OUP user menu

Diet, Echolocation Calls, and Phylogenetic Affinities of the Great Evening Bat(Ia io; Vespertilionidae): Another Carnivorous Bat

Adora Thabah, Gang Li, Yinan Wang, Bing Liang, Kailiang Hu, Shuyi Zhang, Gareth Jones
DOI: http://dx.doi.org/10.1644/06-MAMM-A-167R1.1 728-735 First published online: 1 June 2007


We studied the great evening bat (la io) in India (Meghalaya) and China (Guizhou), and present the 1st account of its feeding behavior. We analyzed droppings collected from 119 bats between November and May 2001–2002 and 2002–2003 from India; 28 included bird feathers, with most of these containing 90–100% feathers by volume. The main constituent of the diet overall was Coleoptera, although Lepidoptera and traces of Diptera, Orthoptera, and Hemiptera also were found. In China, bats were captured in early November, and fresh droppings also were collected from underneath the roost. Bird feathers comprised 82% of the droppings of bats by volume. I. io emits relatively low–frequency echolocation calls and sometimes produces 2–toned calls, a characteristic of species that echolocate distant targets. I. io has a high wing loading (15.4 Nm−2), average aspect ratio (6.9), and a high tip shape index (1.1), features associated with fast efficient flight. Phylogenetic analysis of a concatenation of mitochondrial ND1 and cytochrome-b genes indicated that I. io is phylogenetically close to Scotomanes ornatus, in a clade (Nycticeiini) distinct from bats in the genus Pipistrellus with which it has been previously allied. Although I. io converges in wing shape and echolocation call design with the carnivorous, fast–flying Nyctalus lasiopterus, its carnivorous behavior likely evolved independently because the 2 species are not close relatives. It is unlikely that I. io captured birds nesting or roosting at the study caves, and its morphology and echolocation behavior seem well adapted for the capture of large aerial insects and flying birds.

Key words
  • carnivory
  • China
  • diet
  • echolocation
  • flight morphology
  • Ia io
  • India

With a forearm length of 71–77 mm (Bates and Harrison 1997), the great evening bat (Ia io) is one of the largest aerial-feeding microchiropteran bats. The species occurs in southern Asia, with records from China, Laos, Vietnam, Thailand, India, and Nepal (Bates and Harrison 1997; Bates et al. 2005; Csorba 1998). I. io roosts in caves (Bates and Harrison 1997), and although it has a relatively wide distribution, it is uncommon throughout its range and its biology is little known.

To further knowledge of this enigmatic bat, we studied its echolocation behavior and wing shape. We also present the 1st description of its diet, and show that it eats birds both in India and China. Bird-eating is uncommon in bats, and the recent finding that the fast-flying vespertilionid Nyctalus lasiopterus feeds on birds (Dondini and Vergari 2000; IbAnez et al. 2001) has attracted great interest, especially because it is likely that this species captures its prey in flight, especially during avian migration periods (IbAnez et al. 2001). We relate the specialized diet of I. io to its echolocation behavior and wing morphology. Although the echolocation calls of this species have been described in the laboratory (Feng et al. 2001), there have been no studies of its echolocation behavior in the wild. In particular, we determined whether I. io shares features of its echolocation behavior and wing morphology with N. lasiopterus, to assess whether feeding on birds involves similar adaptations in the 2 species.

Ia io is in a monospecific genus, and its phylogenetic affinities have long been debated. Miller (1907) believed that I. io was closely related to Scotozous. Bourret placed the genus in synonymy with Parascotomanes in 1942 (cited in Corbet and Hill 1992). Ellerman and Morrison-Scott (1951) placed Ia as a subgenus of Pipistrellus, whereas Menu (1987) considered the genus to be synonymous with Eptesicus. Our purpose was to attempt to resolve the evolutionary history of L io within the Vespertilionidae by using a molecular phylogenetic approach. By clarifying its evolutionary affinities, we aimed to confirm that bird-eating evolved independently in I. io, and that any morphological and acoustical similarities with N. lasiopterus are the result of convergent evolution.

Materials and Methods

Study sites.—In India, I. io was captured at Phlangkaruh Cave, situated at Nongtrai within the Sheila confederacy (25°11 ′N, 91°37′E) at an elevation of 170 m above sea level on the southern fringes of the Shillong plateau in the East Khasi Hill District of Meghalaya. A subterranean river flows through the cave and emerges down the valley as Phlangkaruh River. Data were collected in November-May 2001–2002 and 2002–2003. Bats were captured in a 2.4 × 1.8-m harp trap (Austbat, Victoria, Australia), or in mist nets placed at cave entrances. Most bats were captured as they returned to the cave after foraging, generally 2-A h after sunset. The cave functioned as a maternity site. Pregnant females were captured in February-April and lactating females in April-May. One juvenile was captured in May. We caught adult bats of both sexes, including 2 males with distended epididymal tubules in November. Over the course of the study, we captured 88 female and 71 male bats (droppings were not collected from all bats). It was difficult to estimate the number of bats at the site because more than 1 exit was present. It is likely that <40 adult bats occupied the cave, assuming that most animals left by the exit where the traps were placed. In China, fieldwork was conducted in November 2005 at Double Dragon Cave (24°58.426′N, 104°52.687′E) in Xingyi District, Guizhou Province. This is a mountainous region ranging in elevation from 1,500 to 2,200 m above sea level. The surrounding vegetation is dominated by broad-leaved evergreen forests, and trees of the families Leguminosae, Sapindaceae, Meliacea, and Mor-aceae are common.

Morphological measurements.—Body mass was measured using a Pesola scale (±0.1 g; Pesola AG, Baar, Switzerland) and forearm length was measured to the nearest ±0.1 mm using dial calipers. Age was established on the basis of epiphysial fusion of the finger joints (Anthony 1988), and reproductive status of females was determined by palpation and examination of nipples (Racey 1988). All morphological data we report are from Indian bats. Wing tracings were drawn by placing the ventral side of the bat on a plain sheet of paper, extending 1 wing (including the uropatagium) and half the body, and drawing around it. Wing parameters were measured from the tracing using a Summagraphic SummaSketch III digitizing tablet (GTCO Calcomp Inc., Columbia, Maryland). Wing-shape parameters were described according to the definitions of Norberg and Rayner (1987).

Recording and analysis of echolocation calls.—Echoloca-tion calls were recorded from 4 bats released from the hand in India using an S-25 bat detector (Ultra Sound Advice, London, United Kingdom) attached to a Portable Ultrasound Processor (PUSP; 448-kHz sampling rate; Ultra Sound Advice) and a Sony Professional Walkman WMD6C (Sony, Tokyo, Japan). The ultrasound detected by the bat detector was time expanded 10 times by the PUSP and the output recorded on to a metal tape in the Walkman. Ultrasound was analyzed digitally using BatSound version 3 (Pettersson Elektronik, Uppsala, Sweden) to assess start frequency, end frequency, interpulse interval, call duration, and the frequency of maximum energy (kHz) from power spectra (512 point fast Fourier transform, Hanning window). The surroundings of the cave represented a relatively cluttered environment, and bats were recorded at <3 m height while flying toward the microphone. One call per individual, recorded as long after release as possible, was selected for analysis; if 2-toned calls were emitted, 1 call of each type was analyzed per individual.

Dietary analysis.—Bat droppings were collected directly from individuals captured after emergence. Droppings were dried and stored in airtight containers and 1 pellet from each individual captured in India was analyzed. The droppings were teased apart with needles and forceps after being softened by soaking in water. Insect fragments were observed under a low-power binocular microscope (10 × magnification) and identified to order using keys (e.g., McAney et al. 1991) and by comparisons with reference material collected in the field. Prey composition was estimated based on percentage volume. In India, 1 dropping was analyzed from each bat captured. We captured bats once per month, with up to 25 bats captured per session. It is possible that droppings from the same individual were analyzed from different capture sessions, but by analyzing 1 dropping per individual per capture session we minimized pseudoreplication. In China, 75 droppings were analyzed from 3 captured bats and from fresh material deposited under the roosting site.

Molecular phylogenetics.—We took 3-mm biopsy punches of wing membranes from I. io captured in China. DNeasy Tissue Kits (Qiagen, Inc., Valencia, California) were used to isolate genomic DNA. The polymerase chain reaction for amplifying the complete mitochondrial nicotinamide adenine di-nucleotide dehydrogenase subunit I (ND1) gene sequences of I. io followed the program: 94°C (50 s), 50°C (40 s), 72°C (45 s), 72°C (8 min). We used the primers L2985 (5′-CCT CGATGTTGGATCAGG-3 ′) and H4419 (5 ′ -GTATGGGCCC GATAGCTT-3′—Petit 1998) for amplification and sequencing. To amplify the complete Cytochrome-b (Cytb) gene sequences via polymerase chain reaction, we designed the primer “Bat_Cytb” (5′-aaa TCA CCG TTG TAC TTC AAC -3′) and used it alongside the published primer “Bat_Cytb_l” (5′-TAG AAT ATC AGC TTT GGG TG -3′—Li et al. 2006) to obtain the target fragment. The control conditions of polymerase chain reaction were 94°C (5 min); 40 cycles at 94°C (50 s), 50°C (40 s), and 72°C (50 s); 72°C (5 min). Sequencing was done using an ABI PRISM 3730 sequencer (Applied Biosystems, Foster City, California). We also amplified and sequenced the Cytb and ND1 gene sequences from Scotomanes ornatus, Nyctalus plancyi (both from Sichuan, China), and Murina (probably an undescribed species from Beijing, China), the last of which was used as an outgroup.

To assess the phylogenetic status of I. io, we also downloaded published sequences of vespertilionids from GenBank. These were Eptesicus serotinus (Cytb: AF376837, ND1: AF401472), Eptesicus fuscus (AF376835, AY033968), Vespertilio murinus (AF376834, AF401470), Vespertilio superans (now known as V. sinensis [Simmons 2005]; AB085738, AB079823), Chalino-lobus tuberculatus (NC002626, NC002626), Nyctalus leisleri (AF376832, AF401431), Pipistrellus pipistrellus (AJ504443, AF401393), Pipistrellus pygmaeus (AJ504442, AF401413), Pipistrellus kuhlii (AJ504445, AF401416), Myotis nattered (AF376863, AY033984), Myotis schaubi (AF376868, AY033955), Myotis daubentonii (AF376862, AY033954), md Myotis bechsteinii (AF376843, AY033978).

We used ClustalX1.81 (Thompson et al. 1997) for sequence alignment with default parameters and MEGA3 (Kumar et al. 2004) to estimate the sequence variation and divergence under the Kimura 2-parameter model. Before all phylogenetic analyses we used MODELTEST 3.6 (Posada and Crandall 1998) to estimate the most appropriate evolutionary model for our sequences. MrBayes 3.1 (Huelsenbeck and Ronquist 2001) was employed to construct the maximum posterior probability tree used to search for the relationships between Ia and other bats. We set 5,000,000 generations for 6 simultaneous Markov chains and the trees were sampled after 1 million generations. Other parameters were set according to suggestions for vertebrate mitochondrial sequences. We also used PAUP 4.0*b (Swofford 2002) to reconstruct phylogenetic trees by conducting maximum-likelihood analyses. We used heuristic searches using the tree-bisection-reconnection branch swapping method with random addition of taxa for maximum-likelihood tree reconstruction. For bootstrapping, we used 100 replicates for maximum likelihood and 2,000 replicates for neighbor-joining with the Kimura 2-parameter model and unweighted maximum-parsimony trees using PAUP 4.0*b. Mean values ± 1 SD are presented unless stated otherwise.


Wing morphology.—Ia io is 1 of the largest vespertilionid bats, with an average body mass of 58 g and an average forearm length of 77 mm (Table 1). To compare the forearm length and body mass between sexes, data were logarithmically transformed to achieve normality before employing t-tests. Males (58.8 ± 5.0 g, n = 60) were significantly heavier than females (56.2 ± 7.9 g, n = 58; t = 2.09, d.f. = 1, P < 0.05) and had significantly longer forearms (77.9 ±1.5 mm, n = 60; t = 6.06, d.f. = 1, P < 0.001) than females (75.9 ± 1.9 mm, n = 58; Table 1). The wings of male I. io tended to be larger than those of females, but the difference was not significant. The wings are large and long, with a wingspan averaging 0.51 m. I. io has a high wing loading of 15.3 Nm−2 but an average aspect ratio of 6.9 (Table 1).

View this table:
Table 1

Wing shape parameters of male and female Ia io. Abbreviations: Fa = forearm length, M = body mass, S = wing area, B = wing span, WL = wing loading, AR = aspect ratio, Shw = hand wing area, Saw = arm wing area, Lhw = hand wing length, Law = arm wing length, and I = tip area ratio. There was no significant sexual dimorphism in wing-shape parameters except for forearm length and body mass (see text), n = 15 except for Lhw, Law, and I, where n = 7.

variableX ̄ ± SDRange
Fa (mm)76.7 ± 1.374.2-78.9
M(g)58.0 ± 5.149.0-63.0
S(m2)0.04 ± 0.0020.03-0.04
B (m)0.51 ± 0.020.47-0.5
WL (Nm−2)15.4 ± 1.512.7-18.2
AR (A)6.9 ± 0.46.3-7.4
Shw (cm2)61.3 ± 2.856.6-65.8
Saw (cm2)94.3 ± 9.481.8-110.2
Lhw (mm)126.4 ± 3.1122.0-130.0
Law (mm)103.0 ± 6.292.0-111.0
I1.1 ± 0.190.86-1.4

Echolocation calls.—Ia io emits relatively narrowband calls of low frequency (Fig. 1). Echolocation calls sweep from about 40.1 ± 4.5 to 18.4 ± 2.6 kHz on average, with the frequency of maximum energy averaging 26.9 ±3.3 kHz. Call duration was 3.8 ± 0.9 ms and pulse interval was 89.6 ± 33.3 ms. Calls were sometimes audible to the unaided ear. Bats sometimes produce 2-toned echolocation calls when flying at higher altitudes. There was a difference in the frequency of most energy ranging from 1.6 to 4 kHz between these calls, with an average difference of 2.8 ± 1.03 kHz (Table 2).

View this table:
Table 2

Tonal difference in echolocation calls of 4 individual Ia io (1 call of each type/bat selected for analysis). Results are mean ± SD.

Call parametersHigher toneLower tone
Highest energy (kHz)27.6 ± 1.524.8 ± 1.5
Start frequency (kHz)34.9 ± 5.938.6 ± 2.8
End frequency (kHz)21.9 ± 1.818.0 ± 1.1
Duration (ms)2.4 ± 0.53.8 ± 0.6
Pulse interval (ms)65 ± 20.593 ± 9.3
Fig. 1

Waveform and sonogram of an echolocation call of Ia io. The spectrogram was made using a 512-point fast Fourier transform and a Hanning window.

Diet.—We analyzed a total of 119 fecal pellets from India. Twenty-eight pellets contained feathers, of which 22 had 90–100% feather composition by volume, 2 samples had 70% feather composition, and 4 samples had only traces of feathers (2–5%). The presence of feathers was most frequent in March, when 50% of the samples had feathers in them, constituting 47.5% of the total diet in that particular month, followed by 17.3% in May and 16.1% in November (Table 3). Overall, however, coleopterans were the main constituent of the diet in most months, followed by birds, and Lepidoptera. Traces of other insects were detected. The proportion of the diet composed of Coleoptera was highest in December (97%), January (95%), and April (94.5%) when bird feathers were not present in samples. In November (70.4%) and May (82.4%) the contribution of Coleoptera was still high, although other insects were found in traces. It was only in March (30.8%) that the proportion of coleopterans in the diet was relatively low, at the time when birds were the major dietary component. Lepidop-tera contributed substantially to the diet in November and March, whereas Diptera and Orthoptera were only ever found in trace amounts. The small volumes of Acaria (mites) found were probably ingested while grooming. There were no obvious differences in diet between the sexes.

View this table:
Table 3

Diet composition in percent volume determined by fecal analysis of a single pellet from each individual sampled in Phlangkaruh Cave in Meghalaya, northeastern India, n = number of individuals sampled.

Prey(n = 25)(n = 20)(n = 16)(n = 20)(n = 18)(n = 20)

Birds comprised a significant portion of the diet (82% ± 16.4% by volume) based on 75 pellets analyzed from China. Black, yellow, and white colors observed in feathers suggested that one bird species eaten by bats might be Tickell's leaf-warbler (Phylloscopus affinis). At 0002 h on 4 November 2005, we caught an adult female I. io with feathers in its claws and blood stains on its mouth and claws. Insects of various orders were found in the droppings from China. The identifiable insect parts were from Coleoptera (8% ± 8.9%), Hemiptera (4% ± 6.0%), Lepidoptera (3% ± 4.1%), and Hymenoptera (1 ± 1.7%). Additionally, 1 snail and some ova of insects were found in the feces (about 2% overall by volume).

Phylogenetic diagnosis.—The completed mitochondrial Cytb and NDI gene sequences of I. io were deposited in GenBank with accession numbers DQ302094 and DQ302095, as well as DQ435069 (Cytb) and DQ435070 (NDI) of S. ornatus, DQ435073 (Cytb) and DQ435074 (NDI) of N. plancyi, and DQ435071 (Cytb) and DQ435072 (NDI) of Murina.

MODELTEST showed the GTR+G+I model gave the best fit to our data. We set parameters as follows: proportion of invariable sites (I) = 0.450, rate matrix of substitution model (A-C = 0.665; A-G = 17.088; A-T = 0.522; C-G = 0.375; C-T = 16.156), and the parameter of gamma distribution shape was 0.875. Average pairwise distance calculations using MEGA3 (Kumar et al. 2004; Table 4) showed the genera Scotomanes and Ia as having the smallest genetic distance (0.172; Kimura 2-parameter nucleotide substitution model) compared with other genera included in our analyses based on the Cytb sequences (Table 4). There was also minimum divergence between S. ornatus and I. io (0.208) at NDI. Sequence divergence values at Cytb were generally similar to those at NDI although Cytb values were marginally lower for all pairwise comparisons. We reconstructed the phylogenetic tree of I. io and a range of other vespertilionid species based on a concatenation of mitochondrial Cytb (1.1 kilobases [kb]) and ND1 (0.8 kb) gene sequences. The results (Fig. 2) did not support the monophyly of la and Pipistrellus. Pipistrellus and Nyctalus formed a monophyletic group, indicating that Nyctalus has a close relationship with Pipistrellus, but not with la. Correspondingly, Scotomanes and la formed a clade that had a sister relationship with Eptesicus. The maximum-likelihood trees given by PAUP*4.0b (not shown) had the same topology as the Bayesian tree.

View this table:
Table 4

Pairwise sequence genetic distance matrix based on mitochondrial Cytb (below the diagonal) and NDI (above the diagonal) gene sequences. The program MEGA3 was used to do the calculations under the Kimura 2-parameter substitution model with transitions and transversions included.a

30.2080.2060.1940.2240.2450.2120.2120.2370.2380.2200.220.2550.2320.231 ~0.2360.274
  • a 1 = Ia io, 2 = Scotomanes ornatus, 3 = Eptesicus serotinus, 4 = Eptesicus fuscus, 5 = Vespertilio murinus, 6 = Vespertilio sinensis, 1 = Chalinolobus tuberculatus, 8 = Nyctalus plancyi, 9 = Nyctalus leisleri, 10 = Pipistrellus pipistrellus, 11 = Pipistrellus pygmaeus, 12 = Pipistrellus kuhlii, 13 = Myotis schaubi, 14 = Myotis nattereri, 15 = Myotis bechsteinii, 16 = Myotis daubentonii, 17 = Murina.

Fig. 2

Maximum posterior probability tree based on a concatenation of Cytb and ND1 gene sequences under the GTR-f G+I model. Numbers at the node of clade indicate maximum-likelihood (1st number, 100 replicates), neighbor-joining (2nd number, 2,000 replicates), and maximum-parsimony (3rd number, 2,000 replicates) bootstrap values and Bayesian posterior probabilities (last number).


Wing morphology.—Values for wing loading and aspect ratio are comparable to those of other aerial hawking bats that forage in open space, such as E. serotinus (loading 12.2 Nm−2 and aspect ratio 6.5) and Nyctalus noctula (loading 16.1 Nm−2 and aspect ratio 7.4). The carnivorous N. lasiopterus has a wing loading of 19.7 Nm−2 (Jones and Rydell 1994) and an aspect ratio of 7.2 (IbAnez et al. 2001).

Echolocation calls.—Low-frequency echolocation calls with relatively long wavelengths are best suited for detecting large prey (Barclay 1986; Heller 1989, 1995) at considerable distance. However, a small proportion of small insects belonging to the order Diptera also were consumed despite predicted Rayleigh scattering caused by the wavelengths of sound being longer than the wing lengths of the insects (Houston et al. 2004). The relatively large, fast-flying N. noctula eats many small prey (Jones 1995), as does its congener N. leisleri (Waters et al. 1995). Bats in the genus Nyctalus emit low-frequency calls whose energy is focused in a relatively narrowband frequency while searching for prey. The low frequency is used to avoid atmospheric attenuation, which increases with frequency (Lawrence and Simmons 1982). The calls of I. io were short in duration, but this is probably due to the clutter in the recording area (Jensen and Miller 1999). The calls were relatively low in frequency (end frequency about 18 kHz). Calls are predicted to be of lower frequency and of longer duration in more open space to increase prey detection range, thereby giving the bats time to react to echoes while flying fast (Jacobs 1999). Calls recorded in this study were similar to those reported by Feng et al. (2001) from bats flying in a laboratory, although the pulse interval was longer and the start frequency lower in our field measurements. The calls of I. io contrast with those emitted by gleaning carnivorous bats that use high-frequency, short-duration and low-intensity calls (Norberg and Fenton 1988).

Two-toned echolocation calls were also produced by I. io. Two-tone calls are typically produced by fast-flying, aerial-foraging bats feeding in open habitats (Heller 1995; Kingston et al. 2003). Although the tonal difference in most bats is about 5 kHz (Kingston et al. 2003), the difference we recorded was comparatively small, but may be higher when bats fly in more open habitats. Tonal differences between alternating calls probably allow bats to separate the echoes from the 2 call types on the basis of frequency (Kössl et al. 1999). Lower-frequency calls may allow the bats to detect distant targets (e.g., the ground), whereas higher-frequency calls may be adapted for detecting small prey at short range (Heller 1989).

Diet.—The most surprising finding was that I. io eats large numbers of birds, both in India and in China. Carnivorous bats constitute <1% of the 900 bat species recognized by Koopman (1994). With 1,100 bat species currently described (Simmons 2005), the proportion of carnivorous species may be slightly lower than this, but the figure is probably an underestimate because of the lack of knowledge about the feeding habits of most bats. Bird-eating has been reported mostly from gleaning tropical bats that occasionally capture resting birds (Norberg and Fenton 1988). Eating birds is rare, even in carnivorous bats, and has only been reported from 4 families to date (Dondini and Vergari 2000, 2004;IbAnez et al. 2001; Norberg and Fenton 1988). Recently, this behavior was reported for N. lasiopterus from the temperate regions of Spain and Italy (Dondini and Vergari 2000;IbAnez et al. 2001). Small birds were 1st shown to be exploited by N. lasiopterus at roosting sites (Dondini and Vergari 2000) and it has recently been proposed that the species captures migrating birds while flying at high altitude (Ibánez et al. 2003), although this has not been observed directly. A small number of droppings of N. noctula also have been found to contain feathers (Gloor et al. 1995).

Barclay (1995) proposed that bats would eat feathers and bones if they experienced a deficiency of calcium in their diet (Studier and Sevick 1992; Studier et al. 1991). Adams et al. (2003) found that water holes with a high concentration of soluble calcium were preferred by bats. Other means of ingesting minerals have been reported for Miniopterus schreibersii, which licks the surface of rocks (Codd et al. 1999). I. io occurs in areas with limestone deposits; therefore, water bodies inside or outside the cave should have high calcium concentrations. Moreover, Ibánez et al. (2003) argued that feathers actually have low levels of calcium. It is therefore unlikely that I. io eats birds or feathers to overcome a calcium deficiency.

Although feeding on birds is associated with a robust skull, powerful jaws, and large teeth (Freeman 1979, 1984; Swartz et al. 2003), I. io also fed on insect taxa, especially beetles. Other large vespertilionids such as E. serotinus (Vaughan 1997) and E. fuscus (Freeman 1981) also feed mainly on beetles.

Bird feathers and quills are the main constituents in the droppings of bird-eating bats (Dondini and Vergari 2000; IbAnez et al. 2001; Norberg and Fenton 1988). In India, feathers represent the primary evidence for predation on birds, except for 2 samples containing up to 4 small bone fragments. Ibánez et al. (2003) argued that N. lasiopterus fed mainly on the meat-rich flight muscles of birds, meaning that the ingestion of bones would be minimal. During September and October, Dondini and Vergari (2004) found 14 small fragments of bird bones in droppings from 9 N. lasiopterus. The fragments had an average maximum length of only 1.3 mm, implying that their presence can easily be overlooked.

The wing morphology and echolocation characteristics of I. io clearly support the hypothesis that this species chases flying prey in the open and would not hunt birds with a gleaning strategy as done by several other carnivorous bats (Norberg and Fenton 1988). I. io has wings with a high wing loading, average aspect ratio, and a high wing tip index. Wing loading is positively correlated with minimum flight speed and negatively related to maneuverability. The high wing loading implies that I. io a fast flyer with low maneuverability. The average aspect ratio suggests average flight efficiency (Norberg and Rayner 1987). The high wing loading would not allow these bats to carry relatively heavy loads in the way that carnivorous gleaners do, therefore I. io may attack roosting birds. Alternatively, migrating birds may be attacked at high altitude where there are few obstacles to collide with, as proposed for N. lasiopterus (IbAnez et al. 2001). Although I. io is a large bat, its wings are relatively broader than those of N. lasiopterus, resulting in lower wing loading.

During March in India, 50% of I. io fed on birds. November, March, and May were the months when large quantities of birds were found in the droppings, and these are likely migration periods. However, it is difficult to explain why no bird remains were found in droppings collected in April in India when most migration was expected to occur. The main capture of bats in April took place in rainy weather, and this may have reduced the amount of bird migration. Small numbers (<5 pairs) of passerine birds nested in the Phlangkaruh Cave entrance, and no birds were observed roosting in the caves at Guizhou. It therefore seems highly unlikely that birds were being captured at the cave sites. The hypothesis that bats capture feathers drifting in the air due to an inability to discriminate prey from feathers (Bontadina and Arlettaz 2003) is not supported considering the large quantity of feathers ingested. The wing morphology and echolocation call design of I. io suggest that the species is an aerial predator. The lack of roosting birds in the caves supports the hypothesis that birds are captured on the wing.

Phylogenetic diagnosis.—la io was 1st classified as a species of Pipistrellus (Ellerman and Morrison-Scott 1951), but subsequently placed in a separate genus within the Vesperti-lionidae (Topál 1970). Topál (1970) also suggested that Ia was more closely allied with Eptesicus than Pipstrellus based on comparisons of teeth and baculum morphology, a viewpoint supported by Hill and Harrison (1987).Menu (1987) suggested la was synonymous with Eptesicus. Phylogenetic studies by Hoofer and Van Den Bussche (2003) using the mitochondrial 12s and 16s sequences, places traditional Vespertilionini (Pipistrellus-like bats—Tate 1942) into 3 tribes: Nycticeiini (includes, e.g., Eptesicus, Scotomanes, and Nycticeius) Pipis-trellini (includes, e.g., Pipistrellus and Nyctalus) and Vespertilionini (includes, e.g., Chalinolobus and Hypsugo). In our analyses, based on the Cytb and ND1 genes, la is a member of the Nycticeiini, not the Pipistrellini. Our molecular phylogenetic analysis thus supports the view that this taxon is more closely related to Eptesicus than to Pipistrellus. At the same time, we found that la is closely allied with Scotomanes, which is good evidence against the previously proposed synonymy of Eptesicus and la (Menu 1987). Bourret referred to this species as Parascotomanes (cited in Corbet and Hill 1992), presumably because of similarities with Scotomanes, which is confirmed in our genetic analysis. Overall our sequence divergence values were broadly similar for Cytb and ND1 comparisons among taxa, although ND1 values were consistently slightly higher than those at Cytb. Ruedi and Mayer (2001) found similar sequence divergence values for. these 2 genes across a range of bats in the genus Myotis.

In conclusion, we have shown that I. io, like N. lasiopterus, feeds heavily on birds, especially during the migration season. Both bat species are large, and have wing shapes and echo-location calls adapted for detecting aerial prey in open spaces. It seems likely that birds are captured on the wing. The 2 genera belong to different evolutionary lineages within the vespertilionid bats, and evolved this remarkable behavior by convergent evolution.


AT thanks her husband A. Tyler and Mrs. S. Tyler for supporting and sponsoring this work, and Ma boy for use of his vehicle. Mr. O. Laitmon helped in the field, and his family offered assistance. The People of Sheila (Sohlab) supported AT in her work, and W. Pyrbot provided valuable driving assistance. We thank Z. M. Zhou for fieldwork assistance in China, and J. Y. Zeng for help identifying food items. This study was financed by the National Natural Science Foundation of China (grant 30270169) and the National Geographic Society (grant 7806–05) to SZ, a Darwin Initiative grant (14–008) to GJ, and a Joint Project Grant between the Royal Society (London) and the Chinese Academy of Sciences to SZ and GJ. We thank 2 anonymous referees for constructive comments that improved the manuscript.


  • Associate Editor was R. Mark Brigham.

Literature Cited

View Abstract