Murininae. In contrast, a ence of a triangular posterolateral projection triangular anterior projection is present just on the acromion in some Hassianycteris is ventral and medial to the tip of the acromion apparently a derived condition. process in Mormoopidae, Noctilionidae, Character 100: Dorsal articular facet (for Mystacinidae, Myzopodidae, Thyropteridae, trochiter of humerus) absent from scapula Furipteridae, Molossoidea, Vespertilioninae, (0); or present (1). A secondary articulation Miniopterinae, Myotinae, and Kerivoulinae. between the humerus and scapula occurs in This projection curves slightly ventrally in many bats when the humerus is abducted and some taxa. No anterior projection is present an enlarged trochiter (ϭ greater tuberosity) on the acromion process in the outgroups; contacts a dorsal articular facet on the scap- this suggests that absence of an anteroventral ula. As discussed by Vaughan (1959, 1970b), projection in the Eocene forms may be prim- Hill (1974), Strickler (1978), Hill and Smith itive. We were not able to determine if an (1984), Altenbach and Hermanson (1987), anteroventral process is present in Archaeo- and Schlosser-Strum and Schleimann (1995), nycteris; this taxon is therefore scored ``?'' this secondary articulation forms a critical for this character. part of a ``locking mechanism'' in the shoul- Character 99: Distal acromion process der. Functional explanations for the shoulder- without posterolateral projection (0); or with locking mechanism are diverse and, in some triangular posterolateral projection (1). The cases, contradictory (see discussion in entire lateral surface of the acromion process Schlosser-Strum and Schleimann, 1995). One serves as part of the origin of m. acromiod- possibility is that this mechanism functions eltoideus, a muscle that elevates and rotates in part to arrest the upstroke of the wing, thus the humerus and provides important control facilitating long-range, fast ¯ight (1) by pro- during the upstroke of the wing (Vaughan, viding a longer recovery period for critical 1959, 1970b; Strickler, 1978). Presence of a ¯ight muscles between wing beats, and (2) posterolateral projection on the distal acro- by reducing the amount of force that must be mion may serve to increase the area of at- exerted by many of the same muscles tachment for m. acromiodeltoideus, and may (Vaughan, 1959, 1970b; Hill, 1974; Strickler, affect the moment arm for at least some of 1978; Hill and Smith, 1984). An alternative the ®bers in this muscle. view based on new morphological and ex- The acromion process lacks a posterolat- perimental evidence suggests that the sec- eral projection in Icaronycteris, Archaeonyc- ondary shoulder joint serves to increase the teris, Palaeochiropteryx, and some speci- moment arm of m. pectoralis and reduce mens of Hassianycteris (e.g., SMF Me pronatory movements of the abducted fore- 1500). This condition is similar to that seen arm during the downstroke (Altenbach and in most extant bats. In contrast, a triangular Hermanson, 1987; Schlosser-Strum and posterolateral projection is present on the Schliemann, 1995). distal acromion process in extant Molossidae The dorsal articular facet is situated on the and some specimens of Hassianycteris (e.g., dorsal surface of the scapula immediately an- SMF Me 1540a). Both specimens of Hassi- teromedial to the glenoid fossa. Icaronycteris 72 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 lacks a dorsal articular facet. In contrast, a character cannot be polarized a priori be- dorsal articular facet is present in Hassianyc- cause both outgroups lack a dorsal articular teris and Palaeochiropteryx. Most extant facet. bats have a dorsal articular facet; exceptions Character 102: Infraspinous fossa of include Pteropodidae, some Emballonuridae, scapula narrow, length Ն2 times width (0); Rhinopomatidae, and Noctilionidae. Both or wide, length Յ1.5 times width (1). The outgroups and most other mammals do not infraspinous fossa provides the site of origin have a secondary articulation between the for m. infraspinatus and m. teres major, mus- humerus and scapula, and therefore lack a cles that act to ¯ex, rotate, and (in the case dorsal articular facet on the scapula. This of m. infraspinatus) abduct the humerus suggests that absence of a dorsal articular (Vaughan, 1959, 1970b; Norberg, 1970; facet in Icaronycteris represents a relatively Strickler, 1978; Hermanson and Altenbach, primitive condition. We were unable to de- 1983, 1985). The width of the infraspinous termine if a dorsal articular facet occurs in fossa is measured from the junction of the Archaeonycteris; this taxon is therefore scapular spine and the vertebral border of the scored ``?'' for this character. scapula to the axillary border in a line per- Character 101: Dorsal articular facet pendicular to the long axis of the fossa faces dorsolaterally and consists of small (which is de®ned by the line of maximum groove on anteromedial rim of glenoid fossa length). The infraspinous fossa is narrow (0); or faces dorsolaterally and consists of (length greater than or equal to twice the an oval facet on anteromedial rim of glenoid width) in Icaronycteris (®gs. 22, 30), Ar- fossa (1); or faces dorsally and consists of a chaeonycteris, and Hassianycteris. In con- large, ¯at facet clearly separated from gle- trast, the infraspinous fossa is relatively wide noid fossa (2).InPalaeochiropteryx, the dor- (length Յ1.5 times width) in Palaeochirop- sal articular facet consists of a small, dorso- teryx (®g. 2). Among extant forms, a narrow laterally facing groove on the anteromedial infraspinous fossa is seen in Pteropodidae, rim of the glenoid fossa of the scapula. In Rhinopomatoidea, Rhinolophoidea, some contrast, the dorsal articular facet consists of Mormoopidae, Myzopodidae, Thyropteridae, an oval, dorsolaterally facing facet on the an- some Molossinae, and Murininae. A wide in- teromedial rim of the glenoid fossa in Has- fraspinous fossa occurs in Phyllostomidae, sianycteris. Among extant bats, the ``small Noctilionidae, some Mormoopidae, Mysta- groove'' condition is seen in Emballonuri- cinidae, Furipteridae, Natalidae, Antrozo- dae, Nycteridae, and some Hipposiderinae. idae, Tomopeatinae, some Molossinae, Ves- An oval, dorsolaterally facing dorsal articular pertilioninae, Miniopterinae, Myotinae, and facet on the anteromedial rim of the glenoid Kerivoulinae. The infraspinous fossa in both fossa is found in Craseonycteridae, some outgroups is narrow, suggesting that a nar- Megadermatidae, some Phyllostomidae, row fossa is relatively primitive and that the Mormoopidae, Thyropteridae, Furipteridae, wide infraspinous fossa in Palaeochiropteryx and Natalidae. In contrast, the dorsal articular represents a derived condition. facet is a large, ¯at facet that faces dorsally Character 103: Infraspinous fossa with and is clearly separated from the glenoid fos- one facet (0); or two facets (2); or three fac- sa in some Megadermatidae, Rhinolophinae, ets (2). Faceting of the infraspinous fossa, some Hipposiderinae, some Phyllostomidae, which effectively serves to increase the sur- Mystacinidae, Myzopodidae, Molossoidea, face area without increasing the outline di- and Vespertilionidae. This character cannot mensions of the fossa, is thought to function be scored in taxa that lack a dorsal articular to increase the area of origin for m. infras- facet or those in which presence/absence of pinatus and m. subscapularis (Vaughan, a dorsal articular facet has not been deter- 1959; see discussion of m. subscapularis mined. Both outgroups, Pteropodidae, and function below under character 106). Facet- Icaronycteris lack a dorsal articular facet; ing may also re¯ect compartmentalization or these taxa are scored ``-'' for this character. subdivision of ®bers of these muscles into Archaeonycteris is scored ``?'' to re¯ect in- units with distinct functions, although this suf®cient information for this taxon. This has yet to be investigated. 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 73

The infraspinous fossa has two facets in the posterolateral facet is more extensive in Icaronycteris, Archaeonycteris, Hassianyc- Megadermatidae and Rhinolophidae. In these teris, and Palaeochiropteryx. The smaller of forms (which have three facets), the postero- these (the medial facet) lies posterior to the lateral facet extends into the infraglenoid re- scapular spine and faces posterolaterally; the gion anterolaterally, and it wraps around the larger lateral facet, which is convex, lies caudal end of the intermediate fossa poste- more posteriolaterally and faces dorsally. riorly. This character cannot be evaluated in Among extant bats, two facets are found in taxa that have only one infraspinous facet, some Pteropodidae, Rhinopomatoidea, Nyc- which precludes useful scoring of this char- teridae, some Phyllostomidae, Mystacinidae, acter in the outgroups; these taxa are scored Furipteridae, Natalidae, Antrozoidae, Tomo- ``-'' for this feature. Accordingly, the primi- peatinae, some Vespertilioninae, and Myoti- tive condition for bats cannot be reconstruct- nae. In contrast, the infraspinous fossa has ed a priori. three facets in some Pteropodidae, Emballon- Character 106: Thick lip present along uridae, Megadermatidae, Rhinolophidae, axillary border of scapula (0); or thick lip some Phyllostomidae, Mormoopidae, Nocti- with bladelike lateral edge present (1); or lionidae, Myzopodidae, Thyropteridae, Mo- thick lip absent, axillary border ¯at or slight- lossinae, some Vespertilioninae, Myotinae, ly upturned (2). The axillary border of the Murininae, and Kerivoulinae. In these forms, scapula provides part of the site of origin of the facet just posterior to the scapular spine m. subscapularis, which adducts and extends is called the posteromedial facet; this appears the humerus (although a more important to be homologous to the medial facet found function may be to stabilize and provide ®ne in the same position in bats that have only control of the wing), and m. teres minor, two facets. The larger lateral facet is further which is a weak ¯exor of the humerus subdivided into an intermediate facet and a (Vaughan, 1959, 1970b; Norberg, 1970; posterolateral facet. The posterolateral facet Strickler, 1978; Hermanson and Altenbach, lies along the lateral edge of the scapula and 1983, 1985). Morphology of the edge of the faces posterolaterally; the intermediate facet scapula along its axillary (lateral/posterolat- lies between the postermedial and postero- eral) border varies among bats. A thick lip is lateral facets and faces anteromedially. The present along the axillary border of the scap- infraspinous fossa is not subdivided in the ula in Icaronycteris (®g. 30), Archaeonycter- outgroups. In these forms, there is only a sin- is, Hassianycteris, and Palaeochiropteryx. gle, large convex facet in the infraspionous The bone itself is fairly thin; the appearance fossa. Accordingly, it is not possible to de- of a thickened lip is produced by an abrupt termine a priori the primitive condition for fold near the axillary border. This condition bats (two facets or three facets). is similar to that seen among extant Ptero- Character 105: Lateral or posterolateral podidae, some Emballonuridae, Mystacini- facet of infraspinous fossa restricted, does dae, Thyropteridae, Molossoidea, Vesperti- not extend into infraglenoid region anteriorly lioninae, Miniopterinae, Myotinae, and Ker- or wrap around intermediate facet at poste- ivoulinae. In contrast, the axillary border of rior (caudal) angle of scapula (0); or pos- the scapula is characterized by a thick lip terolateral facet more extensive, extends into with a bladelike lateral edge in some Em- infraglenoid region and wraps around cau- ballonuridae, Noctilionoidea, and Myzopod- dal end of intermediate facet (1).InIcaron- idae. No thickening is seen along the axillary ycteris, Archaeonycteris, Hassianycteris, and border in the remaining lineages of bats; in- Palaeochiropteryx, the extent of the lateral stead, the edge of the scapula is ¯at or slight- facet of the infraspinous fossa is relatively ly upturned, and the abrupt fold is absent. restricted; this facet does not extend into the Both outgroups exhibit a simple, thick lip infraglenoid region or wrap around the me- (state ``0'') on the axillary border, suggesting dial facet at the posterior (caudal) angle of that this is the primitive condition. the scapula. A similar condition is seen in Character 107: Pit for attachment of cla- most extant bats regardless of the number of vicular ligament absent from scapula (0); or facets in the infraspinous fossa. In contrast, present anterior and medial to glenoid fossa 74 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

(1). The clavicular ligament extends between chaeonycteris, and Hassianycteris. In con- the base of the coracoid process and the clav- trast, a triangular anteromedial ¯ange is pres- icle. There is no evidence of a pit for attach- ent on the medial superior (anterior) border ment of the clavicular ligament on the scap- of the scapula in Palaeochiropteryx. Among ula in Icaronycteris and Hassianycteris, and extant bats, the anteromedial ¯ange is absent the same is true in most extant bat lineages. in Pteropodidae, Emballonuridae, Rhino- In contrast, a distinct pit for the clavicular pomatoidea, Yinochiroptera, Noctilionoidea, ligament is present anterior and medial to the Mystacinidae, Myzopodidae, Thyropteridae, glenoid fossa in some Emballonuridae, Rhin- and Furipteridae. A large, triangular antero- olophoidea, some Phyllostomidae, and Mor- medial ¯ange is present in Myzopodidae, Na- moopidae. When a dorsal articular facet is talidae, Molossoidea, and Vespertilionoidea. present, the pit for the clavicular ligament is Absence of this ¯ange in both outgroups located medial to this structure. The pit for (and most other mammals) indicates that ab- the clavicular ligament is absent in both out- sence of this structure is the primitive con- groups, suggesting that absence of this struc- dition. ture represents the primitive condition. We Character 109: Coracoid process stout were unable to determine if a pit for the cla- and of moderate length (0); or very long and vicular ligament is present in Archaeonycter- thin (1). The coracoid process of the scapula is and Palaeochiropteryx; these taxa are is the site of origin of the coracoid head (ϭ therefore scored ``?'' for this character. short head) of m. biceps brachii. M. biceps Character 108: Anteromedial edge of brachii is a two-part muscle that ¯exes and scapula without projections or ¯anges (0); or rotates the forearm, adducts the wing, and with triangular anteromedial ¯ange (1). The helps to hold the forearm rigidly outstretched anteromedial ¯ange is a roughly triangular against the the opposing action of m. triceps ¯ange that projects ventrally from the anter- during much of the downstroke (Vaughan, omedial edge of the scapula (medial to the 1959, 1970b; Norberg, 1970; Strickler, 1978; suprascapular notch) in some bats (Vaughan, Hermanson and Altenbach, 1983, 1985). 1959, 1970a; Norberg, 1970; Strickler, 1978; Length of the coracoid process plays a role Hermanson and Altenbach, 1983, 1985). The in function of m. biceps brachii because a dorsal aspect of the scapular edge in this re- long coracoid process enables the coracoid gion serves as the origin of the transverse head of the biceps to act as an adductor of scapular ligament, while part of the origins the wing by placing its origin below the line of m. subscapularis and the anterior division of the long axis of the humerus; the greater of m. serratus anterior occupy its ventral sur- the displacement from this axis, the greater face (Vaughan, 1959, 1970b; Norberg, 1970; the mechanical advantage of the biceps as an Strickler, 1978; Hermanson and Altenbach, adductor (Vaughan, 1959, 1966). 1983, 1985). Projections from the border of The coracoid process of the scapula is the scapula in this region apparently simul- stout (wider than the clavicle) and of mod- taneously serve to provide an increased area erate length in Icaronycteris, Archaeonycter- of attachment for the transverse scapular lig- is, Hassianycteris, and Palaeochiropteryx. ament and an increased area for muscle ori- This condition is similar to that seen in most gins. Presence of a large, triangular antero- extant bats. In contrast, the coracoid is very medial ¯ange is correlated with relatively long and thin (distal half not as wide as clav- large size of m. subscapularis (Vaughan, icle) in Furipteridae, Natalidae, Tomopeati- 1959), which is one of the principal adduc- nae, and some Molossinae. Both outgroups tors and extensors of the humerus. M. sub- exhibit the ``short and stout'' morphology, scapularis plays an important role in the suggesting that this is the primitive condition downstroke during ¯ight and also helps to for the coracoid process. support the weight of the anterior part of the Character 110: Coracoid process curves body in terrestrial locomotion (Vaughan, ventrolaterally (0); or curves ventrally (1); 1959, 1970b; Strickler, 1978). or curves ventromedially (2). The direction The anteromedial rim of the scapula lacks of curvature of the coracoid process varies projections or ¯anges in Icaronycteris, Ar- considerably among bats (®g. 31). Orienta- 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 75

Fig. 31. Anterior view of the shoulder girdle showing the curvature of the coracoid process of the scapula, and the relative sizes of the coracoid and glenoid heads of m. biceps brachii (redrawn from Vaughan, 1970b: ®g. 10). A. Antrozous pallidus (Antrozoidae). B. Eumops perotis (Molossidae). tion of the coracoid process affects the func- coracoid process takes the origin of m. bi- tion of m. biceps brachii, whose coracoid ceps brachii out of the way of the lesser tu- head originates from this process. Ventrolat- berosity of the humerus, allowing more space eral curvature of the coracoid places the cor- for the origin and a large belly for the muscle acoid head of m. biceps brachii well lateral (®g. 31B; Vaughan, 1959, 1966). The site of to the long axis of the humerus throughout origin is only slightly offset from the long the ®rst half of the downstroke, providing axis of the humeral shaft at the top of the mechanical advantage that facilitates brachial upstroke, which means that the biceps is adduction (Vaughan, 1959, 1966). The great- probably not an effective adductor at the top est mechanical advantage of m. biceps bra- of the upstroke (Vaughan, 1959, 1966). How- chii as a brachial adductor occurs when the ever, the angle between the origin of the cor- humerus reaches a horizontal position (®g. acoid head and its insertion decreases contin- 31A), because the origin of the coracoid head uously through the downstroke, increasing of m. biceps brachii is at its maximum dis- the relative offset of the coracoid process placement from the long axis of the humeral from the axis of the humeral shaft and thus shaft (Vaughan, 1959, 1966). Ef®ciency of increasing mechanical advantage. Ef®ciency the biceps is reduced when the wing is de- of m. biceps as an adductor thus increases pressed because the long axis of the humeral continuously throughout the downstroke and shaft approaches the coracoid process, thus peak ef®ciency is probably reached at the reducing the distance between origin and in- bottom of the downstroke (Vaughan, 1959, sertion of the coracoid head (Vaughan, 1959, 1966). 1966). The coracoid process curves ventrolater- In contrast, ventromedial curvature of the ally in Icaronycteris, Archaeonycteris, Has- 76 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 sianycteris, and Palaeochiropteryx. This ula, cranial to the transverse scapular liga- condition is similar to that seen in most ex- ment. In some bats (e.g., Noctilio), this lig- tant bats. In contrast, the coracoid process ament provides part of the origin for m. ac- curves ventrally (i.e., remains with in a plane romiodeltoideus (Strickler, 1978). perpendicular to that of the blade of the scap- The suprascapular process is absent in Ica- ula) in Furipteridae, Natalidae, some Vesper- ronycteris, Archaeonycteris, Hassianycteris, tilioninae, and Myotinae. The coracoid pro- and Palaeochiropteryx. Among extant bats, cess curves ventromedially in Molossidae, the suprascapular process is similarly absent some Vespertilionidae, and Miniopterinae. in some Pteropodidae, some Emballonuridae, The coracoid process curves ventrolaterally Nycteridae, Megadermatidae, Rhinolophinae, in both outgroups, suggesting that the con- some Phyllostomidae, Myzopodidae, Natali- dition seen in the Eocene bats is primitive. dae, Tomopeatinae, Murininae, and Kerivou- Character 111: Tip of coracoid process linae. In contrast, a suprascapular process is not ¯ared, approximately same width as cor- present in some Pteropodidae, some Embal- acoid shaft (0); or tip distinctly ¯ared (1); lonuridae, Hipposiderinae, some Phyllostom- or bifurcated (2). The tip of the coracoid pro- idae, Noctilionidae, Mormoopidae, Mysta- cess of the scapula is blunt and approximate- cinidae, Thyropteridae, Furipteridae, Antro- ly the same width as the shaft of the coracoid zoidae, Molossinae, Vespertilioninae, Min- in Icaronycteris, Archaeonycteris, Hassi- iopterinae, and Myotinae. The suprascapular anycteris, Palaeochiropteryx, and most ex- process is present in both outgroups, sug- tant bats. There is no evidence of a ¯are or gesting that absence of this process in the bifurcation in the tip in these forms. In con- Eocene bats represents a derived condition. trast, the tip of the coracoid process is dis- Character 113: Clavicle articulates with tinctly ¯ared (wider than the shaft) in some or lies in contact with acromion process (0); Emballonuridae, Phyllostomidae, Mormoop- or is suspended by ligaments between acro- idae, Thyropteridae, Molossinae, some Ves- mion and coracoid processes (1); or articu- pertilioninae, and Miniopterinae. A bifurcat- lates with or lies in contact with coracoid ed coracoid tip occurs in some Vespertilion- process (2). The distal (dorsolateral) clavicle inae. The tip of the coracoid process is not articulates with the scapula in the region be- ¯ared or bifurcated in the outgroups, sug- tween the tip of the acromion process and the gesting that the condition seen in the Eocene base of the coracoid process. The articulation bats is primitive. The tip of the coracoid pro- can be accomplished in several different cess was broken on all skeletons of Mysta- ways, depending on whether the clavicle is cina available to us; Mystacinidae is accord- associated principally with the acromion, ingly scored ``?'' for this character. principally with the coracoid, or is suspended Character 112: Suprascapular process between the two. Strickler (1978) indicated present (0); or absent (1). The suprascapular that differences in the claviculoscapular ar- process is a projection that extends medially ticulation appear to have surprisingly little and somewhat anteriorly from the anterolat- effect on the mobility of the scapula with re- eral border of the scapular notch at the base spect to rotary movements about its longi- of the coracoid process, just medial to the tudinal axis (i.e., the axis that runs roughly anterior edge of the glenoid fossa and just from the glenoid fossa to the posteromedial anterior to the pit for the clavicular ligament corner of the scapula), but he noted that con- (in those forms that have this pit; see char- trol of scapular movements about other axes acter 107 above). This process may result has yet to be studied. Hermanson (1981) from ossi®cation of part of an apparently un- studied clavicular function in Antrozous and named ligament5 that extends between this found that clavicular movements (adduction region and the anteromedial rim of the scap- and abduction relative to the sagittal plane) were synchronous with the wingbeat cycle. 5 This ligament might be considered a subdivision of He suggested that movements of the distal the transverse scapular ligament; previous descriptions end of the clavicle (which describes an ar- of the ligaments in this area (e.g., by Vaughan, 1959; cuate path in the transverse plane) permit the Strickler, 1978) are not clear. scapula to be abducted and adducted across 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 77

Fig. 32. Medial view of the proximal end of the left humerus in selected taxa; redrawn from Smith (1972: ®g. 6). A, Balantiopteryx plicata (Emballonuridae); B, Noctilio albiventris (Noctilionidae); C, Micronycteris sylvestris (Phyllostomidae); D, Desmodus rotundus (Phyllostomidae); E, Thyroptera tri- color (Thyropteridae); F, Natalus stamineus (Natalidae); G, Furipterus horrens (Furipteridae); H, Myotis grisescens (Myotinae); I, Molossus molossus (Molossinae). GT, greater tuberosity ϭ trochiter; H, head of humerus; LT, lesser tuberosity.

the dorsal contour of the thorax. However, seonycteridae, Rhinolophoidea, Phyllostom- this study did not address possible differ- idae, and Mormoopidae. The clavicle artic- ences in function of the claviculoscapular ulates with the acromion process of the scap- joint associated with different morphologies. ula in both outgroups, suggesting that the Assumming little postmortem change in its condition seen in Icaronycteris is relatively position in available specimens, our obser- primitive. We were unable to determine the vations suggest that the clavicle articulated mode of articulation of the clavicle with the with the acromion process of the scapula in scapula in Archaeonycteris, Hassianycteris, Icaronycteris (®gs. 22, 30). Among extant and Palaeochiropteryx; these forms are bats, a similar condition (clavicle articulates therefore scored ``?'' for this character. with or lies in contact with the acromion pro- cess) is seen only in Pteropodidae and Rhin- FORELIMB opomatidae. The clavicle is suspended by ligaments (including the clavicular ligament Character 139: Trochiter does not extend discussed under character 107) that extend to level of proximal edge of head of humerus between the acromion and coracoid process (0); or extends just to level of proximal edge in Noctilionidae, Mystacinidae, Nataloidea, of head (1); or extends proximally well be- Molossoidea, and Vespertilionidae. The clav- yond level of head (2). The trochiter (ϭ icle articulates with or lies in contact with greater tuberosity) of the humerus in bats ex- the coracoid process in Emballonuridae, Cra- tends varying distances proximally relative 78 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Fig. 33. Anterior view of the distal end of the left humerus in selected taxa; redrawn from Smith (1972: ®g. 6). A, Balantiopteryx plicata (Emballonuridae); B, Noctilio albiventris (Noctilionidae); C, Micronycteris sylvestris (Phyllostomidae); D, Desmodus rotundus (Phyllostomidae); E, Pteronotus par- nellii; F. Thyroptera tricolor (Thyropteridae); G, Natalus stamineus (Natalidae); H, Furipterus horrens (Furipteridae); I, Lasiurus borealis (Vespertilioninae); J, Myotis grisescens (Myotinae); K, Tadarida brasiliensis (Molossinae); L, Molossus molossus (Molossinae). CC, central surface of capitulum; E, epitrochlea; LC, lateral surface of capitulum; T, trochlea. to the humeral head (®g. 32). As discussed of the head of the humerus in Emballonuri- under 98 above, the importance of the degree dae, Rhinopomatidae, Nycteridae, Megader- of proximal extension of the trochiter is re- matidae, and Noctilionidae. The trochiter ex- lated to the tendency of this structure to form tends well beyond the head of the humerus a secondary articulation with the scapula in Craseonycteridae, Rhinolophidae, and all during abduction of the humerus, thus facil- Yangochiroptera with the exception of Noc- itating a humeral locking mechanism. The tilionidae. In contrast, the trochiter does not distribution of trochiter morphologies is not extend to the level of the humeral head in correlated with the distribution of dorsal ar- Pteropodidae and both outgroups. This sug- ticular facet morphologies (see character gests that both conditions seen among the 100); accordingly, we treat them as indepen- Eocene bats are derived. dent characters. Character 140: Head of humerus round The trochiter extends proximally just to in outline in medial view (0); or oval or el- the level of the proximal edge of the head of liptical (1). The head of the humerus is round the humerus in Icaronycteris (®gs. 22, 30), in outline when seen in medial view6 in Ica- Archaeonycteris (®g. 3), and Palaeochirop- teryx (®g. 2). In contrast, the trochiter ex- 6 Determination of anatomical directions for bat limb tends well beyond the level of the proximal elements is complicated by the unusual orientation of edge of the humeral head in Hassianycteris the limbs during ¯ight. In speci®ying ``medial view'' in (®g. 4). Among extant bats, the trochiter ex- this character, we have followed Smith (1972) and as- tends just to the level of the proximal edge sumed that the wing is folded and the humerus adducted. 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 79

Fig. 34. Anterior view of the elbow region in A, Artibeus toltecus (Phyllostomidae) and B, Molossus molossus (Molossidae; redrawn from Smith, 1972). Note that the lateral offset of the distal articular facets in A places the long axes of the humerus and radius in separate but roughly parallel planes, whereas these axes lie in approximately the same plane in B.

ronycteris, Archaeonycteris, Hassianycteris, basis of a series of ridges and grooves (®g. and Palaeochiropteryx. A similar condition 33). The combined articular surfaces typical- is seen in extant Pteropodidae, Rhinopoma- ly form a spool-shaped structure, although toidea, Nycteridae, Phyllostomidae, Mysta- morphology of this structure varies consid- cinidae, Nataloidea, Antrozoidae, Vesperti- erably both among and within families of lioninae, Myotinae, Murininae, and Kerivou- bats depending upon the relative proportions linae (®g. 32). In contrast, the head of the of the three component articular surfaces humerus is oval or elliptical in outline in Em- (Miller, 1907; Smith, 1972). ballonuridae, Megadermatidae, Rhinolophi- The structural unit formed by the three dis- dae, Noctilionidae, Mormoopidae, Molossi- tal articular surfaces is laterally displaced dae, and Miniopterinae. The head of the hu- from the line of the humeral shaft in Icaron- merus is round in both outgroups, suggesting ycteris, Archaeonycteris, Hassianycteris, and that this is the primitive condition. Palaeochiropteryx. A similar arrangement is Character 141: Humerus with distal ar- seen among extant Pteropodidae, Yinochi- ticular surfaces displaced laterally from line roptera, Phyllostomidae, some Mormoopi- of shaft (0); or facets in line with shaft, not dae, Noctilionidae, Nataloidea, Myotinae, displaced laterally (1). Following the no- Murininae, and Kerivoulinae (®g. 33). In menclature used by Smith (1972) for describ- each of these forms, the capitulum lies either ing humeral morphology in bats, the distal completely or mostly lateral to the lateral articular surfaces of the humerus include the edge of the humeral shaft, and the medial trochlea, capitulum, and the lateral surface of edge of the trochlea lies well lateral to the capitulum (ϭ lateral epicondyle of Hill, medial edge of the humeral shaft. This ar- 1974). These articular surfaces lie adjacent rangement shifts the elbow joint laterally so to one another and are distinguished on the that the long axes of the humerus and radius 80 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 move in separate but parallel planes when the Myotinae. In contrast, a sesamoid element is joint is ¯exed (®g. 34A; Smith, 1972). In present dorsal to the magnum±trapezium ar- contrast, the distal articular surfaces of the ticulation in Emballonuridae, Nycteridae, humerus lie in line with shaft in Emballon- Phyllostomidae, Mormoopidae, and Molos- uridae, some Mormoopidae, Mystacinidae, sinae. The status of this sesamoid is un- Molossoidea, Vespertilioninae, Miniopteri- known in the outgroups, so this character nae, and Myotinae (®g. 33). The capitulum cannot be polarized a priori. The condition occupies variable positions in these forms, in Archaeonycteris, Hassianycteris, Palaeo- but often lies mostly or entirely medial to the chiropteryx, Rhinopomatoidea, Megaderma- lateral edge of the shaft. The medial edge of tidae, Rhinolophinae, Mystacinidae, Myzo- the trochlea always lies at or medial to the podidae, Thyropteridae, Furipteridae, Antro- medial edge of the humeral shaft. As a result, zoidae, Tomopeatinae, Miniopterinae, Muri- the long axes of the humerus and radius lie ninae, and Kerivoulinae is also unknown; in roughly the same plane (®g. 34B; Smith, these taxa are therefore scored ``?'' for this 1972). The distal articular surfaces are dis- character. placed from the line of the shaft of the hu- Character 147: Sesamoid element dorsal merus in both outgroups, suggesting that the to unciform±magnum articulation absent (0); condition seen in the Eocene bats is primi- or present (1). No sesamoid element is pres- tive. ent dorsal to the unciform±magnum articu- Character 142: Epitrochlea broad (0); or lation in Icaronycteris. This condition is also relatively narrow (1). The epitrochlea (ϭ seen in extant Pteropodidae, Emballonuridae, medial process, medial epicondyle) compris- Nycteridae, Hipposiderinae, Phyllostomidae, es the medial portion of the distal end of the and Vespertilioninae. In contrast, a sesamoid humerus adjacent to the articular facets (®g. element is present dorsal to the unciform± 33). The epitrochlea is broad (Ն40% of magnum articulation in Mormoopidae, Noc- width of the articular facets) in Icaronycteris, tilionidae, Natalidae, and Molossinae. The Archaeonycteris, Hassianycteris, and Pa- status of this sesamoid is unknown in the laeochiropteryx. This condition is also seen outgroups, so this character cannot be polar- in most extant bats. In contrast, the epitro- ized a priori. The condition in Archaeonyc- chlea is relatively narrow (Ͻ25% width of teris, Hassianycteris, Palaeochiropteryx, the articular facets) in Mystacinidae, Molos- Rhinopomatoidea, Megadermatidae, Rhino- soidea, and Vespertilionidae. Morphology of lophinae, Mystacinidae, Myzopodidae, Thy- the epitrochlea varies among the outgroups; ropteridae, Furipteridae, Antrozoidae, To- the epitrochlea is broad in Scandentia and mopeatinae, Miniopterinae, Murininae, and narrow in Dermoptera. Accordingly, the Kerivoulinae is also unknown; these taxa are primitive condition for this character cannot therefore scored ``?'' for this character. be reconstructed a priori. Character 149: Wing digit II with ossi®ed Character 144: Sesamoid element dorsal ®rst phalanx (0); or ®rst phalanx unossi®ed to magnum±trapezium articulation absent or absent (1). Phalanges in both the manus (0); or present (1). Although wrist morphol- and pes are numbered from proximal to dis- ogy has never been intensively surveyed tal. The ®rst phalanx of each wing digit is across all bat families, a recent study by Cy- identi®ed by its articulation with the distal pher (1996) illustrated differences among end of the corresponding metacarpal. The taxa in terms of the number and location of ®rst phalanx in wing digit II (the ``index ®n- sesamoid elements associated with tendons ger'') is ossi®ed in Icaronycteris, Archaeo- in the wrist. Preservation of the wrist of Ica- nycteris, Hassianycteris, and Palaeochirop- ronycteris in PU 18150 allowed us to score teryx. This element is also ossi®ed in extant presence or absence of two of these sesa- Pteropodidae, Rhinopomatoidea, Megader- moids. No sesamoid element is present dor- matidae, Noctilionoidea, Mystacinidae, Mo- sal to the magnum±trapezium articulation in lossoidea, and Vespertilionidae. In contrast, Icaronycteris. This condition is also seen in the ®rst phalanx of wing digit II is unossi®ed extant Pteropodidae, Hipposiderinae, Nocti- in Emballonuridae, Nycteridae, Rhinolophi- lionidae, Natalidae, Vespertilioninae, and dae, and Nataloidea. The ®rst phalanx of dig- 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 81

it II is ossi®ed in both outgroups, suggesting Phyllostomidae, Mormoopidae, Mystacini- that the condition seen in the Eocene bats is dae, Myzopodidae, and Thyropteridae. The primitive. third phalanx is ossi®ed only at the base (and Character 150: Wing digit II with ossi®ed the distal portion of the element remains car- second phalanx (0); or second phalanx unos- tilaginous) in Furipteridae, Natalidae, Molos- si®ed or absent (1). The second phalanx is soidea, and some Vespertilionidae. The third de®ned by its articulation with the distal end phalanx is completely unossi®ed or absent in of the ®rst phalanx of a particular digit. The Pteropodidae, Emballonuridae, Yinochirop- second phalanx in wing digit II is ossi®ed in tera, Noctilionidae, and some Vespertilioni- Icaronycteris, Archaeonycteris, Hassianyc- dae. The third phalanx in digit III of the ma- teris, and Palaeochiropteryx. Among extant nus is fully ossi®ed in both outgroups and bats, the second phalanx is present and os- most other mammals, suggesting that the si®ed only in Pteropodidae and Rhinopoma- condition seen in Icaronycteris is primitive. tidae; this element is unossi®ed or absent in Lack of an ossi®ed third phalanx in Archaeo- all other lineages. Digit II of the manus has nycteris, Hassianycteris, and Palaeochirop- an ossi®ed second phalanx in both out- teryx is relatively derived. groups, suggesting that the condition seen in the Eocene bats is primitive. POSTERIOR AXIAL SKELETON AND PELVIS Character 151: Wing digit II with ossi®ed third phalanx (0); or third phalanx unossi®ed Character 158: No vertebral fusion in or absent (1). The third phalanx (ϭ ungual posterior thoracic and lumbar series (0); or phalanx) is de®ned by its articulation with at least three vertebrae fused (1). The ver- the distal end of the second phalanx, and in tebral column of most mammals contains mammals it typically bears a claw or ho- posterior thoracic and lumbar vertebrae as mologous structure (nail or hoof). The third separate elements that can move relative to phalanx of wing digit II is present and ossi- one another. They articulate at various hy- ®ed in Icaronycteris and Archaeonycteris.In pophyseal joints, but are not fused. This is both taxa, the third phalanx of this digit also true of the posterior thoracic and lumbar clearly bore a claw. Indeed, the speci®c ep- series in Icaronycteris (frontispiece, ®g. 21), ithet ``index'' for the type species of Icaron- Archaeonycteris, Hassianycteris (®g. 4), and ycteris refers to presence of a claw on the Palaeochiropteryx (®g. 3), all of which ex- index ®nger (digit II; Jepsen, 1966). In con- hibit no evidence of vertebral fusion. Among trast, the third phalanx of digit II is unossi- extant bats, vertebral fusion in this area is ®ed or absent in Hassianycteris and Palaeo- similarly absent in Pteropodidae, Emballon- chiropteryx. Among extant bat lineages, the uridae, Rhinopomatidae, Nycteridae, some third phalanx is present and ossi®ed only in Megadermatidae, Rhinolophidae, some Hip- Pteropodidae; it is unossi®ed or absent in all posiderinae, Phyllostomidae, Noctilionidae, other groups. Digit II of the manus has an Mystacinidae, Myzopodidae, Thyropteridae, ossi®ed third phalanx in both outgroups, sug- Molossoidea, and Vespertilionidae. In con- gesting that the condition seen in Icaronyc- trast, at least three vertebrae are fused to- teris and Archaeonycteris is primitive. The gether in the posterior thoracic and lumbar state seen in Hassianycteris and Palaeochi- region in Craseonycteridae, some Megader- ropteryx is apparently derived. matidae, some Hipposiderinae, Mormoopi- Character 152: Wing digit III with third dae, Furipteridae, and Natalidae. Both out- phalanx completely ossi®ed (0); or third pha- groups lack vertebral fusion in the posterior lanx ossi®ed only at the base (1); or third thoracic and lumbar regions, suggesting that phalanx unossi®ed or absent (2). The third the condition seen in the Eocene bats is rel- phalanx of wing digit III is small but fully atively primitive. ossi®ed in Icaronycteris. In contrast, the Character 159: Sacrum terminates ante- third phalanx of this digit is either unossi®ed rior to acetabulum (0); or extends posteri- or absent in Archaeonycteris, Hassianycteris, orly to at least the midpoint of the acetabu- and Palaeochiropteryx. A small but fully os- lum (1). The sacrum (as de®ned in this study) si®ed third phalanx is found among extant includes all vertebrae that articulate with the 82 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 pelvis or are fused with those that do. The vertebral body. In constrast, the sacral lami- posterior end of the sacrum terminates ante- nae are narrow or absent in all other taxa, rior to the acetabulum in Hassianycteris.In extending only a short distance laterally from contrast, the sacrum extends posteriorly to at the bodies of the vertebrae. Accordingly, ver- least the midpoint of the acetabulum in Ica- tebral width is less than or equal to three- ronycteris (®g. 23) and Palaeochiropteryx.7 fourths the length of the body of the vertebra. Among extant bats, the sacrum terminates Among extant bats, the sacral laminae are anterior to the acetabulum in Nycteridae and narrow or are absent in Pteropodidae, Em- Rhinolophidae; it extends posteriorly to a ballonuridae, Rhinopomatoidea, Megader- least the midpoint of the acetabulum in all matidae, Rhinolophidae, Noctilionoidea, other lineages. The sacrum terminates ante- Mystacinidae, Furipteridae, and Natalidae. rior to the acetabulum in both outgroups, Among the outgroups, the sacral laminae are suggesting that the condition seen in Hassi- broad in Scandentia but are absent in Der- anycteris is primitive. The condition seen in moptera. Accordingly, this character cannot Icaronycteris and Palaeochiropteryx appears be polarized a priori. Morphology of the sa- derived. We were unable to determine the ex- cral laminae in unknown in Archaeonycteris tent of the sacrum in Archaeonycteris; this and Hassianycteris; these forms are therefore taxon is scored ``?'' for this character. scored ``?'' for this character. Character 160: Sacral laminae narrow or Character 161: Dorsomedial edge of as- absent, vertebra width (including laminae) cending process of ilium upturned, ¯ares less than or equal to three-fourths vertebral dorsally above the level of iliosacral articu- body length (0); or laminae broad, vertebra lation, iliac fossa large and well de®ned (0); width equal to or greater than vertebral or dorsomedial edge not upturned, does not length. Sacral laminae are thin plates of bone extend dorsally beyond the level of the ilios- that extend laterally from the sacral vertebrae acral articulation, iliac fossa not large or posterior to the iliosacral joint. These struc- well de®ned (1). The iliosacral articulation in tures, which are homologous with the trans- mammals is formed between the ascending verse processes of lumbar vertebrae, do not process of the ilium and the sacral vertebrae. articulate with the pelvis. Instead, they typi- The dorsomedial edge of the ascending pro- cally form joints with one another that may cess forms the origin for m. tensor fascia la- or may not be fully fused. Gaps are frequent- tae, which ¯exes and abducts the femur ly present between successive laminae, and (Vaughan, 1959, 1970b). M. gluteus medius lateral enclosure of such gaps may produce (which ¯exes, abducts, and rotates the femur) a series of sacral foramina between succes- also originates from the dorsomedial edge, sive laminae. although the iliac fossa on the dorsolateral The sacral laminae are broad in Icaron- surface of the ascending process forms the ycteris (®g. 23) and Palaeochiropteryx. principal origin of this muscle (Vaughan, Broad sacral laminae are also present in Nyc- 1959, 1970b). teridae, Myzopodidae, Thyropteridae, Mo- The dorsomedial edge of the ascending lossoidea, and Vespertilionidae. In these process of the ilium is upturned and ¯ares forms, the width of each postischial sacral dorsally above the level of the iliosacral ar- vertebra (including the sacral laminae) is ticulation in Icaronycteris, Hassianycteris, subequal to or greater than the length of the and Palaeochiropteryx. Among extant bats, a dorsomedially ¯ared ascending process is 7 Our observations do not agree with those of Russell seen only in Rhinolophidae. In all taxa this and Sige (1970), who reported that the sacrum in Pa- ¯are is associated with a large, well-de®ned laeochiropteryx was very small and included only a sin- iliac fossa. In contrast, the dorsomedial edge gle vertebra. We attribute this discrepency to an incor- of the ascending process is not upturned and rect reconstruction of this region by Russell and Sige (1970), who were working with relatively poorly pre- the ¯are is absent in all other bats. The dor- served material. Discovery of better preserved speci- somedial edge terminates at the level of the mens since their publication (e.g., HLMD Me 15025) iliosacral articulation, and the iliac fossa is demonstrate that Palaeochiropteryx had a long sacrum relatively small and poorly de®ned. The up- composed of multiple vertebrae. turned edge and well-de®ned iliac fossa are 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 83

present in Scandentia but absent in Dermop- the other (Dermoptera); accordingly, this tera; accordingly, this character cannot be character can not be polarized a priori. polarized a priori. We were unable to deter- Character 163: Pubic spine absent (0); or mine the state of the ilium in Archaeonyc- straight (1); or tip of pubic spine bent sharp- teris; this taxon is therefore scored ``?'' for ly dorsally (2). The pubic spine is an elon- this character. gate projection from the anteroventral corner Character 162: Ischium with large ischial of the pubis (Vaughan, 1959; Walton and tuberosity that projects dorsally from poste- Walton, 1970; Simmons, 1994). Presence of rior horizontal ramus (0); or ischial tuber- a pubic spine is a synapomorphy of Chirop- osity small or absent, does not project dor- tera (Simmons, 1994). sally beyond level of ramus (1). The dorsal Two muscles originate from the pubic ischial tuberosity is a projection that extends spine in bats: m. pectineus from the base and dorsally and/or posteromedially from the lateral surface of the spine, m. gracilis from posterior end of the ischium, originating the entire ventrolateral surface of the spine from the ``corner'' of the pelvis at the junc- to the tip (Vaughan, 1959, 1970). M. pecti- tion between the horizontal and vertical rami neus is an adductor and ¯exor of the femur, of the ischium (Vaughan, 1959, 1970a). Two while m. gracilis is a ¯exor of the lower leg muscles may originate from the ischial tu- and an adductor of the hind limb (Vaughan, berosity: m. semitendinosus and m. semi- 1959, 1970b). Additionally, the pubic spine membranosus, both of which act to extend is the site of insertion of m. psoas minor, the femur and ¯ex the lower leg (Vaughan, which originates from the lumbar vertebrae 1959, 1970b). Vaughan (1959) recognized an (Vaughan, 1959, 1970b). Contraction of m. ischial tuberosity in all of the bats that he psoas minor pulls the ventral part of the pel- dissected, but noted that in some forms (e.g., vis forward, thereby arching the posterior Eumops) the ischial tuberosity is large, pro- lumbar section of the vertebral column. jects dorsally above the level of the horizon- Vaughan (1959) noted that this action may tal ramus, and serves as the site of origin of be useful in doubling up the body when the both muscles. In other taxa (e.g., Myotis, Ma- bat is grooming itself while hanging, or when crotus), the tuberosity is smaller, projects pinning an insect to the uropatagium while somewhat medially (not dorsally), and serves adjusting the grip of the jaws on the prey. as the site of origin for just m. semitendi- Action of this muscle may also help to brace nosus. In the latter taxa, m. semimembrano- the vertebral column when the bat lands, and sus originates from the lateral surface of the also while the bat is ¯ying (i.e., against the caudal border of the ischium (Vaughan, shock of the airstream hitting the uropata- 1959). Our examination of skeletons from gium during sudden maneuvers; Vaughan, numerous families revealed a dichotomy be- 1959). tween forms with a large, dorsally projecting The pubic spine is relatively straight and ischial tuberosity and those in which the is- points anterodorsally in Icaronycteris (®g. chial tuberosity is small or absent. Although 23), Archaeonycteris, and Palaeochiropte- we have no new muscle data, we infer that ryx. Among extant bats, this condition is seen morphology of the pelvis is correlated with in Pteropodidae, Emballonuridae, Rhino- differences in muscle origins as observed by pomatoidea, Nycteridae, and Yangochirop- Vaughan (1959). tera. In these forms, the course of the pubic There is no evidence of a large, dorsally spine is essentially parallel to that of the il- projecting ischial tuberosity on the pelvis of ium, with the long axis of the spine directed Icaronycteris (®g. 23), Hassianycteris, and at the anterior lumbar vertebrae. In contrast, Palaeochiropteryx. In contrast, a large, dor- the tip of the pubic spine is sharply upturned sally projecting ischial tuberosity is present in Megadermatidae, Rhinolophinae, and Hip- in Archaeonycteris. Among extant bats, a posiderinae. The tip of the pubic spine in large dorsal ischial tuberosity is present only these taxa points dorsally toward the cranial in Mystacinidae and Molossidae. A large end of the ilium. In Hipposiderinae, a bony dorsally projecting ischial tuberosity is pres- extension from the pubic spine contacts the ent in one outgroup (Scandentia) but not in anterior ilium, thereby enclosing a preaceta- 84 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 bular foramen. The pubic spine is absent in teryx (®g. 2). This condition is also seen in both outgroups, so the form of the pubic most extant bats. In some bats, however, the spine (straight or with tip upturned) cannot shaft of the femur is bent somewhat so that be polarized a priori. We could not determine the distal end is directed more dorsally (the the form of the pubic spine in Hassianycter- equivalent of a lateral bend if the femur were is, so this taxon is scored ``?'' for this char- in the position typical of nonvolant mam- acter. mals). In bats, which already have a femur Character 165: Obturator foramen nor- that projects laterally from the hip, this bend mal, rim well de®ned (0); or foramen par- serves to raise the knee joint even higher, tially in®lled with thin, bony sheet along pos- well above the hip joint in some forms. A teroventral rim (1). The obturator foramen of bend in the shaft of the femur is seen in the pelvis is enclosed dorsally and posteri- Rhinolophoidea and some Phyllostomidae. orly by the rami of the ischium and anteriorly The shaft of the femur is straight in both out- and ventrally by the pubis. In most mam- groups, indicating that the ``straight'' condi- mals, the rim of the obturator foramen is well tion is primitive. de®ned and rounded in cross section. Mus- Character 170: Fibula complete and well cles originating around the rim of the obtu- developed (0); or thin and threadlike (1); or rator foramen include m. adductor brevis (ex- absent or entirely unossi®ed (2). The ®bula tensor and/or adductor of femur), m. adduc- in Icaronycteris (frontispiece) and Archaeo- tor magnus (extensor and rotator of femur), nycteris is well developed (relatively robust) and m. obturator externus (extensor and ad- and complete (ossi®ed from knee to ankle). ductor of femur; Vaughan, 1959, 1970b). M. In contrast, the ®bula in Hassianycteris and obturator internus is absent in bats, probably Palaeochiropteryx is relatively much thinner, as a result of hip modi®cations associated almost threadlike. Among extant bats, a com- with 90Њ rotation of the hindlimbs (Simmons, plete, well developed ®bula is found only in 1994). Pteropodidae, some Phyllostomidae, Mysta- The obturator foramen in Icaronycteris, cinidae, and Molossidae. The ®bula is thin, Archaeonycteris, Hassianycteris, and Pa- threadlike, and often only partially ossi®ed laeochiropteryx is of the normal mammalian in Emballonuridae, Rhinopomatoidea, Me- morphology with a well-de®ned rim. This is gadermatidae, Rhinolophidae, some Phyllos- similar to the condition seen in most extant tomidae, Mormoopidae, Noctilionidae, Na- bats. In contrast, the obturator foramen is taloidea, Antrozoidae, and Vespertilionidae. partially in®lled by a thin, bony sheet along The ®bula is absent or entirely unossi®ed in the posteroventral rim in Rhinopomatidae, Nycteridae. Both outgroups have a complete, Rhinolophinae, Hipposiderinae, and Thyrop- well developed ®bula, suggesting that the teridae. Such in®lling is absent in both out- condition seen in Icaronycteris and Archaeo- groups, suggesting that the condition seen in nycteris is relatively primitive. The thin, the Eocene bats is primitive. threadlike ®bula seen in Hassianycteris and Palaeochiropteryx apparently represents a HINDLIMB derived condition. Character 171: Calcar absent (0); or Character 169: Shaft of femur straight present (1). The calcar is a bony and/or car- (0); or with bend that directs distal shaft dor- tilaginous rod that extends from the ankle to sally (1). Although all bats share a complex support the trailing edge of the uropatagium. series of modi®cations of the hip and proxi- Presence of a calcar (and m. depressor ossis mal femur that serve to rotate the hindlimbs styliformis, which adducts the calcar toward approximately 90Њ from the typical mam- the lower leg) is considered to be a synapo- malian condition (Simmons, 1994), the fem- morphy of Chiroptera (Simmons, 1994, oral shaft and distal femur remain relatively 1995). However, there is no evidence that a unmodi®ed. As in most other mammals, the calcar was present in Icaronycteris and Ar- shaft of the femur is straight in Icaronycteris chaeonycteris. A calcar is lacking in every (frontispiece, ®g. 21), Archaeonycteris (®g. known specimen, and our examination of the 3), Hassianycteris (®g. 4), and Palaeochirop- calcaneum suggests that no calcar facet is 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 85

present on the calcaneum of the these forms, tant groups. However, inclusion of relatively leading us to conclude that a calcar was ab- incomplete taxaÐbe they fossils or poorly sent in these taxa. In contrast, a calcar is known extant groupsÐcan sometimes dra- present in Hassianycteris and Palaeochirop- matically reduce the degree of phylogenetic teryx. In both taxa, the appearance of the cal- resolution obtained by increasing the number car suggests that it may have been largely of equally parsimonious topologies (Rowe, cartilaginous. Among extant bats, a calcar is 1988; Simmons, 1993a, 1993c). present in most taxa; it is absent only in some Completeness of a taxon may be de®ned Pteropodidae, Rhinopomatoidea, and some as the percentage of characters for which it phyllostomids. Both outgroups (and all other can be scored in a given analysis (Simmons, mammals) lack a calcar, indicating that ab- 1993a, 1993c). In practice, paleontologists sence of the calcar is primitive. Presence of have often sought to maximize completeness a calcar in Hassianycteris and Palaeochirop- by focusing largely on osteological charac- teryx represents a derived condition. ters, omitting from consideration many soft- Character 172: Digits II±V of foot with tissue or molecular characters that cannot be three phalanges (0); or two phalanges (1). scored in extinct organisms (e.g., Beard, Digits II±V of the foot each have three pha- 1993). However, inclusion of fossil OTUs in langes in Icaronycteris (frontispiece, ®g. 23), an analysis with many soft-tissue characters Archaeonycteris (®g. 3), Hassianycteris (®g. does not necessarily lead to decreased reso- 4), and Palaeochiropteryx (®g. 2). This con- lution (Gauthier et al., 1988; Novacek, dition is also seen in most extant bats. In 1992). In some cases, inclusion of fossils can contrast, only two phalanges are present in actually increase resolution by reducing the digits II±V in Hipposiderinae, Myzopodidae, number of equally parsimonious trees, as was and Thyropteridae. Both outgroups (and the case with Novacek's (1992) hyracoid ex- most other mammals) have three phalanges ample. When the opposite occurs and inclu- in digits II±V of the foot, suggesting that the sion of fossils increases the number of opti- condition seen in the Eocene bats is primi- mal trees and decreases resolution, tech- tive. niques such as Adams consensus may be used to identify relationships that remain sta- COMPLETENESS ble in all most-parsimonious trees (Simmons, 1993a). The relative stability of topological One problem that plagues phylogenetic analyses that include fossil taxa is that of placement of a fossil taxon in phylogenetic completeness (or lack of completeness) of trees appears to depend more on the partic- available data. Numerous studies have illus- ular combination of character states that it trated the importance of including fossils in exhibits than on its completeness. For ex- phylogenetic analyses, demonstrating that ample, in an analysis of relationships among fossils sometimes preserve information (in archontan mammals, Simmons (1993a) the form of unique combinations of primitive found that one fossil taxon that was only and derived character states) that may be cru- 19% complete could be placed unambigu- cial to resolving relationships among extant ously relative to extant lineages, while rela- lineages (e.g., Gauthier et al., 1988; Dono- tionships of another fossil taxon that was ghue et al., 1989; Novacek, 1992, 1994). 53% complete could not be determined un- Fossils are also important because they fre- ambiguously. quently affect character optimizations, which In our view, the bene®ts of including fos- in turn may affect conclusions concerning sils and soft tissue and molecular characters the degree of support for various monophy- in a single analysis far outweigh the possible letic groups, taxonomic diagnoses, character problems (see discussion below). Neverthe- independence, homoplasy, and the relative less, we calculated the percent completeness timing of various evolutionary events (Sim- for each OTU in our data set in order to pro- mons, 1993a). It is also obvious that fossils vide a basis for a posteriori considerations of must be included in any study that seeks to the effects of completeness (table 6). Extant determine the relationships of fossils to ex- lineages in our data set were 62.0±100% 86 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

TABLE 6 Statistics on Character Codings for Current Data Set

Characters Characters scored with scored with two or more Characters Characters Percent Terminal taxon single statea statesb scored ``Ϫ'' scored ``?'' completenessc Scandentia 161 (77.4%) 3 (1.4%) 19 (9.1%) 25 (12.0%) 88.0% Dermoptera 172 (82.6%) 0 18 (8.7%) 18 (8.7%) 91.3% Pteropodidae 176 (84.6%) 26 (12.5%) 6 (2.9%) 0 100% Icaronycteris 71 (34.1%) 0 3 (1.4%) 134 (64.4%) 35.6% Archaeonycteris 59 (28.4%) 0 1 (0.5%) 148 (71.2%) 28.9% Hassianycteris 61 (29.3%) 2 (1.0%) 1 (0.5%) 144 (69.2%) 30.8% Palaeochiropteryx 75 (36.1%) 1 (0.5%) 1 (0.5%) 131 (63.0%) 37.1% Emballonuridae 166 (79.8%) 26 (12.5%) 6 (2.9%) 10 (4.8%) 95.2% Rhinopomatidae 185 (89.0%) 1 (0.5%) 5 (2.4%) 17 (8.1%) 91.9% Craseonycteridae 125 (60.1%) 0 5 (2.4%) 78 (37.5%) 62.5% Nycteridae 178 (85.6%) 6 (2.9%) 7 (3.4%) 17 (8.1%) 91.9% Megadermatidae 183 (88.0%) 9 (4.3%) 2 (1.0%) 14 (6.7%) 93.3% Rhinolophinae 185 (89.0%) 5 (2.4%) 5 (2.4%) 13 (6.2%) 93.8% Hipposiderinae 170 (81.8%) 13 (6.2%) 6 (2.9%) 19 (9.1%) 90.9% Phyllostomidae 146 (70.0%) 57 (27.6%) 5 (2.4%) 0 100% Mormoopidae 172 (82.7%) 13 (6.2%) 5 (2.4%) 18 (8.7%) 91.3% Noctilionidae 192 (92.3%) 0 7 (3.4%) 9 (4.3%) 95.7% Mystacinidae 132 (63.5%) 0 6 (2.9%) 70 (33.6%) 66.4% Myzopodidae 129 (62.0%) 0 3 (1.4%) 76 (36.6%) 63.4% Thyropteridae 174 (83.7%) 1 (0.5%) 3 (1.4%) 30 (13.4%) 85.6% Furipteridae 147 (70.7%) 1 (0.5%) 3 (1.4%) 57 (27.4%) 72.6% Natalidae 181 (87.0%) 0 2 (1.0%) 25 (12.0%) 88.0% Antrozoidae 148 (71.2%) 1 (0.5%) 3 (1.4%) 56 (26.9%) 73.1% Tomopeatinae 125 (60.1%) 0 4 (1.9%) 79 (38.0%) 62.0% Molossinae 182 (87.5%) 21 (10.1%) 2 (1.0%) 3 (1.4%) 98.6% Vespertilioninae 167 (80.3%) 22 (10.6%) 3 (1.4%) 16 (7.7%) 92.3% Miniopterinae 145 (69.7%) 0 5 (2.4%) 58 (27.9%) 72.1% Myotinae 187 (89.9%) 9 (4.3%) 2 (1.0%) 10 (4.8%) 95.2% Murininae 132 (63.5%) 0 4 (1.9%) 72 (34.6%) 65.4% Kerivoulinae 145 (69.7%) 1 (0.5%) 3 (1.4%) 59 (28.4%) 71.6% Total 4471 (71.7%) 218 (3.5%) 145 (2.3%) 1406 (22.5%) 77.5% a Includes only characters states 0, 1, 2, etc.; does not include characters scored as inapplicable (``Ϫ'') or missing (``?''). Percentage re¯ects the percent of the total characters (208) used in the analysis. b Includes both cases of uncertainty and taxonomic polymorphism. c Completeness is de®ned as the percentage of characters for which a taxon can be scored based on available data; it was calculated by subtracting the percentage of characters scored ``?'' from 100%. complete; the fossils ranged from 28.9 to sensus methods for resolving systematic 37.1% complete. problems (Kluge, 1989; Barrett et al., 1991, 1993; Swofford, 1991; Bull et al., 1993; de METHODS OF PHYLOGENETIC Queiroz, 1993; Eernisse and Kluge, 1993; ANALYSIS Kluge and Wolf, 1993; Nelson, 1993; Chip- The present study follows Simmons pendale and Weins, 1994; Hulsenbeck et al., (1998) in adopting a character consensus 1994; de Queiroz et al., 1995; Farris et al., (``total evidence'') approach to phylogeny re- 1995; Miyamoto and Fitch, 1995). As dis- construction. In recent years there has been cussed by Simmons (1993a, 1998), character much discussion of the relative merits of congruence is the most sensible approach to character consensus versus taxonomic con- the current data set because the data consist 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 87 principally of discrete-state morphological ANALYSIS OF CHARACTER characters that cannot be reasonably parti- TRANSFORMATIONS tioned. Available restriction-site data are not capable of resolving interfamilial relation- Character transformations were analyzed ships in the absence of other data (Baker et by mapping character-state distributions onto al., 1991a), hence taxonomic congruence is the shortest trees derived from the second set not a useful option. of phylogenetic analyses described above The 208 discrete characters used in the (i.e., those obtained using all characters and present study (see appendix 2) were scored all taxa). Optimizations were calculated us- for phylogenetic analysis, and the resulting ing both the ACTRAN (accelerated transfor- data matrix (appendix 3) was analyzed using mation optimization) and DELTRAN (de- PAUP version 3.1.1 (Swofford, 1993). All layed transformation optimization) options of transformations were unordered. A heuristic PAUP version 3.1.1 (Swofford, 1993). search with a random-addition sequence and MacClade version 3.0 (Maddison and Mad- 1000 repetitions was used to ®nd most-par- dison, 1992) was used to visualize the results simonious trees. Near-most-parsimonious of character mapping. trees (one to six steps longer) were identi®ed As discussed by Simmons (1993a), two in subsequent heuristic searches using the kinds of character transformations may be same parameters, and a decay analysis was recognized during the optimization process: performed following the methods of Bremer unequivocal transformations, which have (1988). Decay values for strongly supported only one parsimonious placement on the op- clades were obtained by using constrained timal tree(s), and equivocal transformations, heuristic analyses to identify the shortest which can be parsimoniously arranged in two trees that did not include a particular clade. or more ways. Both ACTRAN and DEL- A bootstrap analysis using heuristic methods TRAN place unequivocal transformations on (random-addition sequence, 10 repetitions a given tree in the same manner, but they for each of 1000 bootstrap replicates) was treat equivocal transformations differently. also used to evaluate the relative support for ACTRAN forces transformations to the low- various groupings. MacClade version 3.0 est possible points on the tree, and thus fa- (Maddison and Maddison, 1992) was used vors hypotheses of reversal over hypotheses for data entry and examination of character- of convergence. Conversely, DELTRAN state distributions. forces transformations to the highest possible Three complete sets of phylogenetic anal- points on a tree, thus favoring hypotheses of yses were conducted: (1) an analysis includ- convergence over reversal. DELTRAN is of- ing all characters but excluding the fossil ten favored in studies involving fossils be- taxa; (2) an analysis including all characters and all taxa; and (3) an analysis including all cause it places transformations at the mini- taxa but excluding those characters that mal level at which they can be observed (i.e., could not be scored in any of the fossil in the face of missing data it does not con- forms. The ®rst analysis was designed to pro- clude that derived states exist below the level vide a starting point by evaluating relation- at which they can be demonstrated). ships of extant lineages in the context of the In this study, we focused only on trans- revised data matrix; this effectively repre- formations that apply at and below the nodes sents an updated version of Simmons' (1998) where fossil taxa join the tree. We compared analysis. Our second analysis represents the the results of ACTRAN and DELTRAN op- principal goal of the project, a character con- timizations in order to identify all equally gruence study including both Eocene fossil parsimonious arrangements of equivocal genera and extant lineages. The third and ®- transformations, and interpreted our obser- nal analysis was designed to evaluate the ef- vations in the context of what is known about fects of soft-tissue and molecular characters the function of various structures and struc- on the outcome of a phylogenetic analysis tural complexes. Results of these analyses including fossil forms that cannot be scored are presented below under ``Character Evo- for these features. lution in Early Chiropterans.'' 88 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

PHYLOGENETIC RESULTS RESULTS OF ANALYSES Comparisons of the results of our Analysis 1 (®g. 35) with Simmons (1998) tree (®g. 20) Analysis 1: All characters, fossil taxa ex- cluded (®g. 35). As noted above, this anal- reveals several topological differences, all ysis was designed to provide a starting point within the yangochiropteran part of the tree. by evaluating relationships of extant lineages First, relationships within Nataloidea are ful- in the context of the revised data matrix. This ly resolved in our Analysis 1 tree, with Thy- effectively represents an updated version of ropteridae unambiguously placed as the sis- Simmons' (1998) analysis, and illustrates the ter-group of the Furipteridae ϩ Natalidae effects of changes that we made in the data clade. This represents a trivial change, since matrix (e.g., corrections, new characters, in- this topology occurred in one of the two most clusion of hyoid data for Antrozoidae; see parsimonious trees found by Simmons above). Analysis of the revised data set re- (1998) and received higher bootstrap support sulted in a single most parsimonious tree than did any of the alternatives in that study. (661 steps; CI ϭ 0.404, RI ϭ 0.579) shown A potentially more signi®cant change in our in ®gure 35. Analysis 1 tree is placement of Antrozoidae

Fig. 35. Results of Analysis 1, which included all characters but omitted the fossil taxa (see text for discussion). Parsimony analysis resulted in a single most-parsimonious tree (662 steps; CI ϭ 0.403 and RI ϭ 0.578), which is shown here. The numbers below internal branches indicate the percentage of bootstrap replicates in which each clade appeared; numbers above the branches are decay values (the minimum number of additional steps required to collapse each clade). The bootstrap analysis was con- strained to consider only trees in which Chiroptera was monophyletic; this was done because most of the characters that support bat monophyly (table 5) were omitted from this study. 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 89

as the sister-taxon of Vespertilionidae (i.e., lineages are represented by two to three Vespertilioninae ϩ Miniopterinae ϩ Myoti- times more data (table 6). The absence of nae ϩ Murininae ϩ Kerivoulinae) rather than soft-tissue and molecular character data as the sister-taxon of Molossidae. This sug- clearly does not preclude relatively secure gests that Molossoidea as de®ned by Sim- placement of the fossil forms in this study. mons (1998) may not be monophyletic, and Results of our analysis indicate that Ica- that Antrozoidae may form a clade with ves- ronycteris, Archaeonycteris, Hassianycteris, pertilionids as traditionally thought (e.g., by and Palaeochiropteryx represent a series of Koopman, 1993, 1994). However, support consecutive sister-taxa to the microchiropter- for the Antrozoidae ϩ Vespertilionidae clade an crown group (i.e., the group comprised of was very low in Analysis 1 (bootstrap value all extant lineages). They do not form a para- ϭ 12%; minimum of one additional step to phyletic group ancestral to both Megachirop- collapse clade), indicating considerable un- tera and Microchiroptera (e.g., Eochiroptera certainty regarding this grouping. The same sensu Van Valen, 1979), a monophyletic is also true of other clades that represent group within Microchiroptera (e.g., Palaeo- changes from Simmons (1998) topology: (1) chiropterygoidea sensu Smith, 1977), or have placement of Molossidae (Tomopeatinae ϩ special af®nities with various extant micro- Molossinae) as the sister-group of the Antro- chiropteran superfamilies (as suggested by zoidae ϩ Vespertilionidae clade (bootstrap Smith and Storch, 1981). Of the four Eocene value ϭ 33%; minimum of one additional genera, Palaeochiropteryx shares the most step to collapse clade); and (2) placement of derived traits with extant microchiropterans. Mystacinidae as the sister-taxon of Molossi- Hassianycteris and Archaeonycteris are con- dae ϩ Antrozoidae ϩ Vespertilionidae (boot- secutive sister-taxa to the clade including Pa- strap value ϭ 20%; minimum of one addi- laeochiropteryx and the microchiropteran tional step to collapse clade). In essence, all crown group, and Icaronycteris occupies the of these changes represent rearrangements of basalmost branch in the microchiropteran relationships at nodes that were poorly sup- part of the tree. ported in Simmons (1998) analysis and re- Comparisons of the results of Analyses 1 main poorly supported in our analyses of the and 2 (®gs. 35, 36) demonstrate that inclu- updated data set. sion of Icaronycteris, Archaeonycteris, Has- Analysis 2: All characters, all taxa (®g. sianycteris, and Palaeochiropteryx in the 36). This analysis represents the principal analysis produced minor changes in topology goal of our project, a character-congruence of the tree, with the changes again centered study including both Eocene fossil genera on those parts of the tree that were weakly and extant lineages. Encouragingly, parsi- supported in previous analyses. Speci®cally, mony analyses of the complete data set in inclusion of the fossils changed the position Analysis 2 produced a single most-parsimo- of Antrozoidae (which in Analysis 2 now nious tree (®g. 36; 681 steps; CI ϭ 0.392, RI forms a clade with Molossidae, supporting ϭ 0.587). Relationships of Icaronycteris, Ar- monophyly of Molossoidea) and Mystacini- chaeonycteris, Hassianycteris, and Palaeo- dae (which now appears as the sister-taxon chiropteryx were fully resolved, and place- of Nataloidea ϩ Molossoidea ϩ Vespertilion- ment of each was moderately to strongly sup- oidea). Interestingly, all of these relationships ported in the bootstrap and decay analyses appeared in Simmons' (1998) tree (®g. 20), (bootstrap values ϭ 54±93%; minimum of although not in the results of our Analysis 1 one to six additional steps to collapse clade). (®g. 35). These levels of support are comparable to Inclusion of the fossils affected the per- those found for many extant clades in Anal- ceived support for many clades even when ysis 2. Indeed, bootstrap values for branches there was no effect on tree topology. This associated with many extant taxa (e.g., Mys- effect was most noticeable near the base of tacinidae, Vespertilioninae, Miniopterinae, tree. Monophyly of the microchiropteran Myotinae, Murininae, Kerivoulinae) are crown group (Emballonuridae ϩ Yinochirop- much lower than those associated with the tera ϩ Yangochiroptera) received only mod- fossil branches despite the fact that the extant erate support in Analysis 2 (bootstrap value 90 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Fig. 36. Results of Analysis 2, which included all characters and all taxa (see text for discussion). Parsimony analysis resulted in a single most-parsimonious tree (680 steps; CIϭ 0.393; RI ϭ 0.587), which is shown here. The numbers below internal branches are bootstrap values; numbers above the branches are decay values. The bootstrap analysis was constrained to consider only trees in which Chiroptera was monophyletic.

ϭ 79%; minimum of two additional step to ing those suggested by Smith and Storch collapse clade), whereas it received extreme- (1981), appear unlikely given results of the ly high support in Analysis 1 (bootstrap val- bootstrap analysis. For example, Hassianyc- ue ϭ 100%; minimum of 16 additional steps teris formed a clade with Yangochiroptera in to collapse clade). Clearly, many of the de- only 6% of the bootstrap replicates. Palaeo- rived traits that diagnose extant Microchirop- chiropteryx grouped with Yangochiroptera in tera evolved in a sequential pattern over only 5% of the bootstrap replicates, and time; inclusion of the fossil taxa ``spreads formed a clade with some or all Vesperti- out'' these synapomorphies over a larger part lionoidea in fewer than 1% of the bootstrap of the tree, thus reducing perceived support replicates. for any single branch. Nevertheless, mono- As in Simmons (1998) and Analysis 1 (®g. phyly of the microchiroperan crown group 20, 35), results of Analysis 2 placed Embal- remains strongly supported in contrast to oth- lonuridae as the sister-group of the clade con- er hypotheses. Alternative topologies, includ- taining all other extant microchiropteran lin- 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 91

eages (®g. 36). Monophyly of the latter clade grouping Mormoopidae ϩ Phyllostomidae (Yinochiroptera ϩ Yangochiroptera) was (bootstrap value ϭ 19%), and the third alter- only weakly supported (bootstrap value ϭ native (Phyllostomidae ϩ Noctilionidae) re- 42%; minimum of two additional steps to ceived no support (bootstrap value Ͻ1%). collapse clade). However, alternative topol- We suspect that much of the ambiguity con- ogies received considerably less support. cerning relationships among noctilionoids is Placement of Emballonuridae within Yino- a result of the high level of taxonomic poly- chiroptera (as suggested by Koopman, 1985, morphism seen in Phyllostomidae. Almost 1994) was supported in only 28% of the 28% of the characters in this study were bootstrap replicates. Emballonuridae grouped scored as polymorphic in Phyllostomidae (ta- with Rhinopomatoidea in fewer than 4% of ble 6), introducing considerable uncertainty the replicates, again indicating that Embal- with respect to polarities in this part of the lonuroidea as traditionally recognized (i.e., tree. Better resolution of noctilionoid rela- including Emballonuridae ϩ Rhinopomati- tionships in future studies will probably re- dae ϩ Craseonycteridae) is not monophylet- quire splitting of Phyllostomidae into multi- ic. ple OTUs. Monophyly of Yinochiroptera sensu Sim- As in Simmons (1998), Mystacinidae was mons (Rhinopomatoidea ϩ Rhinolophoidea) placed as the sister-group to a clade com- was moderately supported in the current prising Molossoidea ϩ Nataloidea ϩ Vesper- analysis (bootstrap value ϭ 68%; minimum tilionoidea (the latter sensu Simmons, 1998, of two additional steps to collapse clade). not sensu Koopman, 1994) in Analysis 2. Within Yinochiroptera, strong support was However, this grouping received only weak found for monophyly of Rhinopomatoidea support (bootstrap value ϭ 50%; minimum (bootstrap value ϭ 83%; minimum of four of one additional step to collapse clade). Al- additional steps to collapse clade) and Rhin- ternative hypotheses that received limited olophoidea (bootstrap value ϭ 93%; mini- support in the bootstrap analysis included mum of seven additional steps to collapse placement of Mystacinidae either (1) in a clade). Relationships within Rhinolophoidea clade with Noctilionoidea (bootstrap value ϭ were well resolved with high bootstrap and 38%); or (2) in a clade with Molossoidea decay values. Monophyly of Rhinolophidae (bootstrap value ϭ 19%); or (3) in a clade (including Hipposiderinae) was very strongly with Molossoidea and Vespertilionoidea supported (bootstrap value ϭ 100%; mini- (bootstrap value ϭ 19%); or (4) in a clade mum of 16 additional steps to collapse with Molossidae (bootstrap value ϭ 15%); or clade), as was a sister-group relationship be- (5) in a clade with Noctilionoidea and Na- tween Megadermatidae and Rhinolophidae taloidea (bootstrap value ϭ 10%). Clearly, (bootstrap value ϭ 95%; minimum of eight placement of Mystacinidae remains problem- additional steps to collapse clade). Very little atic. support was found for a Nycteridae ϩ Rhin- Weak support was found for the clade olophidae clade (bootstrap value ϭ 3%) or a comprising Nataloidea ϩ Molossoidea ϩ Nycteridae ϩ Megadermatidae clade (boot- Vespertilionoidea (bootstrap value ϭ 47%; strap value ϭ 2%). minimum of one additional step to collapse Monophyly of Yangochiroptera (bootstrap clade). Within this group, similarly weak value ϭ 85%; minimum of seven additional support was found for a Molossoidea ϩ Ves- steps to collapse clade) and Noctilionoidea pertilionoidea clade (bootstrap value ϭ 37%; (bootstrap value ϭ 94%; minimum of seven minimum of one additional step to collapse additional steps to collapse clade) was clade). An alternative hypothesis that re- strongly supported. Within Noctilionoidea, a ceived some support in the bootstrap analysis sister-group relationship between Noctilioni- included a sister-group relationship between dae and Mormoopidae received weak sup- Nataloidea and Vespertilionoidea (bootstrap port (bootstrap value ϭ 58%; minimum of value ϭ 31%) and a clade including Nata- two additional steps to collapse clade). Sub- loidea ϩ Vespertilionoidea ϩ Antrozoidae tantially less support was found in the boot- (bootstrap value ϭ 12%). strap analysis for an alternative hypothesis Strong support was found for monophyly 92 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 of Nataloidea (bootstrap value ϭ 87%; min- a Miniopterinae ϩ Kerivoulinae clade (boot- imum of ®ve additional steps to collapse strap value ϭ 13%). Essentially no support clade), and weak support was found for was found for an Antrozoidae ϩ Vesperti- monophyly of Molossoidea (bootstrap value lioninae clade (bootstrap value ϭ 4%), an ϭ 56%; minimum of one additional step to Antrozoidae ϩ Vespertilioninae ϩ Myotinae collapse clade). Within Nataloidea, the Neo- clade (bootstrap value Ͻ1%), or an Antro- tropical taxa (Thyropteridae ϩ Furipteridae zoidae ϩ Myotinae clade (bootstrap value ϩ Natalidae) formed a weakly-supported Ͻ1%). In short, these results indicate that clade (bootstrap value ϭ 51%; minimum of ``Vespertilioninae'' sensu Koopman (i.e., in- one additional step to collapse clade). Within cluding myotines and antrozoines) is not a this group, Furipteridae ϩ Natalidae formed monophyletic group. Similarly, there was lit- a well-supported clade (bootstrap value ϭ tle support for an Antrozoidae ϩ Vesperti- 82%; minimum of ®ve additional steps to lionidae grouping excluding Molossidae collapse clade). Within Molossoidea, very (bootstrap value ϭ 8%). strong support was found for Molossidae (in- Even though bootstrap and decay values cluding Tomopeatinae; bootstrap value ϭ associated with the molossoid and vesperti- 100%; minimum of nine additional steps to lionoid part of the tree are low, these results collapse clade). indicate that support for most competing hy- No support was found for monophyly of potheses is considerably less. The range of Vespertilionidae as traditionally de®ned (i.e., phylogenetic hypotheses that still appear to including Tomopeatinae and Antrozoidae; represent viable alternatives is thus fairly bootstrap value Ͻ1%). Monophyly of Ves- small. As with noctilionoids, it seems likely pertilionidae (ϭ Vespertilionoidea) sensu that taxonomic polymorphism in Vesperti- Simmons (1998), a clade including Vesper- lioninae (10.6% of all characters scored as tilioninae ϩ Miniopterinae ϩ Myotinae ϩ polymorphic; table 6) and Molossinae Murininae ϩ Kerivoulinae, was weakly sup- (10.1% of characters polymorphic) may be ported (bootstrap value ϭ 51%; minimum of reducing tree stability. Low levels of com- two additional steps to collapse clade). With- pleteness of several taxa (Ͻ75% in Antro- in this group, weak support was found for a zoidae, Tomopeatinae, Miniopterinae, Muri- Murininae ϩ Kerivoulinae clade (bootstrap ninae, and Kerivoulinae) probably contribute value ϭ 41%; minimum of two additional to this effect. steps to collapse clade). Myotinae appeared Analysis 3: All taxa, characters limited to as the sister-group of the latter clade (boot- those that could be scored in fossil genera strap value ϭ 31%; minimum of two addi- (®g. 37). This analysis was designed to eval- tional steps to collapse clade). Miniopterinae uate the effects of soft-tissue and molecular was placed as the sister-group of the Myoti- characters on the outcome of a phylogenetic nae ϩ Murininae ϩ Kerivoulinae clade analysis that includes fossil forms that cannot (bootstrap value ϭ 35%; minimum of two be scored for these features. As previously additional steps to collapse clade). noted, paleontologists interested in relation- Several other ``vespertilionid'' groupings ships of fossil forms to extant taxa frequently received limited support in the bootstrap omit from their analyses any characters that analyses, even though they did not appear in cannot be scored in the fossils, thus ignoring the most parsimonious tree. Most notably, potentially informative soft-tissue and mo- Vespertilioninae ϩ Myotinae formed a clade lecular characters. What effect does this have in 32% of the bootstrap replicates, indicating on the outcome, both on perceived relation- that monophyly of this traditional grouping ships of the fossils and perceived relation- cannot be ruled out. Other groupings that re- ships among extant forms? Although ours is ceived weak support in the bootstrap analysis only a single case study, the results were include a clade comprising Vespertilioninae striking. ϩ Miniopterinae ϩ Myotinae (bootstrap val- Parsimony analysis of our restricted data ue ϭ 17%), a larger clade including Vesper- set of 82 characters (see list under ``Char- tilioninae ϩ Miniopterinae ϩ Myotinae ϩ acters Examined in Fossil Bats'' above) pro- Kerivoulinae (bootstrap value ϭ 16%), and duced two equally parsimonious trees (306 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 93

Fig. 37. Results of Analysis 3, which included all taxa but only those characters that could be scored in the fossils (see text for discussion). The tree shown is a strict consensus of eight equally most- parsimonious trees (305 steps; CI ϭ 0.361; RI ϭ 0.601) that resulted from parsimony analysis. The numbers below internal branches are bootstrap values; numbers above the branches are decay values. The bootstrap analysis was constrained to consider only trees in which Chiroptera was monophyletic.

steps; CI ϭ 0.359, RI ϭ 0.603); a strict con- extensively rearranged. Emballonuridae, sensus of these trees is shown in ®gure 37. placed as the sister-group to all other micro- Comparisons of this consensus tree with that chiropterans in Analysis 2, nested within a derived from Analysis 2 (®g. 37) illustrate monophyletic Yinochiroptera sensu Koop- that both analyses recovered the same rela- man (1994) in Analysis 3, appearing as the tionships among the fossil forms and be- sister-group of Rhinolophoidea. Noctiliono- tween the fossils and the microchiropteran idea, a clade whose monophyly was very crown group. However, comparisons of strongly supported in Analysis 2, appeared to crown-group topology in ®gures 36 and 37 be polyphyletic in Analysis 3. Instead of indicate that excluding the soft-tissue and grouping with Noctilionidae, Phyllostomidae molecular characters signi®cantly affected and Mormoopidae formed a clade that was the perceived relationships among the extant placed as the sister-group of Yinochiroptera. lineages. Not only was resolution consider- Resolution was considerably reduced in the ably reduced, but the branching pattern was yangochiropteran part of the tree, and the po- 94 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 sition of Mystacinidae was shifted to within microchiropteran crown group as hypothe- Molossoidea. sized in our Analyses 1 and 2 is little differ- At least one of the groupings obtained in ent than that reported by Simmons (1998). Analysis 3 (the Phyllostomidae ϩ Mormoop- The only exceptions are the suggestions that idae ϩ Emballonuridae ϩ Yinochiroptera Molossoidea and Vespertilionoidea are sister- clade) appears spurious based on results of taxa (Nataloidea appeared more closely re- Analysis 2, in which it was recovered in few- lated to Vespertilionoidea in the earlier anal- er than 1% of the bootstrap replicates. Con- ysis), and that the Neotropical nataloids versely, Analysis 3 did not recover at least (Thyropteridae, Furipteridae, and Natalidae) two important clades that were strongly sup- form a clade. ported in the analysis of the complete data Results of our experiment in using only setÐYangochiroptera and Nataloidea. characters that could be scored in the fossils Topology of the hard-tissue tree is clearly (Analysis 3) argue strongly against the com- not congruent with any previously proposed mon practice of reducing missing data by phylogeny for Chiroptera (e.g., Smith, 1976; eliminating soft-tissue and molecular char- Van Valen, 1979; Pierson, 1986; Novacek, acters from consideration in phylogenetic 1991; Simmons, 1998), nor is it congruent analyses including fossil organisms. In our with results of our analysis of the entire data study, which admittedly included a high ratio set (Analysis 2). We conclude from this ex- of extant to extinct OTUs, limiting the anal- periment that exclusion of soft-tissue and ysis to hard-tissue characters produced phy- molecular characters resulted, at least in this logenetic results that we consider fallacious. case, in a biased topology that does not pro- These results suggest that soft-tissue and mo- vide an adequate basis for inferring relation- lecular characters can be highly informative ships. even when inclusion of such characters in- troduces considerable amounts of missing SUMMARY data (i.e., empty matrix cells) into the data set. The point at which missing data becomes Results of the phylogenetic analyses de- a serious problem and signi®cantly reduces scribed above support the hypothesis that Pa- resolution probably depends on the relative laeochiropteryx, Hassianycteris, Archaeo- number of fossil taxa, the relative proportion nycteris, and Icaronycteris represent a series of hard-tissue to soft-tissue and molecular of successively more distant sister-taxa to the characters, the degree of structure (i.e., level clade comprised of the extant lineages of mi- of congruence among characters) present crochiropteran bats. Inclusion of these taxa within and among various data subsets, and signi®cantly affected perceived relationships the exact mosaic of primitive and derived among extant forms, although only at nodes character states seen in the fossils included that were poorly supported in the analysis in the analysis. In the present study, we were that included only extant lineages. Inclusion fortunate in that these variables combined fa- of the fossils also affected bootstrap and de- vorably and facilitated recovery of a well- cay values associated with the more basal resolved, relatively well-supported phyloge- nodes of the crown group. Topology of the ny of bats.

CHARACTER EVOLUTION IN EARLY CHIROPTERANS CHARACTER TRANSFORMATIONS ropteryx, and the microchiropteran crown ASSOCIATED group share a variety of derived features that WITH BASAL NODES have previously been interpreted as synapo- The phylogeny shown in ®gure 36 pro- morphies of Chiroptera (Simmons, 1994, vides a framework for evaluating patterns of 1995; see table 5). Features that can be ob- character transformation at the base of the served in the fossil forms include only about chiropteran tree. Pteropodidae, Icaronycteris, one-third of the characters listed in table 5, Archaeonycteris, Hassianycteris, Palaeochi- principally those related to the skull and 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 95

Fig. 38. Presence of the calcar (character 171) mapped on the phylogeny from ®gure 36. Note that a calcar apparently evolved independently in Pteropodidae and the lineage leading to extant microchi- ropterans. Accordingly, presence of a calcar does not appear to be a synapomorphy of Chiroptera. Loss of the calcar is presumed to have occurred independently in Rhinopomatoidea and within two terminal taxa, Pteropodidae and Phyllostomidae (loss in terminal taxa not depicted on this tree). See text for discussion. postcranial skeleton: reduction of the jugal, cussion below), so this feature should be re- modi®cation of the elbow, absence of a su- moved from the list of bat synapomorphies. pinator ridge on and entepicondylar foramen The principal effect of our analysis, however, in the humerus, elongation of digits II±V of has been to identify additional character the forelimb and presence of complex car- states that appear to be synapomorphies of pometacarpal and intermetacarpal joints, ab- bats based on their optimizations on our op- sence of claws on digits III±V, modi®cation timal tree (®g. 36). These newly identi®ed of the hip joint, modi®cation of the ankle chiropteran synapomorphies (and we list joint, presence of a calcar, and elongation of only those that are unequivocal given our the proximal phalanx of digit I of the foot. tree topology) include the following: (1) Virtually all of these features appear related vomeronasal epithelial tube absent (character to the evolution of powered ¯ight and un- 7; well developed in both outgroups); (2) ac- derbranch hanging behavior, both of which cessory olfactory bulb absent (character 8; are presumed to have been present in the present in both outgroups); (3) enlarged fe- most recent common ancestor of Megachi- nestra cochleae (character 32; small or of roptera and Microchiroptera (see discussion moderate size in both outgroups); (4) poste- in Simmons, 1995). rior laminae present on ribs (character 88; Our data do not contradict these functional laminae absent in both outgroups); (5) two hypotheses; however, our observations indi- facets present in infraspinous fossa of scap- cate that some modi®cations need to be made ula (character 103; only one facet in both to Simmons' (1994, 1995) list of synapo- outgroups); (6) sacrum terminates posterior morphies of bats. First, examination of the to midpoint of acetabulum (character 159; fossil forms and their relationships suggests terminates anterior to acetabulum in both that presence of a calcar (character 171) is outgroups); (7) baculum present (character not primitive for Chiroptera (®g. 38; see dis- 174; absent in both outgroups); (8) left cen- 96 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 tral lobe of liver separate from other lobes or not extend to level of head in both outgroups partially fused with right central lobe (char- and Pteropodidae). acter 193; left central lobe fused with left Of these characters, enlargement of the or- lateral lobe in both outgroups); and (9) cae- bicular apophysis, expansion of the tip of the cum absent (character 195; present in both stylohyal, and extension of the trochiter to outgroups). Obviously, only half of these the level of the head of the humerus are syn- features (characters 32, 88, 103, 159) can be apomorphies that unequivocally diagnose the scored in fossil taxa; in each case, however, clade comprising Icaronycteris, Archaeonyc- available data from Icaronycteris, Archaeo- teris, Hassianycteris, Palaeochiropteryx, and nycteris, Hassianycteris, and Palaeochirop- the microchiropteran crown group. The teryx con®rm that these derived traits were change in morphology of the angular process present in the earliest members of the micro- is an equivocal transformation because an chiropteran lineage. A revised summary of elongate angular process occurs in Scanden- the morphological synapomorphies of Chi- tia. Alternative explanations for the pattern roptera is given in table 7. observed include (1) evolution of a short an- Icaronycteris, Archaeonycteris, Palaeo- gular process in the common ancestor of Vol- chiropteryx, and Hassianycteris were clearly itantia, and secondary evolution of an elon- well developed bats capable of powered gate angular process in the clade containing ¯ight (Jepsen, 1966, 1970; Habersetzer and Icaronycteris and extant microchiropterans; Storch, 1987, 1989; Habersetzer et al., 1989, or (2) independent evolution of a short an- 1992, 1994; Norberg, 1989, 1994). By trac- gular process in Dermoptera and Pteropodi- ing character changes associated with the dae, implying that the elongate angular pro- nodes at which these taxa branch from the cess seen in Icaronycteris and other members tree, insight can be gained into patterns of of the microchiropteran lineage is plesiom- character evolution in the lineage leading orphic. from the volant common ancestor of bats to Other equivocal transformations that may extant microchiropterans. These taxa do not diagnose the clade comprising Icaronycteris, provide much information about steps in the Archaeonycteris, Hassianycteris, Palaeochi- evolution of powered ¯ight since those ropteryx, and the microchiropteran crown changes presumably took place long before group include: (1) at least moderate enlarge- Icaronycteris diverged from the lineage lead- ment of the cochlea (character 26; absent in ing to extant microchiropterans. both outgroups and most Pteropodidae); (2) Icaronycteris shares a number of condi- presence of ventral accessory processes on tions with Archaeonycteris, Hassianycteris, cervical vertebrae 2 and 3 (character 75; ab- Palaeochiropteryx, and the microchiropteran sent in both outgroups and some but not all crown group that appear to be derived rela- pteropodids); and (3) absence of a supra- tive to those seen in pteropodids and the out- scapular process on the scapula (character groups. These features include (1) presence 112; present in both outgroups, absent in of an elongate angular process on the lower some pteropodids). In each of these cases, jaw (character 22; not elongate in Pteropod- the derived condition could have either idae and Dermoptera); (2) moderate enlarge- evolved independently within Pteropodidae ment of the cochlea (character 26; not en- and in the lineage leading to extant micro- larged in the outgroups and most Pteropodi- chiropterans, or it may be a synapomorphy dae); (3) presence of a large orbicular apoph- of bats that was reversed within some Pter- ysis on the malleus (character 35; small or opodidae. absent in the outgroups and Pteropodidae); Another character transformation that oc- (4) a stylohyal with an expanded cranial tip curred somewhere in the basal part of the (character 74; no enlargement or other mod- microchiropteran tree involves a change in i®cations in Pteropodidae and Scandentia; the type of attachment of the periotic to the lateral stylohyal swollen along entire length basisphenoid (character 25). In the outgroups in Dermoptera); and (5) the trochiter of the and Pteropodidae, the periotic is ®rmly su- humerus extends proximally to the same lev- tured to the basisphenoid; however, the per- el as the humeral head (character 139; does iotic is only loosely attached in Archaeonyc- 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 97

TABLE 7 A Revised Summary of Morphological Synapomorphies of Chiroptera

1) Deciduous dentition does not resemble adult dentition; deciduous teeth with long, sharp, recurved cusps 2) Palatal process of premaxilla reduced; left and right incisive foramina fused in midsaggital plane 3) Postpalatine torus absent 4) Jugal reduced and jugolacrimal contact lost 5) Two entotympanic elements in the ¯oor of the middle-ear cavity: a large caudal element and a small rostral element associated with the internal carotid artery 6) Tegmen tympani tapers to an elongate process that projects into the middle-ear cavity medial to the epitympanic recess 7) Proximal stapedial artery enters cranial cavity medial to the tegmen tympani; ramus inferior passes anteriorly dorsal to the tegmen tympani 8) Enlarged fenestra rotundum 9) Vomeronasal epithelial tube absent 10) Accessory olfactory bulb absent 11) Posterior laminae present on ribs 12) Modi®cation of scapula: reorientation of scapular spine and modi®cation of shape of scapular fossae; reduction in of height of spine; presence of a well-developed transverse scapular ligament; presence of at least two facets in infraspinous fossa 13) Modi®cation of elbow: reduction of olecranon process and humeral articular surface on ulna; presence of ulnar patella; absence of olecranon fossa on humerus 14) Absence of supinator ridge on humerus 15) Absence of entepicondylar foramen in humerus 16) Occipitopollicalis muscle and cephalic vein present in leading edge of propatagium 17) Digits II±V of forelimb elongated with complex carpometacarpal and intermetacarpal joints, support enlarged interdigital ¯ight membranes (patagia); digits III±V lack claws 18) Modi®cation of hip joint: 90Њ rotation of hindlimbs effected by reorientation of acetabulum and shaft of femur; neck of femur reduced; ischium tilted dorsolaterally; anterior pubes widely ¯ared and pubic spine present; absence of m. obturator internus 19) Sacrum terminates posterior to midpoint of acetabulum 20) Absence of m. gluteus minimus 21) Absence of m. sartorius 22) Vastus muscle complex not differentiated 23) Modi®cation of ankle joint: reorientation of upper ankle joint facets on calcaneum and astragalus; trochlea of astragalus convex, lacks medial and lateral guiding ridges; tuber of calcaneum projects in plantolateral direction away from ankle and foot; peroneal process absent; sustentacular process of calcaneum reduced, calcaneoastra- galar and sustentacular facets on calcaneum and astragalus coalesced; absence of groove on astragalus for tendon of m. ¯exor digitorum ®bularis 24) Entocuneiform proximodistally shortened, with ¯at, triangular distal facet 25) Elongation of proximal phalanx of digit I of foot 26) Embryonic disc oriented toward tubo±uterine junction at time of implantation 27) Differentiation of a free, glandlike yolk sac 28) Preplacenta and early chorioallantoic placenta diffuse or horseshoe-shaped, with de®nitive placenta reduced to a more localized discoidal structure 29) De®nitive chorioallantoic placenta endotheliochorial 30) Baculum present 31) Left central lobe of liver separate from other lobes or partially fused with right central lobe 32) Caecum absent 33) Cortical somatosensory representation of forelimb reverse of that in other mammals teris, Hassianycteris, Palaeochiropteryx, and sure if this character transformation occurred most members of the microchiropteran before or after differentiation of Icaronycter- crown group. Because the type of periotic is from the stock leading to extant microchi- attachment could not be scored in available ropterans. specimens of Icaronycteris, we cannot be The next highest node in the microchirop- 98 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 teran tree unites Archaeonycteris with Has- scapula (character 100). The cryptocochlear sianycteris, Palaeochiropteryx, and the mi- condition is seen in both outgroups and Pter- crochiropteran crown group. As noted above, opodidae; the phanerocochlear condition oc- loosening of the connection between the per- curs in Hassianycteris, Palaeochiropteryx, iotic and basisphenoid (character 25) mini- and primitively in the microchiropteran mally applies at this level in the tree. Other crown group. The stapedial fossa in the out- derived characters that diagnose this clade groups and Pteropodidae is indistinct or shal- include (1) reduction of the number of roots low and broad; this contrasts sharply with the on P3 to two or fewer (character 19; P3 with deep, constricted stapedial fossae seen in Pa- three roots in both outgroups, Pteropodidae, laeochiropteryx and primitively in the micro- and Icaronycteris); (2) presence of a ventral chiropteran crown group. Because external accessory process on cervical vertebra 4 cochlear morphology and form of the stape- (character 76; accessory process absent from dial fossa could not be scored in available C4 in both outgroups, Pteropodidae, and Ica- specimens of either Icaronycteris or Ar- ronycteris); and (3) absence of ossi®ed third chaeonycteris, we cannot be sure where on phalanx in wing digit III (character 152; fully the tree between Pteropodidae and Palaeo- ossi®ed third phalanx present in both out- chiropteryx the transformations to a phaner- groups and Icaronycteris). ocochlear condition and a deep, constricted Of these characters, reduction of the num- stapedial fossa occurred. Nor can we tell if ber of roots on P3 and presence of a ventral these changes were coincident or occurred at accessory process on C4 are synapomorphies different points in the tree. Interpretive dif- that unequivocally diagnose the clade com- ®culties similarly occur with other character prising Archaeonycteris and the lineage lead- transformations as a result of missing data. ing to the microchiropteran crown group. Anterior laminae are absent from the ribs of The absence of an ossi®ed third phalanx on both outgroups, some Pteropodidae, and Ica- wing digit III represents an equivocal trans- ronycteris, while anterior laminae are present formation because this condition also char- in Palaeochiropteryx and primitively within acterizes Pteropodidae. Alternative explana- the microchiropteran crown group. Posterior tions for the observed pattern include (1) in- laminae are narrow in Pteropodidae and Ica- dependent loss of ossi®cation of this phalanx ronycteris, but are broad in Hassianycteris, in pteropodids and the lineage leading to ex- Palaeochiropteryx, and primitively within tant microchiropterans; or (2) loss of ossi®- the microchiropteran crown group. A dorsal cation of the third phalanx in the common articular facet is absent from the scapula in ancestor of bats, and secondary acquisition the outgroups, Pteropodidae, and Icaronyc- of ossi®cation of this element in Icaronyc- teris; however, it is present in Hassianycteris, teris. Both hypotheses are equally parsimo- Palaeochiropteryx, and primitively in the mi- nious, and they also seem equally likely giv- crochiropteran crown group. Because the en the complex pattern seen elsewhere in the conditions of these characters in Archaeo- tree (e.g., three patterns occur within Yan- nycteris could not be determined, we cannot gochiropteraÐfull ossi®cation [secondarily be sure if the transformations in these struc- acquired once], ossi®cation only at phalanx tures occurred before or after differentiation base [acquired twice], and no ossi®cation of Archaeonycteris from the lineage leading [secondarily acquired once]). to the microchiropteran crown group. Other character transformations that occur Moving up the microchiropteran part of somewhere near the base of the microchirop- the tree, the next node unites Hassianycteris, teran tree (we cannot determine exactly Palaeochiropteryx, and the microchiropteran where because of missing data) involve evo- crown group. As noted above, presence of lution of the phanerocochlear condition the phanerocochlear condition (character 27), (character 27), a deep, constricted stapedial a deep, constricted stapedial fossa (character fossa (character 31), presence of anterior 31), anterior laminae on the ribs (character laminae on the ribs (character 86), broad pos- 86), broad posterior laminae on the ribs terior laminae on the ribs (character 89), and (character 89), and a dorsal articular facet on presence of a dorsal articular facet on the the scapula (character 100) may apply at this 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 99

level in the tree. Derived characters that un- laminae absent in both outgroups, some Pter- ambiguously diagnose the clade comprising ododidae, and Icaronycteris). Hassianycteris, Palaeochiropteryx, and the microchiropteran crown group include: (1) FEATURES DIAGNOSING THE nyctalodonty (character 21; primitive tribos- MICROCHIROPTERAN CROWN GROUP phenic condition in Scandentia, Icaronycter- is, and Archaeonycteris); (2) a greatly en- Character transformations that appear to larged cochlea (character 26; either not en- diagnose the microchiropteran crown group larged or only moderately enlarged in the fall into two broad categories: (1) features outgroups, Pteropodidae, Icaronycteris and that can be positively attributed to this par- Archaeonycteris); (3) distal tip of ventral ticular node (by virtue of having been scored process of manubrium laterally compressed with a different state in Hassianycteris and (character 91; tip blunt or rounded in Der- more distal outgroups); and (2) derived fea- moptera, Icaronycteris, and Archaeonycter- tures that minimally diagnose the crown is); (4) absence of an ossi®ed third phalanx group but could not be scored in the fossils, (claw) on wing digit II (index ®nger; char- leaving open the possibility that they may acter 151; ossi®ed claw present in both out- have evolved earlier (closer to the basal groups, Pteropodidae, Icaronycteris, and Ar- node) in the tree. Transformations of the lat- chaeonycteris); (5) ®bula thin and threadlike ter sort may be unambiguous synapomor- (character 170; ®bula well developed in both phies, but it is not clear at what level they outgroups, Pteropodidae, Icaronycteris, and apply. Archaeonycteris); (6) calcar present (char- In the ®rst category, we ®nd that only acter 171; calcar absent in both outgroups, three transformations in hard-tissue charac- Icaronycteris, and Archaeonycteris). Opti- ters appear to diagnose the microchiropteran mization of the latter character (®g. 38) is crown group: (1) free premaxilla (character somewhat surprising given previous hypoth- 9; sutured premaxilla in both outgroups, eses (e.g., Simmons, 1994, 1995; see table 5) Pteropodidae, and all four Eocene genera); that presence of a calcar is a synapomorphy (2) reduction to two lower premolars on each of Chiroptera. Our analysis suggests other- side of the jaw (character 20; three lower pre- wise. A calcar seems to have been absent in molars in Pteropodidae and all four Eocene the most recent common ancestor of Mega- genera); and (3) a xiphisternum with promi- chiroptera and Microchiroptera, and appar- nent median keel (character 95; keel absent ently evolved independently in Pteropodidae in both outgroups, Pteropodidae, Icaronyc- and in the lineage leading to extant micro- teris, Archaeonycteris, and Palaeochiropte- chiropterans. ryx). The last clade directly involving the fossil In comparison to the lists of features di- taxa analyzed in this study is that comprising agnosing more inclusive clades (see above), Palaeochiropteryx plus the microchiropteran these synapomorphies by themselves are not crown group. Only a single derived character compelling. For example, few workers would unequivocally diagnoses this clade: presence agree that a free premaxilla is likely primi- of a ventral accessory process on cervical tive for all extant microchiropterans (as sug- vertebra 5 (character 77; absent in both out- gested by optimization of this character on groups, Pteropodidae, Icaronycteris, Ar- our tree; ®g. 39) because yangochiropteran chaeonycteris, and Hassianycteris). Two oth- bats lack this complex specialization. In- er characters minimially apply at this level, stead, yangochiropterans have a premaxilla but lack of data for Hassianycteris and Ar- that is ®rmly fused to the maxilla in the chaeonycteris make it impossible to deter- adult. The free premaxilla is a feature unique mine the point of transformation on the tree. among mammals, and we agree that it is hard These characters include (1) a deep, con- to imagine either its loss or independent or- stricted stapedial fossa (character 31; fossa igin in two different groups (which are, of indistinct in both outgroups, shallow and course, the two most-parsimonious hypothe- broad in Pteropodidae), and (2) presence of ses given the topology of our optimal tree). anterior laminae on the ribs (character 86; Reduction from three to two lower premolars 100 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Fig. 39. Premaxilla±maxilla articulations (character 9) mapped on the phylogeny from ®gure 36. Presence of a free premaxilla appears to be a synapomorphy of the microchiropteran crown group (although see text for discussion). Fusion of the premaxilla to the maxilla is a synapomorphy of Yan- gochiroptera. The condition for Pteropodidae is depicted as ``uncertain'' because two conditions occur within this group: the premaxilla is sutured to the maxilla in most pteropodids, but these elements are fused in other taxa. This optimization suggests that a suture joint between the premaxilla and maxilla is primitive for Pteropodidae. on each side is a character that is similarly features are numerous and include the fol- troublesome, because presence of three lower lowing: (1) presence of a tragus (character 2; premolars characterizes many groups of ex- absent in outgroups and Pteropodidae); (2) tant microchiropterans that nest well up with- aquaeductus cochleae small or absent (char- in the crown group (e.g., Rhinolophidae, acter 33; large in Dermoptera and Pteropod- Mormoopidae, Nataloidea, many vespertili- idae); (3) presence of sophisticated echolo- onids). This is also true of the xiphisternal cation (character 36; absent in outgroups and keel. Although presence of a keel appears to Pteropodidae); (4) m. styloglossus originates diagnose the microchiropteran crown group, from ventral surface of midpoint of stylohyal many lineages within that clade lack a keel (character 58; originates from expanded tip on the xiphisternum (e.g., Rhinopomatidae, of stylohyal and/or adjacent surface of skull Hipposiderinae, some Phyllostomidae, Noc- in outgroups and Pteropodidae); (5) clavicle tilionidae, Myzopodidae, Molossoidea, some articulates with coracoid process of scapula Vespertilioninae). (character 113; clavicle articulates with acro- Despite the relative weakness of these mion in both outgroups, Pteropodidae, and data, monophyly of the microchiropteran Icaronycteris); (6) m. spinotrapezius clearly crown group is indirectly supported by a differentiated from trapezius complex (char- broad range of other characters that unfor- acter 124; m. spinotrapezius not differenti- tunately have somewhat ambiguous distri- ated from trapezius complex in both out- butions. These transformations minimally di- groups and Pteropodidae); (7) origin of m. agnose the microchiropteran crown group, acromiodeltoideus does not include thoracic but might have evolved earlier in the tree; vertebra 6 (character 127; origin does include we could not score them in fossil sister-taxa T6 in both outgroups and Pteropodidae); (8) of the microchiropteran crown group. Such spinal cord with angle between dorsal horns 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 101

of 0±25Њ (character 190; angle between dor- remain obscure, although Slaughter (1970) sal horns 35±50Њ in both outgroups, 70±80Њ noted that reduction and simpli®cation of the in Pteropodidae); and (9) inferior colliculus premolar dentition in early bats appears cor- larger than superior colliculus (character 191; related with shortening of the face (brachy- inferior colliculus signi®cantly smaller than cephaly). It is possible that these trends may superior colliculus in both outgroups and have been related to the need to focus the Pteropodidae). One other transformation may ears anteriorly and reduce interference with diagnose the microchiropteran crown group, returning echolocation callsÐthe auditory although it has been documented only in equivalent of the type of facial shortening Yinochiroptera and Yangochiroptera (no data generally associated with evolution of bin- for Emballonuridae): m. ¯exor digitorum ocular vision. Another possibility is that loss profundus does not insert on digit II of wing of premolars and reduction in the number of (character 154; muscle does insert on digit II roots on the remaining teeth represent mech- in both outgroups and Pteropodidae). Even if anisms for mass reduction. Flying animals these features evolved in a stepwise fashion must generate adequate lift to remain air- up the tree, at least a few probably represent borne, and lift requirements increase with in- true synapomorphies of the microchiropteran creasing body mass (Norberg, 1985, 1986a, crown group. 1987, 1990; Rayner, 1986; Norberg and Ray- ner, 1987). Distribution of body mass is also CHARACTER TRANSFORMATIONS important because it affects the position of AT BASAL NODES: the center of mass (ϭ center of gravity), A FUNCTIONAL PERSPECTIVE which in turn in¯uences ¯ight ef®ciency. The head is the heaviest part of the body in most In the previous section, we described the vertebrates in part because of the density of character transformations that seem to apply the teeth. In addition to the effect that a at each node in the basal part of the micro- heavy head may have on total body mass and chiropteran tree. An interesting pattern also the location of the center of mass, head mass emerges when these transformations are also affect the size of neck muscles needed viewed from a functional perspective. Rather to support the head and resist torque (BuÈhler, than changes in each organ system being 1992). Dental reduction in birds (most of concentrated at one or two nodes, we see a which lack teeth entirely) is widely regarded pattern of stepwise changes in multiple func- as a specialization that increased ¯ight ef®- tional systems as we move up the microchi- ciency because it simultaneously reduced to- ropteran tree. This suggests a complex, mo- tal mass and concentrated more of the body saic pattern of evolution in which several mass near the center of gravity between the organ systems were being re®ned simulta- wings (e.g., Welty, 1955; Stahl, 1985; BuÈhl- neously. er, 1992; Feduccia, 1996). Dental reduction Overall, the changes associated with the in pterosaurs has also been noted as a pos- facial region and masticatory apparatus are sible adaptation for reducing body mass and relatively minor compared to those seen in increasing ¯ight ef®ciency (e.g., Stahl, 1985; other systems. In order of their appearance BuÈhler, 1992). Because bats are mammals along the backbone of the microchiropteran that rely on their dentition for food process- tree (beginning at the base, prior to diver- ing, extreme dental reduction is rare (limitied gence of Icaronycteris), these transforma- mostly to nectarivorous taxa). However, even tions include (1) evolution of an elongate an- small reductions in mass may contribute to gular process (though this may ultimately ¯ight ef®cency. prove to be plesiomorphic for bats), (2) re- The free premaxilla present in Emballon- duction of the number of roots on P3, (3) uridae and Yinochiroptera (and perhaps in evolution of nyctalodonty, and (4) reduction the most recent common ancestor of extant in the number of lower premolars and mod- microchiropterans; ®g. 39) provides an un- i®cation of the simple sutured connection be- usual degree of mobility in the snout. Utility tween the premaxilla and maxilla. of this feature has not been adequately in- Functional implications of these changes vestigated, but our experience in handling 102 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 live emballonurids suggests that dorsi¯exion whatever the function(s) of this structure, an of the snout is under voluntary control. Dor- enlarged orbicular apophysis evolved to fa- si¯exion of the snout effectively increases cilitate ef®cient function of the ear in echo- the gape of the mouth, which may be im- location. Likewise, the expanded tip on the portant in feeding and/or emission of echo- stylohyal may represent a re®nement of the location calls. With respect to the latter, it is system for producing echolocation calls. The interesting to note that the mode of emission microchiropteran echolocation system de- of echolocation callsÐeither through the pends upon calls produced in the larynx, mouth (oral emission) or through the nasal which is supported in part by muscles asso- passages (nasal emission)Ðvaries among ciated with the hyoid apparatus. The expand- bats that have a free premaxilla. Emballon- ed tip on the stylohyalÐwhich is ®rmly at- uridae, Craseonycteridae, and perhaps Rhin- tached to the bullaÐserves to anchor the hy- opomatidae are oral emitters, while Rhino- oid apparatus to the skull, thus providing a lophoidea and perhaps Rhinopomaidae are stable attachment site for muscles of the nasal emitters (Pederson, 1993). Given to- throat. This attachment may be important in pology of our tree (®g. 36), it appears that supporting the larynx during production of oral emission is primitive and nasal emission echolocation calls. derived as suggested by Van Valen (1979) Enlargement of the cochlea (speci®cally and Pederson (1993). The extent to which the basal turn) in microchiropterans appears dorsi¯exion of the snout may have played a to be a specialization for increased sensitivity role in the evolution of nasal emission in yin- to high-frequency sounds (Ͼ20 kHz), such as ochiropteran bats has not been explored. the returning echoes from vocalizations used Free movement of the premaxilla may also in echolocation (Henson, 1970; Novacek, be important in prey manipulation, particu- 1985a, 1987, 1991; Habersetzer and Storch, larly in those taxa that take large arthropod 1992). In all mammals, the basal turn of the or small vertebrate prey. Fenton (personal cochlea is the region where high-frequency commun.) observed a Nycteris grandis eat- sounds are perceived (Henson, 1970; Dallos, ing a Nycteris thebiaca, and noted that the 1973; Bruns, 1979; Bruns et al., 1983±1984; former used its upper lips extensively and Harris and Dallos, 1984). Enlargement of manipulated its prey in an almost closed- this region, which effectively lengthens the mouth fashion. The extent to which a free basal portion of the basilar membrane, ap- premaxilla may facilitate the capacity to han- parently increases sensitivity to high-fre- dle large prey items has yet to be investigat- quency sounds and slight frequency shifts in ed. this range (Henson, 1970; Dallos, 1973; Changes in the basicranium and ear region Bruns, 1979; Bruns et al., 1983±1984; Harris also appear to have evolved in a stepwise and Dallos, 1984). fashion. The ®rst modi®cations to appear In the lineage leading to extant Microchi- (prior to the divergence of Icaronycteris roptera, cochlear enlargement seems to have from the microchiropteran stem stock) in- evolved in a relatively continuous fashion cluded a moderately enlarged cochlea, en- that we chose to score as a series of steps larged orbicular apophysis on the malleus, (®g. 40). The primitive condition for bats and an expanded cranial tip on the stylohyal. was the presence of an unenlarged cochlea, All of these features are probably related to one comparable in size to those seen in other the evolution of echolocation. As noted ear- mammals of similar body size. A moderately lier, presence of a large orbicular apophysis enlarged cochlea evolved prior to the diver- may improve the ability of the middle ear gence of Icaronycteris, to be followed sub- ossicles to transmit high-frequency sounds sequently by even greater enlargement before with a minimum time delay, or may play a the divergence of Hassianycteris. Moderate role in the avoidance of self-deafening (see enlargement of the cochlea does not by itself discussion under character 35 above). The indicate that a bat could echolocate, because fact that all known extant microchiropterans some nonecholocating pteropodids have a have an enlarged orbicular apophysisÐand moderately enlarged cochlea (see discussion all use echolocationÐleads us to suspect that under character 26 above). However, the 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 103

Fig. 40. Cochlear enlargement (character 26) mapped on the phylogeny from ®gure 36. In the lineage leading to extant microchiropterans, a ``moderately enlarged'' cochlea evolved prior to the divergence of Icaronycteris; a ``greatly enlarged'' cochlea evolved prior to the divergence of Hassianycteris. Re- versals to the former condition apparently occurred independently in Mystacinidae and within Megad- ermatidae and Phyllostomidae (note that the latter taxa are marked as ``uncertain'' as a result of taxo- nomic polymorphism). This optimization also suggests that a moderately enlarged cochlea evolved independently within Pteropodidae (see text for discussion).

combination of a moderately enlarged coch- Icaronycteris. Based on the presumed func- lea, enlarged orbicular apophysis on the mal- tion of this loose attachmentÐcochlear iso- leus, and an expanded cranial tip on the sty- lation, thought to function in reducing bone lohyal are seen in extant bats only in forms conduction of laryngeal vibrations (Henson, that use echolocation. We therefore follow 1970; Van Valen, 1979)Ðwe surmise that previous authors (e.g., Novacek, 1985a, this feature evolved in concert with the early 1987, 1991; Habersetzer and Storch, 1992) stages of cochlear enlargement and facilitat- in inferring that Icaronycteris, Archaeonyc- ed the evolution of echolocation. teris, Hassianycteris, and Palaeochiropteryx Another transformation that took place in used echolocation. We agree with Pye (1968: the microchiropteran lineage prior to the di- 797), who observed that vergence of Hassianycteris was evolution of a phanerocochlear cochlea. Unfortunately, The Eocene brought mammals mean we cannot determine the level of this trans- And bats began to sing; formation in the phylogenetic tree because Their food they found by ultrasound And chased it on the wing. this character could not be scored in Icaron- ycteris or Archaeonycteris. The phanero- The implications of cochlear morphology for cochlear condition was explained by Nova- reconstructing foraging strategies is dis- cek (1991: 84) as ``an accommodation to the cussed in depth below under ``Foraging ecol- problem of ``packing'' middle ear structures ogy of Eocene bats.'' in a space constrained by the expanded coch- Loose attachment of the periotic to the lea.'' As discussed above under character 27, basicranium evolved prior to the divergence this feature is probably linked to cochlear ex- of Archaeonycteris, although we cannot de- pansion, although we cannot be sure at what termine the exact level in the phylogeny be- point the phanerocochlear condition evolved cause this character could not be scored in relative to changes in cochlear size. 104 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Prior to the divergence of Palaeochirop- groove or oval situated on the anteromedial teryx, yet another transformation took place: rim of the glenoid fossa. evolution of a deep, constricted fossa for m. In the forelimb skeleton, we see progres- stapedius. Again, we cannot determine the sive reduction in the number of phalanges in level of this transformation in the phyloge- digit II of the wing (the index ®nger) begin- netic tree because this character could not be ning after the onset of evolutionary changes scored in Icaronycteris, Archaeonycteris,or in the shoulder region (®g. 41). Because the Hassianycteris. The function of this transfor- thumb is relatively small in all bats (at least mation remains somewhat obscure, although in comparison to the other digits of the it is possible that it is in some way linked to hand), digit II forms the leading edge of the the system for avoiding self-deafening that is dactylopatagium (``hand wing'') near the used by many echolocating bats (Henson, wrist. The dactylopatagium plays an impor- 1964, 1965, 1966, 1967a, 1979; Jen and tant role in both lift generation and steering Suga, 1976; Fenton et al., 1995; see discus- during ¯ight (Vaughan, 1959; Norberg, 1969, sion below). 1970, 1972b, 1976; Hill and Smith, 1984). Transformations in features of the postcra- Stiffness of leading edge of the dactylopa- nial skeleton also seem to have evolved in a tagiumÐparticularly the section known as series of steps up the tree. The ®rst change the dactylopatagium minus, which extends in the postcranium was proximal extension between the second and third digitsÐis crit- of the trochiter up to the level of the head of ical to the ability of the wing to resist bend- the humerus, which occurred prior to the di- ing and twisting forces (Norberg, 1969, vergence of Icaronycteris. Several other de- 1970, 1972b). This functional unit is also im- rived traits evolved subsequently, although portant in maintaining wing camber we cannot be sure at what level because they (Vaughan, 1959, 1970c; Norberg, 1970, could not be scored in Archaeonycteris. 1972b, 1976). These traits include (1) presence of a dorsal In bats lacking the distal phalanges on dig- articular facet on the scapula, (2) presence of it II, a ligamentous connection runs between anterior laminae on the ribs, and (3) presence the end of ®rst phalanx of digit II and the of broad posterior laminae on the ribs. Trans- base of the second phalanx in digit III (Nor- formations that occurred subsequent to the berg, 1969, 1970, 1972b). This ligament is divergence of Archaeonycteris but prior to kept under continuous tension by the struc- the divergence of Hassianycteris include (1) ture of digit III, which is bent posterodorsally laterally compressed ventral process of the due to the structure of the metacarpophal- manubrium, (2) loss of an ossi®ed third pha- angeal joint (Norberg, 1969, 1970, 1972b). lanx in digit II of the wing, (3) reduction of This arrangement results in a convex frame ®bula to a thin, threadlike element, and (4) within which the dactylopatagium minus is presence of a calcar. kept stretched under tension, thus forming a Dividing these features by anatomical re- unit that is stiff in the plane of the membrane gion, we see progressive changes occurring (Norberg, 1969, 1970, 1972b). Photographs in different functional units. Modi®cations of of ¯ying bats show that the airstream pro- the shoulder region began with enlargement duces little deformation of the dactylopata- of the trochiter, which preceded evolution of gium minus or its supporting elements during a dorsal articular facet on the scapula. A ¯ight (Norberg, 1969, 1970). The tensile functional shoulder-locking mechanism, strength of the second digit is critical to the which requires a secondary articulation be- funtion of this unit, since the second digit tween the trochiter and dorsal articular facet, must resist the forces placed upon the dac- was clearly present by the time that Hassi- tylopatagium minus during ¯ight. Reduction anycteris diverged from the lineage leading of the number of phalanges in digit II ap- to extant microchiropterans. In terms of mor- parently took place in a sequential fashion in phology, the primitive shoulder-lock appar- bats, with complete loss of ossi®ed phalanges ently consisted of a trochiter that extended evolving independently at least three times in up to but not beyond the humeral head, and the microchiropteran crown group (®g. 41). a dorsal articular facet in the form of a small It seems likely that loss of phalanges served 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 105

Fig. 41. Changes in the number of phalanges on wing digit II (the index ®nger; characters 149, 150, and 151) mapped on the phylogeny from ®gure 36. Presence of three ossi®ed phalanges on digit II is primitive for bats. Reduction to two ossi®ed phalanges occurred just prior to the divergence of Hassi- anycteris from the lineage leading to the microchiropteran crown group. The microchiropteran crown group is diagnosed by further reduction to only one ossi®ed phalanx. Compete phalangeal reduction (no ossi®ed phalanges on wing digit II) apparently evolved independently at least three times in the crown groupÐin Emballonuridae, Nataloidea, and either in the common ancestor of Rhinolophoidea or inde- pendently in Nycteridae and Rhinolophidae. Presence of two ossi®ed phalanges in Rhinopomatidae appears to be a reversal rather than retention of the primitive condition.

to increase the tensile strength per unit mass ilarly, anterior laminae evolved at some point of the distal second digit by replacing artic- prior to the divergence of Palaeochiropteryx, ulated phalanges (which together have a rel- although we cannot be sure of the level be- atively low tensile strength per unit mass) cause we could not score this character in with a continuous ligament that has a higher Archaeonycteris or Hassianycteris. In any tensile strength per unit mass. Mass reduc- case, these modi®cations are absent in Ica- tion, particularly near the distal end of the ronycteris, so they must have occurred with- wing, contributes signi®cantly to ef®cient in the basal microchiropteran lineage. We ex- ¯ight performance (Swartz, 1997). Loss of pect that function of rib laminae is twofold: the distal phalanges may represent a mech- to stiffen the ribcage and to provide larger anism for distal mass reduction as con- areas for muscle attachment. Muscle groups strained by the need to provide the tensile that originate directly from the ribcage in- strength necessary to maintain a stiff dacty- clude those of the m. serratus anterior com- lopatagium minus during ¯ight. plex, which is a critical component of the Modi®cations in the axial skeleton in the ¯ight musculature. The posterior division of basal part of the microchiropteran tree in- m. serratus anterior contributes to the down- cluded evolution of anterior laminae and stroke, and the anterior division serves to an- broad posterior laminae on the ribs. Broad chor the medial edge of the scapula and may posterior laminae evolved prior to the diver- also help to initiate the upstroke of the wing gence of Hassianycteris, although we cannot (Vaughan, 1959, 1970b; Norberg, 1970, be sure of the level because we could not 1972a; Strickler, 1978; Hermanson and Al- score this character in Archaeonycteris. Sim- tenbach, 1983, 1985). 106 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

A close association has been demonstrated from our phylogeny), we would expect that between wingbeat and sound emission in mi- any loss in ef®ciency related to ribcage struc- crochiropterans under laboratory conditions ture would have been compensated for by (Schnitzler, 1968, 1970a, 1970b, 1971, 1973; other mechanical changes in the ¯ight, ven- Suthers et al., 1972; Schnitzler and Henson, tilation, and echolocation systems. 1980; Joerman and Schmidt, 1981; Heblich, Another change in the axial skeleton in- 1986; Lancaster et al., 1992) and in free-¯y- volved evolution of a laterally compressed ing bats foraging in nature (Kalko, 1994). ventral process on the manubrium, which ap- Recent studies of ¯ight, respiration, energy peared prior to the divergence of Hassianyc- expenditure, and echolocation have indicated teris. Changes in this structure, which forms that while production of echolocation calls is a keel when laterally compressed, are likely extremely costly in resting bats, there is little to be functionally related to the ¯ight mus- additional cost for echolocation in ¯ying bats cles that originate from the manubrium, spe- (at least for search-phase calls; Kalko, 1994) ci®cally m. pectoralis. M. pectoralis provides because the same muscles that ¯ap the wings most of the power for the downstroke of the also ventilate the lungs and produce the puls- wings (Vaughan, 1959, 1970b; Strickler, es of breath used to generate echolocation 1978). More anteriorly located ®bers in this calls (Speakman et al., 1989; Rayner, 1991a, complex (e.g., those originating from the ma- 1991b; Speakman and Racey, 1991; Speak- nubrium rather than from the body of the man, 1993). A critical link in this system is sternum) draw the humerus downward and m. serratus anterior, which ties the ribcage to sharply forward, whereas the posterior ®bers the ¯ight mechanism. The increased attach- pull the humerus downward and backward ment area for this muscle complex provided (Vaughan, 1959). Modi®cation of the ventral by anterior and posterior laminae on the process of the manubrium may re¯ect ribsÐand concomitant increase in stiffness changes in the relative size, moment arm, of the ribcage that we hypothesize is created and functional importance of the anterior by these laminaeÐmay simultaneously in- portion of m. pectoralis. crease ef®ciency of the ¯ight mechanism, Moving to the hindlimbs, we ®nd the ®rst ventilation, and echolocation system. M. ser- case in which known modi®cations of an an- ratus anterior plays an important role in ¯ight atomical region seemingly evolved in a sin- during the downstroke and early stages of the gle segment of the tree rather than in a step- upstroke. A larger area of origin may provide wise manner. In this instance, all changes oc- for improved performance of this muscle curred after the divergence of Archaeonyc- complex in ¯ight and may also facilitate ven- teris and prior to the divergence of tilation of the lungs both by increasing the Hassianycteris. Two transformations oc- area of the connection between the ¯ight ap- curred at this level: reduction of the ®bula to paratus and ribcage. Increased stiffness of the a thin, threadlike element, and evolution of ribcage caused by presence of rib laminae a calcar. The functional implications of the could also faciliate exhaling and production former are not clear. Several authors have of echolocation calls by increasing the force noted ®bular reduction in extant microchi- of elastic recoil of the ribcage upon relaxa- ropterans, and have implied that reduction of tion of m. serratus anterior. We note that the the ®bula is somehow associated with lack echolocating bat species studied by Speak- of a need for a robust ®bula in bats that ha- man and his colleagues (Phyllostomus has- bitually hang under branches and do not use tatus, Plecotus auritus, Pipistrellus pipistrel- typical quadrupedal locomotion (Vaughan, lus, and Plecotus auritus) all have some de- 1959, 1970a; Walton and Walton, 1968, gree of development of rib laminae. It would 1970; Howell and Pylka, 1977; Hill and be interesting to determine if the few micro- Smith, 1984). This suggests that the ®bula chiropterans that lack rib laminae (e.g., Mo- may have been reduced by default when it lossus molossus) are as energy-ef®cient was no longer needed to support compressive while ¯ying and echolocating as are taxa forces associated with quadrupedial loco- with rib laminae. If rib laminae have been motion. However, the correlation between secondarily lost in these forms (as we infer hanging behavior and a thin ®bula is not per- 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 107 fect, as pteropodids retain a relatively robust helping to hold the uropatagium in a cupped ®bula. In the latter case, retention of a robust position to capture and hold prey during ae- ®bula may re¯ect the use of the hindlimbs in rial hawking; (3) holding the uropatagium manipulating food, which is a behavior that out of the way of the feet in species that trawl is common in pteropodids (e.g., Epomopho- over water for prey; and (4) helping to cup rus, Rousettus, Dobsonia, and Cynopterus; the uropatagium to hold young as they are B. Fenton, personal commun.). Regardless, a being born (Vaughan, 1959, 1970a, 1970b; more detailed evaluation of hindlimb myol- Webster and Grif®n, 1962; Norberg, 1976; ogy and function will be needed to better as- Hill and Smith, 1984; Schnitzler et al., 1994; sess possible causes for ®bular reduction. Kalko, 1995). Of these functions, there is lit- As noted previously, presence of a calcar tle chance that the latter two (use in trawling was once considered to be a synapomorphy and in parturition) were important factors in calcar evolution. Trawlers are limited to of Chiroptera (e.g., by Simmons, 1994, clades nested well within Microchiroptera 1995), but it now appears that a calcar (i.e., Noctilionidae and Vespertilionidae), and evolved independently in Pteropodidae and a calcar size is not sexually dimorphic. We the microchiropteran lineage (®g. 38). suspect that evolution of the calcar in the mi- Known or suspected functions of the calcar crochiropteran lineage was associated with are diverse, including (1) supporting the trail- transformations in ¯ight habits and/or for- ing edge of the uropatagium and helping to aging strategies. Correlations of calcar evo- control camber of the uropatagium during lution with other postcranial modi®cations ¯ight (which in turn may affect the amount seem to support this hypothesis (see below of lift generated by the uropatagium); (2) for additional discussion).

EVOLUTION OF ECHOLOCATION AND FORAGING STRATEGIES An extensive literature on the evolution of 1990, 1994; Bell, 1982a, 1985; Neuweiler, echolocation and foraging strategies in bats 1984, 1989, 1990; Barclay, 1985, 1986; No- has developed in the decades since Grif®n vacek, 1985a, 1987, 1991; Habersetzer, and Novick's in¯uential studies (e.g., Grif®n 1986; Mein and Tupinier, 1986; Scholey, and Novick, 1955; Grif®n, 1958; Novick, 1986; Aldridge and Rautenbach, 1987; Nor- 1958a, 1958b, 1962, 1963a; Grif®n et al., berg and Rayner, 1987; Schnitzler et al., 1960) ®rst documented the use of echoloca- 1987, 1994; Jones and Rayner, 1988, 1991; tion in microchiropterans. Numerous issues Neuweiler and Fenton, 1988; Speakman et have been discussed, including (1) the timing al., 1989; Kalko and Schnitzler, 1989, 1993; of the origin of echolocation relative to the Rayner, 1991b; Speakman and Racey, 1991; evolution of powered ¯ight; (2) relationships Speakman, 1993; Jones, 1994; Fenton et al., between wing shape, echolocation call struc- 1995; Kalko, 1995; Arita and Fenton, 1997). ture, and foraging behavior; and (3) costs and With few exceptions, these discussions have bene®ts of different ¯ight and foraging strat- focused principally on functional correlations egies (e.g., Vaughan, 1959, 1966, 1970b; and have paid little attention to phylogenetic Struthsaker, 1961; Novick, 1963b, 1965, context. This has been due in part to lack of 1977; Novick and Vaisnys, 1964; Gould, a well-corroborated phylogeny, which has 1970; Schnitzler, 1970b; Vaughan and Bate- hindered evolutionary interpretations of man, 1970; Fenton, 1972, 1974a, 1974b, many types of data. The phylogeny proposed 1980, 1982a, 1982b, 1984, 1990, 1994a, in the present study (®g. 36) provides an op- 1995; Findley et al., 1972; Gillette, 1975; portunity to reexamine previous hypotheses Pirlot, 1977; Fiedler, 1979; Simmons et al., about the evolution of echolocation and for- 1979; Pye, 1980; Schnitzler and Henson, aging strategies in bats, and to draw new in- 1980; Simmons, 1980; Simmons and Stein, ferences based on examination of functional 1980; Norberg, 1981, 1986a, 1986b, 1989, correlations in an explicitly phylogenetic 108 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 context. Because the focus of our study is tera (well developed ¯ight capabilities, ori- relationships of the Eocene bats to extant lin- entation by sophisticated echolocation in all eages, we limit our discussion here to hy- species). potheses relevant to the basal nodes in the Considering the origin of ¯ight and echo- tree. Evolution of echolocation call structure, location in Microchiroptera, three theories wing design, and foraging strategies among have recently emerged: the ``tandem evolu- and within extant lineages will be addressed tion'' theory (Speakman, 1993), the ``¯ight elsewhere (e.g., Kalko and Simmons, ms). ®rst'' theory (Norberg, 1989, 1994), and the ``echolocation ®rst'' theory (Hill and Smith, TIMING OF THE ORIGIN OF FLIGHT AND ECHOLOCATION 1984; Fenton et al., 1995). Speakman's (1993) hypothesis that ¯ight and echoloca- One topic that has drawn considerable at- tion evolved simultaneously in Microchirop- tention in recent years is the origin of ¯ight tera is based on the observed functional link and echolocation in bats (e.g., Padian, 1987; between ¯ight, ventilation, and echolocation Norberg, 1989, 1994; Rayner, 1991b; Speak- in extant microchiropterans (Speakman et al., man, 1993; Fenton et al., 1995; Arita and Fenton, 1997). This has been due in part to 1989; Speakman and Racey, 1991). This the- the bat monophyly controversy (see ``Rela- ory suggests that microchiropterans evolved tionships Among Extant Lineages of Bats'' from small nocturnal insectivores that may above), which promulgated the idea that have used ultrasound for intraspeci®c com- ¯ight may have evolved independently in munication, and were thus able to ``capitalize Megachiroptera and Microchiroptera (Smith on the opportunity to develop in tandem with and Madkour, 1980; Hill and Smith, 1984; ¯ight a sophisticated echolocation system, Pettigrew, 1986, 1991a, 1991b, 1994, 1995; since this would require no major redirection Pettigrew and Jamieson, 1987; Pettigrew et in its sensory specialization'' (Speakman, al., 1989; Rayner, 1991b; Pettigrew and 1993: 56). The details of just how and why Kirsch, 1995). If ¯ight evolved independent- ¯ight and sophisticated echolocation evolved ly in these two groups, then it is appropriate have not been explored under this scenario, to treat Megachiroptera and Microchiroptera although a link between predation and echo- separately in developing evolutionary hy- location has been suggested (Speakman, potheses about the origins of ¯ight and so- 1993). Speakman (1993: 58) noted that an phisticated echolocation. This approach su- alternative model where ¯ight evolved ®rst per®cially simpli®es the problem because it and was followed much later by the evolu- eliminates the need to simultaneously explain tion of echolocation ``seems improbable,'' the conditions seen in Megachiroptera (well but he did not present any arguments in sup- developed ¯ight capabilities, visual orienta- port of this position. tion in all species, use of primitive echolo- The ¯ight-®rst (Norberg, 1989, 1994) and 3, 8 cation in a few species ) and Microchirop- echolocation-®rst (Hill and Smith, 1984; Fenton et al., 1995) theories apparently agree 8 Echolocation is thought to be used by only a very few megachiropteran bats: Rousettus aegypticus, R. am- that microchiropterans probably arose from plexicaudatus, R. leschenaulti, and perhaps Eonycteris gliding, nocturnal insectivores that used a spelaea (MoÈhres and Kulzer, 1956; Grif®n et al., 1958; primitive form of echolocation with short, Novick, 1958b; Gould, 1988; Kingdon, 1974; Sales and broadband (i.e., frequency modulated) clicks Pye, 1974; Roberts, 1975; Hebert, 1986). In Rousettus, to help in orientation. The ¯ight-®rst theory echolocation signals are produced by clicking the tongue proposes that the next step involved the evo- (MoÈhres and Kulzer, 1956; Kulzer, 1958, 1960; Novick, lution of powered ¯ight to improve mobility 1958b; Kingdon, 1974; Sales and Pye, 1974; Roberts, in the arboreal milieu and reduce the amount 1975). Eonycteris produces signals that may be used in of time and energy required for foraging echolocation by slapping the tips of the wings together (Norberg, 1994). At this stage, these animals during ¯ight (Gould, 1988). Both Rousettus and Eon- cyteris apparently use echolocation only in situations would have been able to ¯y, but still lacked where there is little or no ambient light (e.g., in the back the ability to capture airborne prey because of a cave or in a dark room); tongue-clicks and wing- they were not maneuverable enough (Nor- slaps cease when light levels are increased (Kingdon, berg, 1989, 1994). Sophisticated echoloca- 1974; Gould, 1988). tionÐwhich could be used for detecting, 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 109 tracking, and evaluating airborne preyÐ is Knight and Mindell, 1993; Simmons, 1993a, hypothesized to have evolved only after 1994, 1995; Wible and Martin, 1993; Sim- powered ¯ight (Norberg, 1989, 1994; Arita mons and Quinn, 1994; Kirsch et al., 1995; and Fenton, 1997). Contra Speakman (1993), Allard et al., 1996; Hutcheon and Kirsch, Arita and Fenton (1997) noted that the ¯ight 1996; Kirsch, 1996; Miyamoto, 1996; Porter ®rst hypothesis is supported by the mechan- et al., 1996). Given that bats are monophy- ical coupling of ¯ight and echolocation in letic, it seems clear that powered ¯ight extant microchiropterans (Speakman et al., evolved only once in mammals, and that the 1989; Rayner, 1991a, 1991b; Speakman and most recent common ancestor of Megachi- Racey, 1991; Speakman, 1993), which sug- roptera and Microchiroptera was a ¯ying gests that laryngeal echolocation may be mammalÐa bat (Simmons, 1994, 1995). cost-effective only when linked to a well de- Speakman (1993) suggested that bats veloped ¯ight and ventilation system. could be monophyletic but derived from a An alternative to this hypothesis is the non¯ying most recent common ancestor. We echolocation-®rst theory proposed by Fenton reject this hypothesis because it is highly un- et al. (1995). They suggested that the ability parsimonious. By our tally, supposing that to use echolocation to detect, track, and as- ¯ight evolved independently in the two sub- sess airborne prey evolved in the gliding an- orders requires the independent evolution of cestors of microchiropterans as a result of se- more than a dozen derived traits and char- lection for stronger signals to increase the ef- acter complexes of the postcranial musculo- fective range of echolocation (Fenton et al., skeletal and nervous systems (table 7), a sce- 1995; Arita and Fenton, 1997). The echolo- nario that we consider to be extremely un- cation-®rst theory argues that gliding ``pre- likely. There is no compelling evidence that bats'' hunted from perches using echoloca- these traits are not homologous in Megachi- tion to detect, track, and assess airborne prey roptera and Microchiroptera, and we are not in the subcanopy, and that powered ¯ight convinced by Speakman's (1993) suggestion evolved later to increase maneuverability and that ¯ight and echolocation must have simplify returning to the hunting perch (Fen- evolved in tandem in Microchiroptera. Flight ton et al., 1995; Arita and Fenton, 1997). The can clearly evolve in the absence of echolo- latter scenario was ®rst suggested by Hill and cation, as demonstrated by the existence of Smith (1984), although they did not provide Megachiroptera and birds. Accordingly, we any details. acceptÐbecause these represent the best sup- Arita and Fenton (1997: 56) reviewed the ported hypotheses given all the data avail- ¯ight-®rst and echolocation-®rst theories and ableÐthat bats are monophyletic and ¯ight concluded that ``both theories are coherent arose prior to the split between megachirop- with current knowledge on echolocation and terans and microchiropterans. ¯ight.'' They failed to observe, however, that Given these inferences, it seems clear that only oneÐthe ¯ight-®rst theoryÐis realisti- the origin of sophisticated echolocation must cally compatible with bat monophyly. As have occurred after the origin of ¯ight as discussed above under ``Relationships proposed by Norberg (1989, 1994). Sophis- Among Extant Lineages of Bats,'' chiropter- ticated echolocation is unknown in Mega- an monophyly is now supported by an enor- chiroptera, so it is unlikely that this system mous body of evidence, and it indeed rep- evolved earlier than in the basal microchi- resents one of the most strongly supported ropteran lineage. According to Fenton et al. hypotheses in all of higher-level mammalian (1995) and Arita and Fenton (1997), if bats systematics (Luckett, 1980a, 1993; Wible are monophyletic and ¯ight arose only once, and Novacek, 1988; Kovtun, 1989; Adkins then the echolocation ®rst theory could be and Honeycutt, 1991, 1993, 1994; Mindell et adjusted to ®t. Fenton et al. (1995: 230) sug- al., 1991; Thewissen and Babcock, 1991, gested that 1993; Ammerman and Hillis, 1992; Bailey et In this case, echolocation was lost early in the Me- al., 1992; Kay et al., 1992; Novacek, 1992, gachiroptera and reappeared [later] on in the genus 1994; Stanhope et al., 1992, 1993, 1996; Rousettus. The echolocation signals in Rousettus are Beard, 1993; Honeycutt and Adkins, 1993; tongue clicks, not vocalizations, supporting the po- 110 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

sition that echolocation per se is not a common fea- wing-slap echolocation (Eonycteris), and la- ture of the Chiroptera. ryngeal echolocation (Microchiroptera) orig- We agree completely with the latter obser- inated as independent evolutionary events af- vation, but fail to see how this is consistent ter the divergence of Megachiroptera and with the echolocation-®rst theory. In our Microchiroptera. view, the fact that known megachiropteran The ¯ight-®rst theory as articulated by echolocators use systems different from that Norberg (1989, 1994) is not entirely com- seen in Microchiroptera (see footnote 8) pro- patible with the current phylogenetic evi- vides additional evidence that the most re- dence because it suggests that the gliding an- cestors of bats used a primitive form of la- cent common ancestor of extant bats did not ryngeal echolocation (short broadband vocal use echolocation. Because this ancestor was clicks) to help with orientation in their noc- presumably a ¯ying mammal, it follows that turnal, arboreal habitat. Again, our dif®cul- echolocation did not evolve ®rst, but rather ties with this idea stem from the absence of powered ¯ight did. Hill and Smith (1984) such echolocation in Megachiroptera as well and Rayner (1991b) observed that the obvi- as the absence of any evidence that echolo- ous bene®ts of echolocation for nocturnal cation predated the evolution of ¯ight in the orientation and obstacle detection are so lineage leading to bats. Many authors (e.g., great that it seems highly unlikely that me- Jepsen, 1970; Fenton, 1974b; Hill and Smith, gachiropterans would have lost such an adap- 1984; Padian, 1987; Norberg, 1989, 1994; tion only to redevelop it in another form. The Speakman et al., 1989; Speakman, 1993; Al- success and broad distribution of Rousettus, tringham, 1996) have suggested that a prim- as well as the continued use of echolocation itive form of echolocation is probably ple- by frugivorous and nectarivorous microchi- siomorphic for bats because this sort of echo- ropterans (i.e., phyllostomids), indicate that location appears today among small noctur- echolocation can be advantageous even for nal mammals, including many lipotyphlan phytophagous forms (Hill and Smith, 1984; insectivores (e.g., Sorex, Blarina, Hemicen- Rayner, 1991b; Speakman, 1993). We agree tetes, Echinops, Microgale, Centetes; Gould with these authors, and see no reason to pos- et al., 1964; Gould, 1965; Sales and Pye, tulate a loss and reevolution of echolocation 1974; Buchler, 1976; Tomasi, 1979; Foreman in the megachiropteran lineage, especially and Malmquist, 1988) and at least one rodent since a more parsimonious explanation is (Rattus; Rosenzweig et al., 1955; Chase, availableÐthat ¯ight evolved in a common 1980; Henson and Schnitzler, 1980). Al- ancestor of all bats and sophisticated echo- though many workers have assumed that bats location arose later, early in the microchirop- evolved from some branch of what we now teran lineage. recognize as Insectivora (e.g., Jepsen, 1970; The primitive types of echolocation seen Hill and Smith, 1984; Kovtun, 1989), recent in Rousettus and Eonycteris apparently phylogenetic studies of interordinal relation- evolved independently of one another and of ships of mammals have not generally sup- echolocation in Microchiroptera. This inter- ported a close relationship between bats and pretation is supported by the most recent lipotyphlan insectivores, nor between bats phylogenies of Megachiroptera derived from and other small nocturnal mammals such as morphological data, DNA hybridization data, rodents (Wible and Novacek, 1988; Bailey et and 12S rRNA-valine tRNA mitochondrial al., 1992; Novacek, 1992, 1994; Stanhope et gene sequences (Springer et al., 1995; Hollar al., 1992, 1993, 1996; Adkins and Honey- and Springer, 1997). Rousettus and Eonyc- cutt, 1993, 1994; Simmons, 1993a, 1994, teris do not form a clade to the exclusion of 1995; Szalay and Lucas, 1993; Wible, 1993; other megachiropterans in any trees, and nei- Simmons and Quinn, 1994; Vrana, 1994; Al- ther occupies a basal position within Mega- lard et al., 1996; Miyamoto, 1996; Porter et chiroptera (Springer et al., 1995; Hollar and al., 1996). The fact that some living insecti- Springer, 1997). Accordingly, the most par- vores and rodents use a primitive form of simonious interpretation is that orientation echolocation is thus only slightly more rele- using tongue-click echolocation (Rousettus), vant to our understanding of bat evolution 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 111 than is the observation that some birds (e.g., tion, while the echolocation system of micro- oilbirds [Steatornis] and cave swiftlets [Col- chiropterans would represent a major adap- locialia, Aerodramus]) also use echolocation tive shift from a vision-based to an auditory- (Grif®n, 1953; Novick, 1959; Medway, based system (Greenwald, 1990, 1991; Sim- 1959, 1967; Cranbrook and Medway, 1965; mons et al., 1991). This alternative was Grif®n and Suthers, 1970; Fenton, 1975; discussed and dismissed by Pettigrew et al. Henson and Schnitzler, 1980). These cases (1989: the ``blind cave bat'' hypothesis9), but illustrate that a limited form of echolocation can (and has) evolved more than once, and 9 Pettigrew et al. (1989: 548) described this hypothesis that such systems may be employed effec- as follows: tively by small nocturnal mammals and vo- lant organisms. These examples also provide Suppose that all early bats had the primate features some insight into how such echolocation sys- that we have described in the living megabats, but that the microbats lost these features subsequently, tems work. However, in the absence of a sis- perhaps as they passed through an evolutionary bot- ter-group relationship between one of these tleneck involving the cavernicolous niche. We have taxa and Chiroptera, they provide no evi- called the this the ``blind cave bat'' scenario in ref- dence concerning the abilities of the earliest erence to the ease with which some cavernicolous bats. vertebrates appear to be able to ``lose'' visual capa- bilities.'' VISION AND THE EVOLUTION One of the principal arguments that Pettigrew et al. OF ECHOLOCATION (1989) used in dismissing bat monophyly and the ``blind cave bat'' scenario was based on an incorrect interpre- Presently, there are essentially two com- tation of the fossil record. They argued that the known peting hypotheses regarding the relationships ages of the earliest fossils referred to Microchiroptera of bats to other mammalian lineages: (1) the (Early Eocene) and Megachiroptera (Middle Oligocene ``Archonta'' hypothesis, which supports the at the time of that publication) refuted the idea that these idea that bats, dermopterans, primates, and groups might be sister taxaÐand precluded the possi- treeshrews are closely related (i.e., they all bility that visual specializations might have predated echolocationÐbecause ``the origin of microbats was belong to a monophyletic group Archonta); earlier than that of megabats, by at least 20 Ma'' (Pet- and (2) the ``non-Archonta'' hypothesis, tigrew et al. 1989: 549). A similar agrument was pre- which places bats well outside the primate± sented by Rayner (1991b). Unfortunately, these argu- dermoptera clade as the sister-group to one ments were based on ¯awed reasoning. Paleontologists or more other eutherian orders (e.g., Artiod- clearly recognize that the earliest appearance in the fos- actyla, Carnivora). Broadly speaking, the for- sil record provides only a minimum age for any given mer hypothesis is strongly supported by mor- lineage, not an indication of its actual time of origin phological and some molecular evidence, (Schaeffer et al., 1972; Novacek and Norell, 1982; Baker while the majority of the molecular evidence et al., 1991b; Norell, 1992; Norell and Novacek, 1992a, supports alternatives that we have lumped 1992b; Huelsenbeck, 1994; Padian et al., 1994; Smith, 1994; Flynn, 1996; Novacek, 1996; Benton and Hitchin, under the latter hypothesis (see discussion 1997). Although each lineage of a pair of sister-taxa under ``Outgroups''). Although the identity must be the same age since they originated in a single of the sister-group of bats has not been sat- speciation event, it is rare that sister-taxa appear simul- isfactorily resolved, the possibility that bats taneously in the fossil record (Norell, 1992; Huelsen- may belong to a monophyletic Archonta rais- beck, 1994; Padian et al., 1994; Smith, 1994; Benton es the interesting possibility that presence of and Hitchin, 1997). Typically, one taxon of a pair of a well developed visual system is primitive sister-taxa has a longer record than the other due to the for bats, having been inherited from their ar- stochastic nature of fossil preservation and discovery, or chontan ancestors (Greenwald, 1990, 1991; to differences in ecology or distribution that affect the Baker et al., 1991b; Simmons et al., 1991). chances of preservation and recovery of fossils (Norell, 1992). Information from phylogenies is now routinely If so, the well developed visual orientation used to reconstruct ``ghost lineages,'' those parts of evo- system of Megachiroptera (which strongly lutionary lineages that must have existed but for which resembles that of tree shrews, primates, and we have no fossils presently (Norell, 1992; Padian et al., dermopterans; Pettigrew et al., 1989) would 1994; Smith, 1994; Flynn, 1996; Novacek, 1996). A represent a retention of the ancestral condi- good ®t between stratigraphic occurrence and cladistic 112 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 clearly deserves further discussion in light of Suthers and Wallis (1970: 1170±1171) the strong evidence that now supports bat studied eye morphology in six microchirop- monophyly. terans (Phyllostomus hastatus, Anoura geof- Even if bats are not related to archontan froyi, Carollia perspicillata, Desmodus ro- mammals, we suspect that the ®rst bats relied tundus, Myotis sodalis, and Pipistrellus su- on vision for nocturnal orientation and ob- b¯avus) and concluded that stacle detection. Vision is clearly of central importance in Megachiroptera, but is also The eyes of these echolocating bats appear in some ways to be well adapted for nocturnal vision. The used by microchiropterans under many con- most salient of these adaptations are the very large ditions. Microchiropterans use vision for ob- corneas and the presence of relatively densely packed stacle avoidance (Bradbury and Nottebohm, rod retinae. In these respects the Microchiroptera 1969; Chase and Suthers, 1969), and many studied compare favorably with other non-echolocat- ing nocturnal mammals and contrast sharply with di- use vision for predator surveillance at the urnal mammals such as Man. roost site (Suthers, 1970, 1978; Suthers and Wallis, 1970; Vaughan, 1970c; Bradbury and The optical properties of microchiropteran Emmons, 1974). Vision has been shown to eyes suggest that vision is used for detecting play a role in escape responses (e.g., in chos- objects beyond the relatively short range of ing an escape route when threatened) in rep- echolocation (Suthers and Wallis, 1970; resentatives of several microchiropteran fam- Chase, 1972; Suthers, 1978; Fenton, 1985; ilies (Davis and Barbour, 1965; Chase, Pettigrew et al., 1988). Microchiropterans re- 1981), and it may be crucial for long-dis- duce the frequency of repetition of echolo- tance homing (Davis, 1966; Mueller, 1966, cation calls with increasing light levels, and 1968; Williams et al., 1966; Williams and may ultimately cease call production when Williams, 1967, 1970; Suthers, 1970; Childs there is adequate light for orientation by vi- and Buchler, 1981). sion (Bell, 1982b; Fenton, 1985). At least some gleaning microchiropterans (e.g., Ma- crotus californicus) apparently use vision to rank (relative time of splitting) is seen in some groups locate their prey (Bell, 1985). However, ae- of mammals, mostly large-bodied hervivorous forms (Norell and Novacek, 1992a, 1992b; Novacek, 1996). rial prey capture apparently requires echolo- 10 For smaller bodied forms (e.g., primates and rodents), cation in most or all microchiropterans. Mi- the ®t is poor at best; much of the history of these groups crochiropteran bats are routinely heard to is not adequately preserved in the fossil record, so long make feeding buzzes (terminal phase echo- ghost lineages are necessary to explain temporal patterns location calls), even when foraging in high- (Norell and Novacek, 1992a, 1992b; Novacek, 1996). light situations (Pettigrew, 1988; Rydell, This is most likely also true of bats, most of which are 1992). In general, it seems that echolocation small and live in habitats that are not condusive to fossil is probably superior to vision for obtaining preservation (e.g., tropical rainforests). Given bat mono- information about fast-moving insect-sized phyly, Microchiroptera and Megachiroptera must be prey, while vision is more useful than echo- equally old, dating to at least the Early Eocene. In turn, their sister-taxon (e.g., Dermoptera) must be at least location for long-range obstacle avoidance. slightly older. Support for this hypothesis does not re- Although visual acuity is modest in micro- quire that we have a complete chronicle of these lineages preserved in the fossil record. However, we do expect 10 Pettigrew (1988) reported that Craseonycteris uses that continued discoveries of fossils might ®ll in the gaps vision for aerial prey capture, often omitting the terminal and decrease the lengths of ghost lineages that must be phase feeding buzz when there is adequate light to see postulated. This is just what has happened in the case their prey. However, these observations must be viewed of bats. Since the publication of Pettigrew et al. (1989) with caution, as the echolocation calls of Craseonycteris and Rayner (1991b), new fossil discoveries in Thailand are high in frequency and the feeding buzz may have have brought the earliest known records of the relevant been present but not detected (B. Fenton, personal com- groups into closer agreement: the fossil record of Me- mun.; E. Kalko, personal commun.). Factors such as di- gachiroptera and Dermoptera now extends back to the rectionality of the echolocation call and the microphone, Late Eocene (Ducrocq et al., 1992, 1993), considerably distance between the bat and the microphone, and fre- reducing the length of ghost lineages that must be pos- quency sensitivity of the microphone may signi®cantly tulated given the phylogeny supported in the current affect the results of recording sessions, particularly un- study. der ®eld conditions (E. Kalko, personal commun.). 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 113 chiropterans, in many species it is compara- behaviors were present simultaneously. In ble to that of nocturnal muroid rodents (e.g., the case of a transformation from visual ori- Rattus, Peromyscus) and marsupials (e.g., entation to echolocation, we suggest that the Didelphis; Suthers, 1966; Rahmann, 1967; early microchiropteran lineage passed through Suthers et al., 1969; Chase, 1972; Manske a period during which they oriented princi- and Schmidt, 1976; Bell, 1982b; Fenton, pally using vision, but also used echolocation 1985; Bell and Fenton, 1986; Pettigrew, to provide supplementary information about 1988; Pettigrew et al., 1988). In both Micro- their surroundings. This possibility was ®rst chiroptera and Megachiroptera, visual acuity suggested by Pettigrew (1988: 649), although deteriorates with decreasing light levels more in the context of bat diphyly: slowly than in humans, so bats can usually see better than humans in dim light; thresh- I . . . think it unlikely that sonar would have been suf®ciently developed in the ®rst microbat to enable olds are similar to those reported for owls much useful guidance toward an airborne insect. The (Neuweiler, 1967; Suthers, 1970; Manske short range of ultrasound pulses makes it unlikely that and Schmidt, 1976; Fenton, 1985; Bell and it was an echo from an insect that ®rst enticed our Fenton, 1986). We surmise that the earliest early microbat off its branch, as does the required neural processing which seems unlikely to have been bats had at least the same visual capabilities suf®ciently sophisticated on the ®rst try. Vision seems as nocturnal rodents and marsupials, and may a more likely candidate to provide the appropriate have been much better equipped (i.e., if they resolution and range, just as it does in many living possessed the derived visual systems seen in microbats today. The primary role of sonar would primates, dermopterans, and megachiropter- then have been in the detection of obstacles, partic- ularly in the cave roost where there is no alternative ans; for a summary see Pettigrew et al., sensory channel and where three other ¯ying verte- 1989). Reliance on the visual system alone brates have sought shelter, independently inventing for orientation probably precluded aerial in- sonar on each occasion (swiftlets, oilbirds and rou- sectivory and ¯ight within cluttered environ- sette megabats). If this is correct, then the use of so- nar for insect capture occurred as a modi®cation of ments (e.g., within dense vegetation and the avoidance system after ¯ight was achieved. many subcanopy habitats). Nevertheless, these early bats would have bene®ted from Indeed, adequate vision may have been a many of the advantages of powered ¯ightÐ necessary prerequisite for the evolution of rapid, energy-ef®cient transportation over echolocation in bats. The sensory range of relatively long distances and an increased echolocation is relatively short (e.g., Suthers, foraging radius (Smith, 1977; Norberg, 1970, 1978; Grif®n, 1971; Fenton, 1980, 1986a, 1989, 1994; Rayner, 1986; Scholey, 1982a, 1984, 1994a; Kick, 1982; Lawrence 1986; Norberg and Rayner, 1987; Thomas, and Simmons, 1982), and quickly moving, 1987), as well as easy avoidance of predators ¯ying organisms may require more infor- (Pomeroy, 1990). mation about distant obstacles than can be This hypothesis raises the linked questions obtained from echolocation alone (Suthers of how and why one fully functional orien- and Wallis, 1970; Chase, 1972; Fenton, tation system (vision-based orientation) 1985; Pettigrew et al., 1988). This may be would be exchanged during evolution for an- especially true in the warm, moist air of the other (echolocation). While such a transfor- tropics, where atmospheric attenuation of ul- mation is admittedly unlikely if one postu- trasonic frequencies is greatest (Grif®n, lates a simple one-step process, it makes 1971; Suthers, 1978; Lawrence and Sim- considerably more sense when viewed as a mons, 1982). Particularly when orienting multistep process that allowed the micro- over long distances (e.g., when commuting chiropteran lineage to successfully invade an to and from foraging areas), vision appears enormous yet empty set of ecological niches to be crucial for obstacle avoidance and land- for nocturnal aerial insectivores. In a rarely mark recognition (Davis, 1966; Mueller, cited paper on the evolution of feeding strat- 1966, 1968; Williams et al., 1966; Williams egies in bats, Gillette (1975) argued that ma- and Williams, 1967, 1970; Suthers, 1970). jor adaptive shifts might have evolved We ®nd it hard to imagine how echoloca- through what he termed ``duality''Ða period tionÐwhich is essentially a short-range sen- during which both the primitive and derived sory systemÐcould have evolved in a group 114 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 of ¯ying organisms in the absence of vision uations, including female contact, precopu- for longer-range obstacle detection. lation, and hostile male±male interactions Speakman (1993) argued that an early bat (Nelson, 1964; Fenton, 1985). Similar calls which relied on a combination of vision and are sometimes used by microchiropteran echolocation would be a ``sensory general- bats, often in comparable social situations ist'' that might be at a selective disadvantage (e.g., hostile interactions between males; compared with sensory specialists that used Porter, 1979; Fenton, 1985). Young bats of either vision or echolocation. This hypothesis many species (both megachiropterans and was offered as a possible explanation for the microchiropterans) apparently use similar absence of echolocation in Megachiroptera, calls when they are isolated from their moth- but might also apply to the earliest micro- ers (Fenton, 1985). Together, these observa- chiropterans. However, we do not ®nd this tions suggest that broadband or multihar- hypothesis compelling, at least as an argu- monic clicks or buzzes were present in the ment against a transformation from vision to vocal repertoire of the earliest bats, and thus echolocation as the primary sensory system were available as a behavioral substrate from in early Microchiroptera. Early members of which echolocation calls could evolve. the microchiropteran lineage would have Early members of the microchiropteran faced no competition from sensory special- lineage, ¯ying and using vision as well as a ists in echolocation, because specialized ae- primitive echolocation system such as de- rial echolocatorsÐtheir descendantsÐhad scribed above, would have had access to not yet evolved. Rather than being at a se- many habitats and food sources but would lective disadvantage when compared with vi- have been poorly equipped to ¯y in cluttered sion specialists (i.e., megachiropterans), we spaces or capture ¯ying insects (Norberg, have every reason to believe that early mi- 1994). Perfection of echolocation for detect- crochiropterans would have immediately ing, tracking, and assessing airborne prey gained several advantages when they ®rst be- would have increased the foraging options gan to evolve echolocation, including an in- open to these bats, especially given the ab- creased ability to detect obstacles at short sence of competition (Fenton, 1974a, 1974b, range (particularly in low-light situations) 1980, 1982a, 1984, 1994a; Norberg, 1994; and an increased ability to utilize caves as Speakman, 1995). Nocturnal ¯ying insects roosting places. offer an abundant food supply, so much so Previous authors have suggested that the that there is little evidence that prey abun- primitive echolocation system within Micro- dance is a limiting resource for most micro- chiroptera may have consisted of short chiropteran aerial insectivores,11 at least broadband or multiharmonic clicks produced in the larynx (Pye, 1980; Simmons, 1980; 11 Possible exceptions to this general pattern may in- Simmons and Stein, 1980; Fenton, 1984; clude aerial insectivores that summer in high latitude Norberg, 1994; Fenton et al., 1995; Arita and habitats, where the nights are very short and peak aerial insect availability occurs before it gets dark (Rydell, Fenton, 1997). Such a system probably 1992; Speakman, 1995). It has recently been shown that evolved from vocalizations used for other some bats living under these conditions frequently fail purposes, most likely intraspeci®c commu- to meet their energy requirements, instead using torpor nication (Fenton, 1984, 1985; Novacek, to balance their daily energy budgets (Kunz, 1980; Kurta 1985a). This hypothesis is supported by ob- et al., 1987, 1989; Speakman and Racey, 1987; Audet servations that some extant microchiropter- and Fenton, 1988). Speakman (1995) suggested that it ans use echolocation calls simultaneously for is nocturnality (and lack of adequate prey resources at orientation and communication, while others night) that forces the use of torpor in these animals. have distinct calls that are apparently used Although this may be true, we suggest that it is the avail- only in a social context (MoÈhres, 1967a, ability of torpor that has permitted these species to ex- pand their summer ranges beyond the limit of most other 1967b; Habersetzer, 1981; Miller and Degan, bats. In tropical and subtropical environments (where 1981; Brown et al., 1983; Fenton, 1984, most insectivorous microchiropteans live), nights are 1985, 1994b; Guppy et al., 1985). Megachi- longer and prey densities are generally high, thus prey ropteran bats use short broadband or multi- availability is probably not a limiting factor under nor- harmonic FM calls in a variety of social sit- mal conditions (e.g., Fenton et al., in press). 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 115 those in tropical and subtropical habitats turning echoes (Schuller and Pollack, 1976; (Fenton, 1980; Fenton et al., in press). The Bruns, 1979; Neuweiler et al., 1980; Schnitz- potential advantages of evolving adaptations ler and Henson, 1980; Vater et al., 1985; to exploit nocturnal ¯ying insects as a food Emde and Schnitzler, 1986, 1990; Schnitzler, source are obvious. 1987; Vater, 1987; Neuweiler, 1990; Obrist Fenton et al. (1995) noted that transfor- et al., 1993; Fenton, 1994a; Fenton et al., mation to the sophisticated echolocation sys- 1995). tem used by extant microchiropterans to de- Most extant microchiropterans are low- tect, track, and evaluate ¯ying insects re- duty-cycle echolocators; high-duty-cycle quired production of longer, stronger (high- echolocation is used only by rhinolophines, intensity) tonal signals and concomitant hipposiderines, and one mormoopid (Pter- evolution of a system to avoid self-deafen- onotus parnellii; Novick, 1958a, 1962, ing. Among extant bats, two such systems 1963a, 1963b, 1977; Novick and Vaisnys, are known: separation of the pulse and echo 1964; Schnitzler, 1970b, 1973, 1987; Fenton, in time (the ``low-duty-cycle'' approach), 1974a, 1974b, 1980, 1982a, 1982b, 1984, and separation of the pulse and echo in fre- 1990, 1994a, 1995; Schuller et al., 1975; quency rather than time (the ``high-duty-cy- Schuller and Pollack, 1976; Schuller, 1977; cle'' approach; Fenton, 1994a, 1995; Fenton Neuweiler et al., 1980; Schnitzler and Hen- et al., 1995). Low-duty-cycle echolocation son, 1980; Simmons and Stein, 1980; Fenton involves short signal pulses with relatively and Bell, 1981; Neuweiler, 1984, 1989, long gaps between them; high-duty-cycle 1990; Neuwieler and Fenton, 1988; Kalko echolocation involves longer pulses and and Schnitzler, 1989, 1993; Lancaster et al., shorter gaps, with pulses longer than the gaps 1992; Surlykke et al., 1993; Fenton et al., between them (Fenton, 1994a, 1995; Fenton 1995; Grinnell, 1995). It seems clear that et al., 1995). Low-duty-cycle bats prevent low-duty-cycle echolocation is primitive for self-deafening by freezing movement of the extant Microchiroptera based on optimiza- middle-ear ossicles through contraction of tion of echolocation strategies on our phy- the middle-ear muscles during pulse emis- logenetic tree (®g. 36). Optimization indi- sion and by reducing auditory sensitivity in cates that the high-duty-cycle approach the inner ear through changes in the sensory evolved twice, once in the lineage leading to cells along the basilar membrane (Henson, 1964, 1965, 1966, 1967a, 1967b, 1970; Jen Rhinolophidae and once within the genus and Suga, 1976). Returning echoes are re- Pteronotus. This con®rms previous hypoth- ceived in the gaps between pulse emissions, eses (e.g., Pye, 1980; Simmons, 1980; Sim- when the middle-ear muscles relax and au- mons and Stein, 1980; Fenton et al., 1995) ditory sensitivity is maximized. High-duty- that suggested independent origins for the cycle bats produce long constant frequency use of long CF signals and Doppler compen- (CF) echolocation signals that overlap with sation in these two groups. returning echoes. These bats utilize the The transformation from primitive to so- Doppler effect, which shifts the frequency of phisticated low-duty-cycle laryngeal echolo- returning echoes to a frequency different cation likely took place in several stages from that of the original pulses (Schnitzler, (Fenton, 1984), probably facilitated by the 1970b, 1973, 1987; Schuller et al., 1975; mechanical coupling of ¯ight and ventilation Schuller and Pollack, 1976; Schuller, 1977; discussed previously (Speakman et al., 1989; Neuweiler et al., 1980; Schnitzler and Hen- Rayner, 1991b; Speakman and Racey, 1991; son, 1980; Simmons and Stein, 1980; Emde Speakman, 1993). If there was indeed ``no and Schnitzler, 1986, 1990; Neuweiler, 1989, cost of echolocation for bats in ¯ight'' 1990; Grinnell, 1995). Self-deafening is re- (Speakman and Racey, 1991: 421), or (more duced because the emitted pulse is dominat- realistically) a relatively low cost, it is easy ed by frequencies outside the acoustic fovea to imagine how this system might have (zone of maximum hearing sensitivity), evolved quickly. However, re®nement of this while both the external and inner ears are systemÐand increased reliance on aerial in- sharply tuned to the frequencies of the re- sectivoryÐapparently brought with it an im- 116 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 portant series of evolutionary constraints, in- in vertebrates, Harvey and Krebs (1990: 145) cluding limitations on body size and repro- noted that duction (Barclay and Brigham, 1991; Bar- If overall size [of the brain] is constrained . . . it may clay, 1994, 1995; Jones, 1994; Arita and well be that specialized enlargement of one region Fenton, 1997). has to be associated with reduction in size of another Maximum body size may be limited in mi- .... In other words, there may be trade-offs in the crochiropteran aerial insectivores for the sev- evolutionary specialization of the brain. eral reasons, including (1) the small size of Cooper et al. (1993a: 340) similarly conclud- available prey (McNab, 1969; Black, 1974); ed that (2) the metabolic requirements for sustained ¯ight, particularly at night when thermal . . . non-visual systems may ``compete'' with neurons of the visual system for available metabolites .... gliding and soaring are not possible (Rayner, Although the use of the expression competition to 1981); (3) the mechanics of ¯ight and aerial describe these evolutionary events is somewhat unex- prey capture (Norberg, 1986a, 1994; Norberg plicit, the relative expansion of one cerebral structure, and Rayner, 1987); (4) the effective range of restrained within the con®nes of a braincase of lim- echolocation calls (relatively short) and the ited volume, may depend upon concomitant decrease elsewhere, notwithstanding the problem of maintain- concomitant need to be maneuverable and ing the most ef®cient and adaptive neuronal popula- agile enough to catch small prey at short tions. range (Barclay and Brigham, 1991); and (5) Small body size places limitations on the size the coupling of ¯ight and echolocation, of neuron populations and may constrain the which may limit the ability of large bats metabolic energy available for brain func- (which have lower wing-beat frequencies and tions (Martin, 1981; Armstrong, 1983; Wil- thus lower call-repetition rates) to detect liams and Herrup, 1988; Deacon, 1990a, small ¯ying prey (Jones, 1994, although see 1990b; Cooper et al., 1993a). Neurons have Heller, 1995). As a probable result of these high energy requirements; the brain can con- constraints, microchiropteran aerial insecti- sume up to 20% of circulating oxygen and vores are typically very small (Ͻ30 g adult glucose even though the brain constitutes body weight), with only four extant species only a small fraction of total body weight weighing more than 100 g12 (McNab, 1969; (Kety and Schmidt, 1948; Martin, 1981; Black, 1974; Fenton and Fleming, 1976; Armstrong, 1983; Williams and Herrup, Krazanowski, 1977; Barclay and Brigham, 1988; Cooper et al., 1993a). Reviewing re- 1991; Arita and Fenton, 1997). Within this lationships between body size, brain size, size range, most larger aerial insectivores and metabolic rates in small mammals, Coo- seem to be limited to relatively large insect per et al. (1993a: 339) summarized their ®nd- prey (e.g., moths and large beetles; Vaughan, ings as follows: 1977), while smaller bats can exploit either large or small prey (Barclay and Brigham, These arguments emphasize the conclusion that in 1991; Arita and Fenton, 1977). small animals, such as moles and bats, the fraction of Returning to the problem of explaining the metabolism devoted to neurons is great and bioener- getic limits become critical. Supplying the brain with evolution of sophisticated echolocationÐ suf®cient oxygen is a challenge to the body for sur- speci®cally our hypothesis that this form of vival. Super¯uous neurons . . . are thus strongly se- orientation replaced a vision-based system in lected against and their reduction can contribute to microchiropteran batsÐwe note that small the animal's ®tness by improving metabolic ef®ciency body size may have precluded the retention (Ricklefs and Marks, 1984). of derived structures of the visual system In the early evolution of microchiropteran (such as those seen in Primates, Dermoptera, bats, it may not have been ef®cient or fea- and Megachiroptera) once neural modi®ca- sible to retain derived complex structures in tions associated with echolocation began to the visual system (e.g., laminated dorsal lat- evolve. In a review article on encephalization eral geniculate nucleus, large superior collic- ulus; Sanderson, 1986; Pettigrew et al., 1989) 12 Taphozous peli (Emballonuridae), Hipposideros and simultaneously provide the neurons and commersoni (Rhinolophidae), Scotophilus nigrita (Ves- metabolic energy necessary for processing pertilionidae), and Cheiromeles torquatus (Molossidae). increasingly complex auditory information 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 117

(e.g., in a large inferior colliculus; Aitkin, cats (Jacobs et al., 1984). Evolutionary 1986; Casseday and Pollak, 1988; Pollack changes in taxon-speci®c rates of neuron and Casseday, 1988; Pollak and Park, 1995). death in components of the visual system From a dual orientation system employing during development may thus affect the size both vision and limited echolocation, we sur- and interconnections of visual centers in mise that the auditory system of microchi- adults. As auditory information became in- ropteran bats simply ``outcompeted'' the vi- creasingly important to early members of the sual system for the resources available. microchiropteran lineage, we suspect that Evolutionary reduction in brain centers re- brain centers involved in processing visual sponsible for processing visual information information were reduced and/or simpli®ed could have occurred through a variety of on- through developmental changes such as these. togenetic changes, including truncation of de- The hypothesized evolutionary reduction velopment of visual centers at increasingly of the visual system in microchiropteran bats early stages. This hypothesis is supported by is not a unique event; similar transformations observed similarities between the relatively are postulated to have occurred independent- simple adult dorsal lateral geniculate nucleus ly in other mammalian groups that depend (dLGN) in microchiropteran bats (Pettigrew little on vision, including moles (Johnson, et al., 1989) and early developmental stages 1954; Lund and Lund, 1965, 1966; Suthers of the dLGN in Tupaia (Casagrande and and Bradford, 1980; Kudo et al., 1988, Brunso-Bechtold, 1985). If development of 1991), mole rats (Bronchti et al., 1991; Coo- the dLGN in Tupaia was arrested at an early per et al., 1993a, 1993b), and odontocete stage, the adult dLGN would resemble that of whales (Jacobs et al., 1975; Sanderson, 1986; a microchiropteran batÐand eight of the de- Deacon, 1990b). In this context, it is inter- rived traits de®ned by Pettigrew et al. (1989: esting to note that odontocetes are the only 512) would be reversed as the result of a sin- other group of mammals known to use so- gle ontogenetic change. The same would pre- phisticated echolocation comparable to that sumably hold true for the dLGN of megachi- of microchiropterans (e.g., Norris et al., ropteran bats, which is similar to that of Tu- 1961; Norris, 1968; Purves and Pilleri, 1983; paia in most respects (Pettigrew et al., 1989). Pilleri, 1983, 1990). Extensive reversals in Another mechanism that might contribute the visual system are also commonly accept- to reduction in visual centers in the brain is ed as explanations for the morphology of changes in the magnitude of neuron death blind snakes, gymnophionan amphibians, during early development. Neuron death is a and blind cave populations of ®shes (e.g., normal part of brain development in many Wilkens, 1971; Halpern, 1973; RepeÂrant et vertebrates, including all mammals (Finlay et al., 1987; Voneida and Sligar, 1976; Clair- al., 1987; Williams and Herrup, 1988). Not ambault et al., 1980; Fritzsch et al., 1985; surprisingly, the magnitude of neuron death Himstedt and Manteuffel, 1985). Similar within homologous neuron populations var- evolutionary reductions have taken place in ies phylogenetically (Finlay et al., 1987; Wil- other sensory systems in some mammals, liams and Herrup, 1988). For example, nor- such as reduction of the olfactory system in mal development includes no death of retinal whales (OelschlaÈger and Buhl, 1985; Oelsch- ganglion cells in ®sh and amphibians (Wil- laÈger, 1989) and the vomeronasal system in son, 1971; Easter et al., 1981), but 40% of catarhine primates (Meredith, 1991). As not- retinal ganglion cells normally die in chick- ed above, these sorts of changes appear to ens (Rager, 1980), 60±70% die in rats and represent modi®cations for ef®cient use of primates (Rakic and Riley, 1983; Crespo et cellular and metabolic resources. When a al., 1985; Provis et al., 1985), and 80% die system is no longer of critical importance, in cats (Williams et al., 1986). Neuron death regression or reductionÐ-which may be in- is known to be a normal part of development terpreted as reversal in a phylogenetic con- of the dLGN in primates (Williams and Rak- textÐtakes place to avoid wasting cells and ic, 1988) and has been shown to change the metabolic energy that may be better spent proportion of contralateral (crossed) versus elsewhere (Cooper et al., 1993a). ipsilateral (uncrossed) retinal projections in The evolution of ¯ight and echolocation 118 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 required modi®cations of many anatomical ment of the basal turn of the cochlea). Loos- and behavioral systems, and this process un- ening of the attachment of the periotic to the doubtedly took place in several stages. To basicranium evolved at this level or just sub- summarize, we hypothesize that ¯ight sequently (prior to the divergence of Ar- evolved ®rst, prior to the divergence of Me- chaeonycteris). It is interesting to note that gachiroptera and Microchiroptera. The ®rst all of these modi®cations occurred before bats most likely used vision for orientation achievement of what we described as ``great and obstacle detection in their arboreal/aerial enlargement of the cochlea'' (i.e., to the size environment. The evolution of ¯ight was lat- range of most extant microchiropterans), er followed by the origin of low-duty-cycle which evolved somewhat later, just prior to echolocation in basal members of the micro- the divergence of Hassianycteris (®g. 40). chiropteran lineage. This system, which was Postcranial modi®cations that may have in- probably derived from vocalizations origi- creased ef®ciency of the ventilation system nally used for intraspeci®c communication, (e.g., increased breadth of posterior laminae was most likely simple at ®rst, permitting on the ribs) evolved in the microchiropteran orientation and detection of obstacles but not lineage sometime between the divergence of detection or tracking of airborne prey. How- Icaronycteris and the divergence of Palaeo- ever, due to the mechanical coupling of ven- chiropterx. Other derived features potentially tilation and ¯ight, energy costs of echoloca- related to the echolocation system also tion to ¯ying bats were low, and the bene®ts evolved at some point in the early microchi- of aerial insectivory quickly led to develop- ropteran lineage. These include a phanero- ment of a more sophisticated low-duty-cycle cochlear cochlea (probably associated in echolocation system capable of detecting, some fashion with cochlear expansion) and a tracking, and assessing airborne prey. The deep, constricted fossa for m. stapedius (per- need for an increasingly derived auditory haps related to the system for avoidance of system, combined with limits on body size self-deafening). The former was minimally imposed by the mechanics of ¯ight, echolo- present before the divergence of Hassianyc- cation, and prey capture, may have resulted teris, the latter before the divergence of Pa- in reduction and simpli®cation of the visual laeochiropteryx. As noted above, missing system as echolocation became increasingly data for cochlear ossi®cation in Icaronycteris important. and Archaeonycteris and for stapedial fossa The theory presented above is consistent form in Archaronycteris and Hassianycteris with (and is indeed based on) bat monophyly make it impossible to exactly place these and the phylogeny proposed in the current transformations in the phylogenetic tree. study. When considered in a phylogenetic Even given uncertainty about the relative context, the Eocene fossil bats Icaronycteris, timing of some transformations, our obser- Archaeonycteris, Hassianycteris, and Pa- vations con®rm the hypothesis that sophisti- laeochiropteryx are not informative about the cated echolocation evolved in a stepwise origin of ¯ight (which preceded diversi®ca- fashion in the early microchiropteran lineage. tion of the entire microchiropteran lineage), However, evolutionary changes in this sys- but they do provide some information con- tem certainly did not cease with the origin of cerning early steps in the acquisition of so- the microchiropteran crown group. Consid- phisticated echolocation. As noted above, erable cochlear size variation exists among some of the earliest morphological transfor- living clades of Microchiroptera, with most mations in the lineage leading to extant Mi- extant microchiropterans (and all high-duty- crochiroptera (those that took place before cycle echolocators) having a cochlea even the divergence of Icaronycteris) included larger than those seen in Palaeochiropteryx changes in features associated with the pro- and Hassianycteris (Habserstezer and Storch, duction of echolocation calls (enlargement of 1992). Reduction in cochlear size (back to the cranial tip of the stylohyal), transmission the ``moderately enlarged'' condition) has of sounds through the middle ear (enlarge- apparently occurred in some lineagesÐin ment of the orbicular apophysis), and some Mystacinidae, within Phyllostomidae, and ®ne-tuning of the inner ear (initial enlarge- within Megadermatidae (Fig. 40). Signi®cant 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 119 changes (some associated with cochlear en- wings and of the minimum weight per unit largment) have also occurred in some lin- area that the wings must support during ¯ight eages with respect to (1) length, width, and (Findley et al., 1972; Norberg and Rayner, thickness of the basilar membrane, (2) length 1987; Arita and Fenton, 1997). Increased and height of the spiral ligament, (3) total wing loading requires greater ¯ight speeds to number of cochlear neurons, (4) size and dis- generate enough lift to remain airborne; tributions of hair cell populations, (5) size ¯ight speed is proportional to the square root and pattern of ¯uid spaces in the cochlea, and of wing loading (Findley et al., 1972; Nor- (6) form of the frequency map and ``acoustic berg and Rayner, 1987). High wing loading fovea'' (Pye, 1966a, 1966b, 1967; Henson, is typical of fast ¯iers; slow-¯ying, more ma- 1970; Bruns, 1979; Bruns et al., 1981, 1983± neuverable animals usually have lower wing 1984; Burda and Ulehlova, 1983; RuÈbsamen loading, with either lower body weight an- et al., 1988; Neuweiler, 1990). Most or all of d/or increased wing area (Findley et al., these changes may re¯ect modi®cations of 1972; Norberg and Rayner, 1987). the auditory system associated with different Aspect ratio (ϭ wing span2/wing area; echolocation and foraging strategies (Bruns Norberg and Rayner, 1987) describes the et al., 1981, 1983±1984; RuÈbsamen et al., overall shape of the wings by quantifying 1988; Neuweiler, 1990). their length relative to their chord (Findley et al., 1972; Norberg and Rayner, 1987; Arita FORAGING ECOLOGY OF EOCENE and Fenton, 1997). For any given body BATS weight and wing loading, high aspect ratio wings are subject to less drag and thus facil- One aspect of morphology and ecology itate greater ¯ight speeds than do lower as- not mentioned in the preceding discussion is pect ratio wings (Findley et al., 1972). How- the relationship between wing shape and for- ever, high aspect ratio wings also generate aging ecology. There is an extensive litera- less lift than do low aspect ratio wings, and ture on this topic (e.g., Revilliod, 1916; Betz, thus bats with high aspect ratio wings may 1958; Vaughan, 1959, 1966; Struthsaker, require greater wingbeat frequencies and air- 1961; Hartman, 1963; Farney and Fleharty, speeds to remain airborne (Findley et al., 1969; Fenton, 1972; Findley et al., 1972; 1972). Some bats with very high aspect ratio Kopka, 1973; Lawlor, 1973; Pirlot, 1977; wings (e.g., Molossus, Eumops) cannot gen- Smith and Starrett, 1979; Norberg, 1981, erate enough lift to become airborne from a 1986a, 1986b, 1987, 1994; Findley and Wil- ¯at surface (personal obs.), and thus usually son, 1982; Findley and Black, 1983; Al- drop from their elevated roosts to build up dridge, 1986; Baagùe, 1987), with many re- enough speed and lift to initiate ¯ight cent contributions integrating data on echo- (Vaughan, 1959). Low aspect ratio wings location call structure as well (e.g., Simmons generate considerable drag at higher speeds, et al., 1979; Neuweiler, 1984, 1989, 1990; but maximize lift at low speeds (Findley et Habersetzer, 1986; Aldridge and Rautenbach, al., 1972). 1987; Norberg and Rayner, 1987; Neuweiler A variety of different measures of the rel- and Fenton, 1988; Norberg, 1989, 1990, ative size and shape of the wing tip have 1994; Fenton, 1990; Arita and Fenton, 1997). been proposed, including the tip index (Fin- One outcome of this research has been iden- dley et al., 1972), alpha angle (Smith and ti®cation of a series of features of wing de- Starrett, 1979), tip length ratio (Norberg and sign that affect ¯ight performance and are Rayner, 1987), tip area ratio (Norberg and correlated with foraging strategies. Among Rayner, 1987), and tip shape index (Norberg the most important measures of these are and Rayner, 1987). All of these quantify var- wing loading, aspect ratio, and a variety of ious aspects of the relative size and shape of wing tip indices designed to measure the size the dactylopatagium (that portion of the wing and shape of the dactylopatagium (e.g., Fin- distal to digit V), which provides much of dley et al., 1972; Norberg and Rayner, 1987). the propulsion generated by the wing during Wing loading (ϭ body weight/wing area) ¯ight (Findley et al., 1972). provides a measure of the relative size of the The most comprehensive study of wing 120 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 design and foraging strategies in bats was ronycteris and Archaeonycteris have wings that of Norberg and Rayner (1987), who that resemble those of large rhinolophoids or summarized morphometric and behavioral the mean of vespertilionids (Habersetzer and data for more than 250 bat species. In addi- Storch, 1987, 1989). Concerning the combi- tion to the ratios mentioned above, they also nation of low aspect ratio and high wing considered total body weight (mass), wing- loading, Habersetzer and Storch (1989: 216) span, and wing area. They interpreted these noted that ``this parameter combination is data in the context of mechanical and aero- most likely for unspecialized ¯ight charac- dynamic models for different modes of ¯ight. teristics as can be found within the majority Norberg and Rayner (1987: 337) summa- of the vespertilionids (e.g., Myotis myotis).'' rized their results as follows: In contrast, Palaeochiropteryx was found to have a somewhat more specialized wing Some adaptive trends in bat wing morphology are clear from this analysis. Insectivores hunt in a range morphology (Habersetzer and Storch, 1987, of different ways, which are re¯ected in their mor- 1989; Habersetzer et al., 1994). Palaeochi- phology. Bats hawking high-¯ying insects have small, ropteryx tupaiodon was characterized as hav- pointed wings which give good agility, high ¯ight ing very low aspect ratio and very low wing speeds and low cost of transport. Bats hunting for loading, while both values were somewhat insects among vegetation, and perhaps gleaning, have very short and rounded wingtips, and often relatively higher in P. spegeli (Habersetzer and Storch, short, broad wings, giving good maneuverability at 1987, 1989). On the basis of these values, low ¯ight speeds. Many insectivorous species forage both taxa were found to be very similar to by `¯ycatching' (perching while seeking prey) and extant rhinolophine and hipposiderine spe- have somewhat similar morphology to gleaners. In- sectivorous species foraging in more open habitats cies (Habersetzer and Storch, 1987; Haber- usually have slightly longer wings, and hence lower setzer et al., 1994). This similarity led Ha- cost of transport. Piscivores forage over open stretch- bersetzer and Storch (1987, 1989) and Ha- es of water, and have very long wings giving low bersetzer et al. (1994) to conclude that Pa- ¯ight power and cost of transport, and usually long, laeochiropteryx was characterized by slow, rounded tips for control and stability in ¯ight. Car- nivores must carry heavy loads, and thus have rela- highly maneuverable ¯ight close to the tively large wing areas; their foraging strategies con- ground. In the case of P. tupaiodon, aerial sist of perching, hunting and gleaning, and wing capabilities may have included ¯ight close to structure is similar to that of insectivorous species and even within foliage with potentially with similar behavior. Perching and hovering nectar- ivores both have a relatively small wing area; this long-lasting hovering phases, as is seen today surprising result may result from environmental pres- in Hipposideros bicolor, a species with a sure for short wingspan or from the advantage of high comparable combination of aspect ratio and speed during commuting ¯ight; the large wingtips of wing loading values (Habersetzer and Storch, these bats are valuable for lift generation in slow 1987, 1989; Habersetzer et al., 1994). Wing ¯ight. parameters of Palaeochiropteryx spiegali Habersetzer and Storch (1987, 1989) and more closely resemble those of Hipposideros Habersetzer et al. (1994) capitalized on the speoris, a form that hunts near obstacles but observed relationships between wing form always stays in open airspace, never ¯ying and ¯ight behavior in extant bats to recon- amongst foliage (Habersetzer and Storch, struct the possible habits of Icaronycteris, 1987, 1989; Habersetzer et al., 1994). Archaeonycteris, Hassianycteris, and Pa- Habersetzer and Storch (1987) found that laeochiropteryx. Habersetzer and Storch both species of Hassianycteris are character- (1987) estimated wing loading, aspect ratio, ized by a high aspect ratio and high wing and tip index for a series of fossil specimens, loading. On the basis of comparisons with and compared these values with similar data Habersetzer's (1986) data from Indian bats, from an extant bat fauna from tropical India Habersetzer and Storch (1987, 1989) and Ha- (i.e., Habersetzer, 1986). They concluded that bersetzer et al. (1994) observed that Hassi- Icaronycteris and Archaeonycteris have a anycteris species are even more specialized relatively high wing loading and low aspect than high-¯ying tropical vespertilionids and ratio (Habersetzer and Storch, 1987, 1989; rhinopomatids, instead showing a greater re- Habersetzer et al., 1994). Based on compar- semblance to extant high-¯ying molossids isons with extant forms, they found that Ica- and emballonurids. As a result, they con- 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 121

cluded that Hassianycteris messelensis and calculated by Habersetzer and Storch (1987) H. magna ``can be considered to be distinctly may have been too high for several reasons adapted to ¯ight in free spaces'' (Habersetzer (e.g., use of wingspan length regression et al., 1994: 238). equations, straight rather than curved wing- Habersetzer and Storch (1987, 1989) and tips in the fossil reconstructions). Using re- Habersetzer et al. (1994) noted that all four vised wing reconstructions (®g. 42) and re- Eocene genera are characterized by a short vised estimates of body weight based on re- wing tip and a small dactylopatagium that gression equations for radius length against accounts for only about 37% of the entire total mass in recent bats, Norberg (1989) re- wing area. Among extant bats, they noted calculated wing loading and aspect ratio for that this morphology is seen in rhinolo- each of the fossil species, and additionally phoids. Habersetzer et al. (1994) concluded calculated wingtip length ratio, wingtip area that the Messel bats had succeeded in occu- ratio, and wingtip shape index for each tax- pying a diverse set of ecological niches by on. Comparisons with recent bats (®gs. 43, evolving variations of the rhinolophoid wing 44; data from Norberg and Rayner, 1987) led type, and suggested that this wing type (i.e, Norberg (1989: 204±205) to the following low aspect ratio with short wing tip) repre- conclusions regarding Icaronycteris, Ar- sents the primitive wing form for bats. chaeonycteris, Hassianycteris, and Palaeo- To further investigate the correlations de- chiropteryx: scribed above, Habersetzer and Storch Based on my reconstructions, the ancient bats ex- (1989) used multivariate methods developed amined here . . . had low aspect ratio and high or by Norberg and Rayner (1987) to remove the average wing loading . . . (that is, high or average effects of size (the ®rst principal component), wing loadings in relation to body size), as compared resulting in a plot of normalized wing load- with recent bats. This indicates that they had expen- sive and average to fast ¯ight. Their wingtips were ing (second principal component) versus nor- extremely short and with fairly large area, which are malized aspect ratio (third principal compo- adaptations for maneuverable ¯ight. Wingtip length nent). Although they used Norberg and Ray- ratios T1 (handwing length/armwing length) are all ner's (1987) method, Habersetzer and Storch between 0.93 and 1.09 ....T1 Ͻ 1 is rather unusual (1989) did not use their data, but instead among modern bats. The wing shape of the fossil species are similar to compared the fossil forms to Habersetzer's several recent pteropodids and phyllostomids and (1986) reference fauna (with known ¯ight some rhinolophids (Rhinolophus hipposideros and R. and foraging habits) from tropical India. Ha- ferrumequinum) and vespertilionids (Barbastella bar- bersetzer et al. (1989) interpreted the results bastellus, Lasiurus borealis, Rhogeesa tumida and some Myotis, Eptesicus and Pipistrellus species). Pa- of this multivariate analysis as supporting laeochiropteryx and Hassianycteris were also similar their previous conclusions about the ¯ight to Rhinopoma hardwickei ..., both in wing shape behavior of the fossil bats (see above). How- and with their extremely short wingtips. Rhinopom- ever, Icaronycteris, Archaeonycteris trigon- atids are found mainly in deserts and steppes. They odon, and Palaeochiropteryx tupaiodon ap- have been noted to forage for insects in open country as well as in open spaces around tree canopies, and peared to have very similar wing parameters their ¯ight has been described as swift, fast and un- based on the size-normalized analysis (at dulating (alternating ¯utters and glides)(see Norberg least in our judgment). This was not explic- & Rayner 1987 for references) . . . Hassianycteris itly discussed by Habersetzer et al. (1989); messelensis and H. magna had higher wing loadings and slightly higher aspect ratio than the other fossil however, they cautioned that their compari- species, and were thus faster ¯iers. They had though sons might be biased by the fact that they much lower aspect ratio than molossids and embal- calculated body weight of the fossil forms lonurids. They may have been foraging in rather open using a measure of wingspan, which intro- spaces, like rhinopomatids, or along vegetation like duced circularity into the analysis and might noctules and lasiurines. The other ancient bats in- cluded here have lower aspect ratio and lower wing have resulted in the observed clustering. loading. Because their short wings and low aspect Norberg (1989) reanalyzed wing morphol- ratio these ancient bats probably foraged or lived ogy of the Eocene fossil forms in the context among vegetation (which was also suggested by Ha- of the much larger data set collected by Nor- bersetzer & Storch 1987) and may have been perch berg and Rayner (1987). Norberg (1989) hunters. suggested that the estimates of body mass Perch huntingÐmaking short ¯ights out to 122 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Fig. 42. Reconstructions of wing shapes of fossil bats; redrawn from Norberg (1989: ®g. 3). Norberg (1989) modi®ed reconstructions published by Habsersetzer and Storch (1987) by adding curved wingtips such as those seen in extant bats, changing the angles between the phalanges of the fourth digit in Archaeopteropus, and changing the angles between the third, fourth, and ®fth digits in most species. capture prey detected from a ®xed perchÐis berg, 1994). Because ¯ights to capture prey widely regarded as a behavior that reduces are infrequent and of short duration, less en- the energy required for successful foraging ergy is apparently spent than would be re- (Norberg and Rayner, 1987; Fenten, 1990; quired by sustained hawking (Norberg and Fenton et al., 1990; Norberg, 1994). Extant Rayner, 1987; Fenton, 1990; Fenton et al., bats that forage near or within vegetation 1990; Norberg, 1994). Another possible ad- may use perches because their short, low as- vantage of perch hunting is that it may pect ratio wings make ¯ight relatively ex- broaden the available prey spectrum by fa- pensive (Norberg and Rayner, 1987; Nor- cilitating the capture and handling of larger 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 123

Fig. 43. Comparative wing morphology described using principal components analysis; redrawn from Norberg (1989: ®g. 3). This diagram is a scatter plot of the second and third principal components of wing morphology, which were identi®ed by Norberg and Rayner (1987) as measures of wing loading and aspect ratio. The wing loading (WL) and aspect ratio (AR) components were obtained using the following exponential equations: eWL ϭ 3.77 ϫ 10-3M3.02b-2.08S-3.71 and eAR ϭ 1.81 ϫ 10-5M-1.47b14.6S-5.12, where M ϭ mass in grams, b ϭ wingspan, and S ϭ wing area (Norberg, 1989). Points representing fossil bats are circled and identi®ed by number as follows: 1, Icaronycteris index;2,Archaeonycteris trigonodon;3,Palaeochiropteryx tupaiodon;4,Palaeochiropteryx spiegeli;5,Hassianycteris messelen- sis;6,Hassianycteris magna;7,Archaeopteropus transiens. The data used to construct this plot were provided in Norberg and Rayner (1987; extant bats except Pteronotus parnellii) and Norberg (1989; fossil bats and Pteronotus parnellii). 124 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Fig. 44. Comparisons of size and shape of the wingtip as described by plotting wingtip length ratio

Tl (length of handwing/length of arm wing) versus wingtip area ratio Ts (handwing area/armwing area); redrawn from Norberg (1989: ®g. 4). Wingtip index I equals Ts/(Tl-Ts), where I Ͻ 1 indicates pointed wingtips and I Ͼ 1 indicates rounded wingtips. Points representing fossil bats are circled and identi®ed by number as follows: 1, Icaronycteris index;2,Archaeonycteris trigonodon;3,Palaeochiropteryx tupaiodon;4,Palaeochiropteryx spiegeli;5,Hassianycteris messelensis;6,Hassianycteris magna;7, Archaeopteropus transiens. Data on which this graph was based were provided in Norberg and Rayner (1987; extant bats) and Norberg (1989; fossil bats). prey (Vaughan and Vaughan, 1986; Norberg probably some phyllostomids; Vaughan, and Rayner, 1987; Fenton, 1989, 1990). Ex- 1976; Sazima, 1978; Fiedler, 1979; Fenton et amples of bats that use perches to forage for al., 1983, 1990; Tidemann et al., 1985; Nor- ¯ying prey (``¯ycatching'' or ``sally forrag- berg and Rayner, 1987; Fenton, 1990; Nor- ing'' bats) include megadermatids (e.g., Lav- berg, 1994). Most megadermatids and nys- ia frons, Megaderma spasma), nycterids terids that glean prey from surfaces also en- (e.g., Nycteris grandis, Nycteris thebaica), gage in ¯ycatching (Vaughan, 1976; Fenton some rhinolophids (e.g., Rhinolophus hilde- et al., 1983, 1990; Norberg and Rayner, brandti, Rhinolophus rouxi, Hipposideros 1987), demonstrating that these are not mu- commersoni), and some vespertilionids (e.g., tually exclusive habits. Nyctophilus bifax and juvenile Myotis luci- Many (perhaps all) extant perch-hunting fugus) (Vaughan, 1977; Buchler, 1980; Fen- bats switch between perch hunting and con- ton, 1982b, 1990; Fenton and Rautenbach, tinuous ¯ight when searching for prey (e.g., 1986; Vaughan and Vaughan, 1986; Neu- Rhinolophus hildebrandti, R. rouxi, Nycteris weiler et al., 1987; Norberg and Rayner, grandis, N. thebaica, Megaderma lyra, Ma- 1987; Tyrell, 1988; Fenton et al., 1990). Oth- croderma gigas; Tidemann et al., 1985; Fen- er perch hunters take non¯ying prey that they ton and Rautenbach, 1986; Neuwieler et al., glean from the ground or vegetation (e.g., 1987; Audet et al., 1988; Fenton et al., 1987, most megadermatids, some nycterids, and 1990; Fenton, 1990). In at least some cases 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 125

(e.g., Nycteris grandis in Zimbabwe), allo- expenditures than aerial hawking. The evo- cation of time between these two foraging lutionary transition to full-time aerial hawk- strategies may re¯ect prey availability, with ing may have mirrored the ontogenetic trans- more time being spent on hunting from con- formations seen today in bats like Myotis lu- tinuous ¯ight when prey is relatively scarce cifugus: (Fenton, 1990; Fenton et al., 1990). Although . . . the bats progress from: (1) ®rst ¯ights and landing Norberg and Rayner (1987) treated perch attempts, during which the moving bat is probably hunting as a foraging strategy distinct from most attentive to large, stationary objects, through (2) continuous aerial hawking, and categorized . . . a short `¯ycatcher' period during which the bat is species as using either one or the other ap- initially ®xed and the target is in motion, to (3) the proach, Fenton (1990) argued that data now most complex stage of integration involving both a continually moving signal source [the bat] and mov- available preclude regarding perch hunting ing target. (Buchler, 1980: 216) and continuous aerial hawking as mutually exclusive foraging strategies. Even if Icaron- Critical factors in this transformation ap- ycteris, Archaeonycteris, and/or Palaeochi- pear to be degree of development of echo- ropteryx hunted from perches, they may also location abilities and the integration of echo- have foraged from continuous ¯ight under location and ¯ight behavior during foraging. some circumstances. Norberg (1989) considered correlations Hill and Smith (1984) suggested that perch among wing morphologies and echolocation hunting may be the primitive foraging strat- call structure, again building on the work of egy for microchiropteran bats. This hypoth- Norberg and Rayner (1987) to reconstruct esis clearly makes sense in terms of energy the possible behaviors of the Eocene fossil ef®ciency. Observations of perch hunting by bats. Norberg (1989: 209) found that juvenile Myotis lucifugus additionally sug- . . . modern bats with similar wing design to the fossil gest that this form of hunting requires less species Hassianycteris messelensis and H. magna elaborate ¯ight maneuvers than those used have echolocation calls for long-range detection in by adults during continuous aerial hawking open spaces (CF ϩ narrowband FM), which occur for (Buchler, 1980). Perch hunting by juvenile example in rhinopomatids, or calls which include both steep and shallow FM sweeps, which occur in Myotis may also have the effect of removing species that hawk fast relatively close to obstacles these bats from group-foraging situations, (such as Lasiurus borealis, L. cinereus, and Nyctalus thus minimizing distractions, reducing the noctula). The predicted fast foraging ¯ight in H. mes- need for complex evasive behaviors, and selensis and H. magna indicates that they used echo- location calls of similar structure. simplifying the analysis of information from The smaller size, lower wing loading, low aspect returning echoes (Buchler, 1980). As such, ratio and short wings of Icaronycteris, Archaeonyc- perch hunting appears to be less demanding teris and Palaeochiropteryx suggest that they foraged than full-time aerial hawking in terms of the more close to, or among, vegetation. They should ¯ight and echolocation skills required for therefore have bene®tted most from echolocation calls for short-range detection. successful prey capture (Buchler, 1980). In the context of our theories about the Extant bats with similar aspect ratio and evolution of ¯ight and echolocation, perch wing-loading parameters to Icaronycteris, hunting represents a possible intermediate Archaeonycteris, and Palaeochiropteryx use between foraging strategies characteristic of either long CF echolocation calls (e.g., Rhin- arboreal mammals (e.g., scansorial foraging olophus hipposideros, Rhinolophus ferrume- while clinging to surfaces) and full-time ae- quinum) or broadband FM calls (e.g., Pipis- rial hawking. The demands of perch hunting trellus kuhlii, Pipistrellus hesperus, Myotis are probably more complex than those of volans, Myotis yumanensis, Micronycteris scansorial foraging even when passive cues megalotis; Norberg, 1989). Use of long CF are used to locate prey because the hunter calls (high-duty-cycle echolocation) is asso- must simultaneously ¯y and keep track of the ciated with greater cochlear enlargement than prey while approaching it. However, perch is seen in any of the Eocene fossils; there- hunting is apparently less complex (in terms fore, it seems unlikely that any of the Eocene of required ¯ight maneuvers and neural pro- taxa used such a system (Habersetzer and cessing) and probably requires lower energy Storch, 1989, 1992). This suggests that Ica- 126 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 ronycteris, Archaeonycteris, and Palaeochi- located, not the exact position of the prey; ropteryx probably used broadband FM calls, this area is then engulfed with the wing which are useful for target texture discrimi- membranes and the prey is located by touch nation, range ®nding, and short-range detec- or other cues (E. Kalko, personal commun.). tion among clutter (Simmons et al., 1975; Olfaction may be used by some bats (e.g., Norberg, 1989). Myotis myotis, Mystacina tuberculata) for There is a great deal of variation in call detecting prey under leaf litter, in tree holes, structure among extant bats that use broad- and in other concealed locations (Kolb, 1961; band FM calls, with calls being either short B. Lloyd, personal commun.). Passive prey or long, steep or shallow, concave up or con- localizationÐ with or without accompanying cave down, etc. (e.g., Simmons et al., 1975, echolocation callsÐis apparently used by 1980; Fenton and Bell, 1979, 1981; Sim- various nycterids, megadermatids, phyllos- mons, 1980; Simmons and Stein, 1980; Fen- tomids, mystacinids, antrozoids, and vesper- ton et al., 1983; Barclay, 1985). High-inten- tilionids (Kolb, 1959, 1961; Fiedler, 1979; sity calls of varying durations and forms are Barclay et al., 1981; Tuttle and Ryan, 1981; used by aerial insectivores. Surface gleaners Bell, 1982a, 1982b, 1985; Guppy and Coles, typically use very short (Ͻ2 msec), low-in- 1983; Ryan and Tuttle, 1983, 1987; Fenton, tensity multiharmonic calls that are appar- 1984, 1990, 1994a; Bell and Fenton, 1986; ently well-suited for ®ne texture discrimina- Norberg and Rayner, 1987; Belwood and tion of targets on a surface by either temporal Morris, 1987; Schmidt et al., 1988; Tyrell, cues or spectral differences in the echoes 1988; Norberg, 1989; B. Lloyd, personal (MoÈhres and Kulzer, 1957; MoÈhres and Neu- commun.). At least one of these species, the weiler, 1966; MoÈhres, 1967b; Simmons et al., gleaning phyllostomid Micronycteris mega- 1974; Simmons, 1979; Simmons and Stein, lotis, has wing parameters similar to those of 1980; Barclay et al., 1981; Habersetzer and the Eocene fossil bats (Norberg and Rayner, Vogler, 1983; Guppy et al., 1985; Norberg 1987; Norberg, 1989). and Rayner, 1987; Suthers and Wenstrup, Cochlear size and structure provide anoth- 1987; Schmidt, 1988; Norberg, 1989; Audet, er source of data useful for evaluating hy- 1990; Faure and Barclay, 1992; Faure et al., potheses about echolocation behavior. Ha- 1993). These short, low-intensity calls may bersetzer and Storch (1992) found correla- minimize echoes from clutter while simulta- tions between relative cochlear size (width of neously minimizing the chances of alerting the second half of the basal turn plotted possible prey (Barclay et al., 1981; Haber- against basicranial width; see character 26 setzer and Vogler, 1983; Fullard, 1987; Rov- above) and the echolocation strategies adopt- erud, 1987; Fenton, 1990; Kober and ed by various taxa. Bats that use long CF Schnitzler, 1990; Faure et al., 1993). In some calls emitted at high duty cycles (e.g., rhin- cases, calls may be ``switched off'' to facil- olophines) are typically characterized by ex- itate passive prey localization by vision or by ceptionally large cochleae (Habersetzer and listening for sounds produced by the prey Storch, 1992). Most bats that use broadband (e.g., Megaderma lyra, Antrozous pallidus, FM calls and forage by continuous aerial Macrotus californicus; Fiedler, 1979; Bell, hawking (e.g., Rhinopoma) have somewhat 1982a, 1992b, 1985). smaller cochleae, but still fall well above the Prey-generated sounds used by bats in- lower limits of cochlear size in microchirop- clude mating calls (e.g., of frogs and katy- terans (Habersetzer and Storch, 1992). Mi- dids), rustling noises, and the sounds of in- crochiropterans with the smallest cochleae sects landing on or crashing into vegetation (those that fall near or within the zone of (Fiedler, 1979; Tuttle and Ryan, 1981; Gup- overlap between Microchiroptera and Me- py and Coles, 1983; Ryan and Tuttle, 1983, gachiroptera) include forms that use broad- 1987; Fenton, 1984, 1990, 1994a; Belwood band FM echolocation calls but are not typ- and Morris, 1987; Tyrell, 1988; E. Kalko, ical aerial insectivores (Habersetzer and personal commun.). In some instances (e.g., Storch, 1992). This group includes several some phyllostomids), the bat uses these nois- phyllostomids (e.g., Phyllostomus hastatus, es to identify the area in which the prey is Trachops cirrhosus, Leptonycteris nivalis, 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 127

Carollia perspicillata, Desmodus rotundus), using ``rather short, multiharmonic pulses.'' megadermatids (e.g., Megaderma spasma, Based on our examinations of Icaronycteris Megaderma lyra), and nycterids (e.g., Nyc- and estimates of cochlear size (see discussion teris grandis). Several of these hunt from under character 26 above), we suspect that perches much of the time (see above); as far Icaronycteris had echolocation capabilities as is known, all use calls of short duration and calls similar to those of Archaeonycteris. and often low intensity, and many apparently Another source of information regarding use cues other than echolocation to locate the foraging habits of Eocene bats is analysis their food (Fiedler, 1979; Tuttle and Ryan, of fossilized stomach contents. No such data 1981; Fenton, 1984, 1990, 1994a; Bell and are available for Icaronycteris, but Richter Fenton, 1986; Norberg and Rayner, 1987; and Storch (1980) and Habersetzer et al. Ryan and Tuttle, 1987; Tyrell, 1988; Nor- (1992, 1994) reported the results of analyses berg, 1989). of stomach contents of Palaeochiropteryx tu- Habersetzer and Storch (1992) compared paiodon, P. spiegeli, Archaeonycteris trigon- values for cochlear width and basicranial odon, Hassianycteris messelensis, and H. width in six species of Messel bats to those magna. The most recent summary was pro- obtained from extant forms (see discussion vided by Habersetzer et al. (1994), who re- under character 26 above) and used the re- ported on stomach contents of a larger series sults to infer possible echolocation strategies of specimens (33 individuals) than were for the fossil bats. They found that Hassi- available in previous studies. anycteris and Palaeochiropteryx resemble Habersetzer et al. (1994) concluded that taxa at the lower end of the range of variation the two species of Palaeochiropteryx (P. tu- for typical continuous aerial hawkers (®g. paiodon, represented by 20 specimens, and 29). Other aspects of cochlear structure ex- P. spiegeli, 5 specimens) show little or no plored by Habersetzer and Storch (1989) and difference in their diet. Stomach contents in- Habersetzer et al. (1994) included (1) the rel- cluded thick layers of scales, hairs, and cu- ative size of the cross-sectional area of the ticle fragments similar to those known from bony cochlear canal above the spiral laminae, primitive Microlepidoptera (e.g., Micropter- (2) the degree of development of the second- igidae, Hepialidae, and Eriocraniidae), Tri- ary spiral lamina, and (3) the distance be- choptera, and scale-bearing Diptera (e.g., Cu- tween the primary and secondary spiral lam- licidae; Habersetzer et al., 1994). In one case, inae in the basal turn (this dimension pro- details of scale ultrastructure permitted un- vides an approximate measure of the width ambiguous identi®cation of the prey as a mi- of the basilar membrane). On the basis of cropterigid microlepidopteran (Habersetzer comparisons with extant forms and consid- et al., 1994). Microlepidopterans such as eration of body sizes, Habersetzer and Storch these are all small moths that are weak night (1992) proposed echolocation calls within a ¯iers; Tricoptera (caddis ¯ies) are similar in frequency band of 30±90 kHz for Palaeo- size and ¯ight habits (Habersetzer et al., chiropteryx, and intense sound frequencies 1992). Caddis ¯ies are active at night, and at below 30 kHz for Hassianycteris. the time of hatching they often swarm in In comparison to Palaeochiropteryx and large numbers just above the water surface Hassianycteris, species of Archaeonycteris at the edges of lakes and streams (Haber- have distinctly smaller cochleae, falling setzer et al., 1992). Thick cuticles, such as within the zone of overlap between Micro- are characteristic of coleopterans (beetles) chiroptera and Megachiroptera (®g. 29). and blattoids (cockroaches and their kin), are These forms had basal cochlear widths well rare in the gut contents of both species of below the range typical for obligate aerial in- Palaeochiropteryx (Habersetzer et al., 1994). sectivores. Instead, they most closely resem- These results suggest that Palaeochiropteryx bled perch hunting/aerial hawking megader- tupaiodon and P. spiegeli may have fed matids and the phyllostomid gleaner Trach- mainly on small, scale-bearing insects with ops (Habersetzer and Storch, 1992). On this weak exoskeletons that tend to ¯y close to basis, Habersetzer and Storch (1992: 466) the ground or water surface (Habersetzer et proposed that Archaeonycteris echolocated al., 1992, 1994). These bats must have been 128 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 capable of detecting, tracking, and capturing only two specimens of A. trigonodon have ¯ying insects on the wing; it is very hard to thus far been investigated (Habersetzer et al., imagine how so many individuals could have 1992). The gut contents of these specimens ®lled their stomachs with tiny moths and consist predominantly of thick chitin frag- caddis ¯ies if they were not expert aerial ments, among which are some with irides- hawkers. This conclusion is consistent with cent structural colors such as those seen in the presumed ¯ight styles of Palaeochirop- Messel coleopterans (e.g., Lutz, 1992). Frag- teryx as reconstructed from wing parameters ments referable to scale-bearing insects (e.g., (Habersetzer et al., 1992, 1994). lepidopterans, Tricoptera) are completely ab- The fossilized stomach contents of Hassi- sent in A. trigonodon (Habersetzer et al., anycteris messelensis (seven specimens) and 1992). This suggests that Archaeonycteris H. magna (one specimen) similarly includes may have been a beetle specialist. chitinous scales, hairs, and cuticle fragments, In summary, much can be inferred about but the proportions of these items are mark- the ecology of Eocene bats (particularly edly different from those seen in Palaeochi- those from Messel) based on their morphol- ropteryx (Habersetzer et al., 1992, 1994). ogy and stomach contents. Given the ¯ight Habersetzer et al. (1994: 246) noted that and echolocation habits suggested for Has- sianycteris by Habersetzer and Storch (1987, Whereas in Palaeochiropteryx thick layers of closely packed scales clearly prevail, in Hassianycteris scales 1989, 1992), Norberg (1989), and Habersetz- are generally rare, which means that scale-bearing in- er et al. (1992, 1994), it seems very likely sects are not the predominant prey. Moreover, most that members of this genus foraged by con- scales found in Hassianycteris show a different and tinuous aerial hawking comparable to that more elaborate ultrastructure ....This type of scale seen today in many microchiropterans (e.g., is not restricted to higher evolved Lepidoptera, yet it is comparatively rare in primitive moths. So its pre- Rhinopoma, some Lasiurus). Palaeochirop- dominance in the gut contents makes it at least prob- teryx was probably also capable of continu- able that Hassianycteris mainly preyed on Macrole- ous aerial hawking; indeed, the abundance of pidoptera .... There are still more general differ- this taxon at Messel (and analyses of stom- ences in the diet of both genera. Whereas in Palaeo- chiropteryx most cuticulae are thin and hairy, but ach contents) suggests that these bats were with an otherwise smooth surface, in Hassianycteris hawking insects over ancient Lake Messel at thick cuticulae with strongly sculptured surface pre- the time of their death. Most extant bats that vail (Richter and Storch, 1980; Richter, 1987). Cuti- forage by continuous aerial hawking use the culae of a similar design are known from many Mes- uropatagium for catching ¯ying insects sel insects (e.g., Coleoptera and Blatto[i]dea). (Vaughan, 1959, 1970a, 1970b; Webster and Inclusion of coleopteran prey in the diet Grif®n, 1962; Norberg, 1976; Hill and would be consistent with the massive denti- Smith, 1984; Schnitzler et al., 1994; Kalko, tion and deep mandible seen in Hassianyc- 1995). Both Hassianycteris and Palaeochi- teris (®g. 24). The dentition and jaw structure ropteryx have a well developed calcar, sug- of Hassianycteris resemble that seen in many gesting the presence of a large uropatagium extant molossids, particularly those thought appropriate for aerial insect capture. to feed regularly on hard-shelled beetles Wing morphology, body size, cochlear (Freeman, 1979, 1981). Scales referable to morphology, dental morphology, and stom- Trichoptera and Microlepidoptera are very ach contents indicate that Hassianycteris rare or even lacking in stomachs of Hassi- probably foraged by hawking insects (per- anycteris, suggesting that these bats did not haps mostly beetles and cockroaches) while hunt close to the ground, but instead hunted in fast ¯ight well above the ground in forest larger prey in relatively open spaces (Haber- gaps and above the canopy. Echolocation setzer et al., 1994). No obvious differences calls were probably intense, of relatively low were detected between the stomach contents frequency (Ͻ30 kHz), and may been either of H. messelensis and H. magna, although narrowband FM signals or calls that included the latter is known from only a single spec- both steep and shallow FM sweeps. In con- imen (Habersetzer et al., 1994). trast, Palaeochiropteryx may have foraged Little information is available about stom- close to the ground and vegetation, probably ach contents of Archaeonycteris; apparently hunting from perches as well as hawking in- 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 129

sects while in slow ¯ight. Palaeochiropteryx perhaps some fruits and other edible items apparently specialized in preying on small- from foliage, bark, and perhaps the ground. scaled insects (i.e., small moths and caddis Location of potential food items was proba- ¯ies), and its echolocation calls were most bly accomplished by a combination of vision likely broadband calls of moderate to high and listening for prey-generated sounds. Like frequency (30±90 kHz). There is no evidence Archaeopteropus, Icaronycteris, Archaeo- that any of the Messl bats used high-duty- nycteris, and Palaeochiropteryx, these bats cycle CF echolocation. probably had low aspect ratio wings, mod- Wing morphology of Icaronycteris and erate wing loading, and relatively large wing Archaeonycteris suggests that these taxa tips, all of which suggest that they habitually probably foraged close to the ground and ¯ew close to the ground and near vegetation. close to vegetation, as did Palaeochiropteryx Low-duty-cycle echolocation evolved sub- (Habersetzer and Storch, 1987, 1989, 1992; sequently, probably from communication Norberg, 1989; and Habersetzer et al., 1992, calls that incidentally produced informative 1994). However, presence of only moderate echoes. As we suggested earlier, this system enlargement of the cochleaÐtogether with was probably simple at ®rst, permitting only absence of a calcarÐsuggests that these orientation and obstacle avoidance but not forms may have been perch hunters that spe- detection, tracking, or evaluation of airborne cialized in gleaning their prey from surfaces prey. Basicranial modi®cations presumed to rather than catching it on the wing. Archaeo- be associated with increased ef®cacy of nycteris may have been a beetle specialist; echolocation began prior to the divergence of we have no record of the preferred prey of Icaronycteris from the microchiropteran lin- Icaronycteris. Echolocation calls in both taxa eage. However, wing morphology, a moder- were probably short (Յ2 msec) broadband ately enlarged cochlea, and absence of a cal- FM signals of moderate to high frequency car in this taxon suggest to us that Icaron- (30±90 kHz) or short, multiharmonic calls. ycteris was a perch-hunting gleaner rather In either case, these echolocation calls may than a predator on aerial insects. This for- have been of low intensity, and may have aging strategy would have had the advantage been ``turned off'' at times to facilitate lo- of being relatively energy-ef®cient while at cation of prey by passive means (e.g., listen- the same time requiring only moderate au- ing for prey-generated sounds or looking for ditory data-processing capabilities to suc- prey movements). It is unlikely that Icaron- cessfully sort the information from returning ycteris and Archaeonycteris used echoloca- echoes. Echolocation calls were most likely tion for detection, tracking, or evaluation of short (Յ2 msec) broadband FM signals or prey. Echolocation was problably used only multiharmonic calls, probably of relatively for orientation and obstacle detection, while low intensity. These calls may have been prey detection and tracking were accom- ``turned off'' at times to facilitate passive plished by passive means. prey localization. Indeed, prey detection, tracking, and evaluation were probably not EVOLUTION OF FORAGING done with echolocation, but rather by vision STRATEGIES: or listening for prey-generated sounds. Pas- A PHYLOGENETIC PERSPECTIVE sive acoustic cues may have been particular- ly important if these bats were strictly noc- The phylogeny generated in our study (®g. turnal. In essence, the only major change in 36) provides a framework for interpreting the foraging method at this level would have morphological data and behavioral infer- been the addition of echolocation as a tool ences presented above. Given the topology for orientation and obstacle detection. The of this tree, we suggest that the earliest mem- basic foraging strategyÐgleaning from a bers of the microchiropteran lineage (forms perchÐwould have been the same as seen in currently unknown from fossils) probably the nonecholocating chiropteran ancestors of used vision for orientation and obstacle de- Icaronycteris. tection in their arboreal/aerial environment, The derived morphological transforma- and probably foraged by gleaning insects and tions that diagnose the node linking Archaeo- 130 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 nycteris with Hassianycteris, Palaeochirop- tion of aerial hawking do not seem to have teryx, and the microchiropteran crown group been correlated with any appreciable modi- (e.g., reduction of the number of roots on P3, ®cations in wing shape or wing loading presence of a ventral accessory process on (based on the results of Norberg,1989; see C4, absence of an ossi®ed third phalanx on ®gs. 43, 44). Similarities in aspect ratio, wing wing digit III) are relatively small changes tip indices, and wing loading in Palaeochi- that do not indicate any major shifts in for- ropteryx, Archaeonycteris, Icaronycteris, and aging ecology. This pattern suggests to us Archaeopteropus lead us to conclude that the that Archaeonycteris retained much the same body size and wing form of Palaeochirop- foraging strategy as Icaronycteris. Archaeo- teryx is plesiomorphic, and that this mor- nycteris and Icaronycteris share a similar phology was present in the most recent com- wing morphology, moderately enlarged mon ancestor of Palaeochiropteryx and Has- cochlea, and lack of a calcar, features we in- sianycteris. Accordingly, the body size and terpret as indicating that these bats foraged wing morphology seen in Hassianycteris by gleaning prey that were detected from a (discussed below) are autapomorphic fea- perch using passive means rather than echo- tures. location. Like Icaronycteris and Archaeonycteris, The next node as one moves up the treeÐ Palaeochiropteryx had low aspect ratio that which links Hassianycteris with Palaeo- wings, moderate wing loading, and relatively chiropteryx and the microchiropteran crown large wing tips, all of which indicate that groupÐis associated with a much more ex- these bats habitually ¯ew close to the ground tensive suite of morphological changes (see and near vegetation. However, the greatly en- discussion above under ``Character Transfor- larged cochlea and well developed calcar mations at Basal Nodes''). Most notable suggest that Palaeochiropteryx was fully ca- among these are: (1) a greatly enlarged coch- pable of aerial hawking. This hypothesis is lea, (2) a dorsal articular facet on the scapula consistent with the analyses of stomach con- (although this may have evolved earlier), (3) tents and taphonomy of Palaeochiropteryx at a laterally compressed ventral process on the Lake Messel. Echolocation signals used by manubrium of the sternum, (4) increased de- Palaeochiropteryx were probably short- to velopment of rib laminae (this may have moderate-length broadband FM calls. evolved somewhat earlier), (5) a threadlike As pointed out by Fenton (1990) and Fen- ®bula, and (6) presence of a calcar. Taken ton et al. (1990), many extant bats that hunt together, these features suggest that a major from perches also hawk insects in continuous shift in foraging strategy occurred in the mi- ¯ight, with allocation of time between these crochiropteran lineage just prior to the di- two foraging strategies re¯ecting prey avail- vergence of Hassianycteris: the evolution of ability. Given that gleaning from a perch aerial hawking. This behavior involves using probably represents the primitive foraging echolocation to detect, track, and assess prey, strategy for the microchiropteran lineage, we and the use of the uropatagium (supported suspect that Palaeochiropterx may have used and controlled by a calcar) to capture prey a combination of perch hunting (including on the wing. Changes in the postcranial skel- ¯ycatching) and slow aerial hawking to cap- eton, particularly the pectoral girdle, suggest ture its prey, much like modern nycterids. that some ``®ne tuning'' of the ¯ight mech- This would be consistent with cochlear size anism may have accompanied the behavioral in Palaeochiropterx (as estimated by Haber- change to aerial hawking. This seems rea- setzer and Storch, 1992; ®g. 29), which plac- sonable given the demands of aerial foraging es this genus just at the lower end of the behavior, particularly if these bats occasion- range of variation seen among forms that are ally foraged in group situations (e.g., when continuous aerial hawkers (e.g., most ves- caddis ¯ies were hatching along the shore of pertilionids) and at the upper end of the Lake Messel) where complex capture and range of variation in extant perch-hunting evasive maneuvers may have been required. forms that use low-duty-cycle echolocation Interestingly, the morphological and be- (e.g., nycterids and megadermatids). havioral changes associated with the evolu- The transition from gleaning stationary 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 131 prey detected by passive means (vision or lis- note that all four of the fossil bats considered tening for prey-generated sounds) to aerial here (Icaronycteris, Archaeonycteris, Hassi- hawking using echolocation to detect and anycteris, and Palaeochiropteryx) have been track prey must have been a complex process collected from lake deposits that are pre- that involved intermediate steps. It seems un- sumed to have been surround by subtropical likely that echolocation was used for detect- or tropical forests at the time of deposition ing stationary prey during this transition, be- (MacGinitie, 1969; Grande, 1980; cause detection of hard targets resting on Schaarschmidt, 1992). The excellent preser- hard or irregular backgrounds is one of the vation of many of the bat fossils found at more dif®cult tasks faced by echolocators Messel and Fossil Basin indicates that these (Simmons et al., 1980; Fenton, 1990, 1994a, animals were not transported far after death, 1995). Instead, we hypothesize that bats that suggesting that they may have foraged regularly foraged by gleaning stationary prey around the edges and over the lakes in which (detected by passive means) increasingly they were ultimately preserved. came to hunt moving insects, perhaps leaping Species of Hassianycteris are larger in after prey that had been startled by move- many dimensions than most of the other Eo- ments or attacks by the bat. At this point, cene bats, and analyses of wing morphology echolocation (which they were already using by Norberg (1989) indicated that Hassianyc- for orientation and obstacle avoidance) could teris was characterized by signi®cantly high- provide more information about prey loca- er wing loading than were Palaeochiropte- tion and movement than could be obtained ryx, Archaeonycteris,orIcaronycteris. In- by passive means. To fully exploit this form deed, Hassianycteris apparently had a higher of data collection, the animals would have wing loading than most extant bats (®g. 43). had to increase signal strength in order to These observations, taken together with a maximize range and provide the necessary reasonably high aspect ratio, a greatly en- time to track and evaluate targets. However, larged cochlea, and presence of a calcar, sug- environmental clutter (which produces many gest that Hassianycteris foraged by fast aerial distracting echoes) would have presented a hawking, most likely well above the ground major impediment to the evolution of effec- in forest gaps or above the canopy. Hassi- tive ¯ycatching behavior. It therefore seems anycteris thus represents another shift in for- likely that ¯ycatching using echolocation to aging strategy, away from combined perch detect and track prey probably evolved in hunting and aerial hawking near the ground bats that frequented forest gaps and the edges or vegetation to a fast-¯ying, continuous ae- of forests along lakes and rivers, places rial hawking foraging strategy similar to that where vegetation (with potential perches) lies seen in extant rhinopomatids and some ves- adjacent to relatively clutter-free open spac- pertilionids. As noted above, however, this es. Once ¯ycatching from perches was well shift apparently occurred after the lineage established, it is easy to imagine a progres- leading to Hassianycteris diverged from the sive transition to spending more time on the lineage leading to the microchiropteran wing, ultimately leading to the evolution of crown group. taxa that relied exclusively on foraging by Continuing to move up the phylogenetic continuous aerial hawking. It is not clear tree, we ®nd that the clade comprising Pa- when bats began to use the uropatagium for laeochiropteryx plus the microchiropteran prey capture, but evolution of the calcar ap- crown group is diagnosed by only one un- parently preceded or was coincident with the ambiguous synapomorphy, the presence of a evolution of aerial hawking. ventral accessory process on C5. This feature Our hypothesis that the evolutionary tran- is not indicative of any change in foraging sition from gleaning (using passive prey de- habits, but rather a continuation of neck tection) to aerial hawking (using echoloca- modi®cations associated with roosting be- tion) took place in habitats associated with havior (see discussion under character 77). forest gaps or forest edges along bodies of Transformations that diagnose the microchi- water cannot be tested given the sparse fossil ropteran crown group (e.g., modi®cation of record of bats. However, it is interesting to the premaxilla articulation, reduction in the 132 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 number of lower premolars) indicate changes (®g. 36) indicates that passive prey detection in the masticatory apparatus, but no major evolved independently at least ®ve or six behavioral shifts. timesÐin phyllostomids, mystacinids, antro- In summary, we propose that foraging be- zoids, vespertilionids, and either indepen- havior in the microchiropteran lineage dently in nycterids and megadermatids or in evolved in a series of steps: (1) gleaning food the common ancestor of Rhinolophoidae. objects during short ¯ights from a perch us- Similarly, gleaning must have evolved inde- ing vision for orientation and obstacle detec- pendently at least four times within Micro- tion; prey detection by passive means, in- chiropteraÐin phyllostomids, antrozoids, cluding vision and/or listening for prey-gen- vespertilionids, and rhinolophoids. Perch erated sounds (no known examples in fossil hunting apparently evolved at least three record); (2) gleaning stationary prey from a timesÐin phyllostomids, vespertilionids, and perch using echolocation and vision for ori- rhinolophoids. Although passive prey detec- entation and obstacle detection; prey detec- tion, gleaning, and perch hunting are linked tion by passive means (Icaronycteris; Ar- in some taxa (e.g., most megadermatids and chaeonycteris); (3) perch hunting for both nycterids), they are decoupled in other forms. stationary and ¯ying prey using echolocation For example, Mystacina apparently uses pas- and vision for orientation and obstacle detec- sive prey detection but does not glean or hunt tion; prey detection and tracking using echo- from perches; instead, it approaches its prey location for ¯ying prey and passive means ``on foot'' (B. Lloyd, personal commun.). for stationary prey (no known example, al- Rhinolophids sometimes hunt from perches though Icaronycteris and/or Archaeonycteris and glean, but apparently do not use passive may have done this at times); (4) combined cues to detect their prey (Norberg and Ray- perch hunting and continuous aerial hawking ner, 1987). Lavia frons hunts from perches using echolocation and vision for orientation but apparently does not glean or use passive and obstacle detection; prey detection and prey detection (Vaughan and Vaughan, tracking using echolocation for ¯ying prey 1986). These observations suggest that pas- and passive means for stationary prey; cal- sive prey detection, gleaning, and perch car-supported uropatagium used for prey hunting evolved in different ways in different capture (common ancestor of Hassianycteris microchiropteran lineages. and Palaeochiropteryx; retained in Palaeo- The relative timing of the evolutionary chiropteryx); and (5) exclusive reliance on and behavioral changes in the lineage leading continuous aerial hawking using echoloca- to extant Microchiroptera can be estimated tion and vision for orientation and obstacle from the fossil record. Taxa representing detection; prey detection and tracking using stages 2, 4, and 5 as de®ned above were ap- echolocation (Hassianycteris; common an- parently present simultaneously at Messel, cestor of microchiropteran crown group). suggesting that evolutionary transformations Given the topology of the tree we derived in foraging strategies may have occurred earlier (®g. 36), it seems most likely that the very rapidly in early members of the micro- latter foraging strategyÐreliance on contin- chiropteran lineage. This was probably facil- uous aerial hawkingÐwas primitive for the itated by the mechanical coupling of venti- microchiropteran crown group. This is con- lation and ¯ight, which meant that the energy sistent with optimization of foraging strate- costs of echolocation to ¯ying bats were low, gies both within the microchiropteran crown particularly in comparison to the energetic group and among the fossil stem group bene®ts of aerial hawking for insects in the forms. absence of any competitors. The conclusion that aerial hawking is the The evolution of continuous aerial hawk- primitive foraging strategy for the microchi- ing may have been the ``key innovation'' ropteran crown group suggests that gleaning, (sensu Liem, 1973) responsible for the burst passive prey detection, and perch hunting of diversi®cation in microchiropteran bats among extant taxa represent secondarily de- that occurred in the Eocene. Fossils referable rived specializations rather than retentions of to six major extant lineages are known from primitive habits. If so, topology of our tree Middle±Late Eocene deposits (table 1): (1) 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 133

Emballonuridae, (2) Megadermatidae, (3) Noctilionoidea, (10) Mystacinidae, (11) An- Hipposiderinae, (4) Rhinolophinae, (5) Na- trozoidae, (12) Tomopeatinae, and (13) Ves- taloidea, and (6) Molossinae. Reconstruction pertilionidae. All of these must have di- of ghost lineages following the methods of verged during the Eocene given the Eocene Norell (1992) leads to the conclusion that ages of their sister-taxa. It thus appears that many more extant lineages were minimally Icaronycteris and the Messel bats provide an present by the end of the Eocene, including unprecedented view of the steps leading up (7) Rhinopomatoidea, (8) Nycteridae, (9) to a major adaptive radiation of mammals.

CLASSIFICATION OF EOCENE BATS The phylogenetic results of this study in- dae Jepsen, 1966; Archaeonycteris in Ar- dicate that many groupings of Eocene taxa chaeonycteridae Revilliod, 1917b; Hassi- previously recognized in formal classi®ca- anycteris in Hassiancyteridae Habersetzer tions (e.g., Eochiroptera, Palaeochiroptery- and Storch, 1987; and Palaeochiropteryx in goidea, Archaeonycterididae [including Ica- Palaeochiropterygidae Revilliod, 1917b. This ronycteris]) are not monophyletic. Beginning arrangement preserves monophyly of higher at the lowest taxonomic level, it seems most taxonomic groups and serves to highlight the appropriate to place each of the four genera morphological and presumed behavioral dif- considered in this study in its own monophy- ferences among genera. letic family13: Icaronycteris in Icaronycteri- As we view them, Icaronycteridae, Ar- chaeonycteridae, and Hassiancyteridae each 13 There is considerable disagreement in the literature currently contain only the nominate genus. concerning the formation of family-group names based The situation is slightly more complex in the on generic names ending in -nycteris. The family name in common usage for extant Nycteris is Nycteridae case of Palaeochiropterygidae. Two addition- (Koopman and Jones, 1970; Hill and Smith, 1984; al genera, Cecilionycteris and Matthesia, Koopman, 1984, 1993, 1994). Revilliod (1917b) used a have been referred to Palaeochiropterygidae similiar formation when he named Archaeonycteridae, by previous authors based on dental features as did Jepsen (1966) when he named Icaronycteridae. (e.g., Sige and Russell, 1980; Hill and Smith, However, Russel and Sige (1970) argued that the proper 1984; Beard et al., 1992). In the absence of formation of family-group names from -nycteris re- additional data, we provisionally accept this quired the spelling -nycterididae. They thus changed the assessment, although we note that discovery spelling of Archaeonycteridae to Archaeonycterididae. Some authors have followed this usage (e.g., Habersetz- of more complete material may ultimately er and Storch, 1987, 1989; Habsersetzer et al., 1992, demonstrate that Palaeochiropterygidae as 1994), but many have continued to use Archaeonycter- thus de®ned (including Palaeochiropteryx, idae (e.g., Hill and Smith, 1984; Hand et al., 1994; Mc- Cecilionycteris, and Matthesia) is a paraphy- Kenna and Bell, 1997). Habsersetzer and Storch (1987) letic assemblage. followed Russel and SigeÂ's (1970) usage when they Ageina, which is known only from dental named Hassianycterididae, and subsequent authors have fragments, has been referred variously to used this spelling (e.g., Habersetzer et al., 1992; Mc- ``family uncertain'' (Russell et al., 1973), Pa- Kenna and Bell, 1997). Despite Russell and SigeÂ's (1970) argument about the laeochiropterygoidea (Smith, 1977; Smith nature of the greek root of -nycteris, we think that it is and Storch, 1981), Eochiroptera (Sige and counterproductive to spell some family-group names in Legendre, 1983), Archaeonycteridae (Hill one fashion and others differently. Because Nycteridae and Smith, 1984), and possibly Natalidae is widely accepted as the spelling of the family-group sensu Van Valen (Beard et al., 1992). We name based on Nycteris, we argue that all family-group consider the assessment of Beard et al. names based on genera ending in -nycteris should be (1992), which is based on the most recent spelled in the same fashion. Accordingly, we recognize the Icaronycteridae, Archaeonycteridae, and Hassianyc- evidence (including comparisons with the teridae as the most appropriate spellings for family- new taxon Honrovits), to represent the best group names based on Icaronycteris, Archaeonycteris, current working hypothesis. Accordingly, we and Hassianycteris. follow suggestions made by Beard et al. 134 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

(1992) and refer both Ageina and Honrovits stem-modi®ed node-based, and other types to Nataloidea incertae sedis. of phylogenetic de®nitions. We concur with Australonycteris, which is also poorly most recent authors in concluding that phy- known, was placed in Microchiroptera, fam- logenetic de®nitions are essential and that all ily incertae sedis by Hand et al. (1994). They named taxa should be monophyletic (or at noted that this form has a dentition similar least potentially monophyletic). We consider to that of Archaeonycteris (at least in terms issues related to rank to be of secondary im- of the absence of derived traits), although the portance compared with the de®nition of taxa petrosal referred to Australonycteris appears (see discussion below). The central issue, in relatively more derived and is somewhat ves- our opinion, is the method(s) used to de®ne pertilionoid-like (Hand et al., 1994). These the limits of taxa, particularly those to which and other observations suggest to us that we may apply well-known taxonomic names Australonycteris probably ®ts in the tree (e.g., Microchiroptera). somewhere between Archaeonycteris and the Given the phylogenetic relationships hy- microchiropteran crown group. Given this pothesized here, three general options for de- hypothesis, any decision about how to clas- ®ning Microchiroptera seem appropriate. Us- sify Australonycteris requires consideration ing a node-based de®nition that emphasizes of broader issues surrounding the classi®ca- the importance of the extant crown group tion of stem-group forms. (i.e., a crown-clade-restricted de®nition), we The principal nomenclatural problem might de®ne Microchiroptera as the clade faced in this study concerns higher-level stemming from the most recent common an- classi®cation of the Eocene groups that fall cestor of Emballonuridae and Yangochirop- outside the microchiropteran crown groupÐ tera. If de®ned this way, Yinochiroptera in other words, where to put Icaronycteridae, would be included within Microchiroptera Archaeonycteridae, Hassiancyteridae, and regardless of whether Emballonuridae falls Palaeochiropterygidae. Debates have raged outside Yinochiroptera (as suggested by this in the systematic literature for decades con- study) or inside Yinochiroptera (as suggested cerning the relative pros and cons of different by Koopman, 1985, 1994). A less explicit approaches to de®ning and naming taxonom- crown-clade de®nition might de®ne Micro- ic groups (e.g., Ghiselin, 1966, 1984; Nelson, chiroptera as the clade stemming from the 1972, 1974; Bock, 1974; Duncan and Esta- most recent common ancestor of all extant brook, 1976; Estabrook, 1978, 1986; Jeffer- bats that use sophisticated echolocation. ies, 1979; Wiley, 1979, 1981; Duncan, 1980; When de®ned either way, Microchiroptera Phillips, 1984; Rowe, 1987, 1988; Gauthier would be equivalent to the microchiropteran et al., 1988; Heywood, 1988; de Quieroz and crown group as discussed earlier, and would Gauthier, 1990, 1992, 1994; Minelli, 1991; therefore exclude Palaeochiropterygidae, Lucas, 1992; Meier and Richter, 1992; Lucas Hassiancyteridae, Archaeonycteridae, and and Luo, 1993; Wyss and Flynn, 1993; Bry- Icaronycteridae. The main advantages of ant, 1994, 1996; de Queiroz, 1994; Smith, these de®nitions are that they are congruent 1994; Wyss and Meng, 1996; McKenna and with general usage of the name Microchirop- Bell, 1997). Four principal issues have been tera by most biologists, and that they can be debated: (1) whether taxa should be recog- expected to remain relatively stable (e.g., dis- nized on the basis of shared characters (i.e., covery of new fossils will not affect the lim- diagnoses) or de®ned phylogenetically (i.e., its of Microchiroptera; fossil species would on the basis of their relationships); (2) either fall inside or outside this clade). The whether all named groups above the species latter option (de®ning Microchiroptera as the level must be monophyletic (some workers clade stemming from the most recent com- have argued that convex paraphyletic groups mon ancestor of all extant echolocating bats) may be usefully employed in classi®cations); depends on persistence of extant forms, and (3) the pros and cons of recognizing formal thus might be considered potentially unstable ranks (e.g., ``order,'' ``family'') in classi®- because major extinctions could affect group cations; and (4) the relative merits of node- contents in the future (see discussion of this based, crown-clade-restricted, stem-based, problem in Lucas [1992] and Lucas and Luo 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 135

[1993]). However, we consider this to be of This de®nition would eliminate the problem minor importance in the case of bats since of classifying future stem-group fossils, but the basal lineages (e.g., Emballonuridae, still leaves us without a formal name for the Yinochiroptera, and Yangochiroptera) are all microchiropteran crown group. Such a de®- speciose. nition is also problematic because it blurs the The principal disadvantage of using a distinction between Megachiroptera and Mi- crown-clade-restricted de®nition is that it ef- crochiroptera. Sophisticated laryngeal echo- fectively excludes some Eocene taxa from location, the single trait most often associ- Microchiroptera simply because they are ex- ated with living microchiropterans, was ap- tinct. Given their morphology, relationships, parently absent in the most recent common and presumed ecologies (see above), it is ancestor of Megachiroptera and Microchi- likely that Icaronycteris, Archaeonycteris, roptera. This implies that echolocation was Hassianycteris, and Palaeochiropteryx also absent in the very earliest members of would be considered microchiropterans were the microchiropteran stem group. Use of a they alive today. Without a formal name stem-group de®nition for Microchiroptera linking these forms to extant Microchirop- would thus force us to include nonecholo- tera, considerable phylogenetic (and inferred cating relatives of extant microchiropterans behavioral and ecological) information in Microchiroptera, a solution that many would be lost from any classi®cation scheme. workers would not ®nd acceptable. An alternative node-based de®nition of Still another issue involves what names (if Microchiroptera could include the Eocene any) should be applied to nodes along the groups (Icaronycteridae, Archaeonycteridae, backbone of the tree below the microchirop- Hassianycteridae, and Palaeochiropterygi- teran crown group. For example, there are dae) in Microchiroptera if the latter was de- considerable morphological (and inferred ®ned as the clade stemming from the most ecological) differences between the basal recent common ancestor of Icaronycteridae branches (e.g., Icaronycteridae and Archaeo- and Yangochiroptera (or some other extant nycteridae) and the more derived clade in- clade). Advantages of this de®nition include cluding Hassianycteridae, Palaeochiroptery- concurrence with current usage as applied by gidae, and the microchiropteran crown most paleontologists (e.g., Habersetzer and group. Monophyly of the latter clade is well Storch, 1987, 1988, 1992; Novacek, 1987, supported, and a formal name for this group 1991; Carroll, 1988; Storch and Habersetzer, would draw attention to its existence and fa- 1988; Hand et al., 1994; McKenna and Bell, cilitate discussion of this group, objectives 1997), and congruence of this de®nition with that we consider desirable. Although provid- ideas about the ``key character'' of Micro- ing names for every node in a tree is exces- chiropteraÐsophisticated laryngeal echolo- sive, some previously unnamed clades do, in cation, which we infer was present in Ica- our view, require formal recognition. ronycteris, Archaeonycteris, Hassianycteris, Keeping all of these issues in mind, we pro- and Palaeochiropteryx. However, this de®- pose a new higher-level classi®cation for bats nition leaves us without a formal name for (table 8) that combines both node-based and the microchiropteran crown group, which is stem-based de®nitions in a way that we think the clade of central interest to most biolo- maximizes both utility and stability. This so- gists. Moreover, this de®nition might prove lution is similar to that proposed by Wyss and problematic if additional bat fossils are found Meng (1996) for gliriform mammals (rodents, that fall outside the Icaronycteridae ϩ Yan- lagomorphs, and their extinct relatives). In our gochiroptera clade, yet stem from the branch classi®cation, we apply stem-based names to leading to Microchiroptera rather than that the two principal branches of the chiropteran leading to Megachiroptera. tree. Microchiropteramorpha is de®ned as all This problem raises yet another possibility, chiropterans sharing a more recent common namely a stem-based approach. Microchirop- ancestor with Microchiroptera than with Me- tera might be de®ned as all chiropterans shar- gachiroptera; Megachiropteramorpha is de- ing a more recent common ancestor with ®ned as all chiropterans sharing a more recent Yangochiroptera than with Pteropodidae. common ancestor with Megachiroptera than 136 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

TABLE 8 TABLE 8Ð(Continued) A Higher-level Classi®cation of Bats Including Selected Fossil Genera Discussed in the Texta Family Thyropteridae Family Natalidae Order Chiroptera Superfamily Molossoidea Megachiropteramorpha, new taxon Family Antrozoidae ²Archaeopteropus Family Molossidae Suborder Megachiroptera Subfamily Tomopeatinae Family Pteropodidae Subfamily Molossinae Microchiropteramorpha, new taxon ²Wallia, ²Cuvierimops ²Australonycteris Superfamily Vespertilionoidea ²Family Icaronycteridae Family Vespertilionidae ²Icaronycteris Subfamily Vespertilioninae ²Family Archaeonycteridae Subfamily Miniopterinae ²Archaeonycteris Subfamily Myotinae Microchiropteraformes, new taxon Subfamily Murininae ²Eppsinycteris Subfamily Kerivoulinae ²Family Palaeochiropterygidae a See text for a discussion of classi®cation below the ²Palaeochiropteryx, ²Matthesia, ²Cecilionycteris level of subfamily. Genera that are not placed in a spe- ²Family Hassianycteridae ci®c family or subfamily are considered incertae sedis ²Hassianycteris within the next higher-level taxon listed above (e.g., Suborder Microchiroptera Stehlinia, which is listed under superfamily Nataloidea ²Vampyravus without reference to family or subfamily, is considered Superfamily Emballonuroidea incertae sedis within Nataloidea). Family Emballonuridae Subfamily Taphozoinae ²Vespertiliavus with Microchiroptera. As so de®ned, Micro- Subfamily Emballonurinae Infraorder Yinochiroptera chiropteramorpha includes Icaronycteridae, Superfamily Rhinopomatoidea Archaeonycteridae, Hassiancyteridae, Palaeo- Family Craseonycteridae chiropterygidae, and Microchiroptera, as well Family Rhinopomatidae as a number of poorly known fossil forms Superfamily rhinolophoidea (e.g., Australonycteris) that are more closely Family Nycteridae related to extant Microchiroptera than to Me- Family Megadermatidae gachiroptera. ²Necromantis Within Microchiropteramorpha, Microchi- Family Rhinolophidae ropteraformes is de®ned as the clade stem- ²Vaylatsia ming from the most recent common ancestor Subfamily Rhinolophinae Subfamily Hipposiderinae of Hassianycteridae, Palaeochiropterygidae, ²Hipposideros (Pseudorhinolophus), and Microchiroptera; this is a node-based def- ²Palaeophyllophora, ²Paraphyllophora inition. Any fossil forms that nest within Infraorder Yangochiroptera these groups or fall between them on the tree ²Family Philididae would be placed in Microchiropteraformes. ²Philisis, ²Dizzya Australonycteris may someday be shown to Family Mystacinidae belong to this group, but for now it must be Superfamily Noctilionoidea considered Microchiropteramorpha incertae Family Phyllostomidae sedis. Family Moromoopidae Finally, we de®ne (and thereby restrict) Family Noctilionidae Superfamily Nataloidea Microchiroptera to the clade stemming from ²Honrovits the most recent common ancestor of Embal- ²Aegina lonuridae and Yangochiroptera. This node- ²Stehlinia based, crown-group-restricted de®nition ²Chadronycteris equates Microchiroptera with what we have Family Myzopodidae thus far in this paper called the ``microchi- Family Furipteridae ropteran crown group''Ðthe smallest clade that comprises all of the extant lineages of 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 137 microchiropteran bats. As noted above, the This compromise allows the formal recog- advantages of this de®nition of Microchirop- nition of names for clades at numerous levels tera are that it is congruent with general us- in the tree, while at the same time promoting age of the name by biologists, and it provides stability by retaining most familiar names at stability because future fossil discoveries or familiar ranks (e.g., Microchiroptera does not extinctions will not affect the limits of this have to be reduced to the rank of superfamily taxon. The principal disadvantage of this def- to accommodate recognition of more inclu- initionÐexclusion of the Eocene fossil taxa sive taxa within Chiroptera). Although this largely because they are extinctÐis mitigat- approach will probably be criticized by ad- ed by our recognition of Microchiroptera- herents to both the ``no rank'' and ``always morpha and Microchiropteraformes, which rank'' schools of classi®cation, we think that focuses attention on the microchiropteran af- it is the only reasonable solution for the tax- ®nities of these taxa without squeezing them onomic problems presented by Chiroptera. into Microchiroptera itself. To avoid proliferation of redundant names, This system of classi®cation, which rec- we currently limit the use of traditional in- ognizes two named groups more inclusive fraordinal and superfamilial rank names to than Microchiroptera and less inclusive than within Microchiroptera. We follow Simmons Chiroptera, presents problems from the point (1998) in recognizing seven monophyletic of view of ranks, since no ranks are recog- superfamilies and two infraorders within Mi- nized to exist between order and suborder. crochiroptera. All of these taxa are recog- This sort of problem has haunted systema- nized on the basis of node-based de®nitions. tists for decades, most particularly since the The superfamilies we recognize are Embal- advent of cladistic methods for phylogenetic lonuroidea (Emballonuridae), Rhinopomato- analysis. Some workers have advocated idea (Rhinopomatidae ϩ Craseofycteridae), elimination of all ranks in favor of a simple Rhinolophoidea (Nycteridae ϩ Megaderma- hierarchical system (e.g., de Queiroz and tidae ϩ Rhinolophidae), Noctilionoidea Gauthier, 1992), while others have main- (Noctilionidae ϩ Mormoopidae ϩ Phyllos- tained fully ranked classi®cations despite tomidae), Nataloidea (Myzopodidae ϩ Thy- methodological problems that arise from too ropteridae ϩ Furipteridae ϩ Natalidae), Mo- many names and too few ranks (e.g., Mc- lossoidea (Antrozoidae ϩ Molossidae), and Kenna, 1975; Wilson and Reeder, 1993; Mc- Vespertilionoidea (Vespertilionidae). Yino- Kenna and Bell, 1997). We adopt an inter- chiroptera contains Rhinopomatoidea ϩ mediate position. While recognizing that for- Rhinolophoidea; Yangochiroptera contains mal ranks are extremely useful for commu- Mystacinidae ϩ Noctilionoidea ϩ Nataloidea nication, arrangement of systematic ϩ Molossoidea ϩ Vespertilionoidea. Not all collections, and other bookkeeping chores families are referred to superfamily-level (e.g., bibliographic projects), we also think taxa (e.g., Mystacinidae is left incertae sedis that the formal ranks available (even the rel- in Yangochiroptera), and not all superfami- atively large number recognized by McKen- lies are referred to infraorders (e.g., Embal- na and Bell, 1997) cannot suf®ce to describe lonuroidea) because we see no purpose in the complex hierarchies now being recovered providing redundant names. The reason that from phylogenetic analyses. Accordingly, we we recognize two monotypic superfamilies choose to recognize both ranked and unran- (Emballonuroidea and Vespertilionoidea) is ked names. We retain ranks for many taxo- that we anticipate that future workers may nomic names, particularly those that apply subdivide these groups into multiple families principally to extant taxa and are in broad as phylogenetic resolution increases in these use in the neontological literature (e.g., Order parts of the tree. Chiroptera, Suborder Microchiroptera, Fam- Classi®cation of most of the remaining ily Phyllostomidae). For other clade names, Eocene taxa is relatively straightforward giv- particularly those that we anticipate will be en what is known about these forms and the used principally for discussions of fossils classi®cation proposed above. Following the (e.g., Microchiropteramorpha, Microchirop- conclusions of previous authors (see the In- teraformes), we do not propose formal ranks. troduction), we place Vespertiliavus in Em- 138 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 ballonuridae: Taphozoinae, and refer Necro- form, and our survey of dental variation sug- mantis to Megadermatidae (table 8). Similarly, gests that a bilobed p4 is common in many we follow the consensus that Hipposideros families of bats. In our opinion, no feature (Pseudorhinolophus), Palaeophyllophora, of Eppsinycteris (as it is currently known) and Paraphyllophora should be classi®ed as clearly link it with any extant family. How- Rhinolophidae: Hipposiderinae. Rather than ever, position of the hypoconulid on the low- placing Vaylatsia in Hipposiderinae as did er molars indicates that Eppsinycteris was ei- Sige (1990), we follow McKenna and Bell ther nyctalodont or myotodont (the teeth are (1997) and refer Vaylatsia to Rhinolophidae too worn to determine which morphology is incertae sedis, although we note that this tax- present), indicating that this genus probably on may subsequently be shown to represent belongs somewhere within the microchirop- a basal member of Rhinolophinae (see dis- teraform clade. On this basis, we place Epps- cussion in Hand and Kirsch, 1998). inycteris in Microchiropteraformes incertae As noted by Simmons (1998), recognition sedis. of Hipposiderinae and Rhinolphinae as sub- In the yangochiropteran part of the tree, family-level taxa (rather than as distinct fam- we refer Philisis and Dizzya to the extinct ilies) is preferred because it facilitates dis- family Philisidae following Sige (1985, cussion of the larger clade comprising these 1991a), which we in turn place incertae sedis taxa (i.e., Rhinolophidae). We do not agree within Yangochiroptera pending additional with McKenna and Bell (1997) that the name study. As noted by Sige (1985), Philisidae Hipposiderinae should be replaced with Rhi- may eventually prove to be closely related to nonycterinae. Although Rhinonycterinae (ϭ Nataloidea if Vampyravus is shown to be a Rhinonycterina Gray, 1866) has priority over synonym of Philisis. As discussed previous- Hipposiderinae Flower and Lydekker, 1891 ly, we refer Honrovits and Aegina to Nata- as a family-group name, no author other than loidea incertae sedis following suggestions Gray (1866) used the former name until it made by Beard et al. (1992). We also refer was resurrected by McKenna and Bell Stehlinia and Chadronycteris to Nataloidea (1997). Miller (1907) used the name Hip- incertae sedis based on recommendations of posideridae for this group in his in¯uential the same authors. We place Wallia and Cu- monograph on the families and genera of vierimops in Molossidae: Molossinae incer- bats because Hipposideros Gray, 1831 has tae sedis following McKenna and Bell priority over Rhinonycteris Gray, 1866 (ϭ (1997) based on arguments developed in Le- Rhinonicteris Gray, 1847). All subsequent gendre and Sige (1983) and Legendre authors have followed Miller's (1907) usage (1985). Finally, we follow Sige (1985) and of Hipposideridae/-inae, and we think that McKenna and Bell (1997) in referring Vam- little would be gained by replacing it with an pyravus to Microchiroptera incertae sedis un- unknown name. Resolution of this problem til more material of this form is discovered. will require a petition to the International Because relationships among extant mega- Commission on Zoological Nomenclature; chiropterans are still relatively poorly under- meanwhile, we retain Hipposiderinae for the stoodÐin part due to incongruence among sake of stability. phylogenies derived from different data sets Although we agree with Hooker (1996) (Hood, 1989; Kirsch et al., 1995; Colgan and that Eppsinycteris is a bat, we do not agree Flannery, 1995; Springer et al., 1995; Hollar that it should be referred to Emballonuridae. and Springer, 1997)Ðwe de®ne Megachirop- Hooker (1996) listed two synamporphies tera using a stem-modi®ed node-based de®- linking Eppsinycteris to emballonurids: pro- nition as recommended by Wyss and Meng gressive mesial elongation of the trigonid (1996). Megachiroptera is thus de®ned as the from m3 to m1, and bilobation of p4. We ®nd clade stemming from the most recent com- neither of these to be convincing evidence of mon ancestor of Pteropus and all Recent emballonurid af®nities. Published photo- mammals more closely related to Pteropus graphs of Eppsinycteris (Hooker, 1996: pls. than to Microchiroptera or any other mam- 1 and 2) illustrate that mesial elongation of malian order or suborder. Megachiroptera as the trigonid is barely distinguishable in this thus de®ned includes all extant Pteropodidae, 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 139

but it may not include Archaeopteropus. Al- placed securely either inside or outside Me- though Archaeopteropus seems to be more gachiroptera as de®ned here. Accordingly, we closely related to living Pteropodidae than to place Archaeopteropus incertae sedis in Me- Microchiroptera, it is too poorly known to be gachiropteramorpha.

CONCLUSIONS AND SUMMARY In an essay on the origin and evolution of Megachiropteramorpha is named for all bats bats, Jepsen (1970: 40) observed that more closely related to Megachiroptera than to Microchiroptera. A skeleton of an old dead bat doesn't give much di- Taken together with an assumption of bat rect information about the details of genic history al- monophyly, the phylogeny developed in our though it may be richly informative about broad evo- lutionary generalities. study provides a basis for evaluating previ- ous hypotheses concerning the evolution of We disagree. As demonstrated in the current ¯ight, echolocation, and foraging strategies. study, careful comparisons of fossil and ex- Through a combination of character mapping tant specimens provide data that yield a well- and examination of correlations between resolved, relatively well-supported hypothe- morphology and behavior, we ®nd support sis of relationships when analyzed using ap- for several linked hyoptheses, many of which propriate cladistic methods. Our analyses in- were derived from the work of earlier au- dicate that Icaronycteris, Archaeonycteris, thors. Based on the data presented here, it is Hassianycteris, and Palaeochiropteryx rep- clear that ¯ight evolved before echolocation; resent a series of consecutive sister-taxa to the ®rst bats most likely used vision for ori- extant microchiropteran bats (®g. 36). This entation in their arboreal/aerial environment. conclusion stands in sharp contrast to previ- The evolution of ¯ight was subsequently fol- ous suggestions that these fossil forms rep- lowed by the origin of low-duty-cycle laryn- resent either a primitive grade ancestral to geal echolocation in early microchiroptera- both Megachiroptera and Microchiroptera morphs. This system was probably derived (e.g., Van Valen's [1979] Eochiroptera), or a from vocalizations originally used for intra- separate clade within Microchiroptera (i.e., speci®c communication. The microchiropter- Smith's [1977] Palaeochiropterygoidea). amorh echolocation system was most likely To better re¯ect observed similarities, dif- simple at ®rst, permitting only orientation ferences, and phylogenetic relationships and obstacle detection and not detection or among both fossil and extant lineages, we tracking of airborne prey. However, the en- have proposed a new higher-level classi®ca- ergy costs of echolocation to ¯ying bats were tion of bats (table 8). This classi®cation com- low due to the mechanical coupling of ven- bines both node-based and stem-based de®- tilation and ¯ight, and the bene®ts of aerial nitions in a way that we think maximizes insectivory apparently led to rapid evolution both utility and stability (see discussion of a more sophisticated low-duty-cycle echo- above). Critical features of this classi®cation location system capable of detecting, track- include restriction of Microchiroptera to the ing, and assessing airborne prey. The need smallest clade including all extant bats that for an increasingly derived auditory sys- use sophisticated echolocation, and formal temÐcombined with limits on body size im- recognition of two more inclusive groups: posed by the mechanics of ¯ight, echoloca- Microchiropteraformes (Microchiroptera ϩ tion, and prey captureÐmay have resulted in Palaeochiropterygidae ϩ Hassianycteridae), reduction and simpli®cation of the visual and Microchiropteramorpha (Microchiropter- system as echolocation became increasingly aformes ϩ Archaeonycteridae ϩ Icaronyc- important. teridae). Megachiroptera is similarly restrict- Our examinations of morphology of the ed to the the smallest clade including all ex- basicranium and auditory region con®rm pre- tant megachiropteran bats (see Simmons vious suggestions by Novacek (1985a, 1987, [1994] for a list of diagonistic features), and 1991) and Habersetzer and Storch (1992) 140 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 that Icaronycteris, Archaeonycteris, Hassi- ronycteris and Archaeonycteris, and in mi- anycteris, and Palaeochiropteryx were ca- crochiropteraform bats. As in ancestral mi- pable of echolocation. Judging from our phy- crochiropteramorphs, detection and tracking logeny, the earliest morphological transfor- of prey would have been accomplished prin- mations in Microchiropteramorpha included cipally by a combination of vision and lis- changes in features associated with produc- tening for prey-generated sounds. tion of echolocation calls, transmission of The earliest microchiropteraforms, Hassi- sounds through the middle ear, and some tun- anycteris and Palaeochiropteryx, differ from ing of the inner ear. Subsequent modi®ca- Icaronycteris and Archaeonycteris in having tions at the level of Microchiropteraformes a larger cochlea and several derived postcra- involved continued tuning of the inner ear, nial features (including a calcar). However, postcranial modi®cations that may have in- these changes do not seem to have been cor- creased ef®ciency of the ventilation system related with any appreciable modi®cations in and its connections to the ¯ight mechanism, wing shape or wing loading. These obser- and evolution of a calcar, which may have vations indicate that a major shift in foraging been linked to aerial insect capture. strategy occurred at the level of Microchi- Foraging strategies of Icaronycteris, Ar- ropteraformes, but that this shift took place chaeonycteris, Hassianycteris, and Palaeo- without any large-scale changes in the habi- chiropteryx were reconstructed based on tat types exploited. The greatly enlarged postcranial osteology and wing form, coch- cochlea and well developed calcar suggest lear size, and stomach contents (table 9). that Hassianycteris and Palaeochiropteryx Comparisons with megachiropterans and out- were fully capable of aerial hawking, using groups also permitted the reconstruction of echolocation to detect, track, and assess prey hypothetical ancestral habits. It seems most and utilizing the uropatagium to capture prey likely that the earliest bats (including the ear- on the wing. This hypothesis is consistent liest microchiropteramorphs) used vision for with the analyses of stomach contents and orientation, and foraged by gleaning insects taphonomy of Hassianycteris and Palaeochi- and perhaps some fruits and other edible ropteryx at Lake Messel. items from foliage, bark, and perhaps the Many extant bats that hunt from perches ground. Detection of potential food items also hawk insects in ¯ight, with allocation of was probably accomplished by a combina- time between these two foraging strategies tion of vision and listening for prey-gener- re¯ecting prey availability. Given that glean- ated sounds. These bats probably had low as- ing from a perch probably represents the pect ratio wings, moderate wing loading, and primitive foraging strategy for the microchi- relatively large wing tips, all of which sug- ropteran lineage, we suspect that Palaeochi- gest that they habitually ¯ew close to the ropterx may have used a combination of ground and near vegetation. This wing mor- perch hunting and aerial hawking to capture phology was subsequently retained in Ar- its prey, much like modern nycterids. Passive chaeopteropus, Icaronycteris, Archaeonyc- means would have been used to locate sta- teris, Palaeochiropteryx, and some extant tionary prey; echolocation would have been megachiropteran and microchiropteran bats used for detecting and tracking ¯ying prey. (see ®gs. 43, 44). This hypothesis is consistent with observa- Wing morphology, moderate cochlea size, tions of cochlear size and wing morphology and absence of a calcar suggest that Icaron- in Palaeochiropterx. ycteris and Archaeonycteris were perch- Hassianycteris, which diverged from the hunting gleaners rather than predators on ae- microchiropteran lineage prior to Palaeochi- rial insects. The ®rst major change in forag- ropteryx, is characterized by relatively large ing methods in Microchiropteramorpha was body size and signi®cantly higher wing load- therefore the addition of echolocation as a ing than any of the other Eocene bats. These tool for orientation and obstacle detection; observations, along with a moderately high the basic strategyÐgleaning from a perchÐ aspect ratio, a greatly enlarged cochlea, and would have been the same as seen in the no- presence of a calcar, suggest that Hassianyc- necholocating chiropteran ancestors of Ica- teris foraged by fast aerial hawking, most TABLE 9 Foraging Ecologies of Eocene Bats: Observations and Hypotheses a

Icaronycteris Archaeonycteris Hassianycteris Palaeochiropteryx Wing morphology Low aspect ratio; low wing tip Low aspect ratio; low wing tip Low aspect ratio; low wing tip Low aspect ratio; low wing tip indices; moderate wing indices; moderate wing indices; high wing loading indices; moderate wing loading loading loading Calcar Absent Absent Present Present Cochlear size Moderate Moderate Large Large Stomach contents Unknown Coleoptera Macrolepidoptera; Coleoptera; Microlepidoptera; Tricoptera; Blattoidea scale-bearing Diptera Proposed mode of Echolocation and vision Echolocation and vision Echolocation and vision Echolocation and vision orientation and obstacle detection Proposed foraging Near ground and vegetation; Near ground and vegetation; Well above ground in forest Near ground and vegetation; habitat along edges of lakes along edges of lakes clearings, above forest along edges of lakes canopy, and over lakes Proposed principal Gleaning prey from surfaces Gleaning prey from surfaces Fast aerial hawking; prey Perch hunting for both ¯ying foraging style(s) during short ¯ights from a during short ¯ights from a capture using calcar-supported and non¯ying prey; slow perch perch uropatagium aerial hawking; prey capture using calcar-supported uropatagium Proposed mode of Vision and/or listening for Vision and/or listening for prey- Echolocation Echolocation (¯ying prey); detection and prey-generated sounds generated sounds vision and/or listening for tracking prey prey-generated sounds (non¯ying prey)

Proposed echolocation Low dutyՅ cycle2 msec) multiharmonic LowShort duty ( cycleՅ2 msec) multiharmonic Low dutyNarrowband cycle FM calls or FM Low dutyBroadband cycle FM calls in 30±90 duty cycle calls or broadband FM calls calls or broadband FM calls calls with both steep and kHz range Proposed type of echo- Shortin ( 30±90 kHz range in 30±90 kHz range shallow sweeps; calls under location call 30 kHz a See text for discussion of the data and chain of inferences used to develop these hypotheses. 142 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 likely well above the ground in forest gaps adjacent to relatively clutter-free open spac- or above the canopy. Hassianycteris thus es. The next evolutionary stage (Stage 4) represents another shift in foraging strategy, brought continued re®nement of the echolo- away from combined perch hunting and ae- cation and ¯ight systems as well as evolution rial hawking near vegetation to a fast-¯ying, of a calcar-supported uropatagium used for continuous aerial hawking foraging strategy capturing prey on the wing. Microchiropter- similar to that seen in extant rhinopomatids aform bats at this stage probably used a com- and some vespertilionids. This shift appar- bination of perch hunting and continuous ae- ently occurred after the lineage leading to rial hawking. Stationary prey would have Hassianycteris diverged from the lineage been dectected by passive means, and aerial leading to the microchiropteran crown group. prey by echolocation. We expect that this In summary, we propose that foraging be- combination of foraging stategies was used havior in the microchiropteran lineage by the common ancestor of Hassianycteris evolved in a series of steps. The earliest mi- and Palaeochiropteryx, and was subsequent- crochiropteramorph bats (Stage 1 in our hy- ly retained in Palaeochiropteryx. Finally, pothesis) probably used vision for orientation Stage 5 brought exclusive reliance on contin- and located prey by a combination of vision uous aerial hawking using echolocation for and listening for prey-generated sounds. prey detection, tracking, and evaluation. This Food objects were probably obtained by foraging strategy was apparently used by gleaning during short ¯ights from a perch. Hassianycteris and the common ancestor of The next evolutionary step (Stage 2) in- extant microchiropterans. volved a switch to the use of echolocation The presence of representatives of three of for orientation but not for prey detection. stages in the evolution of foraging behavior These bats (which included Icaronycteris and (Stages 2, 4, and 5) at Messel suggests that Archaeonycteris) retained the primitive hab- evolutionary transformations in foraging its of locating and tracking their prey by pas- strategies occurred rapidly in early members sive means and gleaning prey from surfaces of the microchiropteran lineage. Given the during short ¯ights from a perch. They may topology of our phylogenetic tree (®g. 36), it have captured ¯ying prey at times, but relied seems likely that reliance on continuous ae- principally on vision and/or prey-generated rial hawking was primitive for the microchi- sounds for locating and tracking prey. Sub- ropteran crown group. This suggests that sequent evolutionary re®nements of the gleaning, passive prey detection, and perch echolocation system began to make detec- hunting among extant microchiropterans rep- tion, tracking, and evaluation of ¯ying prey resent secondarily derived specializations possible, and we expect that this correspond- rather than retentions of primitive habits. ed with a third stage in the evolution of for- Each of these behaviors evolved indepen- aging behaviors, one in which bats continued dently several times in Microchiroptera. to hunt from perches for stationary prey, but The evolution of continuous aerial hawking also used echolocation for detecting and may have been the ``key innovation'' respon- tracking ¯ying prey. To fully exploit this sible for the burst of diversi®cation in micro- method, these bats would have had to in- chiropteran bats that occurred in the Eocene. crease signal strength to maximize range and Fossils referable to six major extant lineages provide the necessary time to track and eval- are known from Middle±Late Eocene deposits uate targets. Environmental clutter (which (table 1), and reconstruction of ghost lineages produces many distracting echoes) would leads to the conclusion that at least seven have presented dif®culties at this stage, so we more extant lineages were minimally present suggest that ¯ycatching using echolocation to by the end of the Eocene. It thus appears that detect and track prey probably evolved in Icaronycteris, Archaeonycteris, Palaeochi- bats that frequented forest gaps and the edges ropteryx, and Hassianycteris provide an un- of forests along lakes and rivers, places precedented view of steps leading to a major where vegetation (with potential perches) lies adaptive radiation of mammals. 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 143

ACKNOWLEDGMENTS

Special thanks go to S. Bell, K. Bhatnagar, Darmstadt), G. Habersetzer and G. Storch B. Blood, W. Bogdanowicz, J. Cypher, B. (Senckenbergmuseum, Frankfurt am Main), Fenton, T. Grif®ths, S. Hand, J. Kirsch, K. M. Hafner (Louisiana State Museum of Zo- Koopman, B. Lloyd, M. McKenna, M. No- ology), P. Jenkins (British Museum of Nat- vacek, S. Parsons, A. Pef¯ey, W. Schutt, S. ural History), B. Patterson (Field Museum of Swartz, J. Wible, H. Whidden, and G. Wil- Natural History), and F. Schwab (Hessisches kinson for sharing unpublished data and Landesmuseum, Darmstadt, Germany) for manuscripts with us. T. Grif®ths was es- permission to study specimens in their care. pecially kind in modifying his research pro- Our sincere thanks to G. Storch for permis- gram to collect data relevant to this project; sion to use the photographs in ®gures 2, 3, without this help, our data set would have 4, and 24, and to M. Novacek for permission been much smaller and our results less sat- to use the photographs in ®gures 23, 25, and isfactory. We also thank W. Clemens, J. Gau- 27. We also thank P. Wynne for her ®ne il- thier, K. Koopman, M. McKenna, T. Rowe, lustrations, and L. Meeker and C. Tarka for and R. Voss for many stimulating discussions their expert preparation, photographs, and regarding principles of classi®cation, and B. drawings of the Wyoming specimen of Ica- Fenton and E. Kalko for extensive discus- ronycteris. P. Brunauer spent long hours in sions of the evolution of echolocation and the library locating references cited in this foraging strategies. Although we are well contribution, for which we offer our sincere aware that these individuals may not agree thanks. Finally, special thanks go to R. Voss, with our conclusions, their viewpoints were without whose support and patience this critical to the development of our ideas on monograph could not have been written. these subjects. T. Conway, B. Fenton, E. Kal- This study was supported by NSF Re- ko, K. Koopman, M. McKenna, A. Pef¯ey, search Grant DEB-9106868 to NBS, a NSF W. Schutt, S. Swartz, and J. Wible reviewed Graduate Fellowship to JHG, a Graduate Fel- earlier versions of this manuscript, and we lowship from the Department of Earth and thank them for their comments and willing- Environmental Science at Columbia Univer- ness to review revised text. Thanks also go sity to JHG, and Summer Support from the to B. Breithaupt (University of Wyoming), P. AMNH Department of Vertebrate Paleontol- Bultynck and A. Dhondt (Institut Royal des ogy to JHG. NSF Research Experiences for Sciences Naturelles de Belgique), M. Carle- Undergraduates Grant BIO-9200351 to M. ton, C. Handley, and L. Gordon (United Stiassny provided funding to JHG during the States National Museum), B. Engesser (Na- ®rst year of this project; we thank M. Stiass- turhistorisches Museum Basel, Switzerland), ny and D. Bynum for their help and support E. Frey (Landesmuseum fuÈr Naturkunde, of the AMNH REU program.

REFERENCES

Adkins, R. M., and R. L. Honeycutt oxidase subunit II gene. J. Mol. Evol. 1991. Molecular phylogeny of the superorder 38: 215±231. Archonta. Proc. Natl. Acad. Sci. U.S.A. Agrawal, V. C., and Y. P. Sinha 88: 10317±10321. 1973. Studies on the bacula of some Oriental 1993. A molecular examination of archontan bats. Anat. Anzeiger 133: 180±192. and chiropteran monophyly. In R. D. E. Aitkin, L. M. MacPhee (ed.), Primates and their rel- 1986. Sensory processing in the mammalian atives in phylogenetic perspective, pp. auditory systemÐsome parallels and 227±249. Advances in Primatology Se- contrasts with the visual system. In J. ries. New York: Plenum. D. Pettigrew, K. J. Sanderson, and W. 1994. Evolution of the primate cytochrome R. Leviak (eds.), Visual neuroscience, 144 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

pp. 223±235. Cambridge: Cambridge Audet, D., D. Krull, G. Marimuthu, S. Sumithran, Univ. Press. and J. B. Singh Aldridge, H.D.J.N. 1988. Foraging strategies and use of space by 1986. Manoeuvrability and ecology in British the Indian false vampire, Megaderma bats. Myotis 23±24: 157±160. lyra (Megadermatidae). Bat Res. News Aldridge, H.D.J.N., and I. L. Rautenbach 29: 43. 1987. Morphology, echolocation and resource Baagùe, H. J. partitioning in insectivorous bats. J. 1987. The scandanavian bat fauna: adaptive Anim. Ecol. 56: 763±778. wing morphology and free ¯ight behav- Allard, M. W., B. E. McNiff, and M. M. Miya- ior in the ®eld. In M. B. Fenton, P. A. moto Racey, and J. M. V. Rayner (eds.), Re- 1996. Support for interordinal eutherian rela- cent advances in the study of bats, pp. tionships, with an emphasis on Pri- 57±74. Cambridge: Cambridge Univ. mates and their archontan relatives. Press. Mol. Phylogenet. Evol. 5: 78±88. Bailey, W. J., Slightom, J. L., and Goodman, M. Altenbach, J. S., and J. W. Hermanson 1992. Rejection of the ``¯ying primate'' hy- 1987. Bat ¯ight muscle function and the sca- pothesis by phylogenetic evidence from pulo-humeral lock. In M. B. Fenton, P. the ⑀-globin gene. Science 256: 86±89. Racey, and J. M. Rayner (eds.) Recent Baker, R. J., R. L. Honeycutt, and R. A. Van Den advances in the study of bats, pp. 100± Bussche 118. Cambridge: Cambridge Univ. 1991a. Examination of monophyly of bats: re- Press. striction map of the ribosomal DNA Altringham, J. D. cistron. Bull. Am. Mus. Nat. Hist. 206: 1996. Bats: biology and behavior. Oxford: 42±53. Oxford Univ. Press. Baker, R. J., M. J. Novacek, and N. B. Simmons Ammerman, L. K., and D. M. Hillis 1991b. On the monophyly of bats. Syst. Zool. 1992. A molecular test of bat relationships: 40: 216±231. monophyly or diphyly? Syst. Biol. 41: Barclay, R. M. R. 222±232. 1985. Long- versus short-range foraging Andersen, K. strategies of hoary (Lasiurus cinereus) 1912. Calalogue of the Chiroptera in the col- and silver-haired (Lasionycteris nocti- lection of the British Museum. 2nd ed. vagans) bats and the consequences for Volume I: Megachirptera. London: prey selection. Can. J. Zool. 63: 2507± British Museum. [Reprinted by John- 2525. son Reprint Corporation, New York.] 1986. The echolocation calls of hoary (Lasi- Arita, H. T., and M. B. Fenton urus cinereus) and silver-haired (La- 1997. Flight and echolocation in the ecology sionycteris noctivagans) bats as adap- and evolution of bats. Trends Ecol. tations for long- versus short-range for- Evol. 12: 53±58. aging strategies and the consequences Armstrong, E. for prey selection. Can. J. Zool. 64: 1983. Relative brain size and metabolism in 2507±270. mammals. Science 220: 1302±1304. 1994. Constraints on reproduction by ¯ying Aubry, M.-P. vertebrates: energy and calcium. Am. 1983. Biostratigraphie du PaleÂogeÁne eÂpicon- Nat. 144: 1021±1031. tinental de l'Europe du Nord-ouest eÂtu- 1995. Does energy or calcium availability de fondeÂe sur les nannofossiles calcai- constrain reproduction by bats? In P. A . res. Docum. Lab. GeÂol. Lyon 89: 1± Racey and S. M. Swift (eds.), Ecology, 317. evolution and behavior of bats. Symp. Audet, D. Zool. Soc. London 67: 245±258. 1990. Foraging behavior and habitat use by a Barclay, R. M. R., and R. M. Brigham gleaning bat, Myotis myotis (Chirop- 1991. Prey detection, dietary niche breadth, tera: Vespertilionidae). J. Mammal. 71: and body size in bats: why are aerial 420±427. insectivorous bats so small? Am. Nat. Audet, D., and M. B. Fenton 137: 693±703. 1988. Heterothermy and the use of torpor by Barclay, R. M. R., M. B. Fenton, M. D. Tuttle, the bat Eptesicus fuscus (Chiroptera: and M. J. Ryan Vespertilionidae): a ®eld study. Physi- 1981. Echolocation calls produced by Trach- ol. Zool. 61: 197±204. ops cirrhosus (Chiroptera, Phyllosto- 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 145

matidae) while hunting for frogs. Can. dae). Behav. Ecol. Sociobiol. 16: 343± J. Zool. 59: 750±753. 347. Barghoorn, S. F. Bell, G. P., and M. B. Fenton 1977. New material of Vespertiliavus Schlos- 1986. Visual acuity, sensitivity and binocular- ser (Mammalia, Chiroptera) and sug- ity in a gleaning insectivorous bat, Ma- gested relationships of emballonurid crotus californicus (Chiroptera: Phyl- bats based on cranial morphology. Am. lostomidae). Anim. Behav. 34: 409± Mus. Novitates 2618: 29 pp. 414. Barkley, L. J. Belwood, J. J., and G. K. Morris 1984. Evolutionary relationships and natural 1987. Bat predation and its in¯uence on call- history of Tomopeas ravus (Mammalia: ing behavior in Neotropical katydids. Chiroptera). Master's thesis, Louisiana Science 238: 64±67. State Univ., Baton Rouge, Lousiana. Benton, M. J., and R. Hitchin Baron, G., H. Stephan, and H. D. Frahm 1997. Congruence between phylogenetic and 1996a. Comparative neurobiology in Chirop- stratigraphic data on the history of life. tera. Vol. 1: Macromorphology, brain Proc. R. Soc. London B 264: 885±890. structure, tables, and atlases. Basel: Berggren, W. A., D. V. Kent, J. J. Flynn, and J. BirkhaÈuser Verlag. A. Van Couvering 1996b. Comparative neurobiology in Chirop- 1985. Cenezoic geochronology. Geol. Soc. tera. Vol. 2: Brain characteristics in tax- Am. Bull. 96: 1407±1418. onomic units. Basel: BirkhaÈuser Verlag. Berggren, W. A., D. V. Kent, C. C. Swisher, and M.-P. Aubry 1996c. Comparative neurobiology in Chirop- 1995. A revised Cenozoic geochronology and tera. Vol. 3: Brain characteristics in chronostratigraphy. In: W. A. Berggren, functional systems, ecoethological ad- D. V. Kent, M. -P. Aubry, and J. Har- aptation, adaptive radiation and evolu- denbol (eds.), Geochronology, time tion. Basel: BirkhaÈuser Verlag. scales and global stratigraphic correla- Barrett, M., M. J. Donoghue, and E. Sober tion. Tulsa, Oklahoma: Society for Sed- 1991. Against consensus. Syst. Zool. 40: imentary Geology Special Publication 486±493. 54: 129±212. 1993. Crusade? A reply to Nelson. Syst. Biol. Betz, E. 42: 216±217. 1958. Untersuchungen uÈber die Korrelation Beard, K. C. der Flugmechanismus bei den Chirop- 1993. Phylogenetic systematics of Primato- teren. Zool. Jahrb. Abt. Anat. 77: 491± morpha, with special reference to Der- 526. moptera. In F. S. Szalay, M. J. Nova- Bhatnagar, K. P., and J. R. Wible cek, and M. C. McKenna (eds.), Mam- 1994. Observations on the vomeronasal organ mal phylogeny: Placentals, pp. 129± of the colugo Cynocephalus (Mamma- 150. New York: Springer-Verlag. lia, Dermoptera). Acta Anat. 151: 43± Beard, K. C., B. SigeÂ, and L. Krishtalka 48. 1992. A primitive vespertilionoid bat from Bhiwgade, D. A., A. B. Singh, A. P. Manekar, and the early Eocene of central Wyoming. S. N. Menon C. R. Acad. Sci. Paris 314: 735±741. 1992. Ultrastructural development of chorio- Bell, G. P. allantoic placenta in the Indian miniop- 1982a. Behavioral and ecological aspects of terus bat, Miniopterus schreibersii fu- gleaning by a desert insectivorous bat, liginosus (Hodgson). Acta Anat. 145: Antrozous palllidus (Chiroptera: Ves- 248±264. pertilionidae). Behav. Ecol. Sociobiol. Black, H. L. 10: 217±223. 1974. A north temperate bat community: 1982b. Prey location and sensory ecology of structure and prey populations. J. Mam- two species of gleaning, insectivorous mal. 55: 138±157. bats, Antrozous pallidus (Vespertilioni- Blood, B. R. dae) and Macrotus californicus (Phyl- 1987. Convergent hind limb morphology and lostomatidae). Ph.D. dissertation, the evolution of ®sh-catching in the Carleton Univ., Ottawa, Canada. bats Noctilio leporinus and Myotis (Pi- 1985. The sensory basis of prey location by zonyx) vivesi (Mammalia, Chiroptera). the California leaf-nosed bat Macrotus Ph.D. dissertation, Univ. of Southern californicus (Chiroptera: Phyllostomi- California. 146 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Bock, W. J. (Myotis lucifugus). Behav. Ecol. Socio- 1974. Philosophical foundations of classical biol. 6: 211±218. evolutionary taxonomy. Syst. Zool. 22: BuÈhler, P. 375±392. 1992. Light bones in birds. In K. E. Campbell Bradbury, J. W., and L. H. Emmons (ed.), Papers in avian paleontology 1974. Social organization of some Trinidad honoring Pierce Brodkorb, pp. 385± bats. I. Emballonuridae. Z. Tierpsychol. 394. Los Angeles Co. Nat. Hist. Mus. 36: 137±183. Sci. series 36. Bradbury, J. W., and F. Nottebohm Bull, J. J., J. P. Hulsenbeck, C. W. Cunningham, 1969. The use of vision by the Little brown D. L. Swofford, and P. J. Waddell bat, Myotis lucifugus, under controlled 1993. Partitioning and combining data in phy- conditions. Anim. Behav. 17: 480±485. logenetic analysis. Syst. Biol. 42: 384± Bremer, K. 397. 1988. The limits of amino acid sequence data Burda, H., and L. Ulehlova in angiosperm phylogenetic reconstruc- 1983. Cochlear hair-cell populations and lim- tions. Evolution 42: 795±803. its of resolution of hearing in two ves- Bronchti, G., R. Rado, J. Terkel, and Z. Wolberg pertilionid bats, Nyctalus noctula and 1991. Retinal projections in the blind mole Eptesicus serotinus. J. Morphol. 176: rat: a WGA-HPR tracing study of a nat- 221±224. ural degeneration. Devel. Brain Res. Butler, P. M. 58: 159±170. 1978. Insectivora and Chiroptera. In V. J . Brown, P. E., T. Brown, and A. D. Grinell Maglio and H. B. S. Cooke (eds.), Evo- 1983. Echolocation, development and vocal lution of African mammals, pp.56±68. communication in the lesser bulldog Cambridge: Harvard Univ. Press. bat, Noctilio albiventris. Behav. Ecol. 1980. The tupaiid dentition. In W. P. Luckett Sociobiol. 13: 287±298. (ed.), Comparative biology and evolu- Brown, R. E., H. H. Genoways, and J. K. Jones tionary relationships of tree shrews, pp. 1971. Bacula of some Neotropical bats. 171±204. New York: Plenum. Mammalia 35: 456±464. 1984. Macroscelidea, Insectivora and Chirop- Bruns, V. tera from the Miocene of East Africa. 1979. Functional anatomy as an approach to Palaeovertebrata 14: 117±200. frequency analysis in the mammalian Carroll, R. L. cochlea. Verhandl. Deutch. Zool. Ge- 1988. Vertebrate paleontology and evolution. sellsch. 1979: 141±154. New York: Freeman. Bruns, V., J. Fiedler, and H.-J. Kraus Cartmill, M., and R. D. E. MacPhee 1983±1984. Structural diversity of the inner 1980. Tupaiid af®nities: the evidence of the ear of bats. Myotis 21±22: 52±61. carotid arteries and cranial skeleton. In Bruns, V., M. M. Henson, H.-J. Kraus, and J. Fei- W. P. Luckett (ed.), Comparative biol- dler ogy and evolutionary relationships of 1981. Vergleichende und funktionelle Mor- tree shrews, pp. 95±132. New York: phologie der Fledermaus-Cochlea. My- Plenum. otis 18±19: 90±105. Casagrande, V. A., and J. K. Brunso-Bechtold Bryant, H. N. 1985. Development of lamination in the lat- 1994. Comments on the phylogenetic de®ni- eral geniculate nucleus: critical factors. tion of taxon names and conventions In: R. N. Aslin (ed.), Advances in neu- regarding the naming of crown clades. ral and behavioral development, pp. Syst. Biol. 43: 124±130. 33±78. Norwood, New Jersey: Ablex. 1996. Explictness, stability, and universality Casseday, J. H., and G. D. Pollak in the phylogenetic de®nition of taxon 1988. Parallel auditory pathways: IÐstruc- names: a case study of the phylogenetic ture and connections. In P. E. Nachtigill taxonomy of the Carnivora (Mamma- and P. W. B. Moore (eds.), Animal so- lia). Syst. Biol. 45: 174±189. nar: processes and performance, pp. Buchler, E. R. 169±196. New York: Plenum. 1976. The use of echolocation by the wan- Chase, J. dering shrew (Sorex vagrans). Anim. 1972. The role of vision in echolocating bats. Behav. 24: 858±873. Ph.D. thesis, Univ. of Indiana, Bloo- 1980. The development of ¯ight, foraging, mington. and echolocation in the little brown bat 1980. Rat echolocation: correlations between 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 147

object detection and click production. France) d'un gisement aÁ verteÂbreÂs con- In: R. G. Busnel, and J. F. Fish (eds.), tinentaux d'aÃge EoceÁne moyen. Cour. Animal sonar systems, pp. 875±877. Forsch.-Inst. Senkenberg 107: 419± New York: Plenum. 434. 1981. Visually guided escape responses of Crochet, J.-Y., J.-L. Hartenberger, J.-C. Rage, J. microchiropteran bats. Anim. Behav. A. ReÂmy, B. SigeÂ, J. Sudre, and M. Vi- 29: 708±713. aney Liaud Chase, J., and R. D. Suthers 1981. Les nouvelles faunes de verteÂbreÂs an- 1969. Visual obstacle avoidance by echolo- teÂrieures a la ``Grande Coupure'' deÂ- cating bats. Anim. Behav. 17: 201±207. couvertes dans les phosphorites du Childs, S. B., and E. R. Buchler Quercy. Bull. Mus. Nation. Hist. Nat., 1981. Perception of simulated stars by Eptes- Paris 4e seÂr., 3 sec. C, 3: 245±266. icus fuscus (Vespertilionidae): a poten- Cuvier, G. tial navigational mechanism. Anim. 1822. Recherches sur les ossements fossiles. Behav. 29: 1028±1035. 2eÁeÂd. Paris: Dufour. Chippindale, P. T., and J. J. Weins Cypher, J. L. 1994. Weighting, partitioning, and combining 1996. Phylogenetic analysis of the order Chi- characters in phylogenetic analysis. roptera using carpal morphology. Mas- Syst. Biol. 43: 278±287. ter's thesis, Eastern Michigan Univ., Clairambault, P., M.-J. Cordier-Picouet, and C. Ypsilanti, Michigan. Pairault Dallos, P. 1980. PremieÁres donneÂes sur les projections 1973. The auditory periphery: biophysics and visuelles d'un amphibiens apode (Ty- physiology. New York: Academic phlonectes compressicauda). C. R. Press. Acad. Sci. Ser. D 291: 283±286. Daniel, M. J. Colgan, D. J., and T. F. Flannery 1976. Feeding by the short-tailed bat (Mys- 1995. A phylogeny of Indo-west Paci®c Me- tacina tuberculata) on fruit and possi- gachiroptera based on ribosomal DNA. Syst. Biol. 44: 209-220. bly nectar. New Zealand J. Zool. 3: Cooper, H. M., M. Herbin, and E. Nevo 391±398. 1993a. Visual system of a naturally blind mi- 1979. The New Zealand short-tailed bat, Mys- crophthalmic mammal: the blind mole tacina tuberculata: a review of present rat, Spalax ehrenbergi. J. Comp. Neu- knowledge. New Zealand Jour. Zool. 6: rol. 328: 313±350. 357±370. 1993b. Ocular regression conceals adaptive 1990. Family Mystacinidae. In C. M. King progression of the visual system in a (ed.), The handbook of New Zealand blind subterranean mammal. Nature mammals, pp. 122±136. Auckland: Ox- 361: 156±159. ford Univ. Press. Corbet, G. B., and J. E. Hill Davis, D. D. 1992. The mammals of the Indomalayan re- 1938. Notes on the anatomy of the treeshrew gion: A systematic review. Oxford: Ox- Dendrogale. Zool. Ser. Field Mus. Nat. ford Univ. Press. Hist. 20: 383±404. Cranbrook, Earl of, and Lord Medway Davis, R. 1965. Lack of ultrasonic frequencies in the 1966. Homing performance and homing abil- calls of swiftlets. Ibis 107: 259. ity in bats. Ecol. Monogr. 36: 201±237. Crerar, L. M., and M. B. Fenton Davis, W. B. 1984. Cervical vertebrae in relation to roost- 1970. Tomopeas ravus Miller (Chiroptera). J. ing posture in bats. J. Mammal. 65: Mammal. 51: 244±247. 395±403. Davis, W. H., and R. W. Barbour Crespo, D., D. D. M. O'Leary, and W. M. Cowan 1965. The use of vision in ¯ight by the bat 1985. Changes in the number of optic nerve Myotis sodalis. Am. Midl. Nat. 74: ®bers during late prenatal and postnatal 497±499. development in the albino rat. Dev. Deacon, T. W. Brain Res. 19: 129±134. 1990a. Problems of ontogey and phylogeny in Crochet, J.-Y., M. Godinot, J.-L. Hartenberger, J. brain size evolution. Int. J. Primatol. A. Remy, B. SigeÂ, and J. Sudre 11: 237±282. 1988. DeÂcouverte dans le bassin de Saint- 1990b. Rethinking mammalian brain evolu- martin-de-Londres (HeÂrault, Sud de la tion. Am. Zool. 30: 629±705. 148 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 de Queiroz, A. Easter, S. S., A. C. Rusoff, and P. E. Kish 1993. For consensus (sometimes). Syst. Biol. 1981. The growth and organization of the op- 42: 368±372. tic nerve and tract in juvenile and adult de Queiroz, A., M. J. Donoghue, and J. Kim gold®sh. J. Neurosci. 1: 793±811. 1995. Separate versus combined analyses of Eernisse, D. J., and A. G. Kluge phylogenetic evidence. Ann. Rev. Ecol. 1993. Taxonomic congruence versus total ev- Syst. 26: 657±681. idence, and amniote phylogeny inferred de Queiroz, K. from fossils, molecules, and morphol- 1994. Replacement of an essentialistic per- ogy. Mol. Biol. Evol. 10: 1170±1195. spective on taxonomic de®ntions as ex- Eisenberg, J. F. empli®ed by the de®nition of ``Mam- 1981. The mammalian radiations: an analysis malia.'' Syst. Biol. 43: 497±510. of trends in evolution, adaptation, and de Queiroz, K., and J. Gauthier behavior. Chicago: Univ. Chicago 1990. Phylogeny as a central principle in tax- Press. onomy: phylogenetic de®nitions of tax- Emde, G. von der, and H. U. Schnitzler on names. Syst. Biol. 39: 307±322. 1986. Fluttering target detection in hipposi- 1992. Phylogenetic taxonomy. Annu. Rev. derid bats. J. Comp. Physiol. A 159: Ecol. Syst. 23: 449±480. 765±722. 1994. Toward a phylogenetic system of bio- 1990. Classi®cation of insects by echolocat- logical nomenclature. Trends Ecol. ing greater horseshoe bats. J. Comp. Evol. 9: 27±31. Physiol. A 167: 423±430. Dobson, G. E. Estabrook, G. F. 1978. Some concepts for the estimation of 1875. Conspectus of the suborders, families evolutionary relationships in systematic and genera of Chiroptera arranged ac- botany. Syst. Bot. 3: 146±158. cording to their natural af®nities. Ann. 1986. Evolutionary classi®cation using con- Mag. Nat. Hist. Ser. 4, 16: 345±357. vex phenetics. Syst. Zool. 35: 560±570. 1878. Catalogue of Chiroptera in the collec- Fahlbusch, V. tion of the British Museum. London: 1976. Report on the International Symposium British Museum. on Mammalian Stratigraphy of the Eu- Donoghue, M. J., J. A. Doyle, J. Gauthier, A. G. ropean Tertiary (MuÈnchen, April 11± Kluge, and T. Rowe 14, 1975). Newsl. Stratigr. 5: 160±167. 1989. The importance of fossils in phylogeny Farney, J., and E. Fleharty reconstruction. Annu. Rev. Ecol. Syst. 1969. Aspect ratio, loading, wing span and 20: 431±446. membrane area of bats. J. Mammal. 50: Doran, A. H. G. 362±367. 1878. Morphology of the mammalian ossi- Farris, J. S., M. Kallersjo, A. G. Kluge, and C. cula auditus. Trans. Linn. Soc. London, Bult 2nd ser., 1: 371±497. 1995. Testing signi®cance of incongruence. Ducrocq, S., E. Buffetaut, H. Buffetaut-Tong, J.- Cladistics 10: 315±319. J. Jaeger, Y. Jongkanjanasoontorn, and Faure, P. A., and R. M. R. Barclay V. Suteethorn 1992. The sensory basis of prey detection by 1992. First fossil ¯ying lemur: a dermopteran the long-eared bat, Myotis evotis and from the Late Eocene of Thailand. Pa- the consequences for prey selection. leontology 35: 373±380. Anim. Behav. 44: 31±39. Ducrocq, S., J.-J. Jaeger, and B. Sige Faure, P. A., J. H. Fullard, and J. W. Dawson 1993. Un meÂgachiropteÁre dans l'EoceÁne su- 1993. The gleaning attacks of the northern peÂrieur de ThaõÈlande: incidence dans la long-eared bat, Myotis septentrionalis, discussion phylogeÂnique du groupe. N. are relatively inaudible to moths. J. Jahrb. Geol. PalaÈontol. Mh. 9: 561± Exp. Biol. 178: 173±189. 575. Feduccia, A. Duncan, T. 1996. The origin and evolution of birds. New 1980. Cladistics for the practicing taxono- Haven: Yale Univ. Press. mistÐan eclectic view. Syst. Bot. 5: Felton, H., A. Helfricht, and G. Storch 136±148. 1973. Die Bestimmung der europaÈischen Fle- Duncan, T., and G. F. Estabrook dermaÈuse nach der distalen Epiphyse 1976. An operational method for evaluating des Humerus. Senckenbergiana Biol. classi®cations. Syst. Bot. 1: 373±382. 54: 291±297. 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 149

Fenton, M. B. their echolocation calls. J. Mammal. 1972. The structure of aerial-feeding bat fau- 62: 233±243. nas as indicated by ears and wing ele- Fenton, M. B., D. H. M. Cumming, J. M. Hutton, ments. Can. J. Zool. 50: 287±296. and C. M. Swanepoel 1974a. Feeding ecology of insectivorous bats. 1987. Foraging and habitat use by Nycteris Bios 45: 3±15. grandis (Chiroptera: Nycteridae) in 1974b. The role of echolocation in the evolu- Zimbabwe. J. Zool. Soc. London 211: tion of bats. Am. Nat. 108: 386±388. 709±716. 1975. Acuity of echolocation in Collocalia Fenton, M. B., and T. H. Fleming hirundinacea (Aves: Apodidae), with 1976. Ecological interactions between bats comments on the distribution of echo- and nocturnal birds. Biotropica 8: 104± locating swiftlets and molossid bats. 110. Biotropica 7: 1±7. Fenton, M. B., C. L. Gaudet, and M. L. Leonard 1980. Adaptiveness and ecology of echolo- 1983. Feeding behavior of the bats Nycteris cation in terrestrial (aerial) systems. In grandis and Nycteris thebaica (Nycter- R.-G. Busnel and J. F. Fish (eds.), An- idae) in captivity. J. Zool. London 200: imal sonar systems, NATO Adv. Study 347±354. Inst. Ser. A 28: 427±448. Fenton, M. B., and I. L. Rautenbach 1982a. Echolocation, insect hearing, and the 1986. A comparison of the roosting and for- feeding ecology of insectivorous bats. aging behavior of three species of Af- In T. H. Kunz (ed.), Ecology of bats, rican insectivorous bats (Rhinolophi- pp. 261±285. New York: Plenum. dae, Vespertilionidae, and Molossidae). 1982b. Echolocation calls and patterns of hunt- Can. J. Sci. 64: 2860±2867. ing and habitat use of bats (Microchi- Fenton, M. B., I. L. Rautenbach, D. Chipese, M. roptera) from Chillagoe, north Queens- land. Australian J. Zool. 30: 417±425. B. Cumming, M. K. Musgrave, J. S. 1984. Echolocation: implications for the ecol- Taylor, and T. Volpers ogy and evolution of bats. Q. Rev. Biol. 1993. Variation in foraging behavior, habitat 59: 33±53. use, and diet of large slit-faced bats 1985. Communication in the Chiroptera. (Nycteris grandis). Z. SaÈugetierkunde Bloomington: Indiana Univ. Press. 58: 65±74. 1989. Head size and the foraging behavior of Fenton, M. B., I. L. Rautenbach, J. Rydell, H. T. animal-eating bats. Can. J. Zool. 67: Arita, J. Ortega, S. Bouchard, M. D. 2029±2035. Hovorka, B. Lim, E. Odgran, C. V. 1990. The foraging behavior and ecology of Portfors, W. M. Scully, D. M Syme, animal-eating bats. Can. J. Zool. 86: and M. J. Vonhof 411±422. In press. Emergence, echolocation, diet and 1992. Bats. New York: Facts on File. foraging behavior of Molossus ater 1994a. Echolocation: its impact on the behav- (Chiroptera: Molossidae). Biotropica. ior and ecology of bats. EÂ coscience 1: Fenton, M. B., C. M. Swanepoel, R. M. Brigham, 21±30. J. E. Cebek, and M. B. C. Hickey 1994b. Assessing signal variability and reli- 1990. Foraging behavior and prey selection ability: ``to thine ownself be true.'' by large slit-faced bats (Nycteris gran- Anim. Behav. 47: 757±764. dis; Chiroptera: Nycteridae). Biotropica 1995. Natural history and biosonar signals. In 21: 2±8. A. N. Popper and R. R. Fay (eds.), Fiedler, J. Hearing in bats, pp. 37-86. New York: 1979. Prey catching with and without echo- Springer-Verlag. location in the Indian false vampire bat Fenton, M. B., D. Audet, M. K. Obrist, and J. (Megaderma lyra). Behav. Ecol. Socio- Rydell biol. 6: 155±160. 1995. Signal strength, timing, and self-deaf- Filhol, H. ening: the evolution of echolocation in 1872. Recherches sur les mammifeÁres fossiles bats. Paleobiology 21: 229±242. des depots de phosphate de chaux dans Fenton, M. B., and G. P. Bell les DeÂpartmentes du Lot, du Tarn et de 1979. Echolocation and feeding behavior in Tarne-et-Garonnne. Ann. Sc. Geol., J. four species of Myotis (Chiroptera). de Zool. 3(7): 1±31. Can. J. Zool. 57: 1271±1277. Findley, J. S., and H. Black 1981. Recognition of insectivorous bats by 1983. Morphological and dietary structuring 150 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

of a Zambian insecivorous bat com- Friant, M. munity. Ecology 64: 625±630. 1963. Les Chiroptera (Chauves-Souris). ReÂ- Findley, J. S., E. H. Studder, and D. E. Wilson vision des Rhinolophidae de l'eÂpoque 1972. Morphological properties of bat wings. tertiaire. Acta Zool. 64: 161±178. J. Mammal. 53: 429±444. Fritzsch, B., W. Himstedt, and M. D. Capron de Findley, J. S., and D. E. Wilson Caprona 1982. Ecological signi®cance of chiropteran 1985. Visual projections in larval Icthyophis morphology. In T. H. Kunz (ed.), Ecol- kohtaoensis (Amphibia: Gymno- ogy of bats, pp. 243±260. New York: phiona). Dev. Brain Res. 23: 201±210. Plenum. Fullard, J. H. Finlay, B. L., K. C. Wikler, and D. R. Sengelaub 1987. Sensory ecology and neuroethology of 1987. Regressive events in brain development moths and bats: interactions in a global and scenerios for vertebrate brain evo- perspective. In M. B. Fenton, P. A. Ra- lution. Brain Behav. Evol. 30: 102± cey, and J. M. V. Rayner (eds.), Recent 117. advances in the study of bats, pp. 244± Fleischer, G. V. 272. Cambridge: Cambridge Univ. 1978. Evolutionary principles of the mam- Press. malian middle ear. Adv. Anat. Em- Gauthier, J. bryol. Cell Biol. 55: 1±70. 1986. Saurischian monophyly and the origin 1980. Morphological adapations of the sound of birds. In K. Padian (ed.), The origin conducting apparatus in echolocating of birds and the evolution of ¯ight. mammals. In R.-G. Busnel and J. F. Mem. Cal. Acad. Sci. 8: 1±35. Fish (eds.), Animal sonar systems, Gauthier, J., A. G. Kluge, and T. Rowe NATO Adv. Study Inst. Ser. A 28: 895± 1988. Amniote phylogeny and the importance 898. of fossils. Cladistics 4: 105±209. Flower, W. H., and R. Lydekker Gillette, D. L. 1891. An introduction to the study of Mam- malia, living and extinct. London: 1975. Evolution of feeding strategies in bats. Adam and Charles Black. Tebiwa (J. Idaho St. Univ. Mus.) 18: Flynn, J. J. 39±48. 1996. Carnivore phylogeny and rates of evo- Ghiselin, M. T. lution: morphological, taxic, and mo- 1966. An application of the theory of de®ni- lecular. In J. L. Gittleman (ed.), Carni- tions to systematic prinicples. Syst. vore behavior, ecology, and evolution. Zool. 15: 127±130. 2: 542±581. Ithaca: Cornell Univ. 1984. ``De®nition,'' ``character,'' and other Press. equivocal terms. Syst. Zool. 33: 104± Forsman, K. A., and M. G. Malmquist 110. 1988. Evidence for echolocation in the com- Gingerich, P. mon shrew, Sorex araneus. J. Zool. 1987. Early Eocene bats (Mammalia, Chirop- London 216: 655±662. tera) and other vertebrates in freshwater Franzen, J. L., and H. Haubold limestones of the Willwood Formation, 1987. The biostratigraphic signi®cance of the Clark's Fork Basin, Wyoming. Contrib. Middle Eocene locality Geiseltal near Mus. Paleontol. Univ. Michigan 27: Halle (German Democratic Republic). 275±320. In N. Schmidt-Kittler (ed.), Internation- Godinot, M. al symposium on mammalian biostra- 1981. Les mammifeÁres de Rians (EÂ oceÁne In- tigraphy and paleoecology of the Eu- feÂrieur, Provence). Palaeovertebrata 10: ropean PaleogeneÐMainz, February 44±126. 18thÐ21st, 1987, pp. 93±99. MuÈn- GoÈpfert, M. C., and L. T. Wasserthal chner Geowiss. Abh. A 10: 1±312. 1995. Notes on the echolocation calls, food Freeman, P. W. and roosting behavior of the Old World 1979. Specialized insectivory: beetle-eating sucker-footed bat, Myzopoda aurita and moth-eating molossid bats. J. (Chiroptera, Myzopodidae). Z. Sauge- Mammal. 60: 467±479. tierk. 60: 1±8. 1981. A multivariate study of the family Mo- Gopalakrishna, A., and N. Badwaik lossidae (Mammalia, Chiroptera): mor- 1992. Systematic position of MolossidaeÐan phology, ecology, evolution. Fieldiana embryological analysis. J. Bombay Zool. 7: 173 pp. Nat. Hist. Soc. 89: 4±8. 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 151

Gopalakrishna, A., and G. C. Chari Grif®n, D. R., and R. A. Suthers 1983. A review of the taxonomic position of 1970. Sensitivity of echolocation in cave Minopterus based on embryological swiftlets. Biol. Bull. 139: 495±501. characters. Curr. Sci. 52: 1176±1180. Grif®n, D. R., F. A. Webster, and C. R. Michael Gould, E. 1960. The echolocation of ¯ying insects by 1965. Evidence for echolocation in the Ten- bats. Anim. Behav. 8: 141±154. recidae of Madagascar. Proc. Am. Phil. Grif®ths, T. A. Soc. 109: 352±360. 1982. Systematics of the New World nectar- 1970. Echolocation and communication in feeding bats (Mammalia, Phyllostomi- bats. In B. H. Slaughter and W. D. Wal- dae), based on the morphology of the ton (eds.), About bats. Fondren Sci. hyoid and lingual regions. Am. Mus. Ser. 11: 144±161. Dallas: Southern Novitates 2742: 45 pp. Methodist Univ. Press. 1983. Comparative laryngeal anatomy of the 1988. Wing-clapping sounds of Eonycteris Big brown bat, Eptesicus fuscus, and spelaea (Pteropodidae) in Malaysia. J. the Mustached bat, Pteronotus parnel- Mammal. 69: 378±379. lii. Mammalia 47: 377±394. Gould, E., N. Negus, and A. Novick 1994. Phylogenetic systematics of Slit-faced 1964. Evidence for echolocation in shrews. J. bats (Chiroptera, Nycteridae), based on Exp. Zool. 156: 19±38. hyoid and other morphology. Am. Mus. Grande, L. Novitates 3090: 17 pp. 1980. Paleontology of the Green River For- Grif®ths, T. A., and A. L. Smith mation, with a review of the ®sh fauna. 1991. Systematics of emballonuroid bats Bull. Geol. Surv. Wyoming 63: 1±333. (Chiroptera: Emballonuridae and Rhin- Gray, J. E. opomatidae), based on hyoid morphol- ogy. Bull. Am. Mus. Nat. Hist. 206: 1831. Zoological miscellany. London: Treut- 62±83. tel. 86 pp. Grif®ths, T. A., A. Truckenbrod, and P. J. Spon- 1847. Characters of six new genera of bats holtz not hitherto distingushed. Proc. Zool. 1992. Systematics of megadermatid bats Soc. London 15: 14±16. (Chiroptera, Megadermatidae), based 1866. A revision of the genera of Rhinolo- on hyoid morphology. Am. Mus. Nov- phidae, or horseshoe bats. Proc. Zool. itates 3041: 21 pp. Soc. London 1866: 81±83. Grinnell, A. D. Greenwald, N. S. 1995. Hearing in bats: an overview. In A. N. 1990. Are bats monophyletic? Phylogenetic Popper and R. R. Fay (eds.), Hearing interpretation of neuroanatomical and in bats, pp. 1±36. New York: Springer- other morphological data. J. Vertebr. Verlag. Paleontol. 10: 25A±26A. Guppy, A., and R. B. Coles 1991. Cladistic analysis of chiropteran rela- 1983. Feeding behaviour of the Australian tionships: why megachiropterans are ghost bat, Macroderma gigas (Chirop- not primates. Bat Res. News 31: 79±80 tera: Megadermatidae) in captivity. [dated 1990 but issued in 1991]. Australian Mammal. 6: 97±99. Grif®n, D. R. Guppy, A., R. B. Coles, and J. D. Pettigrew 1953. Acoustic orientation in the oilbird, 1985. Echolocation and acoustic communi- Steatornis. Proc. Nat. Acad. Sci. 39: cation in the Australian ghost bat, Ma- 884±893. croderma gigas (Chiroptera: Megader- 1958. Listening in the dark. New Haven: Yale matidae). Australian Mammal. 8: 299± Univ. Press. 308. 1971. The importance of atmospheric attenu- Gupta, B. B. ation for the echolocation of bats (Chi- 1984. Phylogeny of bats. Uttar Pradesh J. roptera). Anim. Behav. 19: 55±61. Zool. 4: 37±42. Grif®n, D. R., and A. Novick Habersetzer, J. 1955. Acoustic orientation of Neotropical 1981. Adaptive echolocation sounds in the bats. J. Exp. Zool. 130: 251±299. bat Rhinopoma hardwickei, a ®eld Grif®n, D. R., A. Novick, and M. Korn®eld study. J. Comp. Physiol. 144: 559±566. 1958. The sensitivity of echolocation in the 1986. Vergleichende ¯uÈgelmorphologische fruit bat Rousettus. Biol. Bull. 115: Untersuchungen an einer Fledermaus- 107±113. gellschaft in Maurai. In W. Nachtigall 152 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

(ed.), Bat ¯ightÐFledermaus¯ug. conservation biology. Smithsonian In- Bionia Rept. 5: 75±104. Stuttgart: G. stitution Press. Fischer. Hand, S. J., M. Novacek, H. Godthelp, and M. Habersetzer, J., G. Richter, and G. Storch Archer 1989. Paleoecology of the Middle Eocene 1994. First Eocene bat from Australia. J. Ver- Messel bats. Abstr. 5th Int. Theriolog. tebr. Paleontol. 14: 375±381. Congr., Rome, 2: 629±630. Handley, C. O. 1992. Bats: already highly specialized insect 1955. Nomenclature of some Tertiary Chirop- predators. In S. Schall and W. Ziegler tera. J. Mammal. 36: 128±130. (eds.), Messel: an insight into the his- Harris, D. M., and P. Dallos tory of life and of the Earth, pp. 181± 1984. Ontogenetic changes and frequency 191. Oxford: Clarendon Press. [English mapping of the mammalian ear. Sci- edition. Originally published in Ger- ence 225: 741±743. man in 1988 by Waldemar Kramer, Hartman, F. A. Frankfurt; English translation by M. 1963. Some ¯ight mechanisms of bats. Ohio Shaffer-Fehre.] J. Sci. 63: 59±65. 1994. Paleoecology of Early Middle Eocene Harvey, P. H., and J. R. Krebs bats from Messel, FRG. Apsects of 1990. Comparing brains. Science 249: 140± ¯ight, feeding and echolocation. Hist. 146. Biol. 8: 235±260. Hebert, H. Habersetzer, J., and G. Storch 1986. Korrelation zwischen FluÈgelschlag und Ortungslautaussendung bei ¯iegenden 1987. Klassi®kation und funktionelle FluÈgel- und landenden Flughunden Rousettus morphologie palaÈogener FledermaÈuse aegypticus. In W. Nachtigall (ed.), Bat (Mammalia, Chiroptera). Cour. ¯ightÐFledermaus¯ug. Bionia Rept. 5: Forsch.-Inst. Senkenberger 91: 11±150. 157±168. Stuttgart: Gustav Fischer. 1988. Grube Messel: akustische Orientierung Heblich, K. der aÈltesten FledermaÈuse. Spektrum der 1986. FluÈgelschlag und Lautaussendung bei Wissenschaft 7: 12-14. ¯iegenden und landenden FledermaÈu- 1989. Ecology and echolocation of the Eo- sen. In W. Nachtigall (ed.), Bat ¯ightÐ cene Messel bats. In V. Hanak, T. Hor- Fledermaus¯ug. Bionia Rept. 5: 139± acek, and J. Gaisler (eds.), European 156. Stuttgart: Gustav Fischer. bat research 1987, pp. 213±233. Heller, F. Prague: Charles Univ. Press. 1935. FledermaÈuse aus der eozaÈn Braunkohle 1992. Cochlea size in extant Chiroptera and des Geiseltales bei Halle a. S. Nov. Middle Eocene Microchiroptera from Acta Leopold. Neue Folge 2: 301±314. Messel. Naturwissenchaften 79: 462± Heller, K.-G. 466. 1995. Echolocation and body size in insectiv- Habersetzer, J., and B. Vogler orous bats: the case of the giant naked 1983. Discrimination of surface-structured bat Cheiromeles torquatus (Molossi- targets by the echolocating bat, Myotis dae). Le Rhinolophe 11: 27±38. myotis, during ¯ight. J. Comp. Physiol. Henson, O. W. A 152: 275±282. 1964. Echolocation and hearing in bats with Halpern, M. special reference to the function of the 1973. Retinal projections in blind snakes. Sci- middle ear muscles. Ph.D. dissertation, ence 182: 390±391. Yale Univ.. Hand, S. J. 1965. The activity and function of the middle 1990. First Tertiary molossid (Microchirop- ear muscles in echolocating bats. J. tera: Molossidae) from Australia: its Physiol. London 180: 871±887. phylogenetic and biogeographic impli- 1966. Comparative physiology of middle ear cations. Mem. Queensland. Mus. 28: muscle activity during echolocation in 175±192. bats. Am. Zool. 6: 603. Hand, S. J., and J. A. W. Kirsch 1967a. The perception and analysis of bio-so- 1998. A southern origin for the Hipposideri- nar signals by bats. In R. G. Busnel dae (Microchiroptera)? Evidence from (ed.), Les systeÁmes sonars animaux, the Australian fossil record. In T. H . vol. II, pp 949±1003. Lab. Physiol. Kunz and P. A Racey (eds.), Bats: phy- Acoust., INRA-CNRZ, Jouy-en-Josas logeny, morphology, echolocation, and 78, France. 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 153

1967b. Auditory sensitivity in Molossidae characters and phylogenetic hypothe- (Chiroptera). Anat. Rec. 157: 363±364. ses. Annu. Rev. Ecol. Syst. 24: 279± 1970. The ear and audition. In W. A. Wimsatt 306. (ed.), Biology of bats, vol. II, pp. 181± Hood, C. S. 263. New York: Academic Press. 1989. Comparative morphology and evolu- Henson, O. W., and H. -U. Schnitzler tion of the female reproductive tract in 1980. Performance of airborne animal sonar macroglossine bats (Mammalia, Chi- systems. II. Vertebrates other than Mi- roptera). J. Morphol. 199: 207±221. crochiroptera. In R. G. Busnel and J. F. Hood, C. S., and J. D. Smith Fish (eds.), Animal sonar systems, pp. 1982. Cladistical analysis of female reproduc- 183±195. New York: Plenum. tive histomorphology in phyllostomoid Hermanson, J. W. bats. Syst. Zool. 31: 241±251. 1981. Functional morphology of the clavicle 1983. Histomorphology of the female repro- in the Pallid bat, Antrozous pallidus.J. ductive tract in phyllostomoid bats. Oc- Mammal. 62: 801±805. cas. Pap. The Museum, Texas Tech Hermanson, J. W., and J. S. Altenbach Univ. 86: 1±38. 1983. The functional anatomy of the shoulder Hooker, J. J. of the Pallid bat, Antrozous pallidus.J. 1991. The sequence of mammals in the Tha- Mammal. 64: 62±75. netian and Ypresian of the London and 1985. Functional anatomy of the shoulder and Belgian Basins. Location of the Paleo- arm of the fruit-eating bat Artibeus ja- cene±Eocene Boundary. Newslett. maicensis. J. Zool. London A 205: Stratigr. 25: 75±90. 157±177. 1992. British mammalian paleocommunities Heywood, V. H. across the Eocene±Oligocene transition 1988. The structure of systematics. In D. L. and their environmental implications. Hawksworth (ed.), Prospects in system- In D. R. Prothero and W. A. Berggren atics, pp. 44±56. Systematics Society (eds.), Eocene±Oligocene climatic and Special Volume 36. Oxford: Clarendon biotic evolution, pp. 494±515. Prince- Press. ton: Princeton Univ. Press. Hill, J. E. 1996. A primitive emballonurid bat (Chirop- 1974. A new family, genus, and species of bat tera, Mammalia) from the earliest Eo- (Mammalia: Chiroptera) from Thai- cene of England. In Jubil. par D. E. land. Bull. Br. Mus. Nat. Hist. (Zool.) Russell, M. Godinot and P. D. Ginger- 27: 301±336. ich (eds.), Palaeovertebrata 25 (2±4): Hill, J. E., and D. L. Harrison 287±300. 1987. The baculum in the Vespertilioninae HoraÂcek, I. (Chiroptera: Vespertilionidae) with a 1986. Kerivoula (Mammalia, Chiroptera), systematic review, a synopsis of Pip- fossil in Europe? Acta Univ. Caroli- istrellus and Eptesicus, and the descrip- nae±Geologica, Spinar 2: 213±222. tions of a new genus and subgenus. Howell, D. J., and J. Pylka Bull. Br. Mus. Nat. Hist. (Zool.) 52: 1977. Why bats hang upside down: a biome- 225±305. chanical hypothesis. J. Theor. Biol. 69: Hill, J. E., and J. D. Smith 625±631. 1984. Bats: a natural history. Austin: Univ. Huelsenbeck, J. P. Texas Press. 1994. Comparing the stratigraphic record to Himstedt, W., and G. Manteuffel estimates of phylogeny. Paleobiology 1985. Retinal projections in the caecilian 20: 470±483. Icthyophis kohtaoensis (Amphibia, Huelsenbeck, J. P., D. L. Swofford, C. W. Cun- Gymnophiona). Cell Tiss. Res. 239: ningham, J. J. Bull, and P. J. Waddell 689±692. 1994. Is character weighting a panacea for the Hollar, L. J., and M. S. Springer problem of data heterogeneity in phy- 1997. Old World fruitbat phylogeny: evidence logenetic analysis? Syst. Biol. 43: 288± for convergent evolution and an en- 291. demic African clade. Proc. Natl. Acad. Humphry, G. M. Sci. U.S.A. 94: 5716±5721. 1869. The myology of the limbs of Pteropus. Honeycutt, R. L., and R. M. Adkins J. Anat. Physiol. 3: 294±334. 1993. Higher-level systematics of eutherian Hutcheon, J. M., and J. A. W. Kirsch mammals: an assessment of molecular 1996. Inter-familial relationships within the 154 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Microchiroptera: a preliminary study Jones, G., and J. M. V. Rayner using DNA hybridization. Bat Res. 1988. Flight performance, foraging tactics News 36: 73±74. and echolocation in free-living Dau- Hutcheon, J. M., J. A. W. Kirsch, and J. D. Pet- benton's bats Myotis daubentoni (Chi- tigrew roptera: Vespertilionidae). J. Zool. Lon- In press. Base-compositional biases and the don 215: 113±132. bat problem. III. The question of mi- 1991. Flight performance, foraging tactics crochiropteran monophyly. Phil. Trans. and echolocation in the trawling bat, Biol. Sci. Myotis adversus (Chiroptera: Vesperti- Jacobs, D. S., V. H. Perry, and M. J. Hawken lionidae). J. Zool. London 225: 393± 1984. The postnatal reduction of the un- 412. crossed projection from the nasal retina Kalko, E. K. V. in the cat. J. Neurosci. 4: 2425±2433. 1994. Coupling of sound emission and wing- Jacobs, M. S., P. J. Morgane, and W. L. Mc- beat in naturally foraging European Farland pipistrelle bats (Microchiroptera: Ves- 1975. Degeneration of visual pathways in the pertilionidae). Folia Zoologica 43: bottlenose dolphin. Brain Res. 88: 346± 363±376. 352. 1995. Insect pursuit, prey capture and echo- Jefferies, R. P. S. location in pipistrelle bats (Microchi- 1979. The origin of chordatesÐa method- roptera). Anim. Behav. 50: 1±20. ological essay. In M. R. House (ed.), Kalko, E. K. V., and H. U. Schnitzler The origin of major invertebrate 1989. The echolocation and hunting behavior groups, pp. 443±477. Systematics As- of Daubenton's bat, Myotis daubentoni. sociation Special Volume 12. New Behav. Ecol. Sociobiol. 24: 225±238. York: Academic Press. 1993. Plasticity in echolocation signals of Eu- Jen, P.H.-S., and N. Suga 1976. Coordinated activities of middle ear ropean pipistrelle bats in search ¯ight: and laryngeal muscles in echolocating implications for habitat use and prey bats. Science 191: 950±952. detection. Behav. Ecol. Sociobiol. 33: Jepsen, G. L. 415±428. 1966. Early Eocene bat from Wyoming. Sci- Kallen, F. C., and C. Gans ence 154: 1333-1339. 1972. Mastication in the little brown bat, My- 1970. Bat origin and evolution. In W. A. otis lucifugus. J. Morphol. 136: 385± Wimsatt (ed.), Biology of the bats 2: 1± 420. 65. New York: Academic Press. Kay, R. F., J. G. M. Thewissen, and A. D. Yoder Joerman, G., and U. Schmidt 1992. Cranial anatomy of Ignacius graybul- 1981. Echoortung bei der Vampir¯edermaus, lianus and the af®nities of the Plesia- Desmodus rotundus. II. Lautaussen- dapiformes. Am. J. Phys. Anthropol. dung im Flug und Korrelation zum FluÈ- 89: 477±498. gelschlag. Z. SaÈugetierk. 46: 136±146. Kety, S. S., and C. F. Schmidt Johnson, J. I., and J. A. W. Kirsch 1948. The effects of altered arterial tensions 1993. Phylogeny through brain traits: inter- of carbon dioxide and oxygen on ce- ordinal relationships among mammals rebral blood ¯ow and cerebral oxygen including Primates and Chiroptera. In consumption of normal young men. J. R. D. E. MacPhee (ed.), Primates and Clin. Invest. 27: 484±492. their relatives in phylogenetic perspec- Kick, S. A. tive, pp. 293±331. Advances in Prima- 1982. Target-detection by the echolocating tology Series. New York: Plenum. bat, Eptesicus fuscus. J. Comp. Physiol. Johnson, T. N. 145: 431±435. 1954. The superior colliculus and inferior col- Kingdon, J. liculus of the mole (Scalopus aquaticus 1974. East African mammals: an atlas of evo- machrinus). J. Comp. Neurol. 101: lution in Africa. Vol. II, part A (Insec- 765±800. tivores and bats). Chicago: Univ. of Jones, G. Chicago Press. 1994. Scaling of wingbeat and echolocation Kirsch, J. A. W. pulse emission rates in bats: why are 1996. Bats are monophyletic; megabats are aerial insectivorous bats so small? monophyletic; but are microbats also? Funct. Ecol. 8: 450±457. Bat Res. News 36: 78. 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 155

Kirsch, J. A. W., T. F. Flannery, M. S. Springer, 241. Washington, DC.: Smithsonian In- and F.-J. Lapointe stitution Press. 1995. Phylogeny of the Pteropodidae (Mam- 1994. Chiroptera: systematics. Handbook of malia: Chiroptera) based on DNA hy- zoology, vol.8, pt. 60, Mammalia. 217 bridization, with evidence for bat pp. monophyly. Australian J. Zool. 43: Koopman, K. F., and J. K. Jones 395±428. 1970. Classi®cation of bats. In B. H. Slaugh- Kirsch, J. A. W., and J. M. Hutcheon ter and W. D. Walton (eds.), About 1997. Further on the possibility that micro- bats, pp. 22±28. Dallas: Southern chiropterans are paraphyletic. Bat Res. Methodist Univ. Press. News 37: 138. Kopka, T. Kirsch, J. A. W., and J. D. Pettigrew 1973. Beziehungen zwischen FluÈgel¯aÈche In press. Base compostional biases and the und KoÈrpergroÈûe bei Chiropteren. Z. bat problem. II. DNA-hybridization Wiss. Zool., Leipzig 185: 235±284. trees based on tracers enriched for AT- Kovtun, M. F. or GC-content. Philos. Trans. Biol Sci. 1989. On the origin and evolution of bats. In Klaauw, C. J., van der V. HanaÂk, I. HoraÂcek, and J. Gaisler 1931. The auditory bulla in some fossil mam- (eds.), European bat research 1987, pp. mals. Bull. Am. Mus. Nat. Hist. 62: 1± 5±12. Praha: Charles Univ. Press. 352. Krazanowski, A. Kluge, A. G. 1977. Weight classes of palearctic bats. Acta 1989. A concern for evidence and a phylo- Theriol. 22: 365±370. genetic hypothesis of relationships Kudo, M., Y. Nakamura, T. Morizumi, H. Tokuno, among Epicrates (Boidae, Serpentes). and Y. Kitao Syst. Zool. 38: 7±25. 1988. Direct retinal projections to the latero- Kluge, A. G., and A. J. Wolf 1993. Cladistics: what's in a word? Cladistics posterior thalamic nucleus (LP) in the 9: 183±195. mole. Neurosci. Lett. 93: 176±180. Knight, A., and D. P. Mindell Kudo, M., M. Yamamoto, and Y. Nakamura 1993. Substitution bias, weighting of DNA 1991. Suprachiasmic nucleus and retinohy- sequence evolution, and the phyloge- pothalamic projections in moles. Brain netic position of Fea's viper. Syst. Zool. Behav. Evol. 38: 332±338. 38: 7±25. Kulzer, E. Kober, R., and H.-U. Schnitzler 1958. Untersuchungen uÈber die Biologie von 1990. Information in sonar echoes of ¯utter- Flughunden der Gattung Rousettus È ing insects available for echolocating Gray. Z. Morphol. Okol. Tiere 47: 347± bats. J. Acoust. Soc. Am. 87: 882±896. 402. Kolb, A. 1960. Physiologische und Morphologische 1959. Ein Registrierapparat fuÈr FledermaÈuse Untersuchungen uÈber die Erzeugung und einige biologische Ergebnisse. der Orientierungslaute von Flughunden Zool. Anz. 163: 134±141. der Gattung Rousettus. Ziet. Vergl. 1961. Sinnesleistungen einheimischer Fleder- Physiol. 43: 231±268. maÈuse bei der Nahrungssuche und Nah- Kunz, T. H. rungsauswahl auf dem boden und in 1980. Daily energy budgets of free-living der luft. Z. Verleich. Physiol. 44: 550± bats. In D. E. Wilson and A. L. Gardner 564. (eds.), Proceedings Fifth International Koopman, K. F. Bat Research Conference, pp. 369±392. 1984. Bats. In S. Anderson and J. K. Jones Lubbock: Texas Tech Press. (eds.), Orders and families of recent Kurta, A., G. P. Bell, K. A. Nagy, and T. H. Kunz mammals of the world, pp. 145±186. 1989. Energetics of pregnancy and lactation New York: Wiley. in free-ranging little brown bats (My- 1985. A synopsis of the families of bats, part otis lucifugus). Physiol. Zool. 62: 804± VII. Bat Res. News 25: 25±29 [dated 818. 1984 but issued in 1985]. Kurta, A., K. A. Johnson, and T. H. Kunz 1993. Order Chiroptera. In D. E. Wilson and 1987. Oxygen consumption and body temper- D. M. Reeder (eds.), Mammal species ature of female little brown bats (My- of the world, a taxonomic and geo- otis lucifugus) under simulated roost graphic reference, 2nd ed., pp. 137± conditions. Physiol. Zool. 60: 386±397. 156 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

LacepeÁde, B. G. Etienne, Compte de note on dermopteran enamel. Scann- 1799. Tableau des divisions, sous-divisions, ning Microsc. 2: 371±383. ordres, et genres de mammifeÃres, avec Liem, K. l'indication de toutes les espeÁces deÂcri- 1973. Evolutionary strategies and morpholog- tes par Buffon, et leur distribution dans ical innovations: cichlid pharyngeal chacun des genres par F. K. Daudin. jaws. Syst. Zool. 22: 425±441. Vol. 15. Paris: Didot. Loo, S. K., and R. Kanagasuntheram Lancaster, W. C., A. W. Keating, and O. W. Hen- 1972. The vomeronasal organ in the tree son shrew and slow loris. J. Anat. 112: 1992. Ultrasonic vocalizations of ¯ying bats 165±172. monitored by radiotelemetry. J. Exp. Lucas, S. G. Biol. 173: 43±58. 1992. Extinction and the de®nition of the Lawrence, D. D., and J. A. Simmons class Mammalia. Syst. Biol. 41: 370± 1982. Measurements of atmospheric attenua- 371. tion at ultrasonic frequencies and the Lucas, S. G., and Z. Luo signi®cance for echolocation by bats. J. 1993. Adelobasileus from the upper Triassic Acoust. Soc. Am. 71: 585±590. of West Texas: the oldest mammal. J. Lawlor, T. E. Vertebr. Paleontol. 13: 309±334. 1973. Aerodynamic characteristics of some Luckett, W. P. Neotropical bats. J. Mammal. 54: 71± 1980a. The use of fetal membrane characters 78. in assessing chiropteran phylogeny. In Leche, W. D. E. Wilson and A. L. Gardner (eds.), Proceedings Fifth international bat re- 1886. Uber die saÈugthiergattung Galeopithe- search conference, pp. 245±266. Lub- cus: eine morphologische untersu- bock: Texas Tech Press. chung. Kongl. Svenska Vetenskaps- 1980b. The suggested evolutionary relation- Akademiens Handlingar 21: 1±92. ships and classi®cation of tree shrews. 1911. Einige Dauertypen aus der Klasse der In W. P. Luckett (ed.), Comparative bi- SaÈugetiere. Zool. Anzeiger 38: 551± ology and evolutionary relationships of 560. tree shrews, pp. 3±31. New York: Ple- Legendre, S. num.  1984. Etude odontologique des repreÂsentants 1980c. The use of reproductive and develop- actuels du group Tadarida (Chiroptera, mental features in assessing tupaiid af- Molossidae). Implications phylogeÂne- ®nities. In W. P. Luckett (ed.), Compar- tiques, systeÂmatiques et zoogeÂograp- ative biology and evolutionary relation- hiques. Rev. Suisse Zool. 91: 99±442 ships of tree shrews, pp. 245±266. New 1985. MolossideÂs (Mammalia, Chiroptera) York: Plenum. ceÂnezoiques de l'Ancien et du Nouveau 1993. Developmental evidence from the fetal Monde; statut systeÂmatique; inteÂgration membranes for assessing archontan re- phylogeÂnique des donneÂes. N. Jahrb. lationship. In R. D. E. MacPhee (ed.), Geol. PalaÈont. Abh. 170: 205±227. Primates and their relatives in phylo- Legendre, S., and B. Sige genetic perspective. Adv. Primatol. 1983. La place du ``vespertilion de Montmar- Ser., pp. 149±186. New York: Plenum. tre'' dans l'histoire des chiropteÂres mo- Lund, R. D., and J. S. Lund lossides. In E. Buffetaut, J. M. Mazin, 1965. The visual system of the mole, Talpa and E. Salmon (eds.), Actes du sym- europaea. Exp. Neurol. 13: 302±316. posium paleÂontologique Georges Cu- 1966. The central visual pathways and their vier, pp. 347±361. Montbeliard. functional signi®cance in the mole Le Gros Clark, W. E. (Talpa europaea). J. Zool. 149: 95± 1924. The myology of the tree-shrew (Tupaia 101. minor). Proc. Zool. Soc. London 1924: Lutz, H. 461±497. 1992. Giant ants and other rarities: the insect 1926. On the anatomy of the pen-tailed tree- fauna. In S. Schall and W. Ziegler shrew (Ptilocercus lowii). Proc. Zool. (eds.), Messel: an insight into the his- Soc. London 1926: 1179±1309. tory of life and of the Earth, pp. 55± Lester, K. S., S. J. Hand, and F. Vincent 67. Oxford: Clarendon Press. [English 1988. Adult phyllostomid (bat) enamel by edition. Originally published in Ger- scanning electron microscopyÐwith a man in 1988 by Waldemar Kramer, 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 157

Frankfurt; English translation by M. McKenna, M. C. Shaffer-Fehre.] 1975. Toward a phylogenetic classi®cation of Lydekker, R. the Mammalia. In W. P Luckett and F. 1888. Mammalia. In F. E. Beddard (ed.), Zoo- S. Szalay (eds.), Phylogeny of the Pri- logical Record for 1887, pp. 41±56. mates, pp. 21±46. New York: Plenum. London: Gurney and Jackson. McKenna, M. C., and S. K. Bell MacAlister, A. 1997. Classi®cation of mammals above the 1872. The myology of the Cheiroptera. Phil. species level. New York: Columbia Trans. R. Soc. London 162: 125±171. Univ. Press. MacPhee, R. D. E., M. Cartmill, and K. D. Rose McKenna, M. C., P. Robinson, and D. W. Taylor 1989. Craniodental morphology and relation- 1962. Notes on Eocene Mammalia and Mol- ships of the supposed Eocene dermop- lusca from Tabernacle Butte, Wyoming. teran Plagiomene (Mammalia). J. Ver- Am. Mus. Novitates 2102: 33 pp. tebr. Paleontol. 9: 329±349. McNab, B. K. Maddison, W. P., and D. R. Maddison 1969. The economics of temperature regula- 1992. MacClade, version 3.0. [Computer pro- gram distributed on ¯oppy disk by Sin- tion in neotropical bats. Comp. Bio- auer Associates, Sunderland, MA]. chem. Physiol. 31: 227±268. Maddison, W. P., M. J. Donoghue, and D. R. Mad- Medway, Lord dison 1959. Echolocation among Collocalia. Nature 1984. Outgroup analysis and parsimony. Syst. 184: 1352±1353. Zool. 33: 83±103. 1967. The function of echonavigation among Madkour, G. swiftlets. Anim. Behav. 15: 416±420. 1989. Uro-genitalia of Microchiroptera from Meier, R., and S. Richter Egypt. Zool. Anzeiger 222: 337±352. 1992. Suggestions for a more precise usage of Manske, U., and U. Schmidt proper names of taxa: ambiguities re- 1976. Visual acuity of the vampire bat, Des- lated to the stem lineage concept. Z. modus rotundus, and its dependence Syst. Zool. Evolutionsforsch. 30: 81± upon light intensity. Z. Tierpsychol. 42: 88. 215±221. Mein, P., and Y. Tupinier Marandat, B. 1977. Formule dentaire et position systeÂma- 1991. MammifeÁres de l'Ilerdien moyen tique du MiniopteÁre (Mammalia, Chi- (EoceÁne InfeÂrieur) des CorbieÁres et du roptera). Mammalia 41: 207±211. Minervois (Bas-Langedoc, France) Pa- 1986. Histoire des chiropteÁres en relation laeovertebrata 20: 55±144. avec les systeÁmes de localisation acous- Marandat, B., J.-Y. Crochet, M. Godinot, J.-L. tique. Mammalia 50: 99±106. Hartenberger, S. Legendre, J. A. Remy, Menu, H. B. SigeÂ, J. Sudre, and M. Vianey-Liaud 1985. Morphotypes dentaires actuels et fos- 1993. Une nouvelle faune aÁ mammifeÁres siles des chiropteÁres vespertilionineÂs. d'aÃge EoceÁne Moyen (LuteÂtien SupeÂr- PremieÂre partie: E tude des morpholo- ieur) dans les phosphorites du Quercy. gies dentaires. Palaeovertebrata 15: 71± Geobios 26: 617±623. 128. MacGinitie, H. D. Menu, H., and B. Sige 1969. The Eocene Green River ¯ora of north- 1971. western Colorado and northeastern Nyctalodontie et myotodontie, impor- Utah. Univ. Calif. Publ. Geol. Sci. 83: tants caracteÁres de grades eÂvolutifs 140 pp. chez les chiropteÁres entomophages. C. Martin, R. D. R. Acad. Sci. Paris 272: 1735±1738. 1981. Relative brain size and basal metabolic Meredith, M. rate in terrestrial vertebrates. Nature 1991. Sensory processing in the main and ac- 293: 57±60. cesory olfactory systems: comparisons Matthew, W. D. and contrasts. J. Steroid Biochem. Mo- 1917. A Paleocene bat. Bull. Am. Mus. Nat. lec. Biol. 39: 601±614. Hist. 38: 569±571. Meschinelli, L. Matthews, L. H. 1903. Un nuovo chiropterro fossile (Ar- 1941. Notes on the genetalia and reproduction chaeopteropus transiens Mesch.) delle of some African bats. Proc. Zool. Soc. ligniti de Monteviale. Atti lst. Venteto London, ser. B, 111: 289±346. Sci. 62: 1329±1344. 158 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Miller, G. S. Nelson, G. J. 1907. The families and genera of bats. Bull. 1972. Phylogenetic relationship and classi®- U.S. Natl. Mus. 57: 1±282. cation. Syst. Zool. 21: 227±231. Miller, L. A., and H. J. Degan 1974. Classi®cation as an expression of phy- 1981. The acoustic behavior of four species logenetic relationships. Syst. Zool. 22: of vespertilionid bats studied in the 344±359. ®eld. J. Comp. Physiol. 142: 67±74. 1993. Why crusade against consensus? A re- Mindell, D. P., C. W. Dick, and R. J. Baker ply to Barrett, Donoghue, and Sober. 1991. Phylogenetic relationships among me- Syst. Biol. 42: 215±216. gabats, microbats, and primates. Proc. Nelson, J. E. Natl. Acad. Sci. U.S.A. 88: 10,322± 1964. Vocal communication in Australian ¯y- 10,326. ing foxes (Pteropodidae: Megachirop- tera). Z. Tierpsychol. 21: 857±870. Minelli, A. Neuweiler, G. 1991. Names for the system and names for 1967. Interaction of other sensory systems the classi®cation. In D. L. Hawksworth with the sonar system. In R.-G. Bush- (ed.), Improving the stability of names: nel, Animal sonar systems 1: 509±534. needs and options, pp. 183±189. Reg- NATO Advanced Study Institute. num Vegetabile 123. KoÈnigstein, Ger- 1984. Foraging, echolocation and audition in many: Koeltz. bats. Naturwissenschaften 71: 446± Miyamoto, M. M. 455. 1996. A congruence study of molecular and 1989. Foraging ecology and audition in ech- morphological data for eutherian mam- olocating bats. Trends Ecol. Evol. 4: mals. Mol. Phylogenet. Evol. 6: 373± 160±166. 390. 1990. Auditory adaptations for prey capture Miyamoto, M. M., and W. M. Fitch in echolocating bats. Physiol. Rev. 70: 1995. Testing species phylogenies and phy- 615±641. logenetic methods with congruence. Neuweiler, G., V. Bruns, and G. Schuller Syst. Biol. 44: 64±76. 1980. Ears adapted for the detection of mo- MoÈhres, F. P. tion, or how echolocating bats have ex- 1967a. Communicative characters of sonar sig- ploited the capabilities of the mamma- nals in bats. In R.-G. Busnel (ed.), An- lian auditory system. J. Acoust. Soc. imal sonar systems, 2: 936±945. NATO Am. 68: 741±753. Advanced Study Institute. Neuweiler, G., and M. B. Fenton 1967b. Ultrasonic orientation in megadermatid 1988. Behavior and foraging ecology of ech- bats. In R.-G. Busnel (ed.), Animal so- olocating bats. In P. E. Nachtigall and nar systems, 1: 115±127. Lab. Physiol. P. W. B. Moore (eds.), Animal sonar: Acoust. C.N.R.Z., Jouy-en-Josas, processes and performance, pp. 535± France. 549. New York: Plenum. MoÈhres, F. P., and E. Kulzer Neuweiler, G., W. Metzner, U. Heilmann, R. Rub- 1956. UÈ ber die Orientierung der Flughunde samien, M. Eckrich, and H. H. Costa 1987. (ChiropteraÐPteropodidae). Z. Vergl. Foraging behavior and echolocation in the rufus horseshoe bat, Rhinolophus Physiol. 38: 1±29. rouxi, on Sri Lanka. Behav. Ecol. So- 1957. MegadermaÐein konvergenter Zwis- ciobiol. 20: 53±67. chentyp der Ultraschallpeilung bei Fle- Nixon, K. C., and J. M. Carpenter dermausen. Naturwissenschaften 44: 1993. On outgroups. Cladistics 9: 413±426. 21±22. Norberg, U. M. MoÈhres, F. P., and G. Neuweiler 1969. An arrangement giving a stiff leading 1966. Die Ultraschallorientierung der Gross- edge to the hand wing in bats. J. Mam- blatt-Fledermause (Chiroptera±Megad- mal. 50: 766±770. ermatidae). Z. Vergl. Physiol. 53: 195± 1970. Functional osteology and myology of 227. the wing of Plecotus auritus Linnaeus Mueller, H. C. (Chiroptera). Arkiv fuÈr Zoologi 22: 1966. Homing and distance-orientation in 483±543. bats. Z. Tierpsychol. 23: 403±421. 1972a. Functional osteology and myology of 1968. The role of vision in vespertilionid the wing of the Dog-faced bat Rouset- bats. Am. Midl. Nat. 79: 524±525. tus aegyptiacus (E. Geoffroy) (Mam- 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 159

malia, Chiroptera). Z. Morphol. Tiere Norell, M. A., and M. J. Novacek 73: 1±44. 1992a. The fossil record and evolution: com- 1972b. Bat wing structures important for aer- paring cladistic and paleontologic evi- odynamics and rigidity (Mammalia, dence for vertebrate history. Science Chiroptera). Z. Morph. Tiere 73: 45± 255: 1690±1693. 61. 1992b. Congruence between superpositional 1976. Some advanced ¯ight manoeuvers of and phylogenetic patterns: comparing bats. J. Exp. Biol. 64: 489±495. cladistic patterns with fossil records. 1981. Flight, morphology and the echological Cladistics 8: 319±337. niche in echolocating bats. Symp. Zool. Norris, K. S. Soc. London 48: 173±197. 1968. The evolution of acoustic mechanisms 1985. Flying, gliding, and soaring. In M. Hil- in odontocete cetaceans. In E. T. Drake debrand, D. M. Bramble, K. F. Liem, (ed.), Evolution and environment, pp. and D. B. Wake (eds.), Functional ver- 297±324. New Haven: Yale Univ. tebrate morphology, pp. 129±158. Press. Cambridge: Harvard Univ. Press. Norris, K. S., J. H. Prescott, P. V. Asa-Dorian, and 1986a. On the evolution of ¯ight and wing P. Perkins form in bats. In W. Nachtigall (ed.), Bat 1961. An experimental demonstration of ¯ightÐFledermaus¯ug, pp. 13±26. echolocation behavior in the porpoise, Bionia Rept. 5. Stuttgart: Gustav Fi- Tursiops truncatus (Montagu). Biol. scher. Bull. Mar. Biol. Lab. Woods Hole 120: 1986b. Evolutionary convergence in foraging 163±176. niche and ¯ight morphology in insec- Novacek, M. J. tivorous aerial-hawking birds and bats. 1980a. Phylogenetic analysis of the chiropter- Ornis Scand. 17: 253±260. an auditory region. In D. E. Wilson and 1987. Wing form and ¯ight mode in bats. In A. L. Gardner (eds.), Proceedings ®fth M. B. Fenton, P. Racey, and J. M. V. international bat research conference, Rayner (eds.), Recent advances in the pp. 317±330. Lubbock: Texas Tech study of bats, pp. 43±56. Cambridge: Press. Cambridge Univ. Press. 1980b. Cranioskeletal features in tupaiids and 1989. Ecological determinants of bat wing selected Eutheria as phylogenetic evi- shape and echolocation call structure dence. In W. P. Luckett (ed.), Compar- with implications for some fossil bats. ative biology and evolutionary relation- In V. HanaÂk, I. HoraÂcek, and J. Gaisler ships of tree shrews, pp. 35±93. New (eds.), European bat research 1987, pp. York: Plenum. 197±211. Praha: Charles Univ. Press. 1985a. Evidence for echolocation in the oldest 1990. Vertebrate ¯ight: mechanics, physiol- known bats. Nature 315: 140±141. ogy, morphology, ecology and evolu- 1985b. Comparative morphology of the bat au- tion. Zoophysiology, vol. 27. Berlin: ditory region. Fortschr. Zool. 30: 149± Springer-Verlag. 151. 1994. Wing design, ¯ight performance, and 1986. The skull of leptictid insectivorans and habitat use in bats. In P. C. Wainwright the higher-level classi®cation of euther- and S. M. Reilly (eds.), Ecological ian mammals. Bull. Am. Mus. Nat. morphology: integrative organismal bi- Hist. 183: 1±111. ology, pp. 205±239. Chicago: Univ. 1987. Auditory features and af®nities of the Chicago Press. Eocene bats Icaronycteris and Palaeo- Norberg, U., and J. M. V. Rayner chiropteryx (Microchiroptera, incertae 1987. Ecological morphology and ¯ight in sedis). Am. Mus. Novitates 2877: 18 bats (Mammalia: Chiroptera): wing ad- pp. aptations, ¯ight performance, foraging 1990. Morphology, paleontology and the strategy and echolocation. Phil. Trans. higher clades of mammals. In H. H. R. Soc. London, ser. B, 316: 335±427. Genoways (ed.), Current mammalogy Norell, M. A. 2: 507±543. 1992. Taxic origin and temporal diversity: the 1991. Aspects of morphology of the cochlea effect of phylogeny. In M. J. Novacek in microchiropteran bats: an investiga- and Q. D. Wheeler (eds.), Extinction tion of character transformation. Bull. and phylogeny, pp. 89±118. New York: Am. Mus. Nat. Hist. 206: 84±100. Columbia Univ. Press. 1992. Fossils as critical data for phylogeny. 160 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

In M. J. Novacek and Q. D. Wheeler terminalis systems in baleen whales. (eds.), Extinction and phylogeny, pp. Brain Behav. Evol. 34: 171±183. 46±88. New York: Columbia Univ. OelschlaÈger, H. A., and E. H. Buhl Press. 1985. Development and rudimentation of the 1994. Morphological and molecular inroads peripheral olfactory system in the har- to phylogeny. In L. Grande and O. bor porpoise Phocoeana phocoeana.J. Rieppel (eds), Interpreting the hierar- Morphol. 184: 351±360. chy of nature: from systematic patterns Ostrander, G. E. to evolutionary process theories, pp. 1983. New Early Oligocene (Chadronian) 85±131. New York: Academic Press. mammals from the Raben Ranch Local 1996. Paleontological data and the study of Fauna, northwest Nebraska. J. Paleon- adaptation. In M. D. Rose and G. V. tol. 57: 128±139. Lauder (eds.), Adapation, pp. 311±359. Padian, K. New York: Academic Press. 1984. The origin of pterosaurs. In W.-E. Reif Novacek, M. J., and M. A. Norell and F. Westphal (eds.), Third sympo- 1982. Fossils, phylogeny, and taxonomic sium on terrestrial ecosystems: short rates of evolution. Syst. Zool. 31: 369± 378. papers, pp. 163±168. Tuebingen: At- Novacek, M. J., and A. R. Wyss tempto Verlag. 1986. Higher-level relationships of the Recent 1987. A comparative phylogenetic and func- eutherian orders: morphological evi- tional approach to the origin of verte- dence. Cladistics 2: 257±287. brate ¯ight. In M. B. Fenton, P. Racey, Novick, A. and J. M. V. Rayner (eds.), Recent ad- 1958a. Orientation in Paleotropical bats. I. Mi- vances in the study of bats, pp. 3±22. crochiroptera. J. Exp. Zool. 138: 81± Cambridge: Cambridge Univ. Press. 154. Padian, K., D. R. Lindberg, and P. D. Polly 1958b. Orientation in Paleotropical bats. II. 1994. Cladistics and the fossil record: the Megachiroptera. J. Exp. Zool. 137: uses of history. Annu. Rev. Earth Plan- 443±462. et. Sci. 22: 63±91. 1959. Acoustic orientation in the cave swif- Palmer, T. S. tlet. Biol. Bull. 117: 497±503. 1903. Some new generic names of mammals. 1962. Orientation in Neotropical bats. I. Na- Science, N.S. 17: 873±874. talidae and Emballonuridae. J. Mam- Pederson, S. C. mal. 43: 449±455. 1993. Cephalometric correlates of echoloca- 1963a. Orientation in Neotropical bats. II. tion in the Chiroptera. J. Morphol. 218: Phyllostomatidae and Desmodontidae. 85±98. J. Mammal. 44: 44±56. Pettigrew, J. D. 1963b. Pulse duration in the echolocation of 1986. Flying primates? Megabats have the insects by the bat, Pteronotus. Ergeb- advanced pathway from eye to mid- nisse Biol. 26: 21±26. brain. Science 231: 1304±1306. 1965. Echolocation of ¯ying insects by the 1988. Microbat vision and echolocation in an bat, Chilonycteris psilotis. Biol. Bull. evolutionary context. In P. E. Nachtigill 128: 297±314. and P. W. B. Moore (eds.), Animal so- 1977. Acoustic orientiation. In W. A. Wimsatt nar: processes and performance, pp. (ed.), Biology of bats 3: 73±289. New 645±650. New York: Plenum. York: Academic Press. 1991a. Novick, A., and J. R. Vaisnys Wings or brain? Convergent evolution 1964. Echolocation of ¯ying insects by the in the origins of batc. Syst. Zool. 40: bat, Chilonycteris parnellii. Biol. Bull. 199±216. 127: 478±488. 1991b. A fruitful, wrong hypothesis? Response Obrist, M. K., M. B. Fenton, J. L. Egar, and P. A. to Baker, Novacek, and Simmons. Syst. Schlegel Zool. 40: 231±239. 1993. What ears do for bats: a comparative 1994. Flying DNA. Curr. Biol. 4: 277±283. study of pinna sound pressure transfor- 1995. Flying primates: crashed, or crashed mation in Chiroptera. J. Exp. Biol. 180: through? In P. A. Racey and S. M. 119±152. Swift (eds.), Ecology, evolution and be- OelschlaÈger, H. A. havior of bats. Symp. Zool. Soc. Lon- 1989. Early development of the olfactory and don 67: 3±26. 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 161

Pettigrew, J. D., B. Dreher, C. S. Hopkins, M. J. Pollak, G. D., and J. H. Casseday McCall, and M. Brown 1988. Parallel auditory pathways: IIÐFunc- 1988. Peak density and distribution of gan- tional properties. In P. E. Nachtigill and glion cells in he retinae of microchirop- P. W. B. Moore (eds.), Animal sonar: teran bats: implications for visual acu- processes and performance, pp. 197± ity. Brain Behav. Evol. 32: 39±56. 223. New York: Plenum. Pettigrew, J. D., and B. G. M. Jamieson Pollak, G. D., and T. J. Park 1987. Are ¯ying foxes (Chiroptera: Pteropod- 1995. The inferior colliculus. In A. N. Popper idae) really primates? Australian Mam- and R. R. Fay (eds.), Hearing in bats, mal. 10: 119±124. pp. 368±415. New York: Springer-Ver- Pettigrew, J. D., B. G. M. Jamieson, S. K. Robson, lag. L. S. Hall, K. I. McAnally, and H. M. Pomeroy, D. Cooper 1990. Why ¯y? The possible bene®ts for low- 1989. Phylogenetic relations between micro- er mortality. Biol. J. Linn. Soc. 40: 53± bats, megabats and primates (Mamma- 65. lia: Chiroptera and Primates). Phil. Porter, C. A., M. Goodman, and M. J. Stanhope Trans. R. Soc. London B 325: 489± 1996. Evidence on mammalian phylogeny 559. from sequences of exon 28 of the von Pettigrew, J. D., and J. A. Kirsch Willebrand Factor gene. Mol. Phylo- 1995. Flying primates revisited: DNA hybrid- genet. Evol. 5: 89±101. ization with fractionated, GC-enriched Provis, J., D. van Driel, F. A. Billson, and P. Rus- DNA. South African J. Sci. 91: 477± sel 482. 1985. Human fetal optic nerve: overproduc- In press. Base-compositional biases and the tion and elimination of retinal axons bat problem. I. DNA-hybridization during development. J. Comp. Neurol. melting curves based on tracers en- 238: 92±101. riched for AT- or GC- content. Philos. Purves, P. E., and G. Pilleri Trans. Biol Sci. 1983. Echolocation in whales and dolphins. Phillips, R. B. London: Academic Press. 1984. Considerations in formalizing a classi- Pybus, M. J., D. P. Hobson, and D. K. Onderka ®cation. In T. Duncan and T. F. Stuessy 1986. Mass mortality of bats due to probable (eds.), Cladistics: perspectives on the blue-green algal toxicity. J. Wildl. Dis. reconstruction of evolutionary history, 22: 449±450. pp. 257±272. New York: Columbia Pye, A. Univ. Press. 1966a. The structure of the cochlea in Chirop- Pierson, E. D. tera. I. Microchiroptera: Emballonuro- 1986. Molecular systematics of the Microchi- idea and Rhinolophoidea. J. Morphol. roptera: higher taxon relationships and 118: 495±510. biogeography. Ph.D. dissertation, Univ. 1966b. The structure of the cochlea in Chirop- of California, Berkeley, California. tera. II. The Megachiroptera and Ves- Pierson, E. D., V. M. Sarich, J. M. Lowenstein, pertilionoidea of the Microchiroptera. M. J. Daniel, and W. E. Rainey J. Morphol. 119: 101±120. 1986. A molecular link between the bats of 1967. The structure of the cochlea in Chirop- New Zealand and South America. Na- tera. III. Microchiroptera: Phyllosto- ture 232: 60±63. matoidea. J. Morphol. 121: 241±254. Pilleri, G. Pye, D. 1983. The sonar system of dolphins. Endeav- 1980. Adaptiveness of echolocation signals in our 3: 48±56. bats: ¯exibility in behavior and in evo- 1990. Adaptation to water and the evolution lution. Trends Neurosci. 3: 232±235. of echolocation in the Cetacea. Ethol. Pye, J. D. Ecol. Evol. 2: 135±163. 1968. How insects hear. Nature 218: 797. Pirlot, P. Rager, G. 1977. Wing design and the origin of bats. In 1980. Development of the retinotectal projec- M. K. Hecht, P. C. Goody, and B. M. tion in the chicken. Adv. Anat. Em- Hecht (eds.), Major patterns in verte- bryol. Cell Biol. 63: 1±92. brate evolution, pp. 375±410. NATO Rahmann, H. Adv. Stud. Inst. Ser. (Life Sci.) 14. 1967. Die SehschaÈarfe bei Wirbeltieren. Na- New York: Plenum. turwiss. Rundschau 20: 8±14. 162 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Rakic, P., and K. P. Riley Robbins, L. W., and V. M. Sarich 1983. Overproduction and elimination of ret- 1988. Evolutionary relationships in the family inal axons in the fetal rhesus monkey. Emballonuridae (Chiroptera). J. Mam- Science 219: 1441±1444. mal. 69: 1±13. Rayner, J. M. V. Roberts, L. H. 1986. Vertebrate ¯apping ¯ight mechanics 1975. Con®rmation of the echolocation pulse and aerodynamics, and the evolution of production mechanism of Rousettus.J. ¯ight in bats. In W. Nachtigall (ed.), Mammal. 56: 218±220. Bat ¯ightÐFledermaus¯ug, pp. 27±74. Robin, H. A. Bionia Rept. 5. Stuttgart: Gustav Fi- 1881. ReÂcherches anatomiques sur les mam- scher. mifeÁres des chiropteÁres. Annales des 1991a. The cost of being a bat. Nature 350: ScieÂnces Naturelles, 6th ser. 12: 1±180. 383±384. Romer, A. S. 1991b. Complexity and a coupled system: 1966. Vertebrate palaeontology. 3rd ed. Chi- ¯ight, echolocation and evolution in cago: Univ. Chicago Press. bats. In N. Schmidt-Kittler and K. Vo- Rosenzweig, M. R., D. A. Riley, and D. Krech gel (eds.), Constructional morphology 1955. Evidence for echolocation in the rat. and evolution, pp. 173±190. Berlin: Science 121: 600. Springer-Verlag. Roverud, R. C. RepeÂrant, J., D. Miceli, J.-P. Rio, and C. Weidner 1987. The processing of echolocation sound 1987. The primary optic system in a microp- elements in bats: a behavioral ap- thalamic snake (Calabaria reinhardti). proach. In M. B. Fenton, P. Racey, and Brain Res. 408: 233±238. J. M. V. Rayner (eds.), Recent advances Revilliod, P. A. in the study of bats, pp. 152±170. Cam- 1916. A propos de l'adaptation au vol chez bridge: Cambridge Univ. Press. les MicrocheiropteÁres. Verh. Naturf. Rowe, T. Ges. Basel 27: 156±183. 1987. De®nition and diagnosis in the phylo- 1917a. Contribution aÁ l'eÂtude des ChiropteÁres genetic system Syst. Zool. 36: 208± des terrains tertiaires, premieÁre partie. 211. MeÂm. Soc. Pal. Suisse 43: 1±58. 1988. De®nition, diagnosis, and the origin of 1917b. FledermaÈuse aus der Braunkohle von Mammalia. J. Vertebr. Paleontol. 8: Messel bei Darmstadt. Abh. Grossh- 241±264. erz.-hess. Geol. Landesanst 7: 157± RuÈbsamen, R., G. Neuweiler, and K. Sripathi 201. 1988. Comparative cochlear tonotopy in two 1919. L' eÂtat actuel de nos connaissances sur bat species adapted to movement detec- les ChiropteÁres fossiles (note preÂlimi- tion, Hipposideros speoris and Megad- naire). C. R. Soc. Sci. Phys. Nat. Ge- erma lyra. J. Comp. Physiol. 163: 271± neÁve 36: 93±96. 285. 1920. Contribution aÁ l'eÂtude des ChiropteÁres Russell, D. E., and P. D. Gingerich des terrains tertiaires, deuxieme partie. 1981. Lipotyphla, Proteutheria(?), and Chi- MeÂm. Soc. Pal. Suisse 44: 63±128. roptera (Mammalia) from the Early± 1922. Contribution aÁleÂtude des ChiropteÁres Middle Eocene Kuldana Formation of des terrains tertiaires, troisieÁme partie Kohat (Pakistan). Contrib. Mus. Pa- et ®n. MeÂm. Soc. Pal. Suisse 45: 133± leontol. Univ. Michigan 25: 277±287. 195. Russell, D. E., J.-L. Harenberger, Ch. Pomerol, S. Richter, G. Sen, N. Schmidt-Kittler, and M. Vi- 1987. Untersuchungen zur ErnaÈhrung eozaÈner aney-Liaud SaÈuger aus der FossilfundstaÈtte Messel 1982. Mammals and stratigraphy: the Paleo- bei Darmstadt. Courier Forsch.-Inst. gene of Europe. PalaeoVertebr., MeÂm. Senckenberg 91: 1±33. Extraordinaire: 1±77. Richter, G., and G. Storch Russell, D. E., P. Louis, and D. E. Savage 1980. BeitraÈge zur ErnaÈhrungsbiologie eozaÈ- 1973. Chiroptera and Dermoptera of the ner FledermaÈuse aus der ``Grube Mes- French Early Eocene. Univ. Calif. Publ. sel.'' Natur u. Museum 110: 353±367. Geol. Sci. 95: 1±55. Ricklefs, R. E., and H. L. Marks Russell, D. E., and B. Sige 1984. Insensitivity of brain growth to selec- 1970. ReÂvision des chiropteÁres luteÂtiens de tion of four-week body mass in Japa- Messel (Hesse, Allemagne). Palaeov- nese quail. Evolution 38: 1180±1185. ertebrata 3: 83±182. 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 163

Ryan, M. J., and M. D. Tuttle Schlosser, M. 1983. The ability of the frog-eating bat to dis- 1887. Die Affen, Lemuren, Chiropteren, In- criminate among novel and potentially sectivoren, Marsupialier, Creodonten, poisonous frog species using acoustic und Carnivoren des europaÈischer Ter- cues. Anim. Behav. 31: 827±833. tiaÈrs. Beit. Pal. Geol. Oesterr.-Ung. 6: 1987. The role of prey-generated sounds, vi- 227 pp. sion, and echolocation in prey localiza- 1910. UÈ ber einige fossil SaÈugetiere aus dem tion by the African bat Cardioderma OligocaÈn von AÈ gypten. Zool. Anzeiger cor (Megadermatidae). J. Comp. Phys- 35: 500±508. iol. A 161: 59±66. 1911. BeitraÈge zur Kenntnis der oligozaÈnen Rydell, J. LandsaÈugetiere aus dem Fayum. Beitr. È 1992. Occurrence of bats in northernmost z. PalaÈont. Geol. Oesterr.-Ungarns Ori- Sweden (65ЊN) and their feeding ecol- ents 24: 51±167. ogy in summer. J. Zool. London 227: Schlosser-Strum, E., and H. Schliemann 517±529. 1995. Morphology and function of the shoul- Sales, G., and D. Pye der joint of bats (Mammalia: Chirop- 1974. Ultrasonic communication by animals. tera). J. Zool. Syst. Evol. Res. 33: 88± London: Chapman and Hall. 98. Sanderson, K. J. Schmidt, S. 1986. Evolution of the lateral geniculate nu- 1988. Evidence for a spectral basis of texture perception in bat sonar. Nature 331: cleus. In J. D. Pettigrew, K. J. Sander- 617±619. son, and W. R. Levick (eds.), Visual Schmidt, U., G. Joermann, and G. Rother neuroscience, pp. 183±195. Cambridge: 1988. Acoustical vs. visual orientation in Cambridge Univ. Press. Neotropical bats. In P. E. Nachtigall Savage, D. E. and P. W. B. Moore (eds.), Animal so- 1951. A Miocene phyllostomatid bat from nar: processes and performance, pp. Colombia, South America. Bull. Dept. 589±592. New York: Plenum. Geol. Sci., Univ. California Press 28: Schmidt-Kittler, N. (ed.) 357±366. 1987. International symposium on mammali- Savage, D. E., and D. E. Russell an biostratigraphy and paleoecology of 1983. Mammalian paleofaunas of the world. the European PaleogeneÐMainz. MuÈn- Reading: Addison-Wesley. chner Geowiss. Abh. A 10: 1±312. Sazima, I. Schnitzler, H.-U. 1978. Vertebrates as food items of the woolly 1968. Die Ultraschall-Ortungslaute der Hufei- false vampire, Chrotopterus auritus.J. sen FledermaÈuse (Chiroptera±Rhinolo- Mammal. 59: 617±618. phidae) in verschiedenen Orienti- Schaarschmidt, F. erungssituationen. Z. Vergl. Physiol. 1992. The vegetation: fossil plants as wit- 57: 376±408. nesses of a warm climate. In S. Schall 1970a. Echoortung bei der Fledermaus Chilon- and W. Ziegler (eds.), Messel: an in- ycteris rubiginosa. Z. Vergl. Physiol. sight into the history of life and of the 68: 25±39. Earth, pp. 27±52. Oxford: Clarendon 1970b. Comparison of the echolocation behav- Press. [English edition. Originally pub- ior in Rhinolophus ferrumequinum and lished in German in 1988 by Waldemar Chilonycteris rubiginosa. Bijdr. Dierk. Kramer, Frankfurt; English translation 40: 77±80. by M. Shaffer-Fehre.] 1971. FledermaÈuse im Windkanal. Z. Vergl. Schaeffer, B., M. K. Hecht, and N. Eldredge Physiol. 73: 209±221. 1972. Phylogeny and paleontology. In T. 1973. Control of Doppler shift compensation Dobzhansky, M. K. Hecht, and W. C. in the greater horseshoe bat, Rhinolo- Steere (eds.), Evolutionary biology 6: phus ferrumequinum. J. Comp. Physiol. 31±46. New York: Appleton-Century- 82: 79±92. Crofts. 1987. Echoes of ¯uttering insects: informa- Scholey, K. tion for echolocating bats. In M. B. 1986. The evolution of ¯ight in bats. In:W. Fenton, P. A. Racey, and J. M. V. Ray- Nachtigall (ed.), Bat ¯ightÐFleder- ner (eds.), Recent advances in the study maus¯ug, pp. 1±12. Bionia Rept. 5. of bats, pp. 226±243. Cambridge: Cam- Stuttgart: Gustav Fischer. bridge Univ. Press. 164 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Schnitzler, H.-U., and O. W. Henson Nat. Hist. Natur., C, Sci. Terre 34: 1± 1980. Performance of airborne animal sonar 140. systems. I. Microchiroptera. In R. G. 1977. Les insectivoes et chiropteÁres du PaleÂo- Busnel and J. F. Fish (eds.), Animal so- geÁne moyen d'Europe dans l'histoire nar systems, pp. 109±181. New York: des faunes de mammifeÁres sur ce con- Plenum. tinent. Jurij A. Orlov Memorial, J. Pa- Schnitzler, H.-U., E. K. V. Kalko, I. Kaipf, and A. leontol. Soc. India 20: 178±190. D. Grinnell 1985. Les chiropteÁres oligoceÁnes du Fayum, 1994. Fishing and echolocation behavior of Egypte. Geol. Paleontol. 19: 161±189. the greater bulldog bat, Noctilio lepor- 1988. Le gisement du Bretou (Phosphorites inus, in the ®eld. Behav. Ecol. Socio- du Quercy, Tarn-et-Garonne, France) et biol. 35: 327±345. sa faune de verteÂbres de l'EoceÁne su- Schnitzler, H.-U., E. K. V. Kalko, L. Miller, and peÂrieur. Palaeontogr. Abt. A 205: 69± A. Surlykke 102. 1987. The echolocation and hunting behavior 1990. Nouveaux chiropteÁres de l'OligoceÁne of the bat Pipistrellus kuhli. J. Comp. Physiol. 161: 267±274. moyen des phosphorites du Quercy, Scholey, K. France. C. R. Acad. Sci. Paris 310: 1986. The evolution of ¯ight in bats. In W. 1131±1137. Nachtigall (ed.), Bat ¯ightÐFleder- 1991a. Rhinolophoidea et Vespertilionoidea maus¯ug, pp. 1-12. Biona Rept. 5. (Chiroptera) du Chambi (EoceÁne infeÂr- Stuttgart: Gustav Fischer. ieur de Tunisie): aspects biostratigrap- Schuller, G. hique, biogeÂographique et paleÂoeÂcolo- 1977. Echo delay and overlap with emitted gique de l'origine des chiropteÁres mod- orientation sounds and Doppler shift ernes. N. Jahrb. Geol. PalaÈont. Abh. compensation in the bat, Rhinolophus 182: 355±376. ferrumequinum. J. Comp. Physiol. 114: 1991b. Morphologie dentaire lacteÂale d'un chi- 103±114. ropteÁredel'E oceÁne infeÂrieur-moyen Schuller, G., K. Beuter, and R. RuÈbsamen d'Europe. Geobios 13: 231±236. 1975. Dynamic properties of the compensa- SigeÂ, B., and S. Legendre tion system for Doppler shifts in the 1983. L'histoire des peuplements de chirop- bat, Rhinolophus ferrumequinum.J. teÁres du bassin MeÂditerraneÂen: l'apport Comp. Physiol. 97: 113±125. compare des remplissages karstiques et Schuller, G., and G. Pollack des deÂpoÃts ¯uvio-lacustres. MeÂm. Bios- 1976. Disproportionate frequency representa- peÂol. 10: 209±225. tion in the inferior colliculus of Dopp- SigeÂ, B., and D. E. Russell ler-compensating greater horseshoe 1980. CompleÂments sur les chiropteÁres de bats: evidence of an acoustic fovea. J. l'Eocene moyen d'Europe. Les genres Comp. Physiol. A 132: 47±54. Palaeochiropteryx et Cecilionycteris. Segall, W. PalaeoVertebr., MeÂm. Jubil. R. Lavocat 1971. Auditory region in bats including Ica- :91±126. ronycteris index. Fieldiana Zool. 58: Simmons, J. A. 103±108. 1979. Perception of echo phase information SigeÂ, B. in bat sonar. Science 204: 1136±1138. 1968. Les chiropteÁres du MioceÁne infeÂrieur 1980. de Bouzigues. I. E tude systeÂmatique. Phylogenetic adaptations and the evo- Palaeovertebrata 1: 65±133. lution of echolocation in bats. In D. E. 1971. Anatomie du membre anteÂrieur chez un Wilson and A. L. Gardner (eds.), Pro- chiropteÁre molosside (Tadarida sp.) du ceedings Fifth International Bat Re- Stampien de CeÂreste (Alpesde-Haute- search Conference, pp. 267±278. Lub- Provence). Palaeovertebrata 4: 1±38. bock: Texas Tech Press. 1974. DoneÂes nouvelles sur le genre Stehlinia Simmons, J. A., M. B. Fenton, and M. J. O'Farrell (Vespertilionoidea, Chiroptera) du Pa- 1979. Echolocation and pursuit of prey by leÂogeÁne d'Europe. Palaeovertebrata 6: bats. Science 203: 16±21. 253±272. 1980. Acoustic imaging in bat sonar: echo- 1976. Insectivores primitifs de l'EoceÁne su- location signals and the evolution of peÂrieur et OligoceÁne infeÂrieur d'Europe echolocation. J. Comp. Physiol. 135: occidentale. NyctitheÂriideÂs. Mem. Mus. 61±84. 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 165

Simmons, J. A., D. J. Howell, and N. Suga classi®cation of mammals. Bull. Am. 1975. Information content of bat sonar ech- Mus. Nat. Hist. 85: 350 pp. oes. Am. Scientist 63: 204±215. Slaughter, B. H. Simmons, J. A., W. A. Lavender, B. A. Lavender, 1970. Evolutionary trends of chiropteran den- C. A. Doroshow, S. W. Kiefer, R. Liv- titions. In B. H. Slaughter and W. D. ingston, A. C. Scallet, and D. E. Crow- Walton (eds.), About bats, Fondren Sci. ley Ser. 11: 51±83. Dallas: Southern Meth- 1974. Target structure and echo spectral dis- odist Univ. Press. crimination by echolocating bats. Sci- Smith, A. B. ence 186: 1130±1132. 1994. Systematics and the fossil record: doc- Simmons, J. A., and R. A. Stein umenting evolutionary patterns. Ox- 1980. Acoustic imaging in bat sonar: echo- ford: Blackwell. location signals and the evolution of Smith, J. D. echolocation. J. Comp. Physiol. A 135: 1972. Systematics of the chiropteran family 61±84. Mormoopidae. Univ. Kansas Mus. Simmons, N. B. Natl. Hist. Misc. Publ. 56: 132 pp. 1993a. The importance of methods: archontan 1976. Chiropteran evolution. In R. J. Baker, phylogeny and cladistic analysis of J. K. Jones, and D. C. Carter (eds.), Bi- morphological data. In R. D. E. ology of bats of the New World family MacPhee (ed.), Primates and their rel- Phyllostomatidae, part I, pp. 49±69. atives in phylogenetic perspective, pp. Special Publ. The Museum, Texas Tech 1±61. New York: Plenum. Univ., vol. 10. Lubbock: Texas Tech 1993b. Morphology, function, and phylogenet- Univ. ic signi®cance of pubic nipples in bats 1977. Comments on ¯ight and the evolution (Mammalia: Chiroptera). Am. Mus. Novitates 3077: 37 pp. of bats. In M. K. Hecht, P. C. Goody, 1993c. Phylogeny of Multituberculata. In F. S . and B. M. Hecht, Major patterns in ver- Szalay, M. J. Novacek, and M. C. Mc- tebrate evolution, pp. 427±438. NATO Kenna (eds.), Mammal phylogeny: Me- Adv. Stud. Inst. Ser. (Life Sci.) 14. sozoic differentiation, multitubercula- New York: Plenum. tes, monotremes, early therians, and 1980. Chiropteran phylogenetics: introduc- marsupials, pp. 146±164. New York: tion. In D. E. Wilson and A. L. Gardner Springer-Verlag. (eds.), Proceedings Fifth International 1994. The case for chiropteran monophyly. Bat Research Conference, pp. 233±244. Am. Mus. Novitates 3103: 54 pp. Lubbock: Texas Tech Press. 1995. Bat relationships and the origin of Smith, J. D., and G. Madkour ¯ight. In P. A. Racey and S. M. Swift 1980. Penial morphology and the question of (eds.), Ecology, evolution and behavior chiropteran phylogeny. In D. E. Wilson of bats. Symp. Zool. Soc. London 67: and A. L. Gardner (eds.), Proceedings 27±43. Fifth International Bat Research Con- 1998. A reappraisal of interfamilial relation- ference, pp. 347±365. Lubbock: Texas ships of bats. In T. H. Kunz and P. A. Tech Press. Racey (eds.), Bats: phylogeny, mor- Smith, J. D., G. Richter, and G. Storch phology, echolocation, and conserva- 1979. Wie FledermaÈuse sich einmal ernaÈrht tion biology. Washington: Smithsonian haben. Umschau 79: 482±484. Institution Press. Smith, J. D., and A. Starrett Simmons, N. B., M. J. Novacek, and R. J. Baker 1979. Morphometric analysis of chiropteran 1991. Approaches, methods, and the future of wings. In R. J. Baker, J. K. Jones, and the chiropteran monophyly controver- D. C. Carter (eds.), Biology of bats of sty: a reply to J. D. Pettigrew. Syst. the New World family Phyllostomati- Zool. 40: 239±243. dae 3: 427±437. Spec. Publ. The Mus. Simmons, N. B., and T. H. Quinn Texas Tech Univ., no. 16. Lubbock, 1994. Evolution of the digital tendon locking Texas: Texas Tech Univ. Press. mechanism in bats and dermopterans: a Smith, J. D., and G. Storch phylogenetic perspective. J. Mammal. 1981. New Middle Eocene bats from the Evol. 2: 231±254. ``Grube Messel'' near Darmstadt, W- Simpson, G. G. Germany (Mammalian: Chiroptera). 1945. The principles of classi®cation and a Senckenberg. Biol. 61: 153±168. 166 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Smith, R., and D. E. Russell Stanhope, M. J., M. R. Smith, V. G. Waddell, C. 1992. MammifeÁres (Marsupialia, Chiroptera) A. Porter, M. S. Shivji, and M. Good- de l'YpreÂsian de la Belgique. Bull. In- man stit. R. Sci. Nat. Belgique, Sci. Terre 1996. Mammalian evolution and the interpho- 62: 223±227. toreceptor retinoid binding protein Speakman, J. R. (IRBP) gene: convincing evidence for 1993. The evolution of echolocation for pre- several superordinal clades. J. Mol. dation. Symp. Zool. Soc. London 65: Evol. 43: 83±92. 39±63. Storch, G. 1995. Chiropteran nocturnality. In P. A. Ra- 1968. Funktionsmorphologische Untersu- cey and S. M. Swift (eds.), Ecology, chungen an der Kaumuskulatur und an evolution and behavior of bats. Symp. korrelierten SchaÈdelstrukturem der Chi- Zool. Soc. London 67: 187±201. ropteren. Abh. Senckenberg. Natur- Speakman, J. R., M. E. Anderson, and P. A. Racey forsch. Ges. 517: 1±92. 1989. The energy cost of echolocation in pip- 1989. Die eozaÈnen FledermaÈuse von MesseÐ istrelle bats (Pipistrellus pipistrellus). J. fruÈhe Zeugel der Stammesgeschichte. Comp. Physiol. A 165: 679±685. Laichinger HoÈhlenfreund 24: 21±30. Speakman, J. R., and P. A. Racey Storch, G. , and J. Habersetzer 1987. The energetics of pregnancy and lacta- 1988. Archaeonycteris pollex (Mammalia, tion in the brown long-eared bat, Ple- Chiroptera), eine neue Fledermaus aus cotus auritus. In M. B. Fenton, P. A. Ra- dem EozaÈn der Grube Messel bei cey, and J. M. V. Rayner (eds.), Recent Darmstadt. Cour. Forsch.-Inst. Senck- advances in the study of bats, pp. 367± enberg 107: 263±273. 393. Cambridge: Cambridge Univ. Storer, J. E. Press. 1984. Mammals of the Swift Current Creek 1991. No cost of echolocation for bats in ¯ight. Nature 350: 421±423. Local Fauna (Eocene: Uintan, Sasket- Sprague, J. M. chewan). Saskatechewan Culture and 1943. The hyoid region of placental mammals Recreation, Mus. Nat. Hist. Cont. 7: 1- with especial reference to the bats. Am. 144. J. Anat. 72: 385±472. Strait, S. G. Springer, M. S., L. J. Hollar, and J. A. W. Kirsch 1993. Molar morphology and food texture 1995. Phylogeny, molecules versus morpholo- among small-bodied insectivorous gy, and rates of character evolution mammals. J. Mammal. 74: 391±402. among fruitbats (Chiroptera: Megachi- Strickler, T. L. roptera). Australian J. Zool. 43: 557±582. 1978. Functional osteology and myology of Stahl, B. J. the shoulder in the Chiroptera. Contrib. 1985. Vertebrate history: problems in evolu- Vertebr. Evol. 4: 1±198. tion. New York: Dover. Struhsaker, T. T. Stanhope, M. J., W. J. Bailey, J. Czelusniak, M. 1961. Morphological factors regulating ¯ight Goodman, J.-S. Si, J. Nickerson, J. G. in bats. J. Mammal. 42: 152±159. Sgouros, G. A. M. Singer, and T. K. Sudman, P. D., L. J. Barkley, and M. S. Hafner Kleinschimidt 1994. Familial af®nity of Tomopeas ravus 1993. A molecular view of primate supraor- (Chiroptera) based on protein electo- dinal relationships from the analysis of phoretic and cytochrome B sequence both nucleotide and amino acid se- data. J. Mammal. 75: 365±377. quences. In R. D. E. MacPhee (ed.), Sudre, J., B. SigeÂ, J. A. Remy, B. Marandat, J.-L. Primates and their relatives in phylo- Hartenberger, M. Godinot, and J.-Y. genetic perspective, pp. 251±292. Adv. Crochet Primatol. Ser. New York: Plenum. 1990. Une faune du niveau d'Egerkingen Stanhope, M. J., J. Czelusniak, J.-S. Si, J. Nick- (MP 14; Bartionian Inferieur) dans les erson, and M. Goodman phosphorites du Quercy (sud de la 1992. A molecular perspective on mammali- France). Palaeovertebrata 20: 1±32. an evolution from the gene encoding Surlykke, A., L. A. Miller, B. Mhl, B. B. Ander- interphotoreceptor retinoid binding pro- sen, J. Christensen-Dalsgaard, and M. tein, with convincing evidence for bat B. Jùrgensen monophyly. Mol. Phylogenet. Evol. 1: 1993. Echolocation in two very small bats 148±160. from Thailand: Craseonycteris thon- 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 167

glongyai and Myotis siligorensis. Be- the Smithsonian Institution, Washing- hav. Ecol. Sociobiol. 33: 1±12. ton, D.C.] Suthers, R. A. Szalay, F. S., and G. Drawhorn 1966. Optomotor responses by echolocating 1980. Evolution and diversi®cation of the Ar- bats. Science 152: 1102±1104. chonta in an arboreal milieu. In W. P. 1970. Vision, olfaction, taste. In W. A. Wim- Luckett (ed.), Comparative biology and satt (ed.), Biology of bats 2: 265±309. evolutionary relationships of tree New York: Academic Press. shrews, pp. 133±170. New York: Ple- 1978. Sensory ecology of mammals. In M. A. num. Ali (ed.), Sensory ecology: review and Szalay, F. S., and S. G. Lucas perspectives, pp. 253±287. New York: 1993. Cranioskeletal morphology of archon- Plenum. tans, and diagnoses of Chiroptera, Vol- Suthers, R. A., and M. R. Bradford itantia, and Archonta. In R. D. E. 1980. Visual systems and the evolutionary re- MacPhee (ed.), Primates and their rel- lationships of the Chiroptera. In D. E. atives in phylogenetic perspective, pp. Wilson and A. L. Gardner (eds.), Pro- 187±226. Adv. Primatol. Ser. New ceedings Fifth International Bat Re- York: Plenum. search Conference, pp. 331±346. Lub- Thewissen, J. G. M., and S. K. Babcock bock: Texas Tech Press. 1991. Distinctive cranial and cervical inner- Suthers, R. A., J. Chase, and B. Bradford vation of wing muscles: new evidence 1969. Visual form discrimination by echolo- for bat monophyly. Science 251: 934± 936. cating bats. Biol. Bull. 137: 535±546. 1993. The implications of the propatagial Suthers, R. A., S. P. Thomas, and B. J. Suthers muscles of ¯ying and gliding mammals 1972. Respiration, wingbeat and ultrasonic for archontan systematics. In R. D. E. pulse emission in an echo-locating bat. MacPhee (ed.), Primates and their rel- J. Exp. Biol. 56: 37±48. atives in phylogenetic perspective, pp. Suthers, R. A., and N. E. Wallis 91±109. Adv. Primatol. Ser. New York: 1970. Optics of the eyes of echolocating bats. Plenum. Vision Res. 10: 1165±1173. Thomas, S. P. Suthers, R. A., and J. J. Wenstrup 1987. The physiology of bat ¯ight. In M. B. 1987. Behavioral discrimination studies in- Fenton, P. Racey, and J. M. V. Rayner volving prey capture by echolocating (eds.), Recent advances in the study of bats. In M. B. Fenton, P. Racey, and J. bats, pp. 75±99. Cambridge: Cam- M. V. Rayner (eds.), Recent advances bridge Univ. Press. in the study of bats, pp. 122±1521. Tidemann, C. R., D. M. Priddel, J. E. Nelson, and Cambridge: Cambridge Univ. Press. J. D. Pettigrew Swanepeol, P., and H. H. Genoways 1985. Foraging behavior of the Australian 1979. Morphometrics. In R. J. Baker, J. K. ghost bat, Macroderma gigas (Chirop- Jones, and D. C. Carter (eds.), Biology tera: Megadermatidae). Australian J. of bats of the New World family Phyl- Zool. 33: 705±713. lostomidae, Part III. Spec. Publ. Mus. Tiunov, M. P. Texas Tech Univ. 16: 13±106. 1989. The taxonomic implications of differ- Swartz, S. M. ent morphological systems in bats. In 1997. Allometric patterning in the limb skel- V. HanaÂk, I. HoraÂcek, and J. Gaisler etons of bats: implications for the me- (eds.), European bat research 1987, pp. chanics and energetics of powered 67±76. Praha: Charles Univ. Press. ¯ight. J. Morphol. 234: 277±294. Tomasi, T. E. Swofford, D. L. 1979. Echolocation by the short-tailed shrew 1991. When are phylogeny estimates from Blarina brevicauda. J. Mammal. 60: molecular and morphological data in- 751±759. congruent? In M. M. Miyamoto and J. Turnbull., W. D. Cracraft (eds.), Phylogenetic analysis 1970. Mammalian masticatory apparatus. of DNA sequences, pp. 295±333. New Fieldiana Geol. 18: 149±356. York: Oxford Univ. Press. Tuttle, M. D., and M. J. Ryan 1993. Phylogenetic Analysis Using Parsimo- 1981. Bat predation and the evolution of frog ny (PAUP), version 3.1.1. [Computer vocalizations in the Neotropics. Sci- program distributed on ¯oppy disk by ence 214: 677±678. 168 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Tyrell, K. and Astyanax mexicanus. J. Comp. 1988. The use of prey-generated sounds in Neurol. 165: 89±106. ¯ycatcher-style foraging by Megader- Vrana, P. B. ma spasma. Bat Res. News 29: 51. 1994. Molecular approaches to mammalian Van Valen, L. phylogeny. Ph.D. dissertation, Colum- 1979. The evolution of bats. Evol. Theory 4: bia Univ. Graduate School of Arts and 104±121. Sciences, New York. Vater, M. Walton, D. W., and G. M. Walton 1987. Narrow-band frequency analysis in 1968. Comparative osteology of the pelvic bats. In M. B. Fenton, P. Racey, and J. and pectoral girdles of the Phyllosto- M. V. Rayner (eds.), Recent advances matidae (Chiroptera; Mammalia). J. in the study of bats, pp. 200±225. Cam- Grad. Res. Center, Southern Methodist bridge: Cambridge Univ. Press. Univ. 37: 1±35. Vater, M., A. S. Feng, and M. Betz 1970. Post-cranial osteology of bats. In B. H. 1985. An HPR-study of the frequency-place Slaughter and D. W. Walton (eds.), map of the horseshoe bat cochlea: mor- About bats, a chiropteran biology sym- phological correlates of the sharp tun- posium, pp. 93±126. Fondren Science ing to a narrow frequency band. J. Series, vol. 11. Dallas: Southern Meth- Comp. Physiol. A 157: 671±686. odist Univ. Press. Vaughan, T. A. Wassif, K. 1959. Functional morphology of three bats: 1950. The tensor tympani muscle of bats. Eumops, Myotis, Macrotus. Univ. Kan- Ann. Mag. Nat. Hist. 3: 811±812. sas Publ. Mus. Nat. Hist. 12: 153 pp. Wassif, K., and G. Madkour 1966. Morphology and ¯ight characteristics 1963. Studies on the osteology of the rat- of molossid bats. J. Mammal. 47: 249± tailed bats of the genus Rhinopoma 260. found in Egypt. Bull. Zool. Soc. Egypt 1970a. The skeletal system. In W. A. Wimsatt 18: 56±80. (ed.), Biology of bats 1: 98±139. New Webster, F., and D. Grif®n York: Academic Press. 1962. The role of the ¯ight membranes in in- 1970b. The muscular system. In W. A. Wimsatt sect capture by bats. Anim. Behav. 10: (ed.), Biology of bats 1: 140±194. New 332±340. York: Academic Press. Weid, R., and O. V. Helversen 1970c. The transparent dactylopatagium minus 1987. Ortungsrufe europaÈischer FledermaÈuse in phyllostomatid bats. J. Mammal. 51: beim Jagd¯ug im Freiland. Myotis 25: 142±145. 5±27. 1976. Nocturnal behavior of the African false Weithofer, A. vampire bat (Cardioderma cor). J. 1887. Zur Kenntniss der fossilen Cheiropter- Mammal. 57: 227±248. en der franzoÈsischen Phosphorite. 1977. Foraging behavior of the giant leaf- Sitzb. K. Akad. der Wissensch. Wein nosed bat (Hipposideros commersoni). 96: 341±361. East African Wildl. J. 15: 237±249. Welty, C. Vaughan, T. A., and G. C. Bateman 1955. Birds as ¯ying machines. Sci. Am. 192: 1970. Functional morphology of the forelimb 88±96. of mormoopid bats. J. Mammal. 51: Wible, J. R. 217±235. 1993. Cranial circulation and relationships of Vaughan, T. A., and R. P. Vaughan the colugo Cynocephalus (Dermoptera, 1986. Seasonality and the behavior of the Af- Mammalia). Am. Mus. Novitates 3072: rican yellow-winged bat. J. Mammal. 27 pp. 67: 91±102. Wible, J. R., and K. P. Bhatnagar Volleth, M. and K.-G. Heller 1997. Chiropteran vomeronasal complex and 1994. Phylogenetic relationships of vespertil- the interfamilial relationships of bats. J. ionid genera (Mammalia: Chiroptera) Mammal. Evol. 3: 285±314. [dated as revealed by karyological analysis. Z. 1996 but issued in 1997.] Zool. Syst. Evolut.-Forsch. 32: 11±34. Wible, J. R., and J. R. Martin Voneida, T. J., and C. M. Sligar 1993. Ontogeny of the typmanic ¯oor and 1976. A comparative neuroanatomical study roof in archontans. In R. D. E. Mac- of retinal projections in two ®shes, As- Phee (ed.), Primates and their relatives tryanax hubbsi (the blind cave ®sh), in phylogenetic perspective, pp. 111± 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 169

148. Adv. Primatol. Ser. New York: Wilson, J. P., and V. Bruns Plenum. 1983. Middle-ear mechanics in the CF-bat Wible, J. R., and M. J. Novacek Rhinolophus ferrumequinum. Hearing 1988. Cranial evidence for the monophyletic Res. 10: 1±13. origin of bats. Am. Mus. Novitates Wilson, M. A. 2911: 19 pp. 1971. Optic nerve ®ber counts and retinal Wible, J. R., and U. Zeller ganglion cell counts during develop- 1994. Cranial circulation of the Pen-tailed ment of Xenopus laevis (Daudin). Q. J. tree shrew Ptilocercus lowii and rela- Exp. Physiol. 56: 83±91. tionships of Scandentia. J. Mammal. Winge, H. Evol. 2: 209±230. 1892. Jordfundne og nulevende Flagermus Wiley, E. O. (Chiroptera) fra Lagoa Santa, Minas 1979. An annotated Linnean hierarchy, with Geraes, Brasilien, med udsigt over Fla- comments on natural taxa and compet- germusenes indbyrdes Slaegtskab. E. ing systems. Syst. Zool. 28: 308±337. Museo Lundii, vol. 2. 1981. Phylogenetics: the theory and practice 1941. The interrelationships of mammalian of phylogenetic systematics. New genera. Vol. 1: Monotremata, Marsu- York: Wiley. pialia, Insectivora, Chiroptera, Edenta- Wilkens, H. ta. Copenhagen: C. A. Reitzels. [En- 1971. Genetic interpretation of regressive glish translation by E. Deichmann and evolutionary processes: studies on hy- G. M. Allen of H. Winge, 1923, Pat- brid eyes of two Astyanax cave popu- tedyr-Slaegter. 1. Monotremata, Mar- lations (Characidae, Pisces). Evolution supialia, Insectivora, Chiroptera, Ede- 25: 530±544. netata. Copenhagen. This volume was Wilkinson, G. S., F. Mayer, G. Kerth, and B. Petri a collection of papers written between 1997. Evolution of repeated sequence arrays 1887 and 1918, with footnotes dating in the D-loop region of bat mitochon- up to 1922 (see prefaces to 1941 edi- drial DNA. Genetics 146: 1035±1048. tion). Much of the section on bats was Williams, R. W., M. J. Bastiani, B. Lia, and L. M. taken directly from Winge, 1892.] Chalupa Woodburne, M. O. (ed.) 1986. Growth cones, dying axons, and devel- 1987. Cenozoic mammals of North America, opmental ¯uxuations in the ®ber pop- geochronology and biostratigraphy. ulation of the cat's optic nerve. J. Berkeley: Univ. California Press. Comp. Neurol. 246: 32±69. Woodburne, M. O., and C. C. Swisher Williams, R. W., and K. Herrup 1995. Land mammal high-resolution geochro- 1988. The control of neuron number. Annu. nology, intercontinental overland dis- Rev. Neurosci. 11: 423±453. persals, sea-level, climate, vicariance. Williams, R. W., and P. Rakic In W. A. Beggren, D. V. Kent, M. Au- 1988. Elimination of neurons from the lateral bry, and J. Hardenbol (eds.), Geochro- geniculate nucleus of rhesus monkeys nology, time scales and global strati- during development. J. Comp. Neurol. graphic correlation, pp. 335±364. Tul- 272: 424±436. sa, Oklahoma: Society for Sedimentary Williams, T. C., and J. M. Williams Geology (spec. publ. no. 54). 1967. Radio tracking of homing bats. Science Wyss, A. R., and J. J. Flynn 155: 1435±1436. 1993. A phylogenetic analysis and de®nition 1970. Radio tracking of homing and feeding of the Carnivora. In F. S. Szalay, M. J. ¯ights of a Neotropical bat, Phyllosto- Novacek, and M. C. McKenna (eds.) mus hastatus. Anim. Behav. 18: 302± Mammal phylogeny: placentals, pp. 309. 32±52. New York: Plenum. Williams, T. C., J. M. Williams, and D. R. Grif®n Wyss, A. R., and J. Meng 1966. The homing ability of the Neotropical 1996. Application of phylogenetic taxonomy bat Phyllostomus hastatus hastatus, to poorly resolved crown clades: a with evidence for visual orientation. stem-modi®ed node-based de®nition of Anim. Behav. 14: 468±473. Rodentia. Syst. Biol. 45: 559±568. Wilson, D. E., and D. M. Reeder (eds.) Zeller, U. 1993. Mammal species of the World: A tax- 1986. Ontogeny and cranial morphology in onomic and geographic reference. 2nd the tympanic region of the Tupaiidae, ed. Washington: Smithsonian Institu- with special reference to Ptilocercus. tion Press. Folia Primatol. 47: 61±80. 170 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

APPENDIX 1: SPECIMENS EXAMINED The following list includes all specimens ex- 458209, 458210, 459103, 459108); Hypsignathus amined in this study; see Appendix 2 for pub- monstrosus (AMNH 48654, 244255, 244356, lished data sources. Abbreviations are as follows: 244357, 244358); Macroglossus minimus AMNH, American Museum of Natural History, (AMNH 192755, 250080; USNM 543196, New York, New York, USA; BMNH, British Mu- 543197, 543960, 543970, 543997, 544030, seum (Natural History), London, England; 544032, 544046); Megaerops ecaudatus (AMNH HLMD, Hessisches Landesmuseum, Darmstadt, 87285, 216746); Megaloglossus woermanni Germany; LNK, Landessammlungen fuÈr Natur- (AMNH 83803, 236291); Melonycteris melanops kunde, Karlsruhe, Germany; LSU, Louisiana State (AMNH 194333); Melonycteris woodfordi University Museum of Zoology, Baton Rouge, (AMNH 99950); Micropteropus pusillus (AMNH Louisiana, USA; Me (ME), Messel fossil locality, 239383); Myonycteris torquata (AMNH 236244); Germany (used as a supplementary identi®er for Nanonycteris veldkampi (AMNH 241024); Notop- Messel material housed at several museums); teris macdonaldi (AMNH 31588, 119445; USNM MNB, MuseÂum d'Histoire Naturelle de BaÃle, Ba- 260070, 260071, 260072, 260076); Nyctimene al- sel, Switzerland; MNHN, MuseÂum National biventer (USNM 543237, 543246); Nyctimene ce- d'Histoire Naturelle, Paris, France; PU, Princeton laeno (AMNH 10515); Otopterus cartilagonodus University, Princeton, New Jersey, USA (speci- (USNM 573446, 573447); Paranyctimene raptor mens now housed in the Peabody Museum, Yale (AMNH 198606); Penthetor lucasi (AMNH University, New Haven, Connecticut, USA; RSB, 106823); Ptenochirus jagori (AMNH 187085); Institut Royal des Sciences Naturelles de Bel- Pteropus alecto (AMNH 154547, 236510); Pter- gique, Belgium; SMF, Senckenbergmuseum, opus giganteus (AMNH 83947, 252525); Ptero- Frankfurt am Main, Germany; USNM, United pus mariannus (AMNH 249969, 256890); Ptero- States National Museum, Washington, D.C., USA; pus macrotis (AMNH 256551); Pteropus neohi- UW, University of Wyoming Museum of Geolo- bernicus (AMNH 194668); Pteropus rayneri gy, Laramie, Wyoming, USA. (AMNH 79865) Pteropus tonganus (USNM 546347, 546349); Rousettus aegyptiacus (AMNH Scandentia: Tupaia glis (AMNH 103610, 184438, 265619, 265620, 265621, 265622, 213642; USNM 396664); Tupaia javanica 265623, 265627, 265628, 265629, 265630, (AMNH 107595); Tupaia tana (AMNH 106479). 265631, 265632, 265633, 265634, 265635, Dermoptera: Cynocephalus variegatus 265636; USNM 278616, 458462); Rousettus am- (AMNH 14021, 101501, 107136, 120449); Cy- plexicaudatus (USNM 278616, 458455, 458462); nocephalus volans (AMNH 16697, 187860); Cy- Rousettus celebensis (AMNH 224532); Scotonyc- nocephalus sp. (AMNH 207001; USNM 115603). teris zenkeri (AMNH 239381); Sphaerias blan- Pteropodidae: Aethalops alecto (AMNH fordi (AMNH 240004); Styloctenium wallacei 239600); Balionycteris maculata (AMNH (AMNH 153126, 222981); Syconycteris australis 216759); Boneia bidens (AMNH 254542); Cy- (AMNH 108857); Thoopterus nigricens (AMNH nopterus brachyotis (AMNH 235564; USNM 222771). 197237, 399424, 399426; USNM 20214); Cynop- Icaronycteridae: Icaronycteris index (LNK terus sphinx (AMNH 55568, 107922); Dobsonia 124/126; UW 2244, 21481a±b; PU 18150 [cast]). viridis (USNM 543177, 543180, 543778, Archaeonycteridae: Archaeonycteris trigono- 543798); Dobsonia moluccensis (AMNH don (SMF 80/1379, ME 214, ME 663, ME 963a, 198750); Dobsonia pannietensis (AMNH 157371, ME 1789b; HLMD Me 9754, Me 10591). 157372); Eidolon helvum (AMNH 86245, Palaeochiropterygidae: Palaeochiropteryx 236281); Eonycteris major (AMNH 241759); spiegeli (MNB Me 537; RSB Nr. 380a); Palaeo- Eonycteris spelaea (AMNH 31793, 235566; chiropteryx tupaiodon (MNB Me 719, 861; RSB USNM 458131, 458132); Epomophorous wahl- Nr. 165a, 165b, 166a, 166b, 257, 324a, 1034; bergi (AMNH 168100, 187264, 187265, 187266, SMF ME 788a, ME 788b, ME 1033b); Palaeo- 187269, 187270, 187271, 187272, 187273, chiropteryx sp. (HLMD Me 329, Me 330, Me 187274, 187275, 187276, 187277, 187278, 719a, Me 1477a, Me 10000, Me 10586, Me 187279, 187288, 207008; USNM 20215); Epom- 14479, Me 15008, Me 15010a, Me 15018, Me ops buettikoferi (AMNH 207007, 207008, 15019, Me 15021, Me 15021, Me 15022, Me 207009, 207010, 239375); Epomops dobsonii 15025, Me 15026, Me 15220, Me 15477; RSB Nr. (AMNH 88072); Epomops franqueti (AMNH 188a, 188b; SMF ME 505, ME 788a, ME 788b, 241012); Haplonycteris ®scheri (AMNH 187088; ME 1590a, ME 1590b, ME 1035, ME 1127, ME USNM 459100, 459101, 548573, 573419); Har- 1139a, ME 1151b, ME 1204a, ME 1204b, ME pyionycteris whiteheadi (AMNH 196435; USNM 1205, ME 1487a, ME 1487b, ME 1492a, ME 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 171

1492b, ME 1494, ME 1501b, ME 1504, ME simulator (AMNH 257165); Rhinolophus stheno 1507, ME 1580a, ME 1580b, ME 1590a, ME (AMNH 235577). 1590b, ME 1790b, ME 2018b, ME 2028a, ME Rhinolophidae: Hipposiderinae: Asellia tri- 2028b, ME 2348b, ME 2434a, ME 2434b, ME dens (AMNH 175962, 175963, 175964, 175965); 2434c, ME 2885). Aselliscus tricuspidatus (AMNH 159379); Coe- Hassianycteridae: Hassianycteris magna lops frithi (AMNH 107509); Hipposideros abae (HLMD Me 7605a, Me 7605b, Me 7998; RSB (AMNH 49157); Hipposideros diadema (AMNH Nr.100a, 66; SMF ME 1540a, ME 1761, ME 31800, 31812, 194858, 207561, 237819; USNM 2619b); Hassianycteris messelensis (HLMD Me 477728, 477732); Hipposideros ruber (AMNH 333a, Me 1116, Me 1469b, Me 7480; RSB Nr. 265796, 265797, 265798, 265799, 265800, 127a, 127c; SMF ME 1024b, ME 1414a, 1469a, 265804, 267805, 265806); Rhinonycteris auran- 1500). tius (AMNH 197213, 197214, 197215, 197216); Emballonuridae: Balantiopteryx io (AMNH Triaenops persicus (AMNH 216236, 216237, 185765); Coleura afra (AMNH 188263); Cor- 216238). mura brevirostris (AMNH 74103); Diclidurus al- Phyllostomidae: Artibeus jamaicensis (AMNH bus (AMNH 149167, 214183; USNM 120577, 244634); Brachyphylla cavernarum (AMNH 534417); Diclidurus isabellus (USNM 388543, 188225, 188229); Carollia perspicillata (AMNH 388544); Diclidurus scutatus (AMNH 267832); 209374, 209391, 210700, 210717, 246494, Emballonura alecto (USNM 458512, 458543, 246524); Desmodus rotundus (AMNH 14568, 459312, 459313, 459318, 459323); Emballonura 237372, 237373, 237274, 248942); Diaemus raffrayana (AMNH 153232); Peropteryx kappleri youngi (AMNH 209742, 209743); Erophylla se- (AMNH 239068); Rhynchonycteris naso (AMNH zekorni (AMNH 176294, 245691, 245692); Glos- 209190, 209191, 209213, 209214, 209215, sophaga soricina (AMNH 203613, 247936, 248745, 265986); Saccolaimus ¯aviventris 247937, 247960, 247961; USNM 312998); Lep- (AMNH 107759); Saccopteryx bilineata (AMNH tonycteris nivalis (AMNH 180343); Lonchophylla 210461, 265963); Taphozous melanopogon robusta (USNM 306590, 311996, 522757); Lon- (AMNH 235571, 235572, 241802). chophylla thomasi (AMNH 210688, 230211, Rhinopomatidae: Rhinopoma microphyllum 230284); Macrotus californicus (AMNH 180526); (AMNH 212070, 244388, 244389); Rhinopoma Micronycteris brachyotis (AMNH 175628); Mi- muscatellum (AMNH 244391, 244393, 244394; cronycteris hirsuta (AMNH 267093); Micronyc- USNM 327940, 327963). Craseonycteridae: Craseonycteris thonglon- teris megalotis (AMNH 266020); Micronycteris gyai (BMNH 77.2990, 77.2993, 77.2999, minuta (AMNH 267098); Micronycteris microtis 77.3007, 77.3009, 77.3015). (AMNH 267097); Micronycteris nicefori (AMNH Nycteridae: Nycteris gambiensis (USNM 266019); Phyllostomus discolor (AMNH 478656); Nycteris grandis (AMNH 206685); Nyc- 254610); Phyllostomus hastatus (AMNH teris hispida (AMNH 187291, 187293, 187294, 267434); Platyrrhinus lineatus (AMNH 205185); 187325); Nycteris macrotis (AMNH 187296, Sturnira lilium (AMNH 230529, 246568, 260882, 187297, 187298, 187299, 187300, 187302, 260887, 260888); Trachops cirrhosus (AMNH 187303, 187304, 187305, 187309, 187314, 129075); Vampyrum spectrum (AMNH 2644, 187317, 187319, 187320, 187322); Nycteris the- 42805, 42895, 212951, 256825, 267446; USNM baica (AMNH 169163, 169164, 187289, 187321, 346252). 257153). Mormoopidae: Mormoops blainvillii (AMNH Megadermatidae: Cardioderma cor (AMNH 176148); Mormoops megalophylla (AMNH 187331); Lavia frons (AMNH 187339, 187344); 25589, 25602, 190138, 190139; USNM 431641); Macroderma gigas (AMNH 197205, 197206, Pteronotus davyi (AMNH 204960); Pteronotus 197208, 197210, 197211, 236545, 236546); Me- parnellii (AMNH 176183, 176185, 189595, gaderma lyra (AMNH 33130); Megaderma spas- 245607, 267403; USNM 54139, 54140); Pteron- ma (AMNH 109285, 203289; USNM 458574, otus personatus (AMNH 321127). 573475, 573679). Noctilionidae: Noctilio albiventris (AMNH Rhinolophidae: Rhinolophinae: Rhinolophus 95121, 209304, 182706, 209252, 210576, af®nus (AMNH 27378, 216809, 257199, 257200); 210577); Noctilio leporinus (AMNH 91935, Rhinolophus arcuatus (USNM 175795, 175812, 91941; USNM 382924). 304355, 304356, 459445, 459447); Rhinolophus Mystacinidae: Mystacina robusta (AMNH darlingi (AMNH 257157, 257161); Rhinolophus 160269, 214243; USNM 120576); Mystacina tub- ferrumequinum (AMNH 245358); Rhinolophus erculata (AMNH 214245, 265139; MNHN 1983- hildebrandti (AMNH 216206, 216209); Rhinolo- 1464). phus megaphylla (AMNH 157393); Rhinolophus Myzopodidae: Myzopoda aurita (AMNH 172 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

257130; USNM 448883, 448886, 448932, 145485, 145486, 203937; USNM 52874, 578856). 512843). Thyropteridae: Thyroptera tricolor (AMNH Vespertilioninae: Corynorhinus townsendii 67592, 239085, 266356, 266361, 266364, (AMNH 170810, 180094); Eptesicus furinalis 268576; USNM 281939). (AMNH 268581); Eptesicus fuscus (AMNH Furipteridae: Amorphochilus schnablii 190168); Lasiurus borealis (AMNH 139871); (AMNH 28601); Furipterus horrens (AMNH Lasiurus cinereus (AMNH 15128, 214126); Las- 142903, 267214, 265979; USNM 315734, iurus intermedius (AMNH 203930, 253712); Nyc- 549598). ticeus humeralis (AMNH 3955, 3960); Nyctophi- Natalidae: Natalus stamineus (AMNH 186399, lus geoffroyi (AMNH 160300); Nycyophilus ti- 206696; USNM 362003, 362102, 262103). moriensis (AMNH 197281); Pipistrellus tasman- Antrozoidae: Antrozous pallidus (AMNH iensis (AMNH 197222); Scotoecus albofuscus 21593, 31173, 121495, 138339, 207615; USNM (AMNH 237386); Scotophilus kuhli (AMNH 242718); Vespertilio murinus (AMNH 206564). 244481, 564036); Bauerus dubiaquercus (AMNH Minopterinae: Miniopterus schreibersi 256832). (AMNH 219973, 219974, 219975, 219976, Molossidae: Tomopeatinae: Tomopeas ravus 219977) (LSU 25072, 25084, 25086, 25087, 25148, Myotinae: Myotis fortidens (AMNH 243791); 25150, 27170). Myotis lucifugus (AMNH 141128, 141129, Molossidae: Molossinae: Eumops bonariensis 141130, 208977, 208978, 244863); Myotis vivesi (AMNH 234699, 234700, 234701, 234717, (AMNH 180833, 180835). 235412, 235965); Eumops perotis (AMNH 3259/ Murininae: Murina cyclotis (AMNH 234204, 15751, 42379; USNM 270794, 271164); Molos- 234207; USNM 573777); Murina suilla (AMNH sus molossus (AMNH 211379, 211380, 211381, 217013); Murina sp. (USNM 573775). 211382, 211383, 211389); Molossus rufus Kerivoulinae: Kerivoula hardwickei (AMNH (AMNH 268597); Mops mops (USNM 20217); 109122, 234209); Kerivoula pellucida (AMNH Promops centralis (AMNH 183866); Tadarida 241938); Kerivoula papillosa (AMNH 247576, brasiliensis (AMNH 18020, 60791, 131663, 247577, 247579).

APPENDIX 2: CHARACTER DESCRIPTIONS The following character descriptions are based 1985; (32) Hill, 1974; (33) Hill and Harrison, on those presented by Simmons (1998). The char- 1987; (34) Hill and Smith, 1984; (35) Hood and acter descriptions given here have been modi®ed Smith, 1982; (36) Hood and Smith, 1983; (37) to incorporate some new data from extant forms, HoraÂcek, 1986; (38) Humphry, 1869; (39) Jepsen, and to account for character states seen only in 1970; (40) Johnson and Kirsch, 1993; (41) E. Kal- the fossils. The numbers following each character ko, unpubl. data; (42) Koopman, 1994; (43) Le- description refer to sources used to de®ne char- che, 1886; (44) Le Gros Clark, 1924; (45) Le acter states and obtain the data presented in ap- Gros Clark, 1926; (46) Loo and Kanagasunther- pendix 2. Sources are as follows: (1) Agrawal and am, 1972; (47) Luckett, 1980a; (48) Luckett, Sinha, 1973; (2) Andersen, 1912; (3) Baker et al., 1980b; (49) Luckett, 1980c; (50) Luckett, 1993; 1991a; (4) Baron et al., 1996a, 1996b, 1996c; (5) (51) MacAlister, 1872; (52) Madkour, 1989; (53) Beard et al., 1992; (6) Bhatnagar and Wible, Matthews, 1941; (54) Menu, 1985; (55) Menu and 1994; (7) Bhiwgade et al., 1992; (8) Blood, 1987; SigeÂ, 1971; (56) Norberg, 1970; (57) Norberg, (9) Brown et al., 1971; (10) Butler, 1980; (11) 1972a; (58) Novacek, 1980a; (59) Novacek, Corbet and Hill, 1992; (12) Cypher, 1996; (13) 1980b; (60) Novacek, 1985a; (61) Novacek, 1987; Davis, 1938; (14) Doran, 1878; (15) Felton et al., (62) Novacek, 1991; (63) Novick, 1962; (64) A. 1973; (16) Fenton and Bell, 1981; (17) J. Geisler, Pef¯ey, MS; (65) Pierson, 1986; (66) Pierson et personal obs.; (18) GoÈpfert and Wasserthal, 1995; al., 1986; (67) Robin, 1881; (68) Schlosser-Strum (19) Grif®ths, 1982; (20) Grif®ths, 1983; (21) and Schliemann, 1995; (69) W. Schutt, unpub. Grif®ths, 1994; (22) T. Grif®ths, unpubl. data; data; (70) SigeÂ, 1974; (71) Simmons, 1980; (72) (23) Grif®ths and Smith, 1991; (24) Grif®ths et Simmons, 1993b; (73) N. Simmons, personal al., 1992; (25) Habersetzer and Storch, 1987; (26) obs.; (74) Simmons and Quinn, 1994; (75) Smith, Habersetzer and Storch, 1989; (27) Habersetzer 1972; (76) Smith and Madkour, 1980; (77) Smith and Storch, 1992; (28) Habersetzer et al., 1994; and Storch, 1981; (78) Storch and Habersetzer, (29) Hermanson, 1981; (30) Hermanson and Al- 1988; (79) Strickler, 1978; (80) Surlykke et al., tenbach, 1983; (31) Hermanson and Altenbach, 1993; (81) Vaughan, 1959; (82) Vaughan, 1970b; 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 173

(83) Vaughan and Bateman, 1970; (84) Walton Character 16: Two upper incisors in each side and Walton, 1968; (85) Walton and Walton, 1970; of jaw (0); or one incisor (1); or incisors absent (86) Wassif, 1950; (87) Wassif and Madkour, (2). 10, 17, 42, 73, 77. 1963; (88) Weid and Helversen, 1987; (89) H. Character 17: Three lower incisors in each side Whidden, unpubl. data; (90) Wible and Bhatnagar, of jaw (0); or two incisors (1); or one incisor (2); 1997; (91) Wilkinson et al., 1997; (92) Winge, or incisors absent (3). 10, 17, 39, 42, 73, 77, 78. 1941; (93) Zeller, 1986. Character 18: Three upper premolars in each Character 1: Ear pinnae not funnel-shaped (0); side of jaw (0); or two premolars (1); or one pre- or more or less funnel-shaped (1). 11, 42, 73. molar (2). 10, 17, 39, 42, 73, 77, 78. Character 2: Tragus absent (0); or present (1). Character 19: Middle upper premolar with three 11, 42, 73. roots (0); or with two roots (1); or with one root Character 3: Narial structures absent (0); or der- (2). 10, 17, 39, 42, 73, 77, 78. mal ridge present dorsal to nostrils (1); or noseleaf Character 20: Three lower premolars in each present (2); or dermal foliations with central slit side of jaw (0); or two premolars (1). 10, 17, 39, present (3). 11, 32, 73. 42, 73, 77, 78. Character 4: M. occipitofrontalis inserts into Character 21: Lower ®rst and second molars connective tissue and skin over nasal region (0); with primitive tribosphenic arrangement of cusps or inserts onto nasal cartilage via common tendon and cristids; hypoconulid located near anteropos- with contralateral muscle (1). 89, 92. terior midline, postcristid connects hypoconid Character 5: Nasopalatine duct present (0); or with hypoconulid (0); or nyctalodont, hypoconu- absent (1). 6, 46, 90. lid shifted lingually to lie adjacent to entoconid, Character 6: Paraseptal cartilage (ϭ vomero- postcristid connects hypoconid with hypoconulid nasal cartilage) J-shaped, C-shaped, U-shaped, or (1); or myotodont, postcristid bypasses hypocon- O-shaped (0); or bar-shaped (1); or absent (2). 6, ulid to connect with entoconid (2); or teeth mod- 46, 90. i®ed for fruit and/or nectar or blood feeding, Character 7: Vomeronasal epithelial tube well cusps and cristids not distinct (2). 5, 10, 17, 25, developed, neuroepithelium present (0); or tube 54, 55, 61, 73, 78. rudimentary, neuroepithelium absent (1), or epi- Character 22: Lower jaw with elongate angular thelial tube absent (2). 6, 46, 90. process (0); or without elongate angular process Character 8: Accessory olfactory bulb present (1). 17, 61, 73. (0); or absent (1). 6, 46, 90. Character 23: Angular process projects at or be- Character 9: Premaxilla articulates with maxilla low level of occlusal plane of toothrow, well be- via sutures (0); or premaxilla fused to maxilla (1); low coronoid process (0); or angular process pro- or premaxilla articulates with maxilla via liga- jects above level of occlusal plane of toothrow, at ments, premaxilla freely movable (2). 17, 32, 42, same level as the coronoid process (1). This char- 73. acter cannot be scored in taxa that lack an angular Character 10: Nasal branches of premaxillae process. 17, 73. well developed (0); or reduced or absent (1). 17, Character 24: Postorbital process present (0); or 32, 42, 73. absent (1). 17, 42, 73. Character 11: Palatal branches of premaxillae Character 25: Pars cochlearis of petrosal su- well developed (0); or reduced or absent (1). 17, tured to basisphenoid (0); or loosely attached to 32, 42, 73. basisphenoid via ligaments and/or thin splints of Character 12: Palatal branches of premaxillae bone (1). 17, 59, 61, 62, 73. not fused with one another across midline (0); or Character 26: Cochlea not enlarged (0); or fused at midline (1). 32, 73. moderately enlarged (1); or greatly enlarged (2). Character 13: Emargination not present in an- 17, 27, 58, 59, 60, 61, 63, 73. terior palate, medial incisors directly adjacent to Character 27: Cochlea cryptocochlear (0); or one another (0); or emargination present between phanerocochlear (1). 17, 26, 27, 28, 59, 62, 73. medial incisors (1). 42, 73. Character 28: Lateral process of ectotympanic Character 14: Anterior palatal emargination weak or absent (0); or well developed, forms tu- shallow, extends posteriorly no farther than ante- bular external auditory meatus (1). 17, 58, 59, 73. rior edge of canines (0); or deep, extends poste- Character 29: Tympanic annulus inclined (0); riorly at least as far as posterior edge of canines or annulus semivertical in orientation (1). 58, 59, (1). This character cannot be scored in taxa that 73. lack an anterior palatal emargination. 42, 73. Character 30: Epitympanic recess shallow and Character 15: Hard palate extends posteriorly broad (0); or deep, often constricted in area (1). into interorbital region (0); or terminates either at 17, 58, 59, 73. or anterior to level of zygomatic roots (1). 17, 73. Character 31: Fossa for m. stapedius indistinct 174 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

(0); or shallow and broad (1); or deep, constricted basihyal raphe (0); or inserts on basihyal, basihyal in area, often a crescent-shaped ®ssure (2). 17, 58, raphe, and thyrohyal (1). 19, 20, 21, 22, 23, 24. 59, 73. Character 45: M. mylohyoideus aponeurotic an- Character 32: Fenestra cochleae (ϭ fenestra ro- teriorly (0); or ¯eshy for entire width from man- tundum) small or of moderate size, maximum di- dibular symphysis to at least basihyal region (1). ameter Ͻ20% of the external width of the ®rst 19, 20, 21, 22, 23, 24. half turn of the cochlea (0); or enlarged, maxi- Character 46: M. mandibulo-hyoideus (ϭ me- mum diameter Ͼ25% of the external width of the dial part of anterior digastric) absent (0); or pres- ®rst half turn of the cochlea (1). 17, 58, 73. ent (1). 19, 20, 21, 22, 23, 24. Character 33: Aquaeductus cochleae large and Character 47: M. mandibulo-hyoideus with obvious (0); or small or absent, dif®cult to detect muscle ®bers (0); or reduced to tendinous band (1). 58, 73. (1). This character cannot be scored in taxa that Character 34: M. tensor tympani muscle spin- lack m. mandibulo-hyoideus. 19, 20, 21, 22, 23, dle-shaped, inserts via single tendon onto tuber- 24. cular processus muscularis of malleus (0); or two- Character 48: M. mandibulo-hyoideus well de- headed, inserts via two tendons onto processus veloped (0); or reduced to small muscle with ten- muscularis and accessory process (1); or broad don (1). This character cannot be scored in taxa sheet of ®bers, inserts on crest like processus that lack m. mandibulo-hyoideus or have a man- muscularis (2); or absent (3). 14, 58, 86, 87, 93. dibulo-hyoideus that has been reduced to a ten- Character 35: Orbicular apophysis small or ab- dinous band. 19, 20, 21, 22, 23, 24. sent (0), or large (1). 14, 17, 58, 73. Character 49: M. stylohyoideus with slip that Character 36: Laryngeal echolocation absent passes super®cial to digastric muscles (0); or su- (0); or present (1). 16, 18, 41, 63, 71, 80, 88. per®cial slip absent (1). 19, 20, 21, 22, 23, 24. Character 37: One pair of submaxillary glands Character 50: M. stylohyoideus with slip that present (0); or two pairs (1). 45, 67. passes deep to digastic muscles (0); or deep slip Character 38: Right lung divided into four absent (1). 19, 20, 21, 22, 23, 24. lobes (0); or three lobes (1); or two lobes (2); or Character 51: M. geniohyoideus origin from undivided (3). 67. ¯at posterior surface of mandible lateral to sym- Character 39: Left lung divided into two lobes physis (0); or from pronglike process that extends (0); or undivided (1). 67. posteroventrally from symphysis region (1). 19, Character 40: Tracheal rings subequal in di- 20, 21, 22, 23, 24. ameter throughout length of trachea (0); or one Character 52: M. geniohyoideus originates by ring enlarged to form tracheal expansion just pos- long tendon from the mandible (0); or by very terior to larynx (1); or two to eight rings enlarged short tendon (1); or medial ®bers originate by ten- to form tracheal expansion just posterior to larynx don, lateral muscle ®bers arise directly from the (2); or nine or more rings enlarged to form a tra- bone of the mandible (2); or muscle arises entirely cheal expansion that is separated from larynx by by ¯eshy ®bers from bone (3). 19, 20, 21, 22, 23, four or ®ve rings of normal diameter. 19, 20, 21, 24. 22, 23, 24. Character 53: M. genioglossus originates im- Character 41: Midline hyoid strap musculature mediately lateral to mandibular symphysis (0); or with m. geniohyoideus and m. hyoglossus directly origin extended laterally onto medial surface of attached to basihyal via ¯eshy ®bers (0); or mus- mandible, occupying anterior one-fourth to one- cles attached indirectly to basihyal via basihyal third of medial mandibular surface (1). 19, 20, 21, tendon, resulting in ``free-¯oating'' strap muscle 22, 23, 24. condition (1); or basihyal tendon lost, no connec- Character 54: M. genioglossus and m. genioh- tion between m. geniohyoideus and m. hyoglossus yoideus not fused (0); or ventralmost ®bers of m. and basihyal (2). 19, 20, 21, 22, 23, 24. genioglossus fused to ®bers from caudal portion Character 42: M. sternohyoideus directly at- of m. geniohyoideus (1). 19, 20, 21, 22, 23, 24. tached to basihyal via ¯eshy ®bers (0); or attached Character 55: M. genioglosssus inserts into indirectly to basihyal via basihyal tendon (1); or posterior tongue, no ®bers insert onto basihyal no connection between m. sternohyoideus and ba- (0); or ventralmost ®bers insert onto basihyal (1). sihyal (2). 19, 20, 21, 22, 23, 24. 19, 20, 21, 22, 23, 24. Character 43: Deep division of m. mylohyoi- Character 56: M. hyoglossus originates from deus absent (0); or present, runs dorsal to midline entire lateral basihyal and thyrohyal in broad, un- strap musculature, inserts on basihyal (1). 19, 20, broken sheet (0); or originates from lateral basi- 21, 22, 23, 24. hyal and lateral thyrohyal in two sheets separated Character 44: M. mylohyoideus runs ventral to by a space (1); or originates from antimere in part, midline strap musculature, inserts on basihyal and and from lateral basihyal and thyrohyal (2); or 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 175 originates from lateral basihyal, thyrohyal origin ceratohyal reduced to tiny element or completely absent (3). 19, 20, 21, 22, 23, 24. absent (2). 19, 20, 21, 22, 23, 24. Character 57: M. styloglossus with one belly Character 73: Epihyal unreduced, approximate- (0); or with two bellies separated by lateral part ly equal in length to ceratohyal (0); or epihyal of m. hyoglossus (1). 19, 20, 21, 22, 23, 24. reduced to half the size of ceratohyal (1); or epi- Character 58: M. styloglossus originates from hyal reduced to very tiny element or completely expanded tip of stylohyal and/or adjacent surface absent (2). 19, 20, 21, 22, 23, 24. of skull (0); or from ventral surface of midpoint Character 74: Stylohyal occurs as gently curved of stylohyal (1). 19, 20, 21, 22, 23, 24. bar with no enlargement or other modi®cation to Character 59: M. ceratohyoideus inserts at least the lateral edge or cranial tip (0); or with cranial in part onto ceratohyal (0); or does not insert onto tip slightly expanded (1); or with bifurcated tip ceratohyal (1). 19, 20, 21, 22, 23, 24. (2); or with large, ¯at expansion or ``foot'' on Character 60: M. ceratohyoideus inserts at least lateral cranial tip (3); or with very large, ¯at, axe- in part onto epihyal (0); or does not insert onto shaped enlargement at tip (4); or lateral half of epihyal (1). 19, 20, 21, 22, 23, 24. entire stylohyal swollen (5). 17, 19, 20, 21, 22, Character 61: M. ceratohyoideus inserts at least 23, 24. in part onto stylohyal (0); or does not insert onto Character 75: Posteriorly directed ventral ac- stylohyal (1). 19, 20, 21, 22, 23, 24. cessory processes not present on centra of cervi- Character 62: M. thyrohyoideus inserts onto cal vertebrae 2 and 3 (0); or ventral accessory thyrohyal (0); or muscle enlarged, inserts onto processes present on C2 and C3 (1). 17, 61, 73. thyrohyal and basihyal (1). 19, 20, 21, 22, 23, 24. Character 76: Posteriorly directed, ventral ac- Character 63: M. sternohyoideus origin in- cessory processes not present on centrum of cer- cludes entire anterodorsal surface of manubrium vical vertebra 4 (0); or ventral acessory processes present on C4 (1). 17, 61, 73. (0); or manubrial origin restricted to medialmost Character 77: Posteriorly directed, ventral ac- surface of manubrium in vicinity of keel (1). 19, cessory processes not present on centrum of cer- 20, 21, 22, 23, 24. vical vertebra 5 (0); or ventral acessory processes Character 64: M. sternohyoideus origin does present on C5 (1). 17, 61, 73. not extend onto clavicle (0); or origin includes Character 78: Seventh cervical vertebra not medial tip of clavicle (1). 19, 20, 21, 22, 23, 24. fused to ®rst thoracic vertebra (0); or C7 and T1 Character 65: M. sternohyoideus relatively at least partially fused (1). 17, 56, 64, 73. broad (0); or reduced to a narrow strip of muscle Character 79: First and second thoracic verte- (1). 19, 20, 21, 22, 23, 24. brae not fused (0); or T1 and T2 fused (1). 17, Character 66: M. sternothyroideus originates 56, 64, 73. from lateral manubrium (0); or from medial tip of Character 80: Anterior ribs not fused to verte- clavicle (1); or origin includes both lateral ma- brae (0); or ®rst rib fused to vertebrae (1); or at nubrium and medial clavicle (2). 19, 20, 21, 22, least ®rst ®ve ribs fused to vertebrae (2). 17, 56, 23, 24. 64, 73. Character 67: M. omohyoideus originates from Character 81: Width of ®rst rib similar to other scapula (0); or clavicle (1); or muscle absent (2). ribs (0); or ®rst rib at least twice the width of 19, 20, 21, 22, 23, 24. other ribs (1). 17, 56, 73. Character 68: Body of basihyal consists of Character 82: First costal cartilage not ossi®ed transverse, unadorned bar or plate (0); or body of or fused with manubrium or ®rst rib (0); or ®rst basihyal consists of curved bar with apex directed costal cartilage ossi®ed and fused to manubium anteriorly (1). 19, 20, 21, 22, 23, 24. (where it appears to form a winglike lateral pro- Character 69: Curved body of basihyal V- cess of the manubrium) and fused to ®rst rib (1). shaped (0); or U-shaped (1). This character cannot Character 83: Second costal cartilage articulates be scored in taxa that lack a curved basihyal. 19, with sternum at manubrium±mesosternum joint 20, 21, 22, 23, 24. (0), or second rib articulates with manubrium, no Character 70: Entoglossal process of basihyal contact between rib (or costal cartilage) and me- absent (0); or present (1). 19, 20, 21, 22, 23, 24. sosternum (1). 17, 56, 73. Character 71: Entoglossal process small (0), or Character 84: Second rib articulates with ster- very large, resulting in T-shaped basihyal (1). num via costal cartilage (0), or second rib fused This character cannot be scored in taxa that lack to sternum, costal cartilage absent or ossi®ed (1). an entoglossal process. 19, 20, 21, 22, 23, 24. 17, 56, 73. Character 72: Ceratohyal unreduced, approxi- Character 85: Mesosternum articulates with at mately equal in length to epihyal (0); or cerato- least ®ve costal cartilages posterior to second rib hyal reduced to half the length of epihyal (1); or (0); or articulates with four costal cartilages pos- 176 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 terior to second rib (1); or articulates with only chiter of humerus) absent from scapula (0); or three costal cartilages posterior to second rib (2). present (1). 17, 68, 73. 17, 37, 56, 73. Character 101: Dorsal articular facet faces dor- Character 86: Ribs with no anterior laminae (0); solaterally and consists of small groove on anter- or anterior laminae present (1). 17, 73. omedial rim of glenoid fossa (0); or faces dorso- Character 87: Anterior laminae on ribs narrow, laterally and consists of an oval facet on antero- lamina width less than that of main body of rib medial rim of glenoid fossa (1); or faces dorsally (0); or anterior laminae wide, equal to or wider and consists of a large, ¯at facet clearly separated than main body of rib (1). This character cannot from glenoid fossa (2). 17, 68, 73. be scored for taxa that lack anterior laminae on Character 102: Infraspinous fossa of scapula the ribs. 17, 73. narrow, length Ն2 times the width (0); or wide, Character 88: Ribs with no posterior laminae length Յ1.5 times the width (1). 17, 56, 73. (0); or posterior laminae present (1). 17, 73. Character 103: Infraspinous fossa with one fac- Character 89: Posterior laminae on ribs narrow, et (0); or two facets (1); or three facets (2). 17, lamina width less than that of main body of rib 56, 73. (0); or posterior laminae wide, equal to or wider Character 104: Intermediate infraspinous facet than main body of rib (1). This character cannot narrower than posterolateral facet (0); or facets be scored for taxa that lack posterior laminae on subequal (1); or intermediate facet wider than the ribs. 17, 73. posterolateral facet (2). This character cannot be Character 90: Anterior face of manubrium scored in taxa that have only one or two infras- small (0); or broad, de®ned by elevated ridges (1). pinous facets. 17, 56, 73. 17, 56, 73. Character 105: Lateral/posterolateral facet of Character 91: Ventral process of manubrium infraspinous fossa restricted, does not extend into absent (0), or ventral process present, distal tip infraglenoid region anteriorly or wrap around in- blunt or rounded (1); or ventral process present, termediate facet at posterior (caudal) angle of distal tip laterally compressed (2). 17, 56, 73. scapula (0); or posterolateral facet more extensive, Character 92: Angle between axis of ventral extends into infraglenoid region and wraps around process and body of manubrium acute (0); or ap- caudal end of intermediate facet (1). 17, 56, 73. proximately 90Њ (1); or obtuse (2); or ventral pro- Character 106: Thick lip present along axillary cess bilobed with one acute and one obtuse pro- border of scapula (0); or thick lip with bladelike cess (3). This character cannot be scored in taxa lateral edge present (1); or thick lip absent, axil- that lack a ventral process on the manubrium. 17, lary border ¯at or slightly upturned (2). 17, 56, 56, 73. 73. Character 93: Length of manubrium posterior Character 107: Pit for attachment of clavicular to lateral processes Ͼ2.5 times the transverse ligament absent from scapula (0), or present an- width (0); or length Ͻ2 times the transverse width terior and medial to glenoid fossa (1). 17, 73. (1). 17, 56, 73. Character 108: Anteromedial edge of scapula Character 94: Mesosternum narrow, mean without projections or ¯anges (0); or with trian- width less than half the distance between clavicles gular anteromedial ¯ange (1). 17, 56, 73. at sternoclavicular joint (0); or mesosternum Character 109: Coracoid process stout and of broad, mean width greater than three-fourths the moderate length (0); or very long and thin (1). 17, distance between clavicles (1). 17, 37, 56, 73. 56, 73. Character 95: Xiphisternum without keel (0); or Character 110: Coracoid process curves ventro- with prominent median keel (1). 17, 56, 73. laterally (0); or curves ventrally (1); or curves Character 96: Posterior xiphisternum with wide ventromedially (2). 17, 56, 73. lateral ¯are (0); or not laterally ¯ared (1). 17, 56, Character 111: Tip of coracoid process not 73. ¯ared, approximately same width as coracoid Character 97: Acromion process without medial shaft (0); or tip distinctly ¯ared (1); or bifurcated shelf (0); or with shelf that projects medially over (2). 17, 56, 73. supraspinous fossa or medial base of acromion Character 112: Suprascapular process present process (1). 17, 56, 73. (0); or absent (1). 17, 56, 73. Character 98: Tip of acromion process without Character 113: Clavicle articulates with or lies anterior projection (0); or with triangular anterior in contact with acromion process (0); or is sus- projection (1). 17, 56, 73. pended by ligaments between acromion and cor- Character 99: Distal acromion process without acoid processes (1); or articulates with or lies in posterolateral projection (0); or with triangular contact with coracoid process (2). 17, 29, 56, 73, posterolateral projection (1). 17, 56, 73. 79. Character 100: Dorsal articular facet (for tro- Character 114: M. subclavius originates from 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 177

®rst costal cartilage (0); or from ®rst costal car- zius from m. spinotrapezius. 30, 31, 43, 44, 45, tilage and lateral process of manubrium (1). 30, 56, 79, 81, 82. 31, 43, 56, 79, 81, 82. Character 127. Origin of m. acromiotrapezius Character 115: Anterior division of m. pector- does not include thoracic vertebra 6 (0); or does alis profundus originates from ®rst costal cartilage include T6 (1). This character cannot be scored in (0); or from ®rst costal cartilage and clavicle (1); taxa that lack differentiation of m. acromiotrape- or from clavicle only (2). 30, 31, 43, 56, 79, 81, zius from m. spinotrapezius. 30, 31, 43, 44, 45, 82. 56, 79, 81, 82. Character 116: M. dorsi patagialis absent (0); Character 128: M. clavotrapezius not differen- or present (1). 43, 44, 45, 79. tiated from acromiotrapezius (0); or clearly dif- Character 117: M. humeropatagialis absent (0); ferentiated (1). 30, 31, 43, 44, 45, 56, 79, 81, 82. or present (1). 43, 79, 81, 82. Character 129: M. levator scapulae originates Character 118: M. occipitopollicalis absent (0); from atlas (0); or from three to ®ve vertebrae be- or present (1). 31, 43, 44, 45, 56, 79, 81, 82. tween C2 and C7 (1); or from C4 and C5 only Character 119: M. occipitopollicalis with no (2). 30, 43, 44, 45, 56, 79, 81, 82. tendinous attachments to the anterior division of Character 130: M. omocervicalis absent (0); or m. pectoralis profundus (0); or with tendinous at- present (1). 30, 43, 56, 79, 81, 82. tachment to the anterior division of m. pectoralis Character 131: M. omocervicalis originates profundus (1). This character cannot be scored in from ventral arch of C2 (0); or from transverse taxa that lack m. occipitopollicalis. 79, 83. processes of C2 (1); or from transverse processes Character 120: M. occipitopollicalis with no of C3 and C4 (2). This character cannot be scored tendinous attachments to the posterior division of in taxa that lack m. omocervicalis. 30, 43, 56, 79, m. pectoralis profundus (0); or with tendinous at- 81, 82. tachment to the posterior division of m. pectoralis Character 132: M. omocervicalis inserts on profundus (1). This character cannot be scored in acromion process of scapula (0); or on clavicle taxa that lack m. occipitopollicalis. 79. (1). This character cannot be scored in taxa that Character 121: M. occipitopollicalis without lack m. omocervicalis. 30, 43, 56, 79, 81, 82. muscle belly between cranial muscle belly and Character 133: Anterior division of m. serratus band of elastic tissue (0); or with small interme- anterior originates from six or more ribs (0); or diate muscle belly (1). This character cannot be from four or ®ve ribs (1); or from two ribs (2). scored in taxa that lack m. occipitopollicalis. 56, 13, 30, 31, 44, 45, 56, 79, 81, 82. 79. Character 134: M. latissimus dorsi inserts on Character 122: M. occipitopollicalis insertional ventral ridge of humerus (0); or muscle divided, complex includes muscle ®bers distal to band of inserts on ventral ridge plus distal pectoral crest elastic tissue (0); or entirely tendinous distal to (1). 31, 43, 44, 45, 56, 65, 79, 81, 82. band of elastic tissue (1). This character cannot Character 135: M. teres major inserts into ven- be scored in taxa that lack m. occipitopollicalis. tral ridge (0); or pectoral crest (1). 30, 31, 43, 44, 79. 45, 56, 79, 81, 82. Character 123: M. occipitopollicalis inserts into Character 136: M. acromiodeltoideus originates pollex and metacarpal of digit II (0); or inserts on from acromion process plus Յ25% of the length pollex only (1); or inserts into wing membrane of the transverse scapular ligament (0); or acro- anterior to metacarpal of digit II. This character mion plus Ͼ50% of transverse scapular ligament cannot be scored in taxa that lack m. occipitopol- (1). 13, 30, 31, 43, 44, 45, 56, 79, 81, 82. licalis. 56, 79, 81, 82. Character 137: M. spinodeltoideus originates Character 124: M. spinotrapezius not differen- from vertebral border of scapula plus transverse tiated from anterior trapezius complex (0); or scapular ligament (0); or from vertebral border clearly differentiated from acromiodeltoideus (1). only (1); or muscle absent (2). 13, 30, 31, 43, 44, 30, 31, 43, 44, 45, 56, 79, 81, 82. 45, 56, 79, 81, 82. Character 125. Origin of m. acromiotrapezius Character 138: Coracoid head (ϭ short head) includes thoracic vertebra 4 (0); or does not in- of m. biceps brachii less than or equal to approx- clude T4 (1). This character cannot be scored in imately one-third the size of glenoid head (ϭ long taxa that lack differentiation of m. acromiotrape- head) (0); or coracoid head approximately one- zius from m. spinotrapezius. 30, 31, 43, 44, 45, half the size of glenoid head (1); or coracoid head 56, 79, 81, 82. three-quarters size or subequal to glenoid head Character 126. Origin of m. acromiotrapezius (2); or coracoid head one and a half times the size does not include thoracic vertebra 5 (0); or does of glenoid head (3). 56, 79, 81, 82, 83. include T5 (1). This character cannot be scored in Character 139: Trochiter does not extend to lev- taxa that lack differentiation of m. acromiotrape- el of proximal edge of head of humerus (0); or 178 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235 extends just to level of proximal edge of head (1); inserts on digit IV (0); or does not insert on digit or extends proximally well beyond level of head IV (1). 2, 30, 43, 44, 45, 56, 57, 81, 82, 83. (2). 17, 32, 39, 68, 73, 85. Character 156: M. ¯exor digitorum profundus Character 140: Head of humerus round in out- inserts on digit V (0); or does not insert on digit line in medial view (0); or oval or elliptical (1). V (1). 2, 30, 43, 44, 45, 56, 57, 81, 82, 83. 17, 32, 68, 73. Character 157: M. extensor digiti quinti present Character 141: Humerus with distal articular (0); or absent (1). 2, 30, 43, 44, 45, 56, 57, 81, surfaces displaced laterally from line of shaft (0); 82, 83. or facets in line with shaft, not displaced laterally Character 158: No vertebral fusion in posterior (1). 15, 17, 18, 32, 55, 73, 75. thoracic and lumbar series (0); or at least three Character 142: Epitrochlea broad, width Ն40% vertebrae fused (1). 17, 64, 73. of width of the articular facets (0); or relatively Character 159: Sacrum (de®ned as including all narrow, width Ͻ25% width of articular facets (1). vertebrae that articulate with the pelvis or are 15, 17, 32, 73, 75. fused with those that do form an articulation) ter- Character 143: M. brachioradialis present (0); minates anterior to acetabulum (0); or extends or absent (1). 2, 30, 43, 44, 45, 56, 57, 81, 82, posteriorly to at least the midpoint of the acetab- 83. ulum (1). 17, 64, 73. Character 144: Sesamoid element dorsal to Character 160: Sacral laminae narrow or ab- magnum±trapezium articulation absent (0); or sent, vertebra width (including laminae) less than present (1). 12, 17, 56. or equal to three-fourths vertebral body length (0); Character 145: Sesamoid element ventral to un- or laminae broad, vertebra width equal to or great- ciform±magnum articulation absent (0); or present er than vertebral length. 17, 64, 73. (1). 12, 56. Character 161: Dorsomedial edge of ascending Character 146: Sesamoid element dorsal to lu- process of ilium upturned, ¯ares dorsally above nar±radius articulation absent (0); or present (1). the level of iliosacral articulation, iliac fossa large 12, 56. and well de®ned (0); or dorsomedial edge not up- Character 147: Sesamoid element dorsal to un- turned, does not extend dorsally beyond the level ciform±magnum articulation absent (0); or present of the iliosacral articulation, iliac fossa not large (1). 12, 17, 56. or well de®ned (1). 17, 73. Character 148: Sesamoid element dorsal to tra- Character 162: Ischium with large ischial tu- pezium±metacarpal I articulation absent (0); or berosity that projects dorsally from posterior hor- present (1). 12, 56. izontal ramus (0); or ischial tuberosity small or Character 149: Wing digit II with ossi®ed ®rst absent, does not project dorsally beyond level of (proximal) phalanx (0); or ®rst phalanx unossi®ed ramus (1). 17, 73. or absent (1). 17, 32, 56, 79, 85. Character 163: Pubic spine absent (0); or Character 150: Wing digit II with ossi®ed sec- straight (1); or tip of pubic spine bent sharply dor- ond phalanx (0); or second phalanx unossi®ed or sally (2). 17, 73. absent (1). 17, 32, 56, 79, 85. Character 164: Articulation between pubes in Character 151: Wing digit II with ossi®ed third male broad, symphysis long in anteroposterior di- (ungual) phalanx (0); or third phalanx unossi®ed mension (0); or contact restricted to small area, or absent (1). 17, 32, 56, 79, 85. consists of an ossi®ed interpubic ligament or short Character 152: Wing digit III with third (un- symphysis (1). 43, 73. gual) phalanx completely ossi®ed (0); or third Character 165: Obturator foramen normal, rim phalanx ossi®ed only at the base (1); or third pha- well de®ned (0); or foramen partially in®lled with lanx unossi®ed or absent (2). 17, 32, 56, 79, 85. thin, bony sheet along posteroventral rim (1). 17, Character 153: Wings folded by ¯exing all pha- 73. langes in digits III, IV, and V anteriorly toward Character 166: M. psoas minor tendinous for the underside of the wing (0); or proximal phalanx approximately half of length (0); or thick and of digits III and IV folded posteriorly, distal pha- ¯eshy throughout length (1). 13, 43, 44, 45, 51, langes folded anteriorly (1); or distal phalanges of 73, 81, 82. digits III and IV folded anteriorly, proximal pha- Character 167: M. gluteus super®cialis not dif- langes not folded (2). This character cannot be ferentiated (0); or differentiated into m. gluteus scored in taxa that lack wings. 32, 85. maximus and m. tensor fascia femoris (1). 8, 13, Character 154: M. ¯exor digitorum profundus 38, 43, 44, 45, 51, 73, 81, 82. inserts on digit II (0); or does not insert on digit Character 168: M. piriformis present (0); or ab- II (1). 2, 30, 43, 44, 45, 56, 57, 81, 82, 83. sent (1). 13, 38, 43, 44, 45, 51, 73, 81, 82. Character 155: M. ¯exor digitorum profundus Character 169: Shaft of femur straight (0); or 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 179 with bend that directs distal shaft dorsally (1). 17, larization or hypertrophy of yolk sac endoderm 73. (0); or development includes spread of mesoderm Character 170: Fibula complete, well-devel- and expansion of exocoelom over embryonic half oped (0); or ®bula thin and threadlike (1); or ab- of yolk sac only; embryonic half becomes vas- sent or entirely unossi®ed (2). 17, 73, 84, 85. cular and exhibits endoderm hypertrophy (1); or Character 171: Calcar absent (0); or present (1). mesoderm spreads over entire yolk sac, but exo- 2, 17, 73, 85. coelom expands to cover only embryonic half of Character 172: Digits II±V of foot with three yolk sac; hypertrophy occurs in vascular yolk sac phalanges (0); or two phalanges (0). 17, 73, 85. at embryonic pole (2); or mesoderm spreads over Character 173: Digital tendon locking mecha- entire surface of yolk sac, exocoelom expands to nism absent (0); or present, consists of tubercles separate yolk sac completely from chorion; free, on ¯exor tendon and plicae on adjacent tendon vascular yolk sac subsequently collapses and en- sheath (1); or present, with plicae but no tubercles dodermal cells hypertrophy (3). 47, 50, 51. (2). 69, 74. Character 187: Allantoic vesicle large, occupies Character 174: Baculum absent (0); or present at least half of circumference of chorion during (1). 1, 9, 33, 34, 47, 48, 52, 53, 71, 76, 87. limb-bud stage (0); or small, occupies less than Character 175: Baculum saddle-shaped or slip- half the circumference of the chorion (1); or ves- per-shaped (0); or baculum elongated with long icle does not form, allantois remains tubular and central shaft (1). This character cannot be scored vestigial throughout gestation (2). 47, 48, 49, 50. in taxa that lack a baculum. 1, 9, 33, 34, 47, 48, Character 188: Primordial amniotic cavity does 52, 53, 71, 76, 87. not form at any stage in developement; amnioge- Character 176: Pubic nipples absent in females nesis is by folding (0); or primordial amniotic (0); or one pair present (1). 72, 73. cavity forms but is transitory, lost in later devel- Character 177: Female external genitalia with opment (1); or primordial aminotic cavity persists transverse vulval opening (0); or vulval opening as de®nitive amniotic cavity (2). 47, 48, 49, 50. oriented anteroposteriorly (1). 46, 47, 53, 66, 73. Character 189: De®nitive chorioallantoic pla- Character 178: Clitorus small, not elongated centa endotheliochorial (0); or hemochorial (1). 7, anteroposteriorly (0); or clitorus elongated (1). 43, 47, 48, 49, 50. 47, 53, 66, 73. Character 190: Spinal cord with angle between Character 179: External uterine fusion minimal, dorsal horns 70Њ±80Њ (0); or 35Њ±50Њ (1); or 0±25 uterine horns more than 70% length of common (2). 40. uterine body (0); or fusion more extensive, uterine Character 191: Inferior colliculus signi®cantly horns less than 50% length of common uterine smaller than superior colliculus (0); or inferior body (1). 35, 36, 43, 44, 45, 52, 53, 73. colliculus larger than superior colliculus (1). 40. Character 180: Internal uterine fusion absent, Character 192: Olfactory bulb connected to two cervical openings into vagina (0); or common brain via a ``compact connection'' by way of a uterine lumen present (1). 35, 36, 43, 58, 73. short, thick olfactory peduncle containing paleo- Character 181: Common uterine lumen short in cortical cells (0); or via a ``thin connection'' by comparison to length of cornual lumina (0); or way of a longer and thinner olfactory peduncle common uterine lumen large, cornual lumina ei- containing mainly olfactory ®bers with nerve ther join immediately within common uterine cells only in its most caudal part (1). 4. body or are reduced to tubular intramural uterine Character 193: Left central lobe of liver fused cornua (1). This character cannot be scored in with left lateral lobe (0); or separate from other taxa that lack a common uterine lumen. 35, 36, lobes or partially fused with right central lobe (1). 43, 53, 73. 13, 67. Character 182: Uterotubal junction with ovi- Character 194: Gall bladder located in right lat- ductal papillae or complex folds (0), or simple, no eral ®ssue of liver (0); or in umbilical ®ssue (1). papillae or complex folds (1). 35, 36. 13, 67. Character 183: Blastocyst stage attained in uter- Character 195: Caecum present (0); or absent us (0); or attained in oviduct (1). 47. (1). 47, 48, 67. Character 184: Implantation super®cial (0); or Character 196: rDNA restriction site 20 present secondarily interstitial (1). 47. (0); or polymorphic, either present or absent (1). 3. Character 185: Embryonic disc oriented toward Character 197: rDNA restriction site 28 absent tubo-uterine junction (0); or consistently antime- (0); or present (1). 3. sometrial (1). 47, 48, 49, 50. Character 198: rDNA restriction site 29 absent Character 186: Yolk sac develoment includes (0); or present (1). 3. spread of mesoderm and expansion of exocoelom Character 199: rDNA restriction site 37 absent over embryonic half of yolk sac only; no vascu- (0); or present (1). 3. 180 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

Character 200: rDNA restriction site 38 absent Character 205: rDNA restriction site 46 absent (0); or present (1). 3. (0); or present (1). 3. Character 201: rDNA restriction site 40 absent Character 206: rDNA restriction site 47 absent (0); or present (1). 3. (0); or present (1); or polymorphic, either present Character 202: rDNA restriction site 43 absent or absent (2). 3. (0); or present (1). 3. Character 207: rDNA restriction site 50 absent Character 203: rDNA restriction site 44 absent (0); or present (1). 3. (0); or present (1). 3. Character 208: One copy of R1 tandem repeat Character 204: rDNA restriction site 45 absent present in D-loop region of mtDNA (0); or three (0); or present (1). 3. to nine copies of R1 repreat present (1). 91.

APPENDIX 3: DATA MATRIX The data matrix given below contains all of the ????? ???00 ????? 00{01}10 1000? 210?? ?1??1 data used for Analyses 1, 2, and 3. See appendix ????? ????? ????? ????? ????? ????? ????? ????1 2 for character descriptions. This data set is also 10??0 ?0000 ??110 2?10? 000{01}1 101-0 00000 available electronically via the World Wide Web 01??? ????? ????? ????? ????? ???20 00??? ???00 at ftp://ftp.amnh.org/pub/mammalogy. 12??? ??00? 00??0 ???01 10??? ????? ????? ????? ????? ????? ????? ??? Scandentia Palaeochiropteryx 00000 00000 000-0 00000 00000 00000 00?30 ????? ???00 ????0 00010 {12}0001 210?0 21??1 00000 00000 0--00 00000 00000 00000 000-0 ????? ????? ????? ????? ????? ????? ????? ???11 -0000 00000 00000 0-0-0 0-000 00000 -00--00000 11000 00000 10110 22100 00001 011-0 0?100 00000 000-? ---00 00020 --000 02{01}00 000?? 01??? ????? ????? ????? ????? ???10 00??? ???00 ???00 00-00 00001 01000 0{01}000 0000- 12??? ??011 001?0 ???01 10??? ????? ????? {01}?101 0???0 00001 0?001 ????? ????? ??? ????? ????? ????? ????? ??? Dermoptera Emballonuridae 000?0 00000 000-1 00101 ?1-00 00010 00000 01000 {01}{12}120 10{01}1{01} {01}{01}1-1 0???0 00000 0--11 00001 00010 10010 120-0 1000{01} 21000 201{02}1 1120{012} 00011 -0050 00000 00000 0-0-1 12100 00000 -00-- 0--01 00000 {01}010{01} {01}0010 110-0 00000 00011 010-? ---00 00001 00?00 00?00 -10{12}1 11000 00000 {01}01{01}0 2{12}101 011?? ???00 00-00 0?000 10010 00000 0010- {01}000{01} 00210 {01}{01}000 {01}{01}202 00000 -??00 01111 0?001 00000 01000 00? 01110 10{01}10 11{01}01 01200 0{01}111 Pteropodidae 10?10 00011 121?? ??010 10110 ???01 10110 00000 121{01}0 1{01}0-0 {012}{123}{01}10 00000 -0?01 3?102 1{01}100 01000 01000 100 31-00 {01}0{01}00 11000 0{01}000 {01}2010 Rhinopomatidae 10011 00001 10000 10000 000-0 -000{01} 00000 01100 01120 100-1 112-1 10011 21000 21111 0000{01} {01}010{01} 2{123}{01}00 {01}0000 10313 00010 10101 02000 00100 10101 210-1 -0{12}10 00000 0{01}00{12} {01}{01}100 12021 11000 01000 11110 11100 00000 -01-0 00000 00001 21200 0{01}{123}00 00100 00000 20000 01001 00111 01110 00101 01211 00110 02001 10010 10110 00{01}00 {01}0110 00000 00??? ???00 120?? ??010 10111 ???01 00110 -0101 30200 00100 00{01}00 00100 {02}00 10001 0?00? 3010? 1{01}001 00000 00000 00? Icaronycteris Craseonycteridae ????? ???00 0???? ?0000 000?? 1?0?0 ????1 011?? ???20 11101 112-1 11011 21001 21??1 ????? ????? ????? ????? ????? ????? ????? ???11 1???3 00010 0--01 02000 00110 10101 010-1 00000 00000 0-10? 12100 00000 -01-0 00000 12011 11101 01001 10100 21101 00001 101-0 010?? ????? ????? ????? ????? ???10 00?0? ?0?00 21000 002?? ????? ????? ????? ????? ???20 00??? 00??? ??011 001?0 ???00 00??? ????? ????? ???01 122?? ??110 10110 ???01 002?? 100?? ????? ????? ????? ????? ??? ????? ????? 11??? ????? ????? ??? Archaeonycteris Nycteridae ????? ???00 ????0 00010 00001 1?0?? ????1 01301 22121 000-1 002-1 1{01}001 2{01}111 ????? ????? ????? ????? ????? ????? ????? ???11 00021 11311 00001 0--01 01000 2010{01} 10000 100?? ?0000 ????0 11100 10?0? ?01-0 0?000 {01}10-1 00{01}01 11{01}00 10101 11110 01??? ????? ????? ????? ????? ???10 00??? ???00 23011 00001 001-0 21000 01212 01101 01110 02??? ??0?? ?11?0 ???00 00??? ????? ????? ????? 11101 11111 01110 00?11 00011 120?? ??001 ????? ????? ????? ??? 10110 ???13 10111 00000 -???? ????? 11000 Hassianycteris 11100 01100 010 1998 SIMMONS AND GEISLER: RELATIONSHIPS OF EOCENE BATS 181

Megadermatidae 00000 ?01?? ????? ????? ????? ????? ???20 11??? 012?0 01121 10101 21{12}-1 {01}0001 12}0110 ???01 101?? ??010 11110 ???00 1001? 00101 10111 10200 {01}{01}000 10101 03000 00??? ????? ????? ????? ????? ??? 2010{01} 10000 010-1 00001 11100 11101 Myzopodidae 11111 10101 00001 {12}0221 21000 01212 110?? ???10 01100 00020 20111 21000 10101 11101 01111 11111 11211 11211 00??? ???01 1???0 11010 11-01 13111 21100 11000 020-0 120?? ??{01}10 10210 10011 10111 10001 0?00? -0041 11000 00000 10110 22100 10101 20210 2110? 1{01}001 10001 10000 000 1?100 011?? ????? ????? ????? ????? ???20 00??? Rhinolophinae ???11 100?? ??011 10110 ???01 110?? 00001 00200 01121 000-1 111-0 10011 2{01}011 00101 0???? ????? ????? ????? ????? ??? 1{01}31{0123} 00000 10110 00000 20100 10000 Thyropteridae 0{12}0-1 01031 11102 11111 0-111 10101 10001 11011 01110 01100 00010 20111 21000 101?1 20221 21000 01211 10101 01111 11110 --210 1???0 11100 0--11 13111 21100 {01}?000 ?20-1 11221 00??? ???11 120?? ??000 00201 ??011 10211 11110 00001 11110 21111 01101 10200 10111 10001 0000? 30102 1{01}010 00100 00000 10101 10101 11010 01101 01200 00120 01000 000 00??? ???11 100?? ??011 10111 ???01 110?? Hipposiderinae 00011 01?10 21?1? ????? 00000 00000 00? 00200 01121 000-{01} 11{12}-1 {01}0011 Furipteridae 2{01}011 001{01}1 1131{0123} 00000 10110 110?1 12110 1010{01} 001-0 10111 21010 00000 20{01}00 10000 0{012}0-1 0{12}031 10101 1???0 11100 10011 11111 21100 11100 11112 11111 0-111 {12}0100 00001 {02}0221 000-1 10211 11100 00001 11110 23111 11101 21000 00211 ?0??? ???11 111?0 --2?? ?0221 111-0 20011 001?? ????? ????? ????? ????? ???20 00000 10011 12010 11{01}00 00201 10011 00??? ???11 110?? ??110 10110 ???01 100?? 1111{01} 10001 0???? ????2 10010 01000 01000 00011 0???? ????? 10??? 00000 10000 00? 000 Natalidae Phyllostomidae 110?0 12110 01100 00000 10011 21000 00101 0120{01} {01}{01}{01}10 010-{01} 1???0 11100 10011 11111 21100 01100 020-1 {01}{123}{012}{12}{01} {123}0011 00011 11100 00000 11101 20111 01001 111-0 {12}{01}{01}{01}0 21001 11000 20111 01102 10101 11010 00001 21100 10120 {12}{12}10{01} 0--11 0{013}000 00?00 01011 110?? ??110 10110 ???01 10010 {12}0{01}{01}0 {01}00{01}0 1{02}0-{01} 00001 0???? ????? 10??? 00000 10000 20? 0{01}011 11000 0000{01} {01}010{01} Antrozoidae {12}{01}{01}0{01} 00001 {12}1{12}00 011?? ???10 10110 1{01}2-1 20011 21000 20101 1{01}000 1{01}2{01}{12} {01}{01}1{01}1 1???0 00010 10111 00000 01110 11000 12111 0{01}010 {01}{01}101 11{02}0{01} 12011 11000 00001 0-100 12100 00101 211-0 1{01}{23}20 00111 00101 1001{01} 10010 00100 00102 ????? ???10 00001 10100 10320 10110 101{01}{01} {01}000- 01111 111{01}1 11??? ???01 110?? ??011 10110 ???01 10110 12212 10100 00000 00101 000 00001 0???? ????? ????? ????? ????? ??? Mormoopidae Tomopeatinae {01}100{01} {01}{012}{01}10 010-0 011-0 010?? ???10 10110 112-1 20011 21000 201?1 10011 2{01}011 01001 1???2 11100 0--11 00000 1???0 12100 10011 00000 21110 00000 02110 11100 10010 1{02}0-1 0{01}011 11{01}00 -2011 11100 00000 0-100 12000 00111 211-0 00000 11110 2{12}101 00101 1{01}210 11000 00112 011?? ????? ????? ????? ????? ???21 11??? 10202 11111 01010 11101 11201 10121 ???01 110?? ??011 11110 ???00 101?? 000?? {01}0011 01101 10010 00110 10110 ???01 ????? ????? ????? ????? ????? ??? 100{01}? 01111 11??? ????? 10??? 00011 00010 Molossinae 000 01011 {01}{12}110 {01}{01}{01}1{01} Noctilionidae 1{012}{12}-1 {12}0011 2{01}000 201{01}1 01001 12110 010-0 022-1 20011 20011 210?1 11000 1210{01} ???11 00000 21110 00000 11000 11101 0--11 03000 10100 10010 100-1 02110 -2011 11100 00000 {01}0{01}0{01} 02011 11000 00000 11100 22100 00100 -1200 {12}2000 0{01}111 2{01}2{01}0 001{01}2 10000 00102 11101 00010 11101 11200 11111 10102 01111 01010 01001 00100 10421 11110 00?01 11101 122?? ??010 10110 ?0?01 1010- 01001 11111 01011 11110 01100 101{01}0 01111 01101 12202 10100 00011 00011 00? 00001 00000 31112 10100 00000 01000 210 Mystacinidae Vespertilioninae 01001 22?10 010-0 121-1 20011 11001 01101 01{02}10 12110 10110 {01}0{12}-{01} 1???0 11101 10011 03000 11110 10100 020-1 {12}0011 21000 {12}0101 11{01}10 11100 01011 11000 00000 0-0-0 12101 00101 211-0 11-11 01000 11110 11000 12100 -2011 11000 182 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 235

0000{012} 101{01}0 {12}{12}1{01}{01} 00101 0010{01} 101{01}{01} {01}0010 01001 00100 21{12}10 0010{012} {012}010{12} 10110 10{23}20 0110? 0??01 11010 10011 10110 10{12}10 0100{01} 10100 10320 11100 00101 01101 10110 00001 00000 21112 1???? 00000 1{12}010 10011 10110 ???01 101{01}{01} 00000 001 00001 00000 2111? 10100 ????? ????? ??1 Murininae Minopterinae 010?? ???10 10110 001-1 10011 21000 201?1 01010 00010 10110 001-0 10011 21000 20101 1???0 11101 0--11 01000 21110 01010 120-1 11010 11101 11-11 01000 11110 01000 10100 02011 11000 00002 10110 21111 00001 20220 20100 011?? ????? ????? ????? ????? ???20 01??? -2011 11000 00001 11110 23111 01101 211-0 ???01 110?? ??011 10110 ???01 10110 00001 00102 101?? ????? ????? ????? ????? ???21 11??? 0???? ????? 1???? ????? ????? ??0 ???01 112?? ??011 10110 ???01 1000- 00001 Kerivoulinae 0???? ???1? 10100 ????? ????? ??0 110?? ???10 10110 000{12}0 20011 20000 201?1 Myotinae 11??0 11101 0--11 00000 11111 01000 12100 01010 12110 10110 00{01}2{01} {12}0011 -2211 11000 00002 10110 21111 01101 21210 21000 10101 1???0 11101 00100 011?? ????? ????? ????? ????? ???20 01??? 11-11 01000 11110 01000 12100 -2011 11000 ???01 110?? ??011 10110 ???01 10210 00001 0000{01} 10110 21111 00101 21210 00101 0???? ????? 10100 ????? ????? ??0