OUP user menu

Echolocation Calls, Diet, and Phylogenetic Relationships of Stoliczka'S Trident Bat, Aselliscus stoliczkanus (Hipposideridae)

Gang Li , Bing Liang , Yinan Wang , Huabin Zhao , Kristofer M. Helgen , Liangkong Lin , Gareth Jones , Shuyi Zhang
DOI: http://dx.doi.org/10.1644/06-MAMM-A-273R.1 736-744 First published online: 1 June 2007

Abstract

The comparative biology of the hipposiderid genus Aselliscus has been little studied. Here we report studies of echolocation, diet, and phylogeny of Aselliscus stoliczkanus. The phylogenetic relationships of Aselliscus were investigated based on sequence comparisons of mitochondrial cytochrome-b and nicotinamide adenine dinucleotide dehydrogenase subunit 1 genes. Dates of divergence within the hipposiderid radiation also were estimated. The echolocation call frequency of A. stoliczkanus is quite high, with the dominant constant frequency component at 119–120 kHz, and a terminal sweep down to 104.5 kHz. The call duration is about 5.4 ms. The diet of A. stoliczkanus is mainly composed of lepidopterans (78.5%), beetles (14.9%), and hemipteran insects (6.5%) in November. Our results indicate that Aselliscus is monophyletic and is correctly classified in the Hipposideridae, and the divergence time for Aselliscus was estimated at 22 million years ago.

Key words
  • Aselliscus stoliczkanus
  • A. tricuspidatus
  • Chiroptera
  • diet
  • echolocation
  • Hipposideridae
  • phylogeny

The family Hipposideridae comprises 9 currently recognized genera of extant leaf-nosed bats, distributed throughout the Old World from Africa to Australia and Melanesia (Simmons 2005). The hipposiderid genus Aselliscus is found from southern China to the Pacific archipelago of Vanuatu, and is represented by 2 small-bodied (forearm < 45 mm), allopatric species. Stoliczka's trident bat (Aselliscus stoliczkanus (Dobson, 1871)) occurs in southeast Asia, including extreme southeastern China, Myanmar, Thailand, Laos, Vietnam, and the islands of Tioman and Penang fringing the Malay Peninsula (Bates et al. 2000; Corbet and Hill 1992). Temminck's trident bat bat (Aselliscus tricuspidatus (Temminck, 1834)) occurs in eastern Wallacea and throughout Melanesia, including the Moluccas, New Guinea, and associated islands, and in the Bismarck Archipelago, Solomon Islands, and Vanuatu (Corbet and Hill 1992; Flannery 1995a, 1995b; Schlitter et al. 1983). Aselliscus was originally erected by Tate (1941) to accommodate A. tricuspidatus, previously considered a morphologically unique member of the genus Hipposideros by Dobson (1871). Tate (1941) further suggested that A. stoliczkanus (previously classified in the genus Asellia) might also warrant inclusion in Aselliscus, a classification formalized by subsequent reviewers (e.g., Sanborn 1952). The 2 species of Aselliscus are morphologically highly distinctive, easily discriminated on the basis of external, cranial, and dental features (Corbet and Hill 1992). Interestingly, no species of Aselliscus is recorded from the wide intervening area between the geographic ranges of the 2 species (i.e., the Greater Sundas, Sulawesi, and Nusa Tenggara).

The relationships of hipposiderid genera have attracted considerable attention in recent literature (Bogdanowicz and Owen 1998; Jones et al. 2002; Wang et al. 2003) and the family has a rich fossil record from deposits in Queensland, Australia (Hand and Archer 2005; Hand and Kirsch 1998, 2003). Drawing from varying taxon sets and methodologies, various systematists have arrived at strongly divergent interpretations of relationships within the family, and the phylogenetic affinities of Aselliscus remain particularly poorly understood. In his description of the genus, Tate (1941) originally highlighted potential links between Aselliscus and the hipposiderid genera Anthops, Asellia, Triaenops, and Cloeotis, all of which are characterized by a tridentate upper nose-leaf margin. In contrast, cladistic analyses by Hand and Kirsch (1998, 2003), drawing from craniodental characters, have suggested that A. tricuspidatus is a basal lineage within the family, perhaps sister to all other extant and fossil hipposiderids (their analyses did not include A. stoliczkanus). Based on early genetic comparisons, Pierson (1986) even raised the possibility that Aselliscus may be more closely allied to rhinolophids than to other hipposiderids. Further, cladistic analyses of discrete morphological characters have questioned whether the 2 species of Aselliscus truly comprise a monophyletic clade (Jones et al. 2002), or whether Aselliscus may be nested cladistically within the current taxonomic boundaries of Hipposideros (Wang et al. 2003).

In the present paper we rely on sequence data from the mitochondrial Cytochrome-b (Cytb) and nicotinamide adenine dinucleotide dehydrogenase subunit 1 (NDI) genes, widely used in chiropteran phylogenetic studies (e.g., Baker et al. 1994; Hoofer and Van Den Bussche 2001; Hulva and Horacek 2002), to provide an independent test of previously proposed hypotheses regarding the phylogenetic relationship of these little-studied species of Aselliscus, with special reference to A. stoliczkanus. Drawing from our molecular data set, we also estimate selected dates of divergence in the hipposiderid radiation in order to better understand the evolutionary and bio-geographic origins of A. stoliczkanus. We complement these genetic investigations into the diversification of Aselliscus with the 1st detailed discussion of the echolocation calls and diet of A. stoliczkanus. Echolocation call frequencies can be useful phylogenetic characters when used alongside genetic data in understanding the evolutionary history of hipposiderid bats (Guillén-Servent and Francis 2006; Thabah et al. 2006).

Materials and Methods

Sampling.—Individuals of A. stoliczkanus were captured by mistnetting at caves in Sichuan (29°34′N, 103°16′E), Guizhou (25°19′N, 105°05′E), and Yunnan (25°08′N, 102°38′E) provinces in mainland China, in November 2005. Bats were released after 3-mm punches were taken from the wing membrane. A JYT-1 balance (Shanghai Medical Laser Company, Shanghai, China) accurate to 0.1 g was used for weighing bats and vernier calipers accurate to 0.1 mm were used to obtain forearm lengths. Our tissues of A. tricuspidatus are from vouchered specimens (deposited at the Australian Museum in Sydney and the South Australian Museum in Adelaide) collected from a large cave roost in Vatthe Conservation Area, Espiritu Santo, Vanuatu (Helgen 2004). All animals were handled in accordance with guidelines for animal care and use established by the American Society of Mammalogists (Animal Care and use Committee 1998).

Echolocation calls.—Echolocation calls of A. stoliczkanus were recorded in the hand, held approximately 30 cm from the microphone, using a Pettersson D980 bat detector (sampling rate 350 kHz; Pettersson Elektronik AB, Uppsala, Sweden). After downloading 10-times expanded calls onto a PC (soundcard sampling rate 44.1 kHz), the recordings were subsequently analyzed with the software package BatSound Pro, version 3.0 (Pettersson Elektronik AB), using 512 fast Fourier transform and 16-bit precision for the Hanning window (see Li et al. 2006; Zhao et al. 2003). The constant-frequency component of the call was measured from the power peak in the power spectrum. We analyzed 1 call per bat because we measured no intraindividual variation in the frequency of the constant-frequency component. We measured the frequency of most energy in the call, which, typical of hipposiderid bats, was in the 2nd harmonic. We recorded the calls of resting bats because hipposiderid bats can use Doppler shift compensation in flight, whereby they slightly reduce the frequency of calls as their flight speed increases (Hiryu et al. 2005). Recording the frequencies of handheld bats represents a standardized method of recording calls that are not subject to Doppler shift compensation, and is used routinely in analyses of variation in call frequencies of rhinolophoid bats (e.g., Li et al. 2006; Siemers et al. 2005). For logistical reasons we were unable to record the echolocation calls of A. tricuspidatus.

Dietary analysis. —Dietary analysis was undertaken by examining the remains of prey items in fecal pellets from A. stoliczkanus, following methods discussed by Kunz and Whitaker (1983) and Brack and LaVal (1985). Bats were captured after dawn and placed individually into clean cloth bags when they had finished foraging and returned to the roosting cave. Fecal pellets were recovered from the bags and air dried for subsequent laboratory analysis. Individual pellets were analyzed for insect remains by softening the samples in 70% ethanol and teasing them apart under a dissecting microscope, with all the droppings of an individual classified as 1 sample. Insect remains were identified taxonomically to ordinal level. Percentage volume occupied by each insect order was estimated visually to the nearest 5%, and frequency of occurrence of the different categories of prey was estimated for each fecal sample (Whitaker 1988; Zhang et al. 2005). A total of 100 pellets were analyzed from 8 individuals captured in Guizhou.

Molecular data collection.—We used DNeasy Tissue Kits (Qiagen, Shanghai, China) to isolate genomic DNA from wing membrane (A. stoliczkanus) and liver (A. tricuspidatus) tissue samples preserved in 95% ethanol. We amplified and sequenced complete Cytb (1,140 base pairs [bp]) and NDI (957 bp) gene sequences from samples of Aselliscus, and used newly sequenced or previously published sequences for both genes from Coelops frithi (from Taiwan), 4 species of Chinese Hipposideros, and 6 species of Chinese Rhinolophus in our phylogenetic comparisons. We also used some sequences we published previously and downloaded some relevant sequences from GenBank to carry out our phylogenetic investigations of A. stoliczkanus (Appendix I). A megadermatid (Megaderma lyra) and 2 pteropodids (P ter opus scapulatus and Rousettus leschenaulti) were employed as outgroups (Appendix I). Cytb sequences were amplified using the primers L14724 (5′-GGT CTT AGG CAA AAA ATT GGT GCA ACT 0–3′—Kocher et al. 1989), Bat_Cytb_l (5′-TAG AAT ATC AGC TTT GGG TG-3′—Li et al. 2006), and H15915R (5′-TCAGCTTTGGG TGTTGATGG-3′—Irwin et al. 1991). The primers for sequencing were similar to those used for amplifying.

Polymerase chain reaction conditions were as follows: 94°C (5 min); 35 cycles at 94°C (50 s), 50°C (40 s), and 72°C (80 s); 72°C (5 min). Primer pairs for ND1 amplification and sequencing were L16S (5 ′ -CCTCGATGTTGGATCAGG-3 ′) and HtMet (5 ′ -GTATGGGCCCGATAGCTT-3 ′—Cao et al. 1998). The total volume of the polymerase chain reaction mixture was 50 µl, with reagents at a final concentration of 0.4 µM of each primer, 0.2 µM of each deoxynucleoside triphosphate, 1.5 µM MgCl2, and 1 U of Taq DNA polymerase.

Phylogenetic analyses.—We used Modeltest, version 3.6 (Posada and Crandall 1998) to choose the best model of evolution for phylogenetic analyses. The program was used to determine the most appropriate substitution model for the Cytb and ND1 sequence data, respectively, and the optimum maximum-likelihood parameters.

MrBayes 3.1 (Huelsenbeck and Ronquist 2001) and PAUP, version 4.0b (Swofford 2003) were employed to construct the phylogenetic trees. In the control block of each program, the setting for the codon substitution model and maximum-likelihood parameters followed the results of MODELTEST 3.6. The general time reversible model GTR+Γ+I was selected as the most appropriate model of nucleotide substitution (πA =0.350, &#x03C0;c = 0.393, πG = 0.073, πT = 0.185; rAC = 0.400, rAG = 12.885, rAT = 0.556, rCG = 0.250, rCT = 8.730; I = 0.512; α = 0.833). In the Bayesian analyses, 6 Markov chains with 1 million generations were used for simulation. After 400,000 generations, the trees were sampled. Other sets were analyzed according to the options for vertebrate mitochondrial sequences. PAUP, version 4.0b, program used heuristic searches and tree-bisection-reconnection branch swapping options. In PAUP, version 4.0b, we also generated bootstrap values (2,000 replicates) with neighbor-joining and maximum-parsimony methods to test robustness of the tree topologies. We used MEG A3 (Kumar et al. 2004) to calculate the, genetic distances of the different taxa using the Kimura 2-parameter model.

Divergence estimates.—We combined our molecular sequence data with information from the fossil record to estimate divergence times within the genus Aselliscus and among other rhinolophoid bats represented in our data set. We utilized the software packages PAML 3.14 (Yang 1997), EST-BRANCHES, and MULTIDIVITIME (Kishino et al. 2001; Thorne and Kishino 2002; Thorne et al. 1998) for these analyses. For our divergence estimates, we followed Teeling et al. (2005) in using 2 fossil constraints. First, the basal date of divergence for Rhinolophoidea (Hipposideridae and Rhino-lophidae plus Megadermatidae) is held to be no older than 55 million years (Paleocene-Eocene boundary) because no rhinolophoid fossils are known before the middle Eocene (McKenna and Bell 1997; Simmons and Geisler 1998). Second, the date of rhinolophid-hipposiderid divergence is taken to have occurred not less than 37 million years ago (mya) because fossils referable to both Hipposideridae and Rhino-lophidae are known from the middle Eocene (Hand and Archer 2005; Hand and Kirsch 2003; McKenna and Bell 1997; Simmons and Geisler 1998). The split of Megadermatidae from other Rhinolophoidea (Hipposideridae and Rhinolophidae) was placed at 50 mya, and the split between Hipposideridae and Rhinolophidae at 40 mya (Teeling et al. 2005).

Results

Echolocation calls and diet.—The echolocation call frequency of A. stoliczkanus from Sichuan and Guizhou is high, with a constant-frequency component at 120.3 ± 0.3 kHz (n = 10), and a terminal frequency-modulated sweep down to 104.5 ± 2.1 kHz. The call duration is 5.4 ± 0.3 ms. The echolocation call frequency of bats from Yunnan is a little lower (1 individual called with the constant-frequency component at 118.4 kHz and a 2nd calledat 119.3 kHz; Fig. 1).

Fig. 1

Oscillogram (top), sonogram (middle), and power spectrum (below) of typical calls from Aselliscus stoliczkanus recorded in Yunnan in November 2005.

The diet of A. stoliczkanus is mainly composed of lepi-dopterans, beetles, and hemipterans (Table 1). Lepidopterans were the most abundant food items in the samples (79% of the diet in volume), followed by coleopterans (15%), hemipterans (7%), and odonates (<1%).

View this table:
Table 1

Diet composition of Aselliscus stoliczkanus. Data represent percent volume (%) of total diet represented by each insect group (n = 100 fecal samples).

Insect ordersCurrent studyFeng (2001)
Lepidoptera7843
Coleoptera1529
Hemiptera6
Odonata<1
Diptera14
Trichoptera5
Hymenoptera2
Unidentified7

Phylogeny.—Based on the combined sequences, the neighbor-joining, maximum-parsimony, and Bayesian phylogenies were identical in topology but slightly different in branch support (Fig. 2). Within the bounds of our taxon sampling, the different reconstruction methods each supported several main phylogenetic conclusions: each tree supported the generic monophyly of Aselliscus, grouping A. stoliczkanus and A. tricuspidatus as sister lineages; all trees clustered Aselliscus and Coelops into a single lineage with high bootstrap test values (92 for neighbor-joining and 94 for maximum-parsimony) and a posterior probability value of 0.94; and Hipposideridae and Rhinolophidae constituted monophyletic sister clades. Relationships among different species of Rhinolophus included in our study were not always well resolved and are not considered further here. Table 2 presents the genetic distances among species sequenced in this study. Sequence diverge values at the 2 genes are broadly similar. Three individuals of A. stoliczkanus from Guizhou and Sichuan provinces of China have 5–6% sequence divergence compared with the 2 bats that were sampled in Yunnan Province. Sequence divergence between A. stoliczkanus and A. tricuspidatus is between 14% and 16%. The table also showed comparatively large divergence values among Aselliscus and other genera in the Rhinolophoidea, Hipposideros, and Rhinolophus at both genes.

View this table:
Table 2

Sequence divergence matrix based on complete mitochondrial Cytb (1,140 bp, above the diagonal) and NDI (957 bp, below the diagonal) gene sequences for 2 species of Aselliscus and partial species of Hipposideros, Rhinolophus, and outgroup (Megaderma lyra, Rousettus leschenaulti, and P ter opus scapulatus). MEG A3 was used to calculate the genetic distances based on the Kimura 2-parameter model.a

123456789101112131415161718192021
1000.060.0620.160.1630.1390.2020.1940.1990.1970.1890.1960.1930.2130.2380.20.2560.2860.267
2000.060.0620.160.1630.1390.2020.1940.1990.1970.1890.1960.1930.2130.2380.20.2560.2860.267
3000.060.0620.160.1630.1390.2020.1940.1990.1970.1890.1960.1930.2130.2380.20.2560.2860.267
40.0560.0560.0560.0020.1590.1620.1510.2040.1870.1890.1840.1950.1990.1850.2110.2370.1990.2450.2640.251
50.0560.0560.05600.1590.1620.1510.2040.1870.1870.1840.1950.1990.1880.2080.2340.1990.2450.2640.248
60.1420.1420.1420.1510.1510.0020.1560.1980.1960.1830.2030.2040.2110.210.220.2190.2230.2560.2870.258
70.1420.1420.1420.150.150.0040.1570.1990.1980.1850.2040.2070.2140.2130.220.2220.2230.2570.2880.261
80.160.160.160.1610.1610.1580.1580.1740.1730.1690.1870.1830.2010.1790.1930.1920.180.2460.2650.234
90.1820.1820.1820.1840.1840.1640.1640.1720.0880.1330.180.1970.2040.190.210.2080.190.260.2890.272
100.1940.1940.1940.1870.1870.1770.1760.1710.0930.1290.1490.2040.2080.2070.2120.2180.2070.2520.2830.264
110.1710.1710.1710.1730.1730.1510.1510.1720.1220.1150.1640.1960.2080.1810.2050.20.1890.2340.2620.245
120.1740.1740.1740.1870.1870.1610.1610.1640.1560.1720.1550.2120.2060.2090.2240.2330.2180.2540.2850.278
130.1710.1710.1710.190.190.1740.170.1720.190.190.1670.1860.1450.1120.1420.1620.1510.2470.2680.251
140.1960.1960.1960.2050.2050.1830.1780.1890.1970.1990.1890.2020.1110.1490.1590.1810.1580.2690.2640.255
150.1990.1990.1990.1980.1980.1870.1850.1790.1870.1830.1940.1890.1210.1210.1490.1670.1350.2570.2530.244
160.180.180.180.1890.1890.1850.180.1770.1870.1830.1840.20.1170.1230.1210.1640.1370.2670.2360.246
170.1760.1760.1760.1850.1850.1740.1740.1850.190.1960.20.1840.1320.130.1350.1260.1680.2590.2470.261
180.2060.2060.2060.2050.2050.1810.1790.1780.1910.190.1910.1870.1360.1310.140.1250.1290.2650.2510.234
190.230.230.230.2260.2260.220.2160.2360.2330.2330.2170.2360.2010.2130.2090.220.2180.2150.2810.265
200.2480.2480.2480.2520.2520.2480.2460.2230.2410.240.2210.2460.2430.2350.2430.2440.2440.2520.2350.205
210.2240.2240.2240.2350.2350.210.2080.2170.2040.210.1950.2150.2030.2160.2160.2150.2250.227"0.2360.175
  • a 1, Aselliscus stoliczkanus Guizhou; 2, A. stoliczkanus Guizhou; 3, A. stoliczkanus Sichuan; 4, A. stoliczkanus Yunnan; 5, A. stoliczkanus Yunnan; 6, A. tricuspidatus; 7, A. tricuspidatus; 8, Hipposideros larvatus; 9, H. armiger; 10, H. larvatus; 11, H. pratti; 12, H. pomona; 13, Rhinolophuspusillus; 14, R. pearsonii; 15, R. affinis; 16, R.ferrumequinum; 17, R. luctus; 18, R. hipposideros; 19, Megaderma lyra; 20, Rousettus leschenaulti; 21, P ter opus scapulatus.

Fig. 2

Maximum likelihood tree (ln-likelihood = −15,283 [likelihood value]) based on the combined mitochondiral DNA analysis (Cytb plus ND1) fox Aselliscus, other hipposiderids (Coelops and Hipposideros), rhinolophids (Rhinolophus), and 3 outgroups (Megaderma, Pteropus, and Rousettus). Numbers at each node are neighbor-joining bootstrap values, maximum-parsimony bootstrap values, and posterior probability values given by Bayesian analyses, respectively. The bar at the bottom of the phylogenetic tree is a scale bar representing substitutions per site.

Molecular dating.—We estimate the earliest divergences represented in our sampling of hipposiderids at 30 mya, whereas the deepest split within the family Rhinolophidae is estimated at 20 mya (Fig. 3). Interspecific splits within Hipposideros range from 6 to 20.5 mya. We estimate that Coelops diverged from ancestral Aselliscus at approximately 22 mya, and that within Aselliscus, the split between A. tricuspidatus and A. stoliczkanus dates to approximately 20 mya.

Fig. 3

Estimated timescale (in millions of years; means with ranges showing standard errors) for diversification of selected rhinolophoid taxa based on our combined mitochondrial DNA analysis with the imposition of 2 fossil constraints (see text). The x axis represents millions of years ago (mya).

Discussion

Echolocation and diet.—Aselliscus fly at low speeds and are very small-bodied, roosting in caves and foraging in cluttered microhabitats (Feng 2001; Lekagul and McNeely 1977; McKean 1972). Their low wing loading (as in most rhinolophid and hipposiderid species) lends flexibility when hunting for prey in a complicated environment. According to our results, the echolocation calls of A. stoliczkanus are of the typical hipposiderid constant-frequency-frequency-modulated type, characterized by high frequency and short call duration.

Bats of the families Hipposideridae and Rhinolophidae generally forage in forested areas, catching insects either aerially or by gleaning off foliage or the ground in these narrowed, cluttered environments (Bogdanowicz et al. 1999; Denzinger et al. 2004; Schnitzler and Kalko 1998). Previous investigations into the diets of hipposiderids and rhinolophids have demonstrated that moths and beetles dominate in the diets of these bats, and are selected in larger proportions than available in the local environment (Bowie et al. 1999; Churchill 1994; Gom et al. 2004; Jones 1990; Jones et al. 1993; Pavey and Burwell 2000, 2004). Our results show that A. stoliczkanus primarily consumes lepidopterans and coleopterans. Research carried out at the same sites where we collected A. stoliczkanus revealed that the percentage of Coleoptera in the diets of sympatric Hipposideros armiger, H. pratti, and H. larvatus was 42%, 43%, and 38%, respectively, and for Lepidoptera, 22%, 26%, and 31%, respectively (Feng 2001). Our results on A. stoliczkanus obtained in November 2005 are quite different from those reported by Feng (2001) obtained in June 2000 at the same site, although the relative rankings of Lepidoptera and Coleoptera were the same. We suspect that annual or seasonal variation in the availability of various insects might explain these differences.

Because we did not intensively study the relative proportion of biomass of different insect orders at this site, it is impossible to determine whether selection of certain insects by these hipposiderids was disproportionate to their occurrence in the landscape as a whole. Because diets of insectivorous bats can be highly plastic, varying with local environment, seasonality, and food resource availability (Kunz 1982), it is not surprising that the diet of A. stoliczkanus recorded in our study is somewhat different from that recorded by Feng (2001). Our results are consistent with the limited findings of Nabhitabhata (1986), who found moth remains in the stomachs of all 3 bats examined in Thailand, and Diptera in 1 stomach. No comparative data are yet available regarding the diet of A. tricuspidatus (Bonaccorso 1998).

Phylogenetic affinities and divergence date estimates.— Sequence divergence values between A. stoliczkanus sampled in Guizhou and Sichuan versus Yunnan provinces was relatively high (5–6%). However, these values were considerably lower than the divergence between A. stoliczkanus and A. tricuspidatus. In combination with the absence of echolocation call frequency differences among Chinese populations, the sequence divergence estimates suggest that Chinese A. stoliczkanus may represent geographic races, rather than distinct species, given that cryptic species of hipposiderid bats usually diverge in call frequency (Guillén-Servent and Francis 2006; Thabah et al. 2006).

Each of our trees (maximum-likelihood and Bayesian) supported the sister-relationship of Aselliscus and Coelops, suggesting that Coelops is more closely related to Aselliscus than to Hipposideros. Recently, the supertrees of Jones et al. (2002) questioned the monophyly of A. stoliczkanus and A. tricuspidatus, but our trees showed the 2 species of Aselliscus truly comprised a monophyletic group. Our results also reject the idea that Aselliscus is a basal lineage within Hipposideridae (Hand and Kirsch 1998, 2003), that it may be linked phy-logenetically with rhinolophids (Pierson 1986), or that it is nested within Hipposideros (Wang et al. 2003).

In support of traditional taxonomic arrangements (e.g., Koopman 1994) and more recent molecular assessments, examination of our data supports the hypothesis that rhinolophids and hipposiderids are monophyletic sister lineages (Hutcheon et al. 1998; Jones et al. 2002; Levasseur et al. 2003; Springer et al. 2003; Teeling et al. 2002, 2003). Our molecular divergence estimates indicate that Aselliscus is an old genus. The split between A. stoliczkanus and A. tricuspidatus is estimated at 20 mya, which would indicate that the 2 species diverged from each other in the early Miocene. The well-supported topology of our phylogenetic trees, with Aselliscus as sister to Coelops (a generic lineage endemic to eastern Asia and the Sunda Shelf), strongly indicates a mainland Asian origin for Aselliscus. We suggest that the split between A. stoliczkanus (today endemic to eastern Asia and Indochina) and A. tricuspidatus (with distribution centered on New Guinea) ultimately reflects a dispersal event from Asia to emergent areas of Melanesia, perhaps (if our molecular dating is accurate) in the early Miocene, when New Guinea is thought to have comprised a series of small, discrete islands separated from Australia (Aplin et al. 1993; Flannery 1995a). Whatever its precise biogeographic history, A. tricuspidatus is quite likely to be the most ancient endemic rhinolophoid lineage present in Melanesia today, along with the endemic nominal hipposiderid genus Anthops (Flannery 1995a).

Our molecular sampling included 4 species of Hipposideros that are often classified in different “species-groups” within the genus (Hill 1963; Koopman 1994). Our estimates suggest that divergences within the genus date back 20 million years— a similar time frame for divergences within Aselliscus. The oldest known fossil occurrence of Hipposideros is from the Oligocene of Africa, and fossils attributed to Hipposideros are recorded from the Miocene of South Africa and are abundant in the Miocene record of Riversleigh, Australia (Hand and Archer 2005; Hand and Kirsch 1998). Accordingly, our molecular divergence estimates within the genus are compatible with current knowledge regarding the fossil record. Similarly, Rhinolophus species included in our sampling are classified in several different species-group within Rhinolophus; of these, R. luctus is generally classified in the trifoliatus group, considered by some reviewers to be among the most plesiomorphic lineages of horseshoe bats (Bogdanowicz 1992; Guillén-Servent et al. 2003), an interpretation consistent with our results, which indicate luctus to be the most basal lineage sampled. Our results also suggest that the basal split in Rhinolophus occurred in the same time range as the origin and initial diversification of Hipposideros and Aselliscus—that is, about 20 mya, a date likewise consistent with fossil data and previous biogeographic interpretations (Bates et al. 2004; Guillén-Servent et al. 2003).

Acknowledgments

We thank K. L. Hu and Z. M. Zhou in China and A. and M. L. Simpson, L. E. Helgen, and J. Erskine in Vanuatu for fieldwork assistance and logistical support. We thank P. Tavue, S. Tavue, B. Tavue, F. Hickey, D. Kalfatak, and E. Bani of the Environment Unit of Vanuatu for permitting, facilitating, and assisting with fieldwork in Vatthe Conservation Area, Vanuatu. This study was financed by the National Geographic Society (grant 7806–05) and Zijiang Scholarship of East China Normal University to SZ, a Darwin Initiative grant (14–008) to GJ, support from the South Australian Museum and the Smithsonian Institution to KMH, and a Joint Project Grant between the Royal Society (London) and the Chinese Academy of Sciences. Constructive comments from 2 referees improved the text.

Appendix I

View this table:

Collection localities of the bats analyzed, with the corresponding GenBank accession numbers.

Accession no.
SpeciesCollection localityCytbND1
Aselliscus stoliczkanusYunnan, ChinaDQ888668DQ888648
A. stoliczkanusYunnan, ChinaDQ888670DQ888660
A. stoliczkanusSichuan, ChinaDQ888673DQ888654
A. stoliczkanusGuizhou, ChinaDQ888676DQ888665
A. stoliczkanusGuizhou, ChinaDQ888677DQ888667
A. tricuspidatusEspiritu Santo, VanuatuDQ888675DQ888652
A. tricuspidatusEspiritu Santo, VanuatuDQ888679DQ888657
Coelops frithiiDayuanshan, Renting, TaiwanDQ888674DQ888666
Hipposideros armigerGuizhou, ChinaDQ297585DQ888663
H. larvatusGuangdong, ChinaDQ888672DQ888653
H.prattiGuangxi, ChinaDQ297584DQ888651
H. pomonaYunnan, ChinaDQ888671DQ888662
Rhinolophus pearsoniiSichuan, ChinaDQ297587DQ888664
R. ajfinisGuizhou, ChinaDQ297582DQ888661
R. hipposiderosUpper Langford, United KingdomDQ297586DQ888658
R. ferrumequinumYunnan, ChinaDQ297575DQ888656
R. pusillusHubei, ChinaDQ297583DQ888655
R. luctusHubei, ChinaDQ297596DQ888659
Megaderma lyraYunnan, ChinaDQ888678DQ888650
Rousettus leschenaultiYunnan, ChinaDQ888669DQ888649
Pteropus scapulatusAustraliaNC_002619NC_002619

Footnotes

  • Associate Editors were Jesús E. Maldonado and Robert J. Baker.

Literature Cited

View Abstract