OUP user menu

A New Species of Barbastella (Chiroptera: Vespertilionidae) from North China

Jin-Shuo Zhang, Nai-Jian Han, Gareth Jones, Liang-Kong Lin, Jun-Peng Zhang, Guan-Jian Zhu, Da-Wei Huang, Shu-Yi Zhang
DOI: http://dx.doi.org/10.1644/07-MAMM-A-114R2.1 1393-1403 First published online: 1 December 2007


A new species of Barbastella is described, originally discovered in 2001 in Beijing, northern China. The description of the new species is based on both morphological and molecular data. The morphology of the skull and ears of the new bat is more similar to that of the Egyptian barbastelle (B. leucomelas) and B. barbastellus distributed in Europe than to B. leucomelas found in southern China and Taiwan. Projections and notches occur along the posterior margin of each ear, and a small lobe (vaulted process) protrudes from the middle outer edge of each pinna. The skull and body size of the new species are larger than in B. leucomelas. Echolocation calls were of 2 types, a brief frequency-modulated call that was alternated with longer calls with a convex frequency-time course. The calls were very similar to those of B. barbastellus recorded in Europe, although they may be slightly lower in frequency. Molecular phylogenies were reconstructed from cytochrome-b (Cytb) and nicotinamide adenine dinucleotide dehydrogenase subunit 1 (ND1) gene sequences. Cladograms of ND1 indicated that barbastelles from the Beijing area form a monophyletic group, which is the sister to B. leucomelas from Egypt. The clade including the new species and Egyptian barbastelle clusters with B. barbastellus, but not with B. leucomelas from Sichuan, Taiwan, and Japan. The genetic distances (corrected Kimura 2-parameter) between Barbastella sp. nov. and most bats from other localities (including all B. barbastellus) were 14.31–17.69% at the ND1 gene and 15.01–17.36% at the Cytb gene. However, ND1 divergence is 12.79% between Barbastella sp. nov. and B. leucomelas from Egypt. All these results support the hypothesis that the barbastelle from Beijing is a new species. Additionally, because Egypt is the type locality of B. leucomelas, the paraphyletic nature of B. leucomelas suggests that barbastelles from Sichuan, Taiwan, and Japan—which are currently classified as B. leucomelas darjelingensis—should not be considered conspecific with B. leucomelas.

Key words
  • Barbastella
  • China
  • molecular phylogeny
  • morphology
  • new species

Barbastelle bats, genus Barbastella (Vespertilionidae, Chiroptera), currently comprise 2 species: the western barbastelle (B. barbastellus (Schreber, 1774)) and the eastern barbastelle (B. leucomelas (Cretzschmar, 1826)—Nowak 1999; Simmons 2005). B. barbastellus is found in the western Palearctic, from the United Kingdom and western Europe through to the Caucasus Mountains, Bulgaria, Turkey, Crimea (Ukraine), Morocco, Mediterranean islands, and Canary Islands (where it exists as an endemic subspecies B. barbastellus guanchaeJuste et al. 2003; Trujillo et al. 2002). B. leucomelas is widely distributed from the Sinai (Egypt), Eritrea, northern Iran, and the Caucasus to Afghanistan, the Pamirs, India, Nepal, western China, Japan, and possibly Indo-China (Corbet 1978; Nowak 1999; Simmons 2005). B. leucomelas is considered to comprise 2 subspecies—B. l. leucomelas (Sinai, Egypt, northern Iran, and possibly Eritrea) and B. l. darjelingensis (rest of range— Corbet 1978). B. barbastellus and B. leucomelas are reported as being sympatric in the Caucasus (Corbet 1978).

The eastern and western barbastelle species are often distinguished by morphological differences: B. barbastellus has a prominent projecting lobe of the posterior margin of the ear (Corbet 1978), but B. leucomelas lacks the lobe. However, Hackethal et al. (1988) suggested that because of morphological and geographical variation in B. barbastellus, the projecting lobe may not be accurately used to distinguish barbastelle species. B. leucomelas is often described as a larger species, having a greater forearm length (41–45 mm in B. leucomelas and 31–43 mm in B. barbastellus) and condylobasal length (14.2–15.0 mm in B. leucomelas and 12.4–14.1 mm in B. barbastellus) than B. barbastellus. B. leucomelas also has a wider baculum (0.40–0.57 mm) than has B. barbastellus (0.27–0.30 mm—Rydell and Bogdanowicz 1997). However the measurements of B. l. leucomelas from the Middle East are smaller than those of Asian B. leucomelas darjelingensis (forearm length 37.3–39.2 mm in Egypt—Dietz 2005), and the taxonomic identity of bats from the Middle East remains unclear (Rydell and Bogdanowicz 1997). Indeed, bats now considered as B. leucomelas were 1st described as Vespertilio leucomelas Cretzschmar, 1826, from Sinai and Plecotus darjelingensis Hodgson, in Horsfield, 1855, from northern India (synonyms = Vespertilio leucomelas Cretzschmar, 1830–1831; Barbastella blandfordi Bianchi, 1917; and Barbastella darjelingensis Dobson, 1875), and were only later combined into a single species by Tate (1942) and Ellerman and Morrison-Scott (1951).

Uchida and Ando (1972) analyzed the karyotypes of Taiwanese barbastelles and found that they were identical to those of the Japanese barbastelles. Lin et al. (1997) 1st used the name Barbastella formosanus for barbastelles from Taiwan, but without providing an adequate description of these bats as a new species. Lin et al. (2002) designated the Barbastella of Taiwan as a new record of B. leucomelas, which was classified as B. l. darjelingensis through morphological comparison. Subsequently, the name Barbastella leucomelas instead of B. formosanus was presented in the 2nd edition of Bats of Taiwan (Lin et al. 2004). Therefore, the Latin name Barbastella formosanus is nomen nudum (also see Simmons 2005). Yoshiyuki (1989) determined that the average ear length of the Japanese B. leucomelas (15.3 mm) was shorter than that of the Indian B. leucomelas (19.0 mm) and suggested that they may be distinct species because of rather invariant dimensions of the ear. Horáček et al. (2000) also suggested that the western subspecies of B. leucomelas (i.e., B. l. leucomelas) may be conspecific with B. barbastellus, and that Japanese populations may be distinct at the species level.

Mitochondrial DNA genes, such as cytochrome b (Cytb) and nicotinamide adenine dinucleotide dehydrogenase subunit 1 (ND1) are inherited maternally and evolve more rapidly than nuclear genes. These genes are used frequently to study the evolutionary histories of closely related species and investigate the phylogenetic relationships among different animals at species and higher levels (Baker et al. 1994; Li et al. 2006; Ruedi and Mayer 2001). Employing molecular methods permits the detection of cryptic species that have become reproductively isolated but resemble each other morphologically. Morphological distinctions, nevertheless, are still the most convenient means for the identification of new species.

In this paper, we describe a new species of barbastelle through morphological characters and echolocation calls, and perform a molecular phylogenetic analysis of this genus based on Cytb and ND1 gene sequences.

Materials and Methods

Bat sampling and morphological comparisons.—On 14 August 2001, a female barbastelle was captured with a mist net from the San-Qing Cave (39°45′N, 115°45′E) at Wang-Lao-Pu Village, Fangshan District, southwestern Beijing, and the echolocation calls of this bat were recorded. The bat was photographed and later released. On 4 November 2002, we captured 1 hibernating male barbastelle bat at the same cave, and the specimen was kept (specimen IOZ-BRG00065; Bat Research Group, Institute of Zoology, Chinese Academy of Sciences). Another male specimen (IOZ-BRG00054) of Barbastella was caught with a mist net in an abandoned tunnel around the Darwin Bat Research and Conservation Center at San-Liu-Shui Village (39°43′N, 115°45′E), Fangshan District, on 25 September 2003. Both specimens were preserved in 75% ethanol (2 with extracted skulls). On 21 August 2006, 4 individuals of Barbastella were found in the same tunnel. They were set free after measurements were taken and wing-membrane sampling was performed (collection nos. 20060821A0197-20060821A0200). One specimen (IOZ-BRG-FLW007) of B. leucomelas from Sichuan (= Szechwan) Province (preserved in 75% ethanol) and 2 specimens (THU12920(7184) and THU6977 (12454)) of B. leucomelas from Tunghai University of Taiwan (preserved as dry skins with skulls) also were examined.

Captured bats were weighed to the nearest 0.1 g, and a set of 14 external measurements was taken for each specimen. The external measurements were taken to the nearest 0.1 mm with dial calipers. A set of 15 cranial and dental measurements was taken in the laboratory to the nearest 0.01 mm with dial calipers.

The following external measurements were taken: body mass, total length (head and body), tail length, forearm length, ear length, ear width, tragus length, tragus width, foot length (including claws, measured to the distal part of claw), tibia length, calcar length, and length of the metacarpals of the 3rd, 4th, and 5th digits.

The following cranial and dental measurements, followed by abbreviations in parentheses, were taken: greatest length of skull (GLS), condylobasal length (CBL), condylocanine length (CCL), braincase breadth (BB), braincase height (BH; height of braincase posterior to the auditory bullae), zygomatic width (ZW), least interorbital breadth (IOB), rostrum length (RL; rostral length from preorbital foramina to the alveolus of the inner incisor), rostrum width (RW; rostral width at the level of the preorbital foramina), auditory bullae length (ABL; length of the skull at the level of the auditory bullae), mandible length (ML), C-M3 length (C-M3), c-m3 length (C-M3), C-C width (C1-C1), and M3-M3 width (M3-M3).

We compared morphological characters of the specimens we examined with published values for both B. barbastella and B. leucomelas (Harrison and Bates 1991; Rydell and Bogdanowicz 1997; Trujillo et al. 2002).

Recording and analysis of echolocation calls.—We recorded the calls from 1 bat that we captured from the San-Qing Cave on 14 August 2001, released into open space. The calls were recorded using a Pettersson D-980 bat detector (Pettersson Elektronik, Uppsala, Sweden) at 10 × expansion onto a Sony TC-3 DAT recorder (Sony, Tokyo, Japan). The calls were analyzed using BatSound version 3.31 (Pettersson Elektronik). We measured call durations and pulse intervals from waveforms, and frequency measurements from spectrograms except for the frequency of most energy, which was measured from the peak of the power spectrum. Spectral analysis was performed with a 1,024-point fast Fourier transform, using a Hanning window.

Molecular data collection.—We took 3-mm biopsy punches from wing membranes of 4 live individuals (collection nos. 20060821A0197-20060821A0200, but only 2 were used for analysis in view of their absolutely identical nucleotides) and 2 specimens (IOZ-BRG00065 and IOZ-BRG00054) from Beijing, 3 from Taiwan (not the 2 dry specimens mentioned in the previous context), and 1 voucher specimen (IOZ-BRG-FLW007) from Sichuan. The punches were preserved in 99% ethanol and stored at 4°C. The handling of all bats conformed to guidelines for animal care and use established by the American Society of Mammalogists (Gannon et al. 2007) and all voucher specimens are deposited at the Institute of Zoology, Beijing, at present. Some published mitochondrial sequences also were obtained from the GenBank database, including B. leucomelas from Japan and Egypt, and B. barbastellus from much of its range—Canary Islands (Spain), Turkey, Switzerland, Morocco, Germany, Hungary, and Greece (Juste et al. 2003; Kawai et al. 2002; Mayer et al. 2007; Mayer and von Helversen 2001). Plecotus auritus and Plecotus austriacus were used as outgroups because they represent the larger clade of plecotine bats to which Barbastella is the sister (Bogdanowicz et al. 1998). See sampling localities and GenBank accession numbers in Table 1 for details.

View this table:
Table 1

—The collection localities of the bats analyzed, with the corresponding GenBank accession numbers. The sequences obtained from GenBank are AF401365 and AF401376 (Mayer and von Helversen 2001); AB079816 (Kawai et al. 2002); AF513745, AF513749, AF513752, and AF513753 (Juste et al. 2003); AF513771 (Juste et al. 2004); DQ915030-DQ915032 (Mayer et al. 2007); and AY665169 and AY699874 (Tsytsulina et al. 2004). A dash indicates that the corresponding sequences were unsuccessfully amplified or not yet unavailable in GenBank. ND1 = nicotinamide adenine dinucleotide dehydrogenase subunit 1; Cytb = cytochrome b.

GenBank accession no.
Bbei1.BeijingBarbastella from BeijingSan-Liu-Shui Village, Fangshan District, Beijing, ChinaEF534767EF534760Biopsy
Bbei2.BeijingBarbastella from BeijingSan-Liu-Shui Village, Fangshan District, Beijing, ChinaEF534768EF534761Biopsy
Bbei3.BeijingBarbastella from BeijingWang-Lao-Pu Village, Fangshan District, Beijing, ChinaEF534769EF534762IOZ-BRG00065
Bbei4.BeijingBarbastella from BeijingSan-Liu-Shui Village, Fangshan District, Beijing, ChinaEF534770IOZ-BRG00054
Bleu. SichuanB. leucomelasSichuan, ChinaEF534774EF534766IOZ-BRG-FLW007
B leu 1. Tai wanB. leucomelasTaiwan, ChinaEF534772EF534763Biopsy
Bleu2.TaiwanB. leucomelasTaiwan, ChinaEF534771EF534764Biopsy
Bleu3.TaiwanB. leucomelas.Taiwan, ChinaEF534773EF534765Biopsy
Bleu.JapanB. leucomelasJapanAB079816
Bleu.EgyptB. leucomelasEgyptDQ915030
Bbar.GreeceB. barbastellusGreeceDQ915031
Bbar.HungaryB. barbastellusHungaryDQ915032
Bbar.GermanyB. barbastellusGermanyAF401376
Bbar.TurkeyB. barbastellusThrace, TurkeyAF513753
Bbar.SpainB. barbastellusCanary Islands, SpainAF513745
Bbar.SwitzerlandB. barbastellusValais, SwitzerlandAF513749
Bbar.MoroccoB. barbastellusAzrou, MoroccoAF513752
PlecotuslPlecotus auritusAY699874AY665169
Plecotus2Plecotus austriacusAF401365AF513771

We used DNeasy Blood & Tissue Kits (QIAgen, Inc., Basel, Switzerland) to extract mitochondrial genomic DNA. ND1 and Cytb genes were amplified separately using 2 pairs of primers as follows: L16S (5′-CCT CGA TGT TGG ATC AGG-3′) and HtMet (5′-GTA TGG GCC CGA TAG CTT-3′—Cao et al. 1998) for the ND1 gene, and Bat_Cytb_l (5′-TAG AAT ATC AGC TTT GGG TG-3′— Li et al. 2006) and Bat_Cytb_2 (5′-AAA TCA CCG TTG TAC TTC AAC-3′—G. Li, pers. comm.) for the Cytb gene.

Polymerase chain reaction profiles for ND1 and Cytb sequences were the same: 1st denaturation at 95°C for 5 min, followed by 35 cycles at 95°C (30 s for denaturing), 55°C (30 s for annealing), and 72°C (80 s for extending), with the final extension at 72°C for 10 min. Each 50-μl polymerase chain reaction cocktail included 25 μl of 2 × ExTaq polymerase (TAKARA, Inc., Shiga, Japan), 1 μl of each primer (10 μM), and 2 ul of DNA templates (50 μg/μl). The polymerase chain reaction products were purified using the Agarose Gel DNA Purification Kit version 2.0 (TAKARA, Inc.) and sequenced in both directions with polymerase chain reaction primers using Big-Dye Terminator version 3.1 and an ABI3730 automated DNA sequencer (Applied Biosystems, Inc., Foster City, California) by the Sanger method. Base calling and quality trimming of the sequences were carried out using Phred (Ewing et al. 1998). The 2 overlapping fragments of the ND1 or Cytb gene for each individual were compiled using AssemblyLIGN 1.0.9 software (Oxford Molecular Group PLC 1998).

Phylogenetic analyses.—After acquiring the sequences of ND1 and Cytb genes, we aligned them respectively using ClustalX1.81 (Thompson et al. 1997), and edited them manually after a final alignment adjusted by eye. Sequence variation and divergence were calculated by the program MEGA3.1 (Kumar et al. 2004)—in order to be comparable with the database in Baker and Bradley (2006), we chose the Kimura 2-parameter model (Kimura 1980) to calculate sequence divergence for both genes.

The program PAUP 4.0*b10 (Swofford 2002) was used to construct a maximum-likelihood tree, with the best-fit model identified by MODELTEST 3.6 (Posada and Crandall 1998) and the following executive parameters: heuristic search =10 replicates, random addition of taxa, and tree-bisection-recon-nection branch-swapping. Robustness of the topology was assessed with 100 bootstrap replicates. In contrast, 2,000 bootstrap replicates were carried out to assess robustness of the neighbor-joining (with best-fit model distances) and the maximum-parsimony trees obtained with PAUP 4.0*bl0 (Swofford 2002). We constructed separate trees for Cytb and ND1 sequences because some bats could not have their DNA sequenced for both genes because of tissue degradation, and because Cytb and ND1 sequences on GenBank did not always come from the same individual.


Molecular phylogenetic analysis.—We did not obtain a Cytb sequence from 1 Beijing specimen (Bbei4: the holotype, IOZ-BRG00054) because no genomic DNA could be extracted, probably because of tissue degradation. Fortunately, a complete ND1 sequence was obtained soon after the bat was captured. We obtained 7 sequences (4 ND1 sequences and 3 Cytb sequences) from 4 individuals (Bbeil-Bbei4; Table 1). The mitochondrial Cytb and ND1 gene sequences of barbastelle bats are available in GenBank (accession nos. EF534760-EF534774; Table 1).

After sequence alignment and manual editing, we finally obtained ND1 sequences of 795 base pairs (bp) and Cytb sequences of 680 bp, where our data overlapped with published sequences. The program MODELTEST3.6 (Posada and Crandall 1998) selected HKY85+G (base frequencies: A = 0.3349, C = 0.2903, G = 0.1280, T = 0.2468; transition : transversion [Ts/Tv] = 7.5088; gamma shape parameter = 0.2654) and K81uf+G (base frequencies: A = 0.3009, C = 0.2991, G = 0.1270, T = 0.2731; gamma shape parameter = 0.2085) as the best-fit models by hierarchical Likelihood Ratio Tests for the ND1 gene and Cytb gene, respectively.

All phylogenetic trees produced display similar topologies (Figs. 1A and 1B). The genus Barbastella is divided into 2 major groups. One major group includes barbastelles from Sichuan and Taiwan, with intraspecific genetic distances at the ND1 gene of 0.38–1.27% and at the Cytb gene of 0.29–0.59%, indicating their monotypic status. This clade presumably represents bats currently described as B. leucomelas darjelingensis. Barbastella from Beijing and B. barbastellus form the 2nd major clade. Individuals from the same species cluster together, forming separate clades, with genetic distances at the ND1 gene of 14.31–15.31% and at the Cytb gene of 16.90–17.36% (Table 2), separating Barbastella from Beijing from B. barbastella. The intraspecific ND1 sequence divergence of B. barbastellus from Germany, Hungary, and Greece varies between 0.38% and 1.14%, whereas the intraspecific Cytb sequence divergence varies between 0.74% and 2.40% from Morocco, Switzerland, and Turkey. The divergence of the subspecies B. barbastellus guanchae from Canary Islands (Spain) with the mainland barbastelles is larger (3.80–5.69%) than divergence values seen among the mainland barbastelles at the Cytb gene (Juste et al. 2003). As for intraspecific variation in Barbastella from Beijing, the pairwise genetic distances of the 3 individuals (Bbeil-Bbei3) are 0% at partial ND1 and Cytb sequences studied, so we consider them as 1 haplotype. Obviously, Bbei4 is another haplotype, with an ND1 sequence divergence of 1.40% between the 2 haplotypes.

Fig. 1

A) Ciadogram without branch lengths by integrating neighbor-joining, maximum-parsimony, and maximum-likelihood trees based on nicotinamide adenine dinucleotide dehydrogenase subunit 1 (ND1) gene sequences. B) Cladogram without branch lengths by integrating neighbor-joining, maximum-parsimony, and maximum-likelihood trees based on cytochrome b (Cytb) gene sequences. Three numbers at the nodes separated by “/” from left to right are bootstrap values following neighbor-joining (2,000 replicates), maximum-parsimony (2,000 replicates), and maximum-likelihood (100 replicates) approaches, respectively. An asterisk (*) indicates 3 approaches obtained the same bootstrap values. Black squares indicate that the interior topologies of the 3 clades are disparate by 3 methods despite high bootstrap values. The number “0” in the ND1 cladogram tree indicates that the Japanese clade in maximum-likelihood tree locates at other position but with a poor bootstrap support (see text for details). Species abbreviations are defined in Table 1.

View this table:
Table 2

—Corrected genetic divergences of group means (Kimura 2-parameter model) based on partial mitochondrial nicotinamide adenine dinucleotide dehydrogenase subunit 1 sequences (823 bp, below the diagonal) and partial cytochrome-b sequences (680 bp, above the diagonal), calculated by MEGA version 3.1 (Kumar et al. 2004).

Barbastella from BejingB. leucomelas, SichuanB. leucomelas, TaiwanB. leucomelas, JapanB. leucomelas, EgyptB. barbastellus, GreeceB. barbastellus, HungryB. barbastellus, GermanyB. barbastellus, SwizerlanB. barbastellus, MoroccorB. barbastellus, TurkeyB. barbastellus, Island, Spain
Barbastella from
B. leucomelas, Sichuan0.17290.00590.15920.15330.15860.1627
B. leucomelas, Taiwan0.17690.00590.15810.15220.15750.1616
B. leucomelas, Japan0.1690.14350.1474
B. leucomelas, Egypt0.12790.17880.17940.1872
B. barbastellus, Greece0.14480.15970.16480.18340.1461
B. barbastellus,
B. barbastellus,
B. barbastellus,
B. barbastellus,
B. barbastellus, Turkey0.0569
B. barbastellus, Canary
Islands, Spain

Although we obtained only 1 ND1 sequence for Egyptian and Japanese B. leucomelas, their genetic divergence to the other barbastelles and positions in the phylogenetic trees are still remarkable. The Egyptian haplotype has a smaller divergence with B. barbastellus (14.44–15.07%) than with B. l. darjelingensis (17.88–18.72%), and divergence of 12.79% exists between the Egyptian sample and Barbastella from Beijing. These relationships contradict the current classification and are also reflected in the phylogenetic trees—the Egyptian haplotype is nested in the group that includes B. barbastellus and Barbastella from Beijing with firm support from all 3 phylogenetic methods. Moreover, it appears as sister to Barbastella from Beijing with high bootstrap support in the neighbor-joining and maximum-parsimony trees, although the maximum-likelihood tree has a bootstrap value of only 57.

The Japanese haplotype also has a large ND1 divergence from other barbastelles: 16.90% with Barbastella from Beijing, 18.72% with B. leucomelas from Egypt, 18.17–19.01% with B. barbastellus, and 14.35–14.74% with B. leucomelas from Sichuan and Taiwan. It is sister to the Sichuan-Taiwan B. leucomelas with bootstrap support of 88 on the neighbor-joining tree and 67 on the maximum-parsimony tree (Fig. 1A). On the maximum-likelihood tree, however, the Japanese haplotype forms 1 branch of a basal trichotomy within Barbastella, the other branches corresponding to B. I. darjelingensis from Sichuan-Taiwan and the barbastelles from Europe, Egypt, and Beijing. The phylogenetic analysis thus supports the hypothesis that the barbastelle from Beijing is an undescribed species, which is named and compared with other congeners below.

Barbastella beijingensis, new species

Beijing Barbastelle

Holotype.—Institute of Zoology, Chinese Academy of Sciences, IOZ-BRG00054, adult male in alcohol with skull extracted, collected by Drs. Jie Ma and Li-Biao Zhang on 25 September 2003. GenBank accession number for ND1 is EF534770.

Type locality.—Darwin Bat Research and Conservation Center at San-Liu-Shui Village, Fangshan District, Southwestern Beijing, China (39°43′N, 115°45′E), 407.8 m above sea level. We captured the specimen in an abandoned tunnel (approximately 1 km long, 3.5–4 m high) in this village.

Paratype.—Institute of Zoology, Chinese Academy of Sciences, IOZ-BRG00065, adult male in alcohol with skull extracted, collected by Dr. Hui-Hua Zhao and Jin-Shuo Zhang on 4 November 2002.

Referred material.—One specimen (IOZ-BRG-FLW007, male) of B. leucomelas from Sichuan (= Szechwan) Province, was deposited at the Bat Research Group, Institute of Zoology, Chinese Academy of Sciences. Two specimens (THU12920(7184), male and THU6977(12454), female) of B. leucomelas from Taiwan were preserved at the Department of Life Science, Tunghai University.

Distribution and habitat.—So far known only from the type locality and a nearby cave. We found the new barbastelle species 4 times in Fangshan District, about 100 km southwest of Beijing. San-Liu-Shui Village is in a mountainous region with riparian woodland, and the roost site was an abandoned tunnel, more than 1 km in length. Other caves also were important diurnal roost sites. The surrounding vegetation is classified as warm temperate zone forest. Most of the native forest consists of Chinese pine (Pinus tabulaeformis), arbor-vitae (Sabina chinensis), and oak (Quereus mongolica and Q. liaotungensis). In this region at least 4 bat species are sympatric with the Beijing barbastelle: Rhinolophus ferrume-quinum, Myotis ricketti, Myotis blythii, and an unidentified Murina in light of our field surveys. In 2001 and 2002, we also captured 2 barbastelles at Wang-Lao-Pu Village, Fangshan District (39°45′N, 115°45′E).

Etymology.—We nominate this species using the name of the type locality, Beijing (formerly Peking), capital city of the People's Republic of China.

Diagnosis.—This is a relatively small bat among species of the order Chiroptera in China, but relatively large in the genus of Barbastella (forearm length 41.1 mm), with a large skull (greatest length of skull [GLS] 15.7 mm) and a long condylobasal length (CBL 14.5 mm). Interorbital breadth is relatively wider than in B. leucomelas (IOB 3.9 mm). Therefore, its skull measurements fall within the range reported for B. leucomelas rather than those of B. barbastellus. The posterior margin of the ear has projections and notches. There is a small lobe (vaulted process) protruding from the outer edge of each ear, which is less conspicuous than that in many B. barbastellus. This species can be distinguished from the other species of Barbastella based on the ND1 and Cytb sequences.

Description.—The dorsal fur is dark black with brown-gray tips, the ventral fur is lighter than the dorsal pelage. The flat and wide muzzle has pronounced glandular swellings. A few long whiskers and dense hairs line the margin of the upper and lower lips. The relatively large nasal aperture is broad. Ear length is 15.5 mm; the ears are brownish black and have transverse ridges. The outline of the ears is nearly square; as in many B. barbastellus there is a slender and delicate projecting lobe or vaulted process on the outer edge of the pinna. The ears are forward facing and join across the forehead. The tragus is triangular and large, and is more than half the height of the pinna (see Fig. 2). In contrast to B. leucomelas, the skull is relatively large with a condylocanine length of 14.3 mm. The upper surface of the rostrum is smooth, slightly concave, and the rostrum is relatively wider than in B. leucomelas. The postdental extension is poorly developed, with a blunt median spine. The coronoid process of each half mandible is short. The condyle of each mandible is narrow. The angular process projects for a relatively large distance (see Fig. 3).

Fig. 2

Barbastella beijingensis (IOZ-BRG00054), male. Photo by Shu-Yi Zhang.

Fig. 3

Ventral and lateral views of the skull of Barbastella beijingensis (IOZ-BRG00054), male, with lateral and dorsal views of the lower jaw.

Upper toothrow length (C-M3) is 4.7 mm. The 1st upper incisor (12) is bicuspidate and the 2nd upper incisor (13) is very small. The canine is slender with a well-developed cingulum but lacks secondary cusps. The 1st upper premolar (P2) is very small between the canine and the 2nd upper premolar (P4) and is displaced inward. P4 is large and attains two-thirds the height of the canine. P4 is wider than the 1st upper molar (Ml). There are no hypocones on Ml and M2 and the mesostyle is weaker than the parastyle and metastyle. M3 consists of 3 commissures and a metacone. In the lower dentition, the 3 incisors are overlapping. The well-developed anterior cingular cusp of the weak lower canine is higher than the 3rd incisor (i3). The lower canine (c) is higher than p4. p2 is very small, about one-third the height of p4. p4 is similar in height to ml (see Fig. 3).

Echolocation calls.—The bat alternated 2 signal types. One call type followed a convex frequency-time course (n = 10 all data: start frequency 42.7 ±1.6 kHz, end frequency 25.1 ±1.4 kHz, frequency of most energy 39.4 ± 0.7 kHz, duration 8.2 ± 1.7 ms, pulse interval 99.0 ± 28.0 ms). This call type was interspersed with brief, frequency-modulated signals with a lower frequency of most energy (n = 6 all data: start frequency 39.2 ±1.6 kHz, end frequency 26.8 ± 0.5 kHz, frequency of most energy 32.1 ±1.9 kHz, duration 5.1 ± 0.8 ms, pulse interval 72.2 ± 6.2 ms). Representative calls are illustrated in Fig. 4.

Fig. 4

Spectrogram of consecutive echolocation calls produced by Barbastella beijingensis. The spectrogram was made using a 1,024-point fast Fourier transform and a Hanning window. A brief frequency-modulated signal is followed by a longer call with a convex frequency-time course.

Comparisons.—Barbastella beijingensis can be distinguished from B. leucomelas of South China and Taiwan by its larger body size (Table 3) and by its projecting ear lobes (vaulted processes). The ear lobes of this new species are delicate and slender, and do not look as buttonlike as in many western barbastelles. Ear lobes were not present in specimens of B. leucomelas from Sichuan and Taiwan that were examined. Interestingly, although B. leucomelas is larger than B. barbastella (Rydell and Bogdanowicz 1997), B. beijingensis is larger than B. leucomelas. The size of the skull of B. beijingensis is relatively larger than that of B. leucomelas from Taiwan (Table 4). Measurements of p4 from the holotype and paratype are higher than that of the Taiwan specimen—the reason might be that the Taiwan specimen may be an old individual with worn premolars. In contrast to B. leucomelas from Taiwan and Iran (see Bates and Harrison 1997), the canine and P4 of this new species are stronger and better developed. We propose that size of P4 might prove to be a diagnostic dental character in a larger sample of bats. For bat identification, the fur color is not a consistent character to identify different species because of variation related to age and geography. However, the pelage color of Beijing's barbastelle is quite similar to that described for the Egyptian barbastelle (both of them are browner than that of B. barbastellus, see Harrison and Bates 1991). Bats captured in Beijing in 2006 appeared much darker than those captured previously, although all bats captured were adults. Although we did not have access to voucher specimens or sequences from India, Nowak (1999) and Simmons (2005) stated that B. I. darjelingensis is distributed in southwestern China, suggesting that barbastelle bats from Sichuan are the same subspecies as B. leucomelas from India, and this conclusion is strongly supported through comparisons of their morphological data (Tables 3 and 4; Lin et al. 2002). Meanwhile, although the barbastelle from Japan was suggested as a distinct species from B. I. darjelingensis (Horáček et al. 2000; Yoshiyuki 1989), we still considered it as B. I. darjelingensis in our analysis. We note that cranial characters of the new species are quite similar to the eastern and western barbastelles, but body size and projecting ear lobes are keys to distinguishing and identifying B. beijingensis.

View this table:
Table 3

—External measurements, to nearest 0.1 mm, of the holotype and paratype as well as other specimens of Barbastella beijingensis sp. nov. (length in mm, mass in g). Additionally, data from B. leucomelas from Sichuan and Taiwan are presented, together with information collated by Rydell and Bogdanowicz (1997) for B. barbastellus, B. barbastellus guanchae from the Canary Islands (Trujillo et al. 2002), and B. leucomelas from Israel and Sinai (Harrison and Bates 1991).

B. beijingensis sp. nov.B. leucomelas from Sichuan and TaiwanB. l. leucomelasB. barbastellusB. b. barbastellusB. b. guanchae
ParameterHolotype IOZ-BRG00054Paratype IOZ-BRG0006520060821A019720060821A019820060821A019920060821A0200IOZ-BRG-FLW007THU12920 (7184)THU6977 (12454)Cited by Harrison and Bates 1991Cited by Rydell & Bogdanowicz 1997Cited by Trujillo et al. 2002Cited by Trujillo et al. 2002
SexMaleMaleFemaleFemaleFemaleMaleMaleMaleFemaleMales and females
Body mass11.913.913.010.55.6-13.7
Total length52.649.753.453.754.154.341.151.547.945-60
Tail length47.032.740.027.041.733.2–51.036-52
Ear length15.514.615.015.913.115.410.512.811.918.0
Ear width12.89.811.98.4
Tragus length6.
Tragus width3.
Foot length7.–6.8
Tibia length19.123.118.418.119.2
Calcar length5.
3rd digit
4th digit
5th digit
View this table:
Table 4

—Cranial and dental measurements, to nearest 0.1 mm, of the holotype and paratype of Barbastella beijingensis sp. nov. and B. leucomelas and B. barbastellus (length in mm).

B. beijingensis sp. nov.B. leucomelas from TaiwanB. l. leucomelasB. barbastellusB. b. barbastellusB. b. guanchae
ParameterHolotypeParatypeTHU12920 (7184)THU6977 (12454)Cited by Harrison and Bates 1991Cited by Rydell and Bogdanowicz 1997Cited by Trujillo et al. 2002Cited by Trujillo et al. 2002
Greatest length of skull (GLS)15.715.214.614.914.613.4–14.413.9–14.2
Condylobasal length (CBL)14,514.213.714.013.412.4–14.112.8–13.512.8–13.6
Condylocanine length (CCL)14.313.913.213.4
Braincase breadth (BB)–7.2
Braincase height (BH)
Zygomatic width (ZW)–7.57.0–8.27.2–7.77.2–7.7
Interorbital breadth (IOB)–3.9
Rostrum length (RL)
Rostrum width (RW)
Auditory bullae length (ABL)
Mandible length (ML)–9.5


Integrated analysis is important for recognizing cryptic species by using phenotypic, phylogenetic, and other biological data (e.g., echolocation calls and karyology) when there is significant genetic divergence among taxa but relatively little morphological divergence or vice versa. For example, integrated techniques have been used on morphologically similar pipistrelles in Britain (Barratt et al. 1997; Jones and Van Parijs 1993), long-eared bats (Plecotus) in Europe (Kiefer et al. 2002; Spitzenberger et al. 2006), and house bats (Scotophilus) in South Africa (Jacobs et al. 2006). Barbastelle bat populations at different sites have morphological similarities, but large genetic differences of mitochondrial DNA. In this paper we use molecular data to support assignment of species status to B. beijingensis. The echolocation calls of B. beijingensis are very similar to those of B. barbastellus, which also alternated a brief frequency-modulated signal with a longer call with a convex frequency-time course (Denzinger et al. 2001). The frequency values we recorded are similar to those described for B. barbastellus in Europe (Denzinger et al. 2001; Parsons and Jones 2000; Russo and Jones 2002). The echolocation calls of B. leucomelas from Egypt start at 40 kHz and end at 28 kHz for the short (3- to 4-ms) frequency-modulated type, and span 44–27 kHz for the longer call type (8-12 ms—Dietz 2005), values very similar to those we recorded for B. beijingensis. Although barbastelle bats can be readily identified to genus by their distinctive calls, we doubt whether echolocation call parameters will offer diagnostic features for individual species.

Baker and Bradley (2006) argued that species are genetically isolated and should show consistent levels of sequence divergence in Cytb comparisons. For bats, they documented levels of variation at Cytb between 0.2% and 3.8% within populations, 0% and 5.9% within species, 4.8% and 18.7% among species within genera (nonsister species), and 2.3% and 14.7% between sister taxa. In our study, in addition to the Cytb gene, we also chose the ND1 gene as a robust supplementary gene, in consideration of less published molecular data for Cytb gene than ND1 gene, and of their approximately identical rate of evolution in vespertilionid bats (Ruedi and Mayer 2001). The congruence between slight morphological differences and substantial sequence divergence of at least 5% indicates that distinct species may be present.

In our research, the phylogenetic trees constructed by 2 protein coding genes reveal that B. beijingensis clusters with B. barbastellus (also with B. I. leucomelas in view of ND1 gene) rather than with B. I. darjelingensis (Figs. 1A and 1B). The sequence divergences between B. beijingensis and European B. barbastellus (not including B. b. guanchae) are almost 3–4 times those between B. b. guanchae and B. b. barbastellus in southern Europe and Morocco (Juste et al. 2003).

The original proposal that B. leucomelas from Sinai (Egypt) and B. darjelingensis from northern India represent separate species (Horáček et al. 2000) may be correct after all, and therefore the appropriate names for the taxa would be B. leucomelas, type locality in Sinai, Egypt, and B. darjelingensis, type locality in Darjeeling, India. The latter, we also speculate, may contain cryptic species as indicated by the Japanese Barbastella in our study, having an ND1 genetic distance of 14.35-14.74% with the bats from Sichuan and Taiwan, which appears compatible with the findings of Yoshiyuki (1989) and Horáček et al. (2000). However, we withhold firm conclusions because of inadequate data and no voucher specimens of Indian and Japanese barbastelles.

In the late summer of 2006, we caught 4 individuals of B. beijingensis from the same colony in a fissure in the tunnel at San-Liu-Shui Village. We estimated there were 15 or more bats in the fissure because of audible sounds of bats and, upon our approach, 2 bats flew out from the fissure. We had previously captured 1 in 2001, 1 in 2002, and another in 2003 in the same tunnel and in a cave in Fangshan District. DeBlasé (1980) suggested B. leucomelas is generally a nonsocial species roosting and hunting singly or, in the former United Soviet Socialist Republic, roosting in groups of no more than 3 individuals. We hypothesize that the ecology of B. beijingensis might be different from that of B. leucomelas or B. barbastellus, and therefore further study is required. In Europe B. barbastella typically roosts in trees during the summer (Russo et al. 2004), (2005).

Allen (1938) suggested that B. leucomelas is not a common bat species throughout much of its range. Bates and Harrison (1997) noted that there have been no recent population assessments, but outlined the potential threat that deforestation could pose. B. leucomelas was assessed as Lower Risk/Least Concern in 1996 and 2003 listings of the Red List of IUCN and Lower Risk/Least Concern in the IUCN/SSC Action Plan (Hutson et al. 2001). In the China Species Red List, Wang and Xie (2004) listed B. leucomelas as Vulnerable. We conducted many field surveys in Fangshan, Beijing, from 1997 to 2006 and found B. beijingensis only 4 times, suggesting that it might be rare.

In conclusion, barbastelles show morphological similarities but considerable genetic divergence over their range. Barbastelles around Beijing are more similar genetically to European B. barbastellus than they are to Asian B. leucomelas, but are sufficiently different genetically to warrant specific status. Further sampling of populations between Beijing and western Europe will be needed to determine whether the Beijing bats are isolated, or part of a cline of increasing genetic divergence extending eastward, and whether any natural divisions occur between B. leucomelas and B. barbastellus over the species' ranges.


We are indebted to the Darwin Initiative for support in establishing the Bat Research and Conservation Center at Fangshan District, Beijing, where we conducted this fieldwork. We are grateful to Mr. P.-Y. Hua for his help with some DNA sequencing. Many thanks to Ms. Y.-J. An and Mr. C.-Z. Liu for drawing the pictures of skull and Mr. A. Chmura for corrections to English in an early draft of the manuscript. We appreciate anonymous reviewers and the associate editor, C. W. Krajewski, as well as the editor, E. J. Heske, for their helpful comments and reviews. We also thank the American Society of Mammalogists Nomenclature Committee for suggestions on standardized description of new species. This study was supported by a Darwin Initiative grant (14-008) to G. Jones and a grant (7806-05) from the National Geographic Society as well as Zijiang Scholarship of East China Normal University to S-YZ.


  • a Jin-Shuo Zhang and Nai-Jian Han contributed equally to this paper.

  • Associate Editor was Carey W. Krajewski.

Literature Cited

View Abstract