A behaviourai investigation of the sticking mechanisms and non-aerial locomotion of

Spix's disk-winged , Thyroptera tricolor (Microchiroptera: Thyropteridae)

Daniel Ke~ethRiskin, BSc-

A thesis submitted to the Faculty of Graduate Studies in partiai fulfiment of the

requirements for the degree of

Master of Science

Graduate Programme in Biology

York University

Toronto, Ontario

August 2000 Bibliothèque nationale du Canada Acquisitions and Acquisitions et Bibliographie Services services bibliographiques 395 Wellington Street 395. rue Wellington Ottawa ON K1A ON4 Ottawa ON KI A ON4 Canada Canada

The author has granted a non- L'auteur a accordé une licence non exclusive licence allowing the exclusive permettant à la National Library of Canada to Bibliothèque nationale du Canada de reproduce, loan, distnbute or sell reproduire, prêter, distribuer ou copies of this thesis in microfom, vendre des copies de cette thèse sous paper or electronic formats. la fome de microfichelnlm, de reproduction sur papier ou sur format électronique.

The author retallis ownership of the L'auteur conserve la propriété du copyright in this thesis. Neither the droit d'auteur qui protège cette thèse. thesis nor substantial extracts fiom it Ni la thèse ni des extraits substantiels may be printed or otherwise de celle-ci ne doivent être imprimés reproduced without the author's ou autrement reproduits sans son permission. autorisation. A BEHAVIOURAL INVESTIGATION OF THE STICKING MECHANISMS AND NON-AERIAL LOCOMOTION OF SPIX'S DISK-WINGED BAT, ïlyroptera tricolc (MICROCHIROPTERA: THYROPTERIDAE)

by DANIEL KENNETH RISKIN

a thesis submitted to the Facuky of Graduate Studies of York University in partial fulfillment of the requirernents for the degree of Master of Science

O Permission has been granted to the LIBRARY OF YORK UNI- VERSITY to lend or seIl copies of this thesis, to the NATIONAL LIBRARY OF CANADA to microfilm this thesis and to (end or seIl copies of the film, and to UNIVERSITY MICROFILMS to publish an abstract of this thesis. The author reserves other publication rights, and neither the thesis nor extensive extracts from it rnay be printed or other- wise reproduced without the author's written permission. Abstract

Roosting, Spix's disk-winged , Tizyroprera tricolor (Microchiroptera:

Thyropteridae), use the disks on their wrists and ankles to cling to smooth leaves. To determine the mechanisms of sticking, in 123 trials 1 tested the adhering abiiity of 31 T. tricolor on four surfaces, medium egpade sandpaper, lexan polycarbonate, soiid aIurninum sheet, and perforated aluminum sheet. WhiIe dl bats readily adhered to the Iexan polycarbonate and the solid aluminum sheet, they did not stick to perforated aluminum (a surface which makes suction impossible) supporting the hypothesis from previous anatomid studies, that the disks work by suction. The abfity of 54.8% of T. nicolor (n

= 3 1) to stick to the perforated surface at overhanging angles, where friction is eliminated, suggests that in the absence of suction, T. tricolor use wet adhesion as an additional sticking mechanism. Wet adhesion may allow bats to conserve energy while roosting by allowing partial or complete relaxation of the musculature responsible for suction. For cornpcuison to T. tricolor, in 461 trials 1 tested the sticking ability of 121 bats

(18 species; 3 families) without disks on the same four surfaces. T. îri~~loradhered better to smooth surfaces than any other species, but not as well to rough ones. This may represent an evolutionary tradeoff: in adapting to smooth roost surfaces, T. tricolor appear to have suffered a reduction in their ability to use their cIaws for clinging to rough surfaces. T. zricolor retain some ability to cling to rough surfaces, as indicated by the ability of one individual to climb on screen by interloclcing the claws of the thumbs with the surface. 1 observed the non-aerial locomotion of 21 T. tricolor inside a transparent

enclosure of lexan polycarbonate. Oa. horizontai surfaces, 7'. tricolor used two gaits, a crawling one, most sUnilar to that used by vespertilionid bats, and a leaping one, not previously described for bats. On vertical wds, bats used a crawling gait, but in 9û0 corners between two vertical walls they used crawling and leaping gaits. Descent was achieved by a reverse crawfing gait, in wnich the head was kept above the feet, or by droppuig some distance before the bat attached the disks to a wall to stop itseif. Acknowledgements

X am fortmate-to- have- extreemef y supportive-friencts, df of wtrom- deserve-thanks, not only for kgeping me on task while 1 worW on ~s thesis, but dso for providing swh fine distractions. I am matgratefd to Enrico Bernard, Sylvie Bauchard, Laurel

Chaabuxnb cuhaa,-ECIamiaaae gehe~eznaeker~~h.a.L)ec~~zaas

Dume, GEEm md Nigel hyes, Daniela EbmMdini, John Morgan Ratcfiffe, Tobin S.

Rooney, K. Lorraine Staoàing, Greg Tabor, Jason R. Taylor, Thanh Van, Maarten

Vonhof, and Joanna Zigouns. John J. CIare, a fine bidogist but far greater frhd, tafked me into applying for a postgraduate schofarship in the first place, and also made

me throughout my thesis, keeping me on track when my mind wandered, w6iIe my sister,

Raque1 E, Riskin, dso provided trernendous support. 1 am proud of bath of thern, and thaak them ImmenseEy for hdpiag me.-

Mydosest friend, anct gatest source of stren-crtb[ has hem Caitlin Jane Berry.

Despite having to move across the country to Eve in an apartment with no running water, despite being left behind for months cm end while 1 traveled tu work on bats, and despite being awoken at3 AM when I'd brought anraisy bat home even tbough I'd said I

this degree. She makes me a beffer person, and I am gateful ta have her in my He. vi My passion for bats was born in the tenth grade, when 1 read a book by M. Brock

Fenton, and working with him over the past two years has been a privilege of the highest order. He has led by example, exhibithg an enthusiasm and work ethic which combined are the mark of a man who is doing what he loves. Working in the field with Brock has been a most fuffïg experience, His assistance as 1worked tu completion on this project has been immeasurable. 1 am also grateful to other faculty of York University who have provided many constructive comments which have si~~tlyimproved the quality of this thesis, especially Dawn Bazely, Peter Keir, Bernie Ligbnan, Barry Laughton, Don

McQueen, Bridget Stutchbury, a.Norman Yan-

Funding for this project was provided by a Natural Sciences and Enginee~g

Research Council of Canada (NSERC) postboraduate scholarship to myseif, and an

NSERC research &orantto M. B. Fenton.

Finally, 1 thank Mary W. Walters, not only for looking over dozens of thesis clraft', but for inspiring me to do things properly. She has helped me more than any other person in the completion of this degree, urrselfrshlgr giving the, advice, and support. I admire and am ,oratefuf fur her, so I dedicate this thesis to her. Thanks mom.

vii Table of Contents

Abstrac t

Acknow Cedgements

Table of ContentS

List of Tables

List of Figures

Introduction

Materials and meehods

Shidy areas and anllnals

Rotating plafform td te.e-ti6n 6.f

~bservaticmsof locomotion

Rësuits

Rotating platform

Mechariisnils of stickia'g

Locomotion on a horizontal surface

Locomotion on a vertical surface

Use-ofhirrnb daws

Discussion

Mechanims of sticking Roosting posture 20

Roosting ecology 21

Non-aerial locomotion 23

Evohtion of disks 26

Conclusioos 30

References 31

Appendix 1. Last angle of stick for bats on each of four surfaces tbat 50

were rotated from horizontal with the bat on top (O0), through

vertical (W),to horizontal with the bat below (180") at a

velocity of 3's-'. The order in which the surfaces were presented

to each bat are ais0 shown. Emb. = Ernballonuridae, Phy. =

Phyllostomidae, Thy. = Thyropteridae, Ves. = Vespertilionidae. List of Tables

Page

Table 1. Species used in rotating platform experiment. 37

Table 2. Kniskai Wdis tests for differences in stickùig ability among 38 bats grouped by family on each of the four surfaces.

Table 3. Q statistics obtained from multiple cornparisons (two-tailed) of 39 sticking abilities among groups on each of the four surfaces (Ernb. =

Emballonuridae; Phy. = PhyUostomidae; Thy. = Thyropteridae; Ves. =

Vespertilionidae). L'kt oPF~s

Page

Fi9re 1: Thumb disk of Thyroptera ~icolorviewed through a vertical 40 glass surface to which it is attached. Photopph by M. B. Fenton.

Fi me2: The four test surfaces used in the rotathg plaîfonn experiments: sandpaper (a), lexan polycarbonate (b), duminum sheet (c), and perforated duminum sheet (d).

Fipre 3: Percentage of individuals in each family of bats tested that 42 had not fallen from a roiating sandpaper surface (velocity = 3's-'), with changing angle of the surface. 0" is horizontal with a bat on top, 90" is vertical, and 180" is horizontal with the bat below-

Fipure 4: Percentage of individuals in each family of bats tested that 43 had not failen from a rotating aluminum sheet surface (velocity = 3's-'), with changing angle of the surface. 0" is horizontal with a bat on top, 90" is vertical, and 180" is horizontal with the bat below.

Figure 5: Percentage of individuals in each faniily of bats tested that 44 had not failen from a rotating perforated aluminum surface

(velocity = 3's-'1, with changing angle of the surface. 0' is horizontal xi with a bat on top, 90" is vertical, and 180" is horizontal with the bat below. Note that 54.8% of T-iricoZor hold the sudace past 90' (purple he), indicating that those bats used wet adhesion.

Fimire 6: Percentage of individuals in each family of bats tested that 45 had not fallen from a rotating lexan polycarbonate sufice (velocity = 3's-'), with changing angle of the surface:ü' is horizontal with a bat on top, 90" is vertical, and 280" is horizontal with the bat below.

Fime 7: Adult male T-îricolor (foream, 35.1 mm; rnass, 3.7 g) using 46 the crawling apït on a horizontal surface. The camera faces the bat's dorsum from above. Images are separated by 1/30 s intervals, crosshairs measure LU mm by 10 mm. The bat traveled at a net velociîy of 0.082 dsin thk.-

Ftoure 8: Addt male tricolor (forearm, 34.8 mm;mas, 4-0 g) using 47

-thekaping -@t-on a .horizontrd-surface. The-canera faees -thebat's teft side and the bat's antero-posterior axis is horizontal. Images are separated by 1230 s hitervals, crosshairs masure 10 mm by 10 mm. The bat atveled at a net veiocity of 0.284 dsin this sequence.

Fimw 9: Adult female 71 tricolor (foreann, 34.5 mm;mas, 3.6 g)using 48 xii

Bats are most often differeniiated from other by their ability to fly, so the aeriai locomotion of bats and the associated matornical features have received considerable attention (Arita and Fenton, 1997; F'idley et aï., 1972; Norberg, 1%9;

Stnihsaker, 1%1; Vaughan and Bateman, 1970). When roosting, however, al1 bats face situations that require the ability to cling to, and move on, a substrate. This means that the matornical specializations allowing flight must also accommodate non-aeM locomotion and roosting. Despite the overall similarity in body design across bat clades, bats have diversified to make use of a staggering array of roost types (Foster, 1992; Riskin and

Pybus, 1998; Simmons and Voss, 1998).

Disk-like structures appear on the wnsts and ankles of bats in two families: the

Myzopodidae, encompassing only Myzopodu aunta Milne-Edwards and Grandidier 18% of Madagascar, and Thyropteridae, containing Thyroptera nicolor Spix 1823, T. discifera

(Lichtenstein and Peters 1855), and T. lavali Pine 1993, from the Neotropics. Their disks are presumed to give these the ability to grip on smooth surfaces such as the waxy cuticles of furled leaves (Altringham, 19%; Fenton, 1992). The most studied of these species is T. iricolor, Spix's disk-winged bat, This ca 3 -5 g species inhabits lowland forests from Veracruz, México to tropical South America (Wilson and Findley, lm- Disk-winged baîs are only known to roost in foliage. Thyroprera spp. most often have been found roosting diumaliy in the furled leaves of (Musaceae) and

CaIathea (Marantaceae). These leaves are elliptical in shape at maturity, but during their 1 development are rolIed and tube-like for ca. 24 hours, providing roosts for thyropterid

bats (Findley and Wilson, 1974). Most authors presume that M. roost on leaves

(Brosset, 1966), but little is known about which roosts are actually used by this species

(Schliemann, 1974), although Findey and Wilson (1974) cited a personal communication describing one M. auria found in the Malagasy traveler's palm, Ravenala

For a bat to remain stationary on a smooth non-horizontal surface, such as the vertical face of a Heliconia leaf, the must resist slipping or falling by appIying forces that act in static opposition to gravity. In auimals, six mechanisrns for achieving this have been reported, namely intermolecular adhesion, gIuing (Nachtigail, 1974), suction (Smith, 199 la), wet adhesion, interlocking, and friction (Emerson and Diehl,

1980).

lntermolecular adhesion (sometimes called dry adhesion) describes the attraction betweea the closely contiguous surfaces of adjacent materials, and requires a very small distance of separation (Nachtigall, 1974). When two materials are brought into contact with each other, the surface molecules interact, gïving rise to attractive forces that may be physical, chernical or electrostatic (corresponding to adsorption, covalent bonding or van der Waals forces, respectively). The feet of geckos, for exarnple, are made up of billions of setae that attach to smooth substrates by means of intermolecular van der Waals forces

(Autumn et al., 20).

Gluing combines intermolecular adhesion (as defined above) and cohesion (the intermolecular forces among identicai molecules). When a cernent is spread between an 2 animal and a substrate, adhesion between the liquid and the two surfaces beb& immediately. After the cernent dxies, the cohesiveness of the intervening solid holds the surfaces together in conjunction with adhesion at the solid-to-surface interfaces

(Nachtigali, 1974). Gluing is often used in situations where an organism will remain in contact with the substrate indefinitely, as in barnacles which attach to rocks (Edlund and

Koehl, 1998). However, gluing is also used by some animais which will later detach, Iike limpets (Moilusca) (Smith, 1991b), and some rotifers (Rotifera) (Nachtigall, 1974). In s-sh (Echinodermata), the bulk of the sticking ability is due to a sticky mucus secreted by specialized cells of the tube feet. For detachment, dûferent cells, also of the tube feef secrete a second substance which chemically destmys the first (Flammang et al, 1998).

Wet adhesion occurs when two solids are beld together by an intervening layer of liquid, and there are two types, Stefan adhesion and capillarity. Stefan adhesion occurs when the liquid is present both at the surface interface and around it- The two adhering surfaces are so close that the viscosity of the intervening liquid resists the flow that must accompany the sepration of the surfaces (Smith, 1991b). Capillarity occurs when the intervening Iiquid is not present outside the joint. The force in capillarity results from the surface tension of the liquid. Capillarity at the toe pads and the belly of tree frogs

(Chordata) holds them in place on the leaves to which they cling (Emerson and Diehi,

1980). Both types of wet adhesion are positively correlated to the surface area of contact between the two surfaces (Emerson and Diehl, 1980). They require no muscular exertion, and function even in holding wet paper to glass (Hanna and Barnes, 1991). Suction occurs when an animal creates a partial vacuum over some area of the substrate/body interface (Emerson and Diehi, 1980). It is limited by the magnitude of the air pressure differential produced, and is dependent on both a non-porous surface (Smith,

1991a), and an unintermpted seal between the sucking organ and surface (Emerson and

Diehl, 19233. Sucking devices are known from several animal taxa. Aduft tapeworms

(Platyhelminthes), for example, hord themselves in place on the intestinal wdof the host animal by means of suckers, hooks and other adhesive organs whiich make up the scolex.

Parasitic leeches (A~elida)attach themselves to the skin of their hosts with two suckers, one anterior and one posterior, to reniain in place while feeding on blood. Perhaps the animals best known for their use of suction are the cephalopods (Moliusca). Octopus suckers are sessile, musculâr cups which create a pressure differentid by movement of muscles contained within the cup merand Smith 1990). The stalked suckers of decapoàs (cuttlefish and squid), on the other hand, each consist of a muscular piston which reduces the pressure under the sucker by pulling a stalk which connects to the sucker (Smith, 19%).

Interlocking, the intermesbing of projections from two solid surfaces, is limited by the roughness of the interacting solids. For example, Hooke (1665-7)and

Leeuwenhoek (1690) postulated that the adhesive setae of insects could catch on the minute irregularities of apparently smooth surfaces. This theory was dismissed when scanning electron micrograph (SEM) views of glass and Perspex reveded them to be rernarkably smooth surfaces, even at the scale of insect setae (Stork, 198û).The supplied force in interlockhg results from the resistance of the substrate to compression and its 4 ability to resist breaking (Emerson and Diehl, 1980). Interlocking differs hmfriction which is the potential energy analogue to interlocking, and oniy occurs in the presence of a normal force. The greatest functional difference between friction and interlocking is that friction will dow an animal to adhere to an inclined surface, but not a vertical or overhanging one. Using interlocking, animals can adhere to surfaces at aU angles.

Friction and interlocking are both independent of the surface area of contact (Emerson and Diehl, 1980).

Unless other, yet unknown mechanisms of sticking exist, disk-winged bats mut adhere to leaves by means of one or some of these six mechanisms. AIthough we know T. tricolor cling easily to smooîh surfaces (Findley and Wilson, 1974; W~msattand Vilia R.,

1970), Our knowledge of how attachment is maintaineci is inferred only from anatomical investigations. Observations of living anïmds, which have been essentid to our understanding of this ability in other taxa (Emerson and DiehI, lm;Hama and Barnes,

1991; Smith, 1991b), are lacking for mammals altogether (Dobson, 1876; Nachtigall,

1974).

The thumb disk of T. tricolor consists of a hairless, nearl y circular pad ca. 3.5 mm wide (Wmsatt and Villa R., 1970). The face of the pad (Figure l),which is directed ventrally, is divided into four concentric zones. The outer zone is a smooth raiseci edge divided into 6û to 80 compartments. Sudoriparous (sweat) glands open into each cornpartment (Schiiemann, 1974), and secrefe a substance which is non-mucoid and of low viscosity, consisting mostly of water and soluble proteins (Wimsatt and Villa R.,

19'70). Adjacent to the outermost zone is one of irreedar elevations and crevasses. Third 5 is a broad zone of radial grooves, the surface of which slopes toward the centre. The

innermost zone is a network of asymmetric gooves (Thewissen and Etnier, 1995).

Wimsatt and Via R. (1970) described the shape of a T. ~icolordisk as a cup with a

deeper imer cup, and noted that when a bat placed its disk on a transparent surface, the

larger of the two cups was flattened and eliminated, leaving only the smaller inner

concavity. The disks of the hind limbs are centered over the metatarsophalangeal joints of

anchylosed digits III and IV. They are slightly smaller than the thumb disks, measuring

ca 2 mm in diameter (Schliemann, 1974). The density of sweat glands is close to that of

the thumb di&, and the absolute number of glands is correspondingly lower (Wimsatt

and Villa R., 1970). Grooming T. tricolor spend considerable time licking the disks of

both the thumbs and the feet (Carvalho, 1939; Fmdley and Wilson, 1974; Wimsatt and

Villa R., 1970), keeping them free of debris. The combination of saliva and fluid from the

sudoriparous glands keep the disk face moist at al1 times mmsatt and Villa R., 1970),

producing a fluid-filled joint between the disk and any surface it contacts.

Since the species was described in 1823, many researchers have focused on the

disks of 7'. trr*color.Espada (1870)first noted that their disks, when adhering to the skin of the observer, produced a sensation similar to that experienced when the air was drawn out of a thimbie with the mouth, and the tongue was placed over the opening. He stated . - that the suctonal ability of the disks was achieved by $he exertion of musculature intrinsic to the disks. Dobson (1876), using microscopy, discovered that no musculature existed in the disks. He proposed that bats were therefore unable to actively change their disk shape, so evacuation of air imm beneath the disk was achieved when the body of the bat pressed

the disk into position on a surface.

In the past century, researchers have continued to infer suction as the mechanism

of adhesion in T. h-icolor (Schliemann, 1970a; Schliemann, 1970b; Schliemann, 197q

Thewissen and Etnier, 1995; Wmsatt and Villa R., 1970). Beneath eacb thumb disk,

microanatomicd investigations have reveded a cartilaginous plate. The plate is thin

centrally and thickens radially, eventually splitting into two thin cartilaginous an&r

collagenous laminae. Wïmsatt and Villa R. ( 1970) stated that these laminae recombined

at the disk periphery, but Thewissen and Etnier (1995) described the laminae as separate

at the disk edge. Each plate is surrounded by adipose and comective tissues, and at its

centre is attached via tendon to the flexor pollicis brevis profundus muscle

(Schliemann,l974), w hich onDates at the metacarpophalangeai joint of the thumb

(Thewissen and Etnier, 1995). Like the suckers of decapods, this arrangement wouid

permit the manipulation of disk shape without intrinsic musculature. The intemal anatomy of a hind disk is similar to that of a thumb disk, but the cartilaginous plate is

more robust in the hind disk. Two spurs extend from the lower surface of the plate toward the metatarsophalangeal joints III and IV. These spurs each end in a groove whic5

accommodates the tendons of the plantaris muscle and possibly the flexor digitorum fibutaris (Wimsatt and Viila R., 1970). Muscular control of the hind disks is hence also from without, The morphology of T. tricolor disks led Wimsatt and VUa R. (1970) to conclude that their functional mode is suctonal, rather than any other sticking mechanism. 7 The morphology of the wrists and feet, the main points of contact during chkopteran locomotion, is vastly different in disk-winged bats than in diskless bats.

Presumably, this @ves T. M~~iurthe ability to climb on smooth surfaces at angles where most bats carmot, Despite the diversity of bats, with the number of species exceeding

900, non-aerial locomotion has been documented for less than 15 species, none of which are disk-bearing bats (Altenbach, 1979; Dietz, 1973; Vaughan, 1959)-Fiidley and

Wdson (1974) noted bnefly that 7'. micolor climbed easily on glass, but did not describe how the bats moved. The non-aerial locomotion of disk-winged bats on smooth surfaces is an important component of terrestrial locomotion in rnammals which has yet to be descnbed.

Evolutionary biology examines the relationship between form and function. In the case of T. tricoior disks, the functiond mechanism has been inferred from anatomical studies, with limited mecdota1 observations of live animals, By investigating how Live animals use the disks, I tested the accuracy of those inferences. The primary purpose of this study was to investigate the ability of T. tncolor to adhere to different sufaces and thus to determine the underlying mechanisms, while examining their impact on roosting posture, roost selection, and locomotion. Using behavioural data from live bats, I tested the hypotheses that 1) suction is the sole mechanism of sticking (Wimsatt and Villa R.,

IWO), and 2) T. tricolor is specialized for clinging to smooth surfaces, and unable to -gip roua ones (Findley and Wilson, 1974). Materials and methods

Stuay Wear and animals

From 6 May to 12 June 1999,I conducteci field work at the Cano Palma Research

Station near Tortuguero, Costa Rica (10" 35' N, 83" 32' W), where 1 caught most bats after dusk using mist nets. I caught Rhynchonycteris mu(Emballonuildae) with haad nets at a day roost, in the boat house of the station. Also during daylight hours, I searched the forest for plants of any species that exhibited the Young, rolled leaves commonly used by T. tricolor as day roosts. After experimentation, I returned ail T. tricolor to their roosts of capture, and released other bats from the research station-

On the nights of 27 and 28 August 1999,I caught bats with a harp trap (Tuttle,

1974) near Perth Road Village, Ontario, Canada (44' 35' N, '76"19' W), Each night 1 left the trap at the entrance to a mine before sunset, and removed bats from it after midnight. I then brought the bats back to nearby Queens University Biological Station (QUBS) for experimentation and release.

After weighing, rneasuring the forearm of, and sexing each bat, I used the descriptions and keys of Reid (1997) and Van Zyll de Jong (1985) to ciassify captured bats to species. 1 aged bats by looking for epiphyseal closure of the digits, which is seen only in adults (Anthony, 1988). 1 also identified subadult T, ~icolorby their dark brown immature pelage, which is distinguishable from the white ventral fur of adults (Wilson and Findey, 197'7). 1 clipped a patch of hair from every bat before release so that it would be recognized if recaptured. For most species 1 took the hair from the top of the head, but in Micronycterk spp. (PhyiLostomidae), 1 clipped hair from the back to avoid damaging the interaurai band. 1 released recaptured individu& without experimentation.

Rotating platform

To quantify the sticking ability of a bat on a surface, L used the protocol of

Emerson and Diehl (1980). 1 placed each bat on a surface which 1 robted with an electronic motor from O' (horizontal) through 90" (vertical) to 180' at a velocity of 3' s-'.

A scale divided into one-degree increments revealed the angle of the platform at al1 times.

During rotation, the bat could move about on the surface to attain whatever orientation it preferred. A transparent polycarbonate cage was attached to the surface and remained stationary relative to it during rotation. This prevented bats from escaping during experimentation. Since T. îricolor climbed easily on the polycarbonate cage surface, 1 prevented their escape with the methodology used in the rotation of tree frogs

(Hanna and Barnes, 1991), by holding open han& about two inches away from the rotating T. tncolor. This was not aiways entïrely effective, so 1 conducted the rotations inside an enclosed tent. When bats did escape from the surface 1 could quickly recapture them with a hand net.

In some trials, bats fell from the surface during rotation; in others, bats still heId after their rotation was stopped at 180'. In dl cases, 1 recorded the last angle at which a bat maintained its hold on the rotating platform. When bats fell, they did not make any movements of the wings until they were falling or sliding off the surface. However, during some trials, a bat began flapping its wings before it feu, and launched itself off the 10 surface. This often occurred in the fmt few degrees of rotation, when the surface was nearly horizontal, Mena bat of any species jumped from the surface prematurely, or used the contours of the cage for gripping, 1 stopped the trial, placed the bat at the centre of the surface, and repeated the trial. The values obtained for the points at which bats feu are therefore accurate indications of when they lost their grïps. Some bats, after severai attempts, still would not cling to a surface long enough to allow an accurate masurement of when they Iost gip, so 1 could not record data for ail bats on al1 surfaces.

1 rotated each bat on four surfaces: 1) aluminum oxide medium-,-de sandpaper

(emery cloth, ca. 120 diamond grid); 2) smooth transparent lexan polycarbonate; 3) duminum ucility sheet; 4) alunulflum sheet perforated by small (diameter = ca. 1.48 X m) holes evenly distributed across the surface with a frequency of 1.78 X IO4 holes m-*.

These holes permitted air flow from one face of the aluminum to the other, and reduced the surface area of the duniinum by 26% compared to the solid utility sheet. The size and frequency of the holes was such that hind or ttiumb disks at any position on the surface would completely cover at least three holes. These four surfaces were each presented once to every bat in a random order. SEM views of each surface dernonstrate the texture of the surfaces (Figure 2) at the scale of a T. iricolor disk (Figure 1).

Experimenral derection of sticking mechanism

In invertebrates which use gluing, such as limpets, the drying of the cernent takes several hours (Smith, 1991b): Because the entire rotation through 180" of a bat took only

60 s, the secretion and subsequent hardening of a glue beneath the disks could not occur during a trial. Similarly, the volume occupied by the liquid at the disk-surface interface 11 prevents intermolecular adhesion, which functions only at separation distances of ca. 03 nm (Autumn et al., 20).Neither gluing nor intermolecular adhesion was possible for T.

îricolor on any surface in this experiment, so 1 eliminated them from anaiyses. Friction only occurred at angles less than W,so the mechanisms available to bats at angles of 90" and greater were limited to interlocking, wet adhesion, and suction. SEM inspection of the sheet aluminirm and lexan polycarbonate surfaces 1 used reveded that they are srnooth even well below the scde of a T. tncolor disk (FiDpres2b, d), so interlocking is not possible on them. On these two surfaces, bats could only possibly adhere by wet adhesion, suction, and, at angles less than 90°, friction.

To determine whether suction is used by living T. tricolor, I compared the sticking abilities of T. tricolor on sheet aluminum, where suction was possible, to those on perforated aluminum, where the seal necessary for suction could not be created. If bats used only wet adhesion, and not suction, this decreased surface area would result in a lower stickin, ability, but T. tricolor cm bang beneath a smooth surface by only one disk

(Wimsatt and Villa R., IWO), indicating that a decrease in disk surface area of ca. 88%

(the result of removing al1 but one disk from a surface) does not induce failing. The decrease in surface area from the non-porous to porous surface is ca. 26%. Therefore if suction is not used, there should be no difference in sticking ability. Any observed decrease in sticking ability on the porous surface compared with the solid one cm be attributed to the use of suctian. Observations of locomotion

Of the T. tncolor used for rotation experiments, 21 were also used for video analysis of non-aerial locomotion. 1 placed the bats individually in a lexan polycarbonate cage 18 cm x 23 cm x 5 cm, dimensions that provided bats with enough space to crawl about, but not enough to fly. 1 adjusted the orientation of the cage so that bats could be observed moving vertically or horizontaily on the cage wdsfor different trials. 1 induced bat locomotion by touching or blowing on the bats, and recorded them at 30 frames s-l using a Cannon ES2500 HI8 video camera, I uploaded the video footage to a Macintosh

G3 computer for frame-by-frame analysis.

Resuits

Rotating pl/rtform

In 5û4 trials, I put 152 bats from four families and 19 species through the rotating surface experiment (Table 1). On the rotating platform, the postures of bats varied among families, but I observed no intrafamilial variation. At angles approximating W, vespertilionids, phyllostomids and ernballonurids rnaintained a heads-down orientation, but thyropterids were heads-up. Vespertilionids and phyllostomids held only with their hind feet at angles close to lm0,while emballonurids and thyropterids maintained four- point contact.

The angles at which bats, grouped by family, stuck to sandpaper, sheet aluminum, perforated aluminum, and lexan are demonstrated by figures 3,4,5, and 6, respectively.

On all four figures, the performance of thyropterids can be readily distinguished from those of other families. Because trials were artifïcially stopped at 180', there is a.upper 13 limit beyond which bats cannot continue to chg, so the data are not nonnally disûibured.

It was therefore uecessary to employ non-parametric statisticai tests. I performed a

Kniskal Wallis test with tied ranks (Zar, 1999) to determine whether interfamilial differences in sticking ability existed. On every surface, there were siwcant differences in sticking ability arnong the families (P < 0.0005; Table 2). Knowing that differences existed among fades,1 set out to determine where those differences existed: which families differed from each other on a given surface. Testïng each surface separately, 1 made multiple two-tded cornparisons using rank sums (Table 3), which pedtted analysis despite unequal group sizes and tied ranks (Zar, 1999).

On sandpaper, embalIonuids held on better than bats in any other fady (P <

O.OS), while thyropterids feu off sooner than other bats (P < 0.001). 1 detected no difference between the sticking abilities of phyUostomids and vespertilionids on sandpaper. Thyropterids rnaintained a grip through more angles on the sheet aluminum (P

< 0.05) and lexan polycarbonate (P < 0.001). No other families differed in sticking ability on either of those surfaces. On perforated alurninum, thyropterids fell sooner than other bats (P < O.ûûl), which did not differ from one another in their sticking abilities.

Mechanisms of sticking

Evidence that suction was involveci in the adhesion of T. îricolor is provided by a cornparison of these bats' performances on solid and perforated aluminum surfaces. On the former, 93.5% of T. tricalarheld through 1W (n = 3 1; Figure 4), while on the perf'orated surface, T. tricolor adhered only to an average of 90.0" (n = 3 1 S.D. = 25-02; figure 5); no bats held on past 121". 14 On the perforateil sheets, in the absence of suction, 54.8% of T. tricolor (n=3 1)

held on pst90" (Figure 5), indicating the use of interlocking and/or wet adhesion.

Because SEM views of the surface at the scde of a disk revealed very few irreedarities

that might interlock with the disk face (Figure 2c), and since the disks of living T. iricolar

are kept moist (Wimsatt and Villa R., LWO), 1 conclude that wet adhesion was the ody

mechanism of sticking possible on that surface at angles 2 90".

Locomotion on a horizonral surfme

When at rest on a horizontal surface, a bat rested its body on the substrate, with its

head elevated. A position that could facilitate echolocation. The folded wings lay upon

the surface, with the thumb disk dso making contact. The hind disks were planted on the

surface, the legs folded with the knee raised above the bat. In this position, the distal

segment of the leg (tibia and fibula) was held approximately 6û0from horizontal.

On horizontal surfaces, T. tricolor usually moved in short bursts, using one of two

distinct gait cycles, crawling (cf: Vaughan, 1959) and leaping, although some bats

alternated between the two gaits in mid-stride. Crawhg T. tn'color raised the anterior part of the body higher off the substrate than the posterior before simultaneously

extending one forelirnb and the reciprocal hind limb- In this move the thumb disk was extended weli past the head, and the knee was raised anterior to the pelvic grcile, but the ankle was kept posterior to it. The two moving limbs were then placed on the substrate.

Their disks remained attached to the surface and the bat pulled itself forward, while the reciprocal fimbs moved in the same fashion (Figure 7). The horizontal leaping gait began with both forelimbs outstretched anterior to the head. Bats pulled themselves forward by them, and launched into the air ca. 10' to 40" from horizontai. in these trials, the maximum angle of the leaps may have been iimited by the dimensions of the cage, as bats rarely (four of 24 observed leaping gait sequences) contacted the ceiling when leaping. Before falling back to the substrate, the bats rotated their shoulders in a mamer similar to that exhibited in flight (Altenbach, 1979), until they were back in their starting position at touch down. During the rotation of the arms, the wing membranes were partidy opened, and may have assisted in the propulsion of the bats. After landmg, the bats pushed themselves off the surface again, repeating the cycle

(Fibwe 8). Bats using the leaping gait moved more quickly than crawling bats (Figures 7,

8,9, 10).

Locomotion on a vertical surfme

When T. colo or held thernselves on a vertical surface, the head was always onented above the feet. Tbey used a crawling gait when walking up the lexan polycarbonate walls of the cage. There, crawling consisted of the same altemating movements of the forelimbs as seen on horizontal substrats, but the hind limbs moved differentiy. The left hindlimb was extended antenorly at the same tirne as the left forelimb, and the right hindlimb with the right forelimb (Figure 9). 1 saw no vertical analogue of the leaping gait when T. tricolor walked up a wail of the cage.

Bats sometimes manoeuvred into the comer between two perpendicular vertical walls. When ascending corners, they sometimes used the crawling gait, facing ventrally toward the comer, with two left disks on one wall and the right disks on the other. Of the 16 18 thes 1 observed a bat ascending a corner, 1 saw this gait nine thes. 1 also observed a

leaping gait used in these corners (10 of 18 ascents). The bat centered itself between the

perpendicular walls and phted its thumb disks above the head, one on each of the waiis.

Thrusts of the wings propelleci the bat upwards quickiy. The shoulders quickly rotated, so

that upon landing, the disks were again above the head (Figure 10).

When moving down a vertical surface, T. tn;-color maintained the head-up

orientation. Bats used two rnethods of descent. Usually, bats simply let go of the surface, dropping some distance before catching themselves again with the disks. 1 observed this behaviour in 28 of 23 descents. In other trials (5 of 23), bats simply used a reverse of the vertical crawling gait-

I observed one adult male bat (forearm, 34.0 mm; mass, 3.8 g) on the vertical lexan surface which could not establish a grip with its left thumb disk, though it did not demonstrate difficulty uçing its right one. With its right disk planted on the wdl, the bat probed the surface 40 times over 12.4 s, before it began to lick the left thumb disk- After

Iicking for 4.2 s, the bat was able to immediately grasp the surface and climbed upward without any visible difEculty.

Use of thumb claws

Originally I had intended to use the same cage around rotating T. tricobr as 1 had used for the other bats. To stop T. tricolor from crawling on the polycarbonate walis of the cage, 1 covered its inside walls with screen, a surface upon which T. tricolor should not be able to crawl (Findiey and Wilson, 1974). The fmt T. tricolor placed in the cage, an adult male (forearm, 35.3 mm; mas, 43 g), immediately climbed on the walls by 17 interiocking the claws of its thumbs with the holes of the screen. Because the cage was ineffective at preventing bats from ieavhg the test surface, I did not use it in the rotation experiments. No other bats were placed in the screen cage. 1 did not observe any T. n-icolor interlocking the thumb claws with the sandpaper or perf'orated alurninum surfaces.

Discussion

Mechanisms of sticking

My data support the predictions from anatomicai work that T. tricolor use suction to adhere to smooth surfaces (Wimsatt and Villa R., 1970; Schliemann, 1970b;

Thewissen and Etnier, 199S), but also suggest a secondary mle for wet adhesion. Wimsatt and Villa R. (1970) stated that the sudoriparous glands of the disk face probably serve a locomotory function by ensuring the 'adhesional efficiency' of the disks, but did not explore the possibility of a non-suctorial sticking mechanism. Wethe fluid-fiiled joint between the disk face and the substrate probabiy assists in the maintenance of a seal for suction (Wimsatt and Villa R., 1970), my experiment reveals that it also provides the vehicle for wet adhesion.

I could not detennine which of the two possible components of wet adhesion,

Stefan adhesion or capillarity, was used by T. tricolor. They are distinguishable because in Stefan adhesion, the sticking force is exponentially correlatecl to surface ara, and in capillarity the correlation is aithmetic (Emerson and Diehl, 1980). By observing the sticking abilities of several different-sized species of tree frogs on a rota ting surface, researchers deterrnined that capillarity was the component of wet adhesion used by these 18 animals (Emerson and Diehl, 1!3€#3). The similarity in size among the three species of

Thyropterc precludes the use of this approach in these bats.

Because T. tn'colûr are small mammals, energy conservation is an important

component of their ecology (Langman et al., 1995). Wet adhesion is probably beneficiai

to these bats for just this reason. A locking rnechanism (tendon locking rnechanism -

?LM)opposïte the proximal phalanges of each toe and pollex allows some bats to hold

the digits in flexed positions with no muscular effort. The TLM is widespread among

bats, but absent in thyropterids (Quinn and Baurnel, 1993). T. tricolor have a thick

retinaculum on the flexor sheath associated with the proximal phalam, but it lacks the

tubercles and plicae of the TLM.The retinaculum provides some measure of friction when the tendon is forced against it during flexion, but constant muscular exertion is stiil

required. For T. tricolor, adhenng to a leaf over long periods may place sibonificant

energy requirements on the flexor pollicis brevis profundus muscle, which maintaias

suction by pulling the cartilaghous plates of the thumb disks. Wet adhesion may conserve energy by allowing complete or partial relaxation of the musculature controlling disk shape.

Wimsatt and Villa R. (1970) stated that the disks of T. îricolor are always kept moist, and this was supported by dl of my observations. However, after long periods of roosting, the liquid between the adhesive organ and substrate ~ghtdry, as it does in lirnpetç (Smith, 1991b). At this stage, the cohesive properties of the precipitate might enhance the strength of attachment, marking a shift from wet adhesion to &hg. 1 did not test for gluing in this experiment. 19 Roosting Posture

While roosting, bar typically hang head-down, holding on by the daws of their toes. In chùopteran evolution, the hindlimb skeleton became Iighter, probably to facilitate flight. As a consequence, the legs of most modem bats cannot support the body's weight.

The head-down, pendulous posture of bats is the byprodlnct of this evolutionary trend

(Howell and Pylka, 1977). Uniike nearly al1 species of bats, 77Zyrop~eraspp. roost head- up, and this fact is reflected by larger wrist disks which presumably, because of their ara, cm support more weight than the srnalier hind disks. In tbis experïment, the differences in the mechanism of adhering to a surface coincided with differences in roosting postures - Le., heads up for T. tricolor, heads down for the other bats 1 studied, 1 also found that TItricobr used interIocking on screen, but they did so in the hads-up orientation as well.

Espada (1870)stated that while sleeping, T. m'cobr hung head-down by their claws, like other bats, but did not specify how he obtained that information- It is most likely that he did not observe this behaviour, but rather inferred it based on the morphology of the bats, specifically the presence of the hind claws. In my own observations in the field, in my search of Iiterature on TItricolor, and in conversations with M. Vonhof, who has found severai hundred T. tricolar rwsts, (per. corm.), 1 have not been able to find any evidence which supports the Espada contention. While it is possibIe that T. tricolor occasiondy roost head-down, we should postpone acceptance of that theory until T. colo or are observed doing so. Mayernballonwids, including Rhynchonycterk naso, ,psp surfaces with the hind and thumb claws in a characteristic head-down, four-point stance. This unusuai posture, exhibited in this study only by R. nasa (the only embdonurid tested) probably accounts for the greater ability of R- rurro over those of other bats in the study to grip sandpaper. Because R. moheld no better than phyllostomids or vespertilionids on surfaces where interlocking was not possible, the advantage of the four-point stance must be improved interlocking.

Roostz'ng ecology

The morphological specializations of disks and the habit of head-up orientation while roosting may make more conventional roosts such as bark, tree hollows, or caves unusable by T. tricolor, but they use a variety of roosts. Although 27zyoptera species have most often been found in young furled leaves (e.g. Findey and Wilson, 1974), there are records of them roosting elsewhere. In Trinidad, Goodwîn and Greenhall (1%1) found T. tricolor roosting with R. naso in a curied dead Heliconia Id. ln French Guiana, three of 12 roosting groups of T. tricolor found by Simmons and Voss (Lm)roosted in dry, curled, dead leaves of Phedospermum (Strelitziaceae). T. disciftera have been reported roosting under dead banana leaves (Robinson and Lyon, 190 1; Torres et al.,

Findley and Wilson (1974) postulated that on the Osa Peninsula of Costa Rica, the availability of rolled leaves limited the abundance of T. tricolor. However, my data dernonstrate how the thumb claws can be used by 7'. tricolor to attach to rough surfaces, demonstrating how these bats might use other, as yet unknown types of roosts. Pine 21 (1993) has aiready siiggested that the rarity of T. baliin collections may be a result of roost types used by this species which are not known to researchers. Sampling biases may be the cause of the apparent dependence of Thyroptera spp. on furled leaves as roosts.

Our knowledge of the roosîing behaviour of this species may be overestimated.

Recent advances in the field of radio telemetry have greatly increased our knowledge of where bats roost (Lewis, 1995) yet we cannot radio tag many bats because of their small body sizes. A heavy transmitter can change the behaviour of the study animai, rendering any data collected questionable, so when possible, researchers use transmitters that weigh less than 5% of the bat's weight (Aldridge and Brigham, L988).

The smallest transmitters available in 2000 weigh around 0.45 pms. WhiIe this is half the mass of the smallest transmitter available in 1990 (M. B. Fenton, per. comm.), it is still more than 10% of the mass of adult T. tricolor or R. mu.As advances in technoiogy progressively decrease the weights of the srnailest radio transrnitters, we cm look forward to less biased studies of where T- tricolor and other smd bats roost. Non-aerial locomotion

The abiliîy to walk on smooth vertical and overhanging surfaces in mammals is rare, and no survey of this behaviour in bats has previously been undertaken. Presumably, three species of Thyroptera share this abiiity only with M. aurira, and perhaps also

Tylonycteris padzypus, Pipistrellus nanus, and Myotis bocagei (Vespertilionidae), al1 of which appear morphoIogicdy suited to crawling on smooth surfaces (Thewissen and

Etnier, 1995; Brosset, f 90,although the vespertilionids Iack disks.

When manoeuvring on smooth, non-horizontal surfaces, bats must resist slipping or falling. The necessity of sticking ability to T. b-ïcolor ascending vertical lexan was exernplified by the bat which appeared unable to crawl until it had cleaned its disk of debris. The crawling gaits used by T. tricolor on horizontal and vertical surfaces are different, probably refiecting their responses to the relative direction of gravity which is different in those situations-

There have ken relatively few investigations of non-aerial locomotion in bats.

Mmotus caZr;forninrs (Phyllostomidae) do not land on the ground for roosting or feeding, and cannot crawl on horizontal surfaces (Vaughan, 1959). Desmodus rotundus

(Phyllostomidae) are remarkably agile on the ground, readily crawling, ninning, and leaping (Altenbacb, 1979), probabiy adaptations for drinking blood (Schutt et al ., 1997).

These species represent the known lower and upper extremes of terrestrial locomotion in bats. The Iocomotory capabilities of T. tricolor fall somewhere in between these extremes. Die& (1973) surveyed the horizontal walking behaviours of ten bat species

spanning four families (Antrozoidae, Molossidae, PhyUostomidae, Vespertilionidae), but did not discuss the relevance of anatomy to different walking styles. There are more complete studies of anatomy and terresirial locomotion for only four species of bats,

Eumops perotis (Molossidae), Myoiis velver (Vespertilionidae), Mmroîus calijornicus

(Vaughan, 1959), and Desrnodus rocundits (Altenbach, 1979). The range of ways in which bats have solved the problem of what do to after they land remains largely unkoown. This rnakes it difficult to make broad cornparisons of T.tncolor with bats in

While crawling, the disks are the main points of contact bebveen T. tricolor and the substrate, while in other crawling bat species the wrists and feet are in contact with the substrate (Dietz, 1973). As a result, the movements exhibited by T. tricolor are very similar to the pattern described for diskless bats. In their horizontal crawling gait, T.

Éricolor appear most similar to vespertilionids, holding their bodies at an angle while crawling, udike molossids which hold the body parallel to the substrate (Dietz, 1973).

AIso, the tails of T. tricolor are not curved upward during locomotion they are in molossids (Vaughan, 1959).

Until now, descriptions of terrestrial locomotion in bats have been Iimited to the crawling gait (Dietz, 1973; Vaughan, 1959), and descriptions of fIight initiating leaps by

Desmodus (Altenbach, 1979; Schutt et al., 1997). 1 could find no description of a leaping gait for any other bat. The scarcity of locomotion studies precludes any statement regarding whether or not other bats leap, though on smooth surfaces, the disks of T. 24 tricolor rnay provide enough grip to allow them to pull themselves fonvard, making the

gait unavailable to diskless bats, The leaping gait of T-tricolor likely occurs in nature,

within the restrictive dimensions of a fded leaf. Also, thoiigh the foraghg behaviour of

T- triculor has not been documentai, these gai& may aiso be used by T. iricoior while foraehg, or in some other aspect of its ecology.

colo or have a much pater ability to manoeuvre on horizontal surfaces than previously thought. Findley and Wilson (1974) observed that when T. tricolor were placed on a horizontal wood surface, they did not crawl but instead attempted flight by means of strong downward thrusts of the forelirnbs. Although 1 did not observe bats in a cage similar to that in wbich Findley and Wilson did, their description probably represents the horizontal leaping gait observed here. Because bats in my horizontal- locomotion trials rarely contacted the cehgof the enclosure, it is reasonable to infer that they were aware of the cage dimensions. It does, therefore, not appear that the leaping gait simply represents attempts at flight that were intempted when bats hit the roof.

The signifïcance of the crawling and teaping pits of T. îricolor to its ecology were not investigated by this study, but non-aerial locomotion is presumably relevant whenever bats roost For example, the ability to exit a roost rapidly would provide two selective advantages. By shortenhg emergence time, the roost is made less conspicuous to predators, and under predation threat, a rapid emergence decreases the likelihood of being caught (Fenton et al., 1994)- In the young tube-shaped leaves where T. tricolor are most often found, bats enter and exit through the hole at the upward-facing tip, so emergence must be executed by upward movement. This is probably facilitated by the 25 head-up posture of resting bats. The movement exhibited by bats Ieaping up corners between perpendicular walls would also function inside a tube-shaped leaf, allowing a rapid means of exit. Upon emergence from the roost, shoulders would already be rotating together, as they do in flight, allowing a fast transition to aerial locomotion, and rapid escape. The habit of detachhg the disks and dropping provided rapid descent, and may be beneficial to bats exiting the down-facing openings of dned PkeMkospemum roosts.

Evolution of disb

At the ventral surface of the pollex beneath the rnetacaf~ophalangealjoint of most microchiropterans is a pad îhat makes contact with the substrate when a bat rests on a horizontal surface wmsatt and Villa R., 1970). Pads also occur on the feet of some bats at the metatarsophaiângeaI junction (Altenbach, 1979). The disks of T. h-icolor occur at just these positions. According to Schliemann (1970a), thumb pads arose in bats as the result of a latent potency to develop adipose and comective tissue to such an extent that callosities could have fonned on the palrnar surface of the thumb and the sole of the foot.

Such callosities are apparent in bats such as P. nus and M. bocagei, species that also have been found roosting in furled leaves (Brosset, 1966). SchIiernann (1WQa)stated that in Thyroptera species, these pads were morphological~yimproved by natural selection as they provided mechanical protection to tissues while animals moved on the ground. This refinement led to a stage of praedisposition (prospective adaptation) which enabled the animals, &ter a change in function had talcen place, to use the organs as sucking devices.

My results indicate that in the course of this evolution, T. tricolor lost some ability to use its claws for roosting the way its diskless ancestors did. 26 The thumbs of T. tricolor are unusual compared to those of other bats. The first phalanx extends in lhe with the metacarpal in most bats, but in 2'. micolor it is bent back dorsally upon the metacarpal approximately 130'. Also in T. nicolor, the two phalanges of digit 1 are rotated approximately 45' out of line with the bending plane of the metacarpophalangeal joint (Wimsatt and Villa R, 1970). This arrangement, not seen in diskless bats, keeps the thumb clear of the disk on smooth surfaces, and appears to have diminished the functionality of the thumb as a means of gipping rough surfaces. Findley and Wilson (1W4) observed that during locomotion, only disks of the wnsts made contact with the substrate, while the thumb claw did not. JuveI.de T. discifiera use the thumb daws for grîpping the mother during suckling (Robinson and Lyon, 1901), and rny data demonstrate that aduit T. tricolor retain some utility of their thurnb claws as interlockîng devices.

On the sandpaper and perforated aluminum surfaces, diskless bats stuck to the rotating platfonn through a larger range of angles than T. tricolor did. However, this might not be solely due to morphologicd differences between T. tncobr and other bats.

Because I did not observe any attempt by T. frzkolor to interlock the thumb claws with any of those surfaces, 1 could not determine what their performance might have been had the claws been interlocked. The behavioural trait of not using interlocking may have been the factor which decreased their sticking ability, not the morphological inability to do so, suggested by other authors (Findley and Wilson, 1974). 1 can only Say with certainty that some combination of morphology and/or behaviour of T. tricolor render them less effective at clinging to those surfaces than those of other bats. 27 Although eariïer works had aligned thyropterids with phyllostomids (Wimsatt and

Enders, 1980), the Myzopodidae and Thyropteridae are currently placed together in the superfaanlly Nataloidea (Microchiroptera), which also includes Natalidae and

Furipteridae (Simmons and Geisler, 1998). The earliest fossil nataioids occur in Eocene strata, but no fossils have been assigneci to either of the disk-bearing families, so the history of disk evolution is unclear from the fossil record. Morphological analyses by

Schliemann (l970b, 1974) revealed that the disks are synapomorphic in extant f op rem, but that despite the relationship of thyropterids to myzopodicis, their disks bear only a superficial sidarity- The disks of these two families are convergent, having

&sen independently.

Detailed anatomical studies perfomed on T. tricolor have not yet been performed on T. discifera or T. Zavaii. Because their disks are synapomorphic with those of T. tricolor (Schliemann, 1974), it should be assumed that their mechanisms of attachent are the same, i.e. suction and wet adhesion. These predictions can be tested by repeating the experimental protocol used here on those species.

The thumb disks of M. aurita measure ca 6.5 mm in diameter, and the hind disk diameter is ca. 5.0 mm. Schliemann (1974) proposed that they function through suction; however, he did not observe living organisms, so he could make that inference based only on dissections of museum specimens. M. aurita disks are not concave, but flat

Schliemann (1974) suggested that to change the shape of the pad to provide suction, the bat may puU bundles of tendons, which are concentrated at the center of the pad, creating space below the disk. He the~rizedthat the high concentration of veins around the disk 28 periphery might permit the disk edge to sweil, which would enhance suction by providing

a smooth seai on a surface (Schliemann, 1974). More recent studies of museum

specimens suggest that the disks of M. aurita function through gluing (Thewissen and

Etnier, IWS),though the disk faces lack glands (Schliemann, 1974). Like T. mzkolor, M. dtu lack a T'LM (Quina. and Baumel, 1993). The suction (Schliemann, 1974) and

gluhg (Thewissen and Etnier, 1995) hypotheses of M. -tu attachent may be flawed simply because they each assume a lone mechanism of gripping. Multiple attachment mechanisms probably allow bats to manoeuvre on a greater range of surfaces than does a single one, and because 1 detected more than one mechanism in T. tricobr,the possibility of multiple ones in M. aurita should be hvestigated.

In Neotropical rainforests, foliage is an abundant source of poteatial roosts, but the smooth surfaces of leaves are not conducive to interiocking with the hind toes. This challenge has been overcome behaviourally by tent-roosting species, rnainly phyllostomids (Kunz and McCracken, 1996). T. îricolor has arrived at another solution to this problem, their wrists and ankles having converged morphologicaily on the suckers of decapods. Suction allows immediate, firm attachment on leaf surfaces, while wet adhesion permits the bats to conserve energy wbile roosting over longer periods of time.

T. îricolar also retain the capacity to interlock on rough surfaces, though their abiiity to do so is diminished compared with those of bats in other famiries.

The disks of T. tricolor have also provided a means by which to crawl and leap on smooth surfaces. These abilities allow the bats protection in foliage, by providing a means of emerging from a roost quickly to escape predation or other threats. This is 29 facilitated by a head-up posture. While abiding by the morphological restraints necessary for flight, T. triculor has become extremely well suited to an ecology very different from those of other bats.

Conclasions

1. Living T. b-icolor use suction to adhere to smooth surfaces.

2. On surfaces where suction can not be used by T. colo or, they adhere by means of

wet adhesion-

3. T- ficolor can cling to screen by interlocking the thumb claws with the surface.

4. T. hicolor adhere better to lexan polycarbonate and alurninum sheet than bats without

disks do.

5. T. îricoZor adhere Iess well to medium grade sandpaper and perforated aluminum than

bats without disks do.

6. T. iricolur move across smooth horizontal surfaces using either a crawling or leaping

gait.

7. T. tricolor move up smooth vertical surfaces using a crawling gait, and move up the

W corner between two vertical surfaces using either a crawling or leaping gait.

8. WhiIe crawling on horizontal surfaces, T. colo or move a hindlimb anteriorly at the

same time as the reciprocal forelirnb is moved anteriorly, while on vertical surfaces,

each hindlimb moves antenorIy at the same time as the forelimb on that side of the

body. Aldridge, H. D. J. N., Brigham, R M. (1988). Load carrying and maneuverability in an

insectivorous bat: a test of the 5% deof radio-telernetry. J. Mm-69,379-382.

Altenbach, J. S. (1979). Locomotor morphology of the vampire bat, Desmodus rotundus.

Pittsburgh: The American Society of Mamrndogists.

Aitringbam, JOD. (1996). Bats: Biology and Behaviour. Oxford: Oxford University

Press.

Anthony, E. L. P. (1988). Age Determination in Bats. In Ecological and Behavioral

Metho& for the Shldy of Bats, (ed. T.H. Kunz), pp. 533. Washington D.C.:

Srnithsonkm Institution Press.

Arita, R T. a F., M. B. (1997). Flight and echolocation in the ecology and evolution of

bats. Tred in Ecology and Evolution 12,53-58.

Antamn, K., Liang, Y. A, IIsleh, S. T., Zesch, W., Chan, W. P., Kenny, T. W.,

Fearing, R., M,R J. (2000). Adhesive force of a single gecko foot-Kr.

Nature 405,6û 1-685.

Brosset, A. (1966). La Biologie des chiroptères. Pans: Masson.

Carvalho, A. L. d, (1939). Zur biologie einer fledennaus (Thyroptera tricolor Spix) des

Amazonas. Ges. Na~ur-Freunde Berlin ,249-253.

Die@ C. L. (1973). Bat walking behavior- J. Mmm.54,790-792- Dobson, G. E, (1876). On pecuiiar structures in the feet of certain species of mammaIs

which enable them to waik on smooth perpendicular surfaces. Roc. &OZ- Soc.

Lod ,526-535.

EdIund, A. F'. and Koew M. A. (1998). Adhesion and reaüachment of compound

ascidians to various substrats: weak glue can prevent tissue damage. J Exp Bi02

201,2397-2402-

Emerson, S. B., Diehl, D. (1980). Toe pad rnorphology and mechanisms of sticking in

frogs. BioZ. J. Linn. Soc. 13, 199-216.

Espada, J. d. 1. (1870). Aleouno datos nuevos O curiosas acerca de la fauna del alto

Amazonas (Maniif~eros).Bol. Rev. Univ. Madrid, 21-27.

Fenton, M. B. (1992). Bats. New York: Facts On Fie.

Fenton, M. B., Rautenbach, 1. L., Smith, S. E., Swanpoel, C. M., Groseil, J., and van

Jaa~sveld,J. (1994). Raptors and bats: threats and opporhrnities. Anim. Behav.

48,9- 18.

Findley, J. S., Stodïer, E. E,and Wilson, D. E. (1972). Morphologie properties of bat

wings. J. Mmm. 53,429-444.

Findley, J. S., Whn, D. E. (1974). Observations on the neotropical disk-winged bat,

Thyropiera tricolor Spix. J. Mmm.55,562-57 1.

Flammang, P., Micheï, A., Cauwenberge, A. V., Alexandre, R and Jangoux, M.

(1998). A study of the temporary adhesion of the podia in the sea star Asferias

rubens (Echinoderrnata, Asteroidea) through their footprints. J Exp Bi02 ml, Foster, M. S. (1992). Tent Roosts of Macconnell's Bat (Vampyressa macconnelZ&

Biotropica 24,447454.

Goodwirt9G- G., GreenhaIl, A. M. (1961). A review of the bats of Trinidad and Tobago:

descriptions, rabies infections, and ecology. Bull- Amer. Mus. Nat. Hist. 122,187-

342.

Hama., G., Barnes, W. J. P. (1991). Adhesion and detachment of the toe pads of tree

frogs. J. q.Biol. 155,103-125.

Hooke, R. (1665-7). Mcrographica. London: .

Howeil, D. J. and Pyb, J. (1W7). Why bats hang upside down: a biomechanical

hypothesis. J Theor Biot 69,625-631.

Kier, W. m, and Smith, A- M. (19%). The morphology and mechanics of octopus

suckers. Biol. Bull. 178, 126-136.

Kmq T. R,McCracken, GoF. (19%). Tents and harems: apparent defense of foliage

roosts by tent-making bats. J. Trop. Ecot. 12, 121-137.

Langman, V. A., Roberts, T. J., BkkYf., Maloiy, Go W O., Heglmd, N, C., Weber,

J.-mYICram, R. and Taylor, C. R. (1995).Moving cheaply: energetics of

walking in the African elephant. J. exp. Biol. 198,629-632.

Leeuwenhoek, A, (1690). Coliected Works (tram. S. Hooie, 1800- 1807). II.

Lewis, S. E. (1995). Roost fidelity of bats: a review. J. Mmm.76,481-496.

Nachtigall, W. (1974). Biological mechanisms of attachment. The comparative

morphology and bioengineering of organs for linkage, suction, and adhesion. New

York: Springer-Verlag. 33 Norberg, U. M. (1969). An arrangement giving a stiff leading edge to the hand wing in

bats- J. Mmm- 50,766-770.

Pine, R. R (1993). A new species of Thyroptem Spix (Mamrnalia : Chiroptera :

Thyroptendae) from the Amazon Basin of northeastern Peni. Mamdia 57,213-

225.

Quinn, T. & Banmel, J. J- (1993). Chiropterazl tendon Locking mechanism. J. Murph.

216,197-208.

Reid, F. (1997). A field guide to the mammais of Central America & southeast Mexico.

New York: Oxford University Press.

Riskin, D. K., Pybns, M. J. (1998). The use of exposed diumal roosts in Alberta by the

littIe brown bat, Myotis luczjxgus. Cari, J. ZooE. 76,767-770.

Robmson, W.,Lyon, M. W. (1901). An amotated List of marnmals coilected in the

viciniv of La Guaira, Venezuela. Proc. US.Nat. Mus 24, 135- 162.

Scblîemann, H- (1970a). Ban und funktion der haftorgane von Thyroptera und

Myzopoda (Vespertilionoidea, Microchiroptera, Mammalia). Am dem

Zoolgischen Staatsimtitut und ZooIogischen Museum Hamburg ,353-400.

Schliemann, H. (197Ob). Die Haftorgane von Thyroptera und

(Microchiroptera, Mammalia) - Gedanken ni ihrer Entstehung als

Parallelbildungen. 2. zool. Syst. und Evolut .-forsch. 9,61-80.

SchKemann, H. (1974). Haftorgane bei flededmausen. Natur und Museum 104, 15-20.

Schutt, W. A., Altenbach, J. S., hg,Y. R, Colünane, De M., Hermanson, J. W.,

Munadal&F. and Bertram, J. E. A. (1997). The dynamics of flight-initiating 34 jumps in the common vampire bat Desmodus rotundus. J. Exp. Zûol. 2ûû, 3003-

3012.

Simmons, N. B., Vsss, R. S. (1998a). The Mammals of Parawu, French Guiana: A

Neotropicai Lowland Raidorest Fauna Part 1. Bats. Bull- Amer. Mus. Nat. Hist.

237, 1-219.

Simmons, N. B., Geisler, J. H. (1998b). Phylogenetic relationships of Icaronycteris,

Archaeonyctens, Hassianycteris, and Palaeochiropteryx to extant bat lineages,

with comments on the evolution of echolocation and foraging strategies in

Microchiroptera. New York: Amencan Museum of Natural History.

Smith, A. M. (1991a). Negative pressure generated by octopus suckers: a study of the

tensile strength of water in nature. J. exp. Biol. 157,257-271-

Smith, A. M, (1991b). The role of suction in the adhesion of limpets. J. W.Biol. 161,

151-169.

Smith, A. M. (19%). Cephalopod sucker design and the physical limits to negative

pressure. J. exp. Biol. 199,949-958.

Stark, N. E. (1980)- Experimental analysis of adhesion of ChtysoZimpolita

(Chrysornelidae: Coleoptera) on a variety of surfaces. J. q.BioL 88,91- 107.

Strnhsaker, T. T. (1961). Morphological factors replathg flight in bats. J. Mm.42,

152-159.

Thewissen, J. G. M., Etnier, S. A. (1995). Adhesive devices on the thumb of

vespertilionoid bats (Chiroptera). J. Mmm.76,925-936. Toms, U P., Rosas, T., Tiranti, S. 1. (1988). Thyroptera discifera (Chiroptera:

Thyropteridae) in Bolivia J. Mm-69,434435

Tnttie, M, D. (1974). An improved trap for bats. J. Mm,55, 475-477.

Van Zyll de Jong, C. G. and National Museum of Natol;ll Sciences (Canada). (1985).

Handbook of Canadian mammds :bats. Ottawa- National Museum of Natural

Sciences.

Vaughan, T. A. (1959). Functiond morphology of three bats: Eumops, Myotis,

Macrotus. University of KmmPublications Museum ~fNatura1History 12, 1-

153.

Vaughan, T. A., Baternan, GoC. (1970). Functional Morphology of the Forelimb of

Monnoopid Bats. J. 1Wamm. 51,217-235.

Wilson, DmE., Findley, J. S. (1977). Thyroptera tricolor. Mmmulian Species 71, 1-3.

Witt,W. A., Villa-R, B, (1970). Locomotor adaptations in the disc-winged bat

liTzyroptera tricolor 1. Functional organization of the adhesive discs. Am. J. I.-

129,89-120.

Wimsaît, W. A. and Enders, A. CI (1980). Structure and morphogenesis of the uterus,

placenta, and paraplacental organs of the neotropical disc-winged bat Thyroptera

tricolor spix (Microchiroptera: Thyroptendae). Am J Anat 159,209-243.

Zar, J. EL (1999). Biostatistical Analysis. Upper Saddle River, NJ: Prentice HA. Table 1: Species used in mtating platform experiment Family Species Mean Mass (g) +/- n Lodity standard deviation when n 2 5 Rhynchonycteris naso 4.5 +/- 0.43 8 Ca60 Palma Artibeus jamaicensis 50.8 2 CaiioPalma Artibeus watsoni 10.7 +/- 1.40 20 Caîio Palma Carollia brevicauda 16.0 +/- I .49 5 Caiïo Palma CaroUia castanea 14.1 4-2-40 10 Cano Palma Carollia perspicillata 18-1 +/- 2.58 8 Caiio Palma Glossophaga commissarisi 8.1 3 CaiïoPalma Glossophaga soricina 7.8 2 CaÏioPaha Hylonycteris underwoodi 73 1 C6oPaima Micronyctens brachyotis 13 .O 3 CaoPalma Micronycteris microtus 5.7 +/- 0.47 6 Cafio Palma Platyrrhinus helleri 17.2 2 Caiïo Palma Uroderma bilobatum 17.6 1 CaÏïo Palma Vampyressa nymphaea 11.8 1 CaÎioPalma Thyro pteridae Thyroptera tricolor 3.5 +/- 0.79 31 Cano Palma Vespertîlionidae Myotis leibii 4-9 4 QUBS Myotis lucifugus 7.0 +/- 1.00 26 QUBS Myotis nigricans 5.6 3 CailoPaima Myotis septentrionalis 6.0 +/-OS2 16 QUBS Table 2: Kniskal Wallis tests for differences in sticking ability among bats grouped by family on each of the four surfaces:

Sandpaper 2.67 26.009 (0.0005 Sheet Aiurninum 2.67 50.415 <0.0005 Perforated Aluminum 2.67 80343 (0.0005 Polycarbonate 2.67 50.051 (0.0005 Table 3: Q statistics obtained from multiple cornparisons (two-tailed) of sticking abilities among groups on each of the four surfaces (Emb- = Emballonuridae; Phy. = Phyliostomidae; Thy. = Thyroptendae; Ves. = Vespertilionidae): Sandpaper Sheet Perforated Polycarbonate Aiuminum Aluminum Emb, vs. Phy. 3.24** 1-55 0.73 0.41 Emb. vs, Thy. 5.%*** 2.89* 5.5?3*** 4.23 *** Emb. vs. Ves. 2.77* 1.68 0.93 O. 18 Phy. vs. Thy. 5.09*** 7.85*** 8.73*** 837*** Phy. vs. Ves. 0.85 030 0.40 1.19 Thy. vs- Ves. 5.65*** 7.71*** 8.40*** 6-94*** ** P€O-001 * P<0.01 * P < 0.05 Eyre 1: Thumb disk of ï%yropterahicolor viewed through a vertical giass surface to which it is attached. Photograph by M. B. Fenton. Emre 2: The four test surfaces used in the rotahg platform experiments: sandpaper (a), lexan polycarbonate (b), aiuminum sheet (c), and perforated aiuminum sheet (d). Fioure 3: Percentage of individuais in each farnily of bats tested that had not fdlen from a rotating sandpaper surface (velocity = 3"s-'), with changing angle of the surface. 0' is horizontal with a bat on top, 90' is vertical, and 180" is horizontal with the bat below.

O 20 40 60 80 100 120 140 160 180 Angle of rotation (degrees) Fipure 4: Percentage of individuais in each family of bats tested that had not fallen fmm a rotating aluminum sheet surface (velocity = 3"s-'), with changing angle of the surface. 0' is horizontal with a bat on top, 90" is verticai, and 180" is horizontal with the bat below.

-Emb. (n=8) ' Phy.(n=61) -Thy- (n=31) Ves. (n=46)

O 20 40 60 80 100 120 140 160 180 Angle of rotation (degrees) Lure 5: Percentage of individuals in each family of bats tested that had not fdlen from a rotating perforated aluminum surface (velocity = 3"s-'),with changing angle of the surface. 0' is horizontal with a bat on top, 90' is vertical, and 180' is horizontal with the bat below. Note that 54.8%of T. tricolor hold the surface past 90' (purple line), indicating that those bats used wet adhesion.

90% Y -a8096 %O% Y rP

E~O 7 .150% Phy. (n=59) E J niy*(nSl) 340% Ves. (n=48) 'Ps30% 2 g20% P 10%

Wo O 20 40 60 60 100 120 140 160 180 Angle of rotation (degrees) Fioure 6: Percentage of individuals in each fadyof bats tested that had not fdlen from a rotating lexan polycarbonate sudace (velocity = 3's-'), with changing angle of the surface. 0" is horizontal with a bat on top, 90' is vertical, and 180" is horizontal with the bat below.

-Emb. (n=8) Phy.(n=63) -Thy. (nS1) -Ves. (n=47)

O 20 40 60 80 100 120 140 160 180 Angle of rotation (degrees) Fioure 7:Adult male T. tricolor (forearm, 35.1 mm; mas, 3.7 gf using the crawling gaiait on a horizontal surface. The camera faces the bat's dorsum from above. Images are separated by 1/30 s intervals, crosshairs measure 10 mm by 10 mm. The bat traveled at a net velocity of 0.082 dsin this sequence. Fiwe 8: Addt male T- tricolor (forearm, 34.8 mm; mas, 4.0 g) ushg the Leaping gait on a horizontal surface. The carnera faces the bat's left side and the bat's antero-posterior axis is horizontai. Images are separated by 1/30 s intervals, crosshairs measure IO mm by

10 mm. The bat traveled at a net velocity of 0.284 dsin this sequence. Fi we9: Adul t fernale T. tricolor (forearm, 34.5 mm; mass, 3 -6 g) ushg the craw hg gait on a vertical surface. The canera faces the bat's dorsum and left side, and the bat is oriented heads-up. Images are separated by 1/30 s intervais, crosshairs mesure 10 mm by 10 mm. The bat traveled at a net velocity of 0.133 dsin this sequence. Fimire 10: Adult fernale T. hicolor (same individual as in Figure 9) using the leaping gait in the corner between two perpendicular vertical surfaces. The carnera faces the bat's domand the bat is oriented heads-up. Images are separated by 1/30 s intervals, crosshairs measure 10 mm by 10 mm. The bat traveled at net velocity of 0.235 m/s in this Appendix 1: Last angle of stick for bats on each of four surfaces that were rotateci from horizontal with the bat on top (Cl0),through vertical (90°),to horizontai with the bat beiow

(180') at a velocity of 3's-'. The order in which the surfaces were presented to each bat are also shown. Emb. = Emballonuridae, Phy. = Phyllostomidae, Thy. = Thyropteridae,

Ves. = Vespertilionidae.

1 Emb. Rhynchonycteris naso 40 ACBD 2 Ernb. Rhynchonycteris naso 34 CBAD 3 Emb. Rhynchonycteris naso 45 CBDA 4 Emb. Rhynchonyctens naso 30 CBAD 5 Emb. Rhynchonyctefis naso 29 ACBD 6 Phy. Carollia perspicillata 39 CBAD 7 Phy. Caroilia brevicauda 30 ADBC 8 Phy. Carollia perspicillata 25 CDAB 9 Ves. Myotis nigricans ABCD 10 Phy. Glossophaga soricina 15 ACDB 11 Phy. Artibeus watsoni 16 CBDA 12 Ves. Myotis nigricans CDBA 13 Phy. Platyrrhinus helleri 9 CDBA 14 Phy. Artibeus watsoni CBAD 15 Phy. Carollia castanea 46 CDAB 16 Phy. Artibeus watsoni 34 BADC 17 Phy. Artibeus watsoni 33 DACB 18 Phy. Carollia brevicauda 36 DBCA 19 Phy. Carollia perspicillata 24 DCBA 20 Phy. Artibeus watsoni 34 ACBD 21 Phy. Artibeus watsoni 47 CDBA 22 Phy. Artibeus jamaicensis 24 CADB 23 Phy. Artibeus jamaicensis 38 ACDB 24 Phy. Micmnycteris microtus 38 CABD 25 phy. Glossophaga comm issarki 24 OCBA 26 Phv. Artibeus watsoni 41 BCAD -- Phy. Carollia perspicills - 20.2 28 21 DAC6 Phy. Artibeus watsoni 46 BDCA Phy. Carollia castanea 31 ABCD Phy. Hylonyctens underwoodi 60 DBCA Phy. Carollia perspicillata 46 DBAC Phy. Carollia castanea 44 DABC Phy. Artibeus watsoni 20 ACDB Phy. Micronydefls microtus 46 CADB Phy, Carollia castanea 42 CDAB Phy. Glossophaga soncina 21 DBAC Phy. Platyrrhinus helleri 37 BCAD Phy. GIosso phaga comm issarisi 27 CABD Ves. Myotis nigricans 33 CABD Phy. Artibeus watsoni 30 BDAC Phy. Artibeus watsoni 31 BDCA Phy. Carollia perspicillata 29 BCDA Phy* Micronycteris brachyotis 21 DABC Phy. CaroIf ia castanea 39 ADCB Phy. Artibeus watsoni 60 DABC T~Y* Thyroptera tricolor 180 CDAB T~Y- Thyroptera tricolor 64 DCBA T~Y. Thyroptera trÏcolor 180 DCBA Phy. Micronycteris brachyotis 30 BCOA Phy. Mictsnycteris brachyotis 20 DA8C Phy. Carollia castanea 30 ACBD Phy. Carollia castanea 19 DABC Phy. Artibeus watsoni 44 ACDB Phy- Carollia castanea 39 BDAC Phy. Carollia perspicillata 44 CDBA Thy. Thyroptera tricolor 180 ADCB T~Y Thyroptera tricolor 180 ACBD T~Y- Thyroptera tricolor 180 BACD ThY Thyroptera tricolor 180 ADCB T~Y- Thyroptera tricolor 180 CBOA Thy. Thyroptera tricolor 180 BACD T~Y- Thyroptera tnco lor 180 DACB T~Y- Thyroptera tricolor 180 %DCA T~Y. Thyroptera tricolor 180 BCAD T~Y- Thyroptera tricolor 180 CBDA ThY- Thyroptere tricolor 180 CBDA my. Thyroptera tricolor 180 CAO6 T~Y- Thyroptera tricolor 180 DBAC Phy. Artibeus watsoni 41 DABC Phy. Glossophaga commissarisi 34 CBDA Phy. Carollia perspicillata 29 ADCB Phy. Micronycteris microtus 20 DCBA Phy. Carollia castanea 31 ADBC Phy. Artibeus watsoni 40 CDAB 75 ~hy. Thyroptera tricolsr 34.0 3.7 58 180 68 180 BCAD 51 T~Y- Thyroptera tricolor DCBA T~Y- Thyroptera tricolor ADCB Phy. Carollia brevicauda ABCD Phy, Urodena bitobatum BCDA T~Y- Thyroptera tricolor ABDC T~Y- Thyro ptera tricolor ABCD Emb. Rhynchonycteris naso ACBD Emb. Rhynchonycteris naso BACD Ernb. Rhynchonycteris naso DACB Phy. Carollia brevicauda ACBD Phy. Micronycteris rn icrotus DBAC Phy. Carollia brevicauda ACBD Phy. Vampyressa nymphaea ABDC Phy. Artibeus watsoni BDAC Phy- Artibeus watsoni BCAD Phy. Carollia castanea CDBA Phy. Artibeus watsoni CBDA Phy, Artibeus watsoni ACBD ny. Thyroptera tricolor CABD T~Y- Thyro ptera tricolor CDAB T~Y- Thyroptera tricolor BCDA T~Y. Thyroptera tricolor BDCA Phy. Micronycteris microtus BDCA Phy. Artibeus watsoni BDCA Phy. Micronycteris microtus DCBA T~Y- Thyroptera tBcolor DACB T~Y- Thyroptera tncolor - CBDA T~Y- Thyroptera tricolor BCAD T~Y. Thyroptera tricolor ACBD T~Y- Thyroptera tricolor BCDA Thy. Thyroptera tricolor BDCA Ves. Myotis [ucifugus BADC ves. Myotis leibii BDAC ves. Myotis lucifugus BCAD Ves. Myotis lucifùgus ACDB Ves. Myotis septentrionalis BACD Ves. Myotis leibii CDBA Ves. Myotis lucikrgus CAB0 Ves. Myotis lucifugus BCAD Ves. Myotis lucifugus CABD Ves. Myotis lucifugus CDAB ves. Myotis lucifugus ADCB Ves. Myotis lucifugus DBCA Ves. Myotis lucifugus DBCA ves. Myotis Iucifugus ADCB ves. Myotis septentnonalis CDBA Ves. Myotis septentrionalis CADB 123 ves. Myotis septentrionalis 37 BCAD Ves- Myotis fucifugus 36 CABD ves. Myotis septentrionalis 37 BACD Ves. Myoth septentrionalis 44 BCAD ves* .Myotis septentrionalis 42 CBDA ves. Myotis septentrionalis 34 ACDB Ves. Myotis lucifugus 43 DBCA Ves. Myotis septentrionafis 42 DCBA Ves. Myotis septentrionalis 46 DABC Ves. Myotis septentrionalis 37 CDAB ves, Myotis leibii 33 DACB va. Myotis lucifugus 38 BCAD Ves. Myotis septentrionalis 31 DBAC ves. Myotis lucifugus 44 CBDA Ves. Myotis septentrionalis 50 BCAD Ves. Myotis lucifugus 38 ACBD Ves. Myotis lucifugus 42 D8AC Ves. Myotis leibii 33 ACDB Ves. Myotis septentrionalis 34 ACBD Ves, Myotis lucifigus 30 DCBA Ves. Myotis Iucifugus 38 ABCD Ves. Myotis lucifugus 49 DABC Ves. Myotis lucifugus 53 DBAC Ves. Myotis lucifugus 38 ACDB Ves. Myotis lucifugus 4û CDAB Ves. Myotis septentrionalis 41 DBCA Ves. Myotis lucifugus 47 DCAB ves. Myotis lucifugus 42 CDAB Ves. Myotis septentrionalis 38 DCBA 152 ves- Myotis Iucikrgus 47 BCDA