Open Access
How to translate text using browser tools
1 September 2004 Acoustic Identification of Eight Species of Bat (Mammalia: Chiroptera) Inhabiting Forests of Southern Hokkaido, Japan: Potential for Conservation Monitoring
Dai Fukui, Naoki Agetsuma, David A. Hill
Author Affiliations +

Assessing the impact of forest management on bat communities requires a reliable method for measuring patterns of habitat use by individual species. A measure of activity can be obtained by monitoring echolocation calls, but identification of species is not always straightforward. We assess the feasibility of using analysis of time-expanded echolocation calls to identify free-flying bats in the Tomakomai Experimental Forest of Hokkaido University, Hokkaido, northern Japan. Echolocation calls of eight bat species were recorded in one or more of three conditions: from hand-released individuals, from bats flying in a confined space and from bats emerging from their roost. Sonograms of 171 calls from 8 bat species were analyzed. These calls could be categorized into 3 types according to their structure: FM/CF/FM type (Rhinolophus ferrumequinum), FM types (Murina leucogaster, Murina ussuriensis, Myotis macrodactylus and Myotis ikonnikovi) and FM/QCF types (Eptesicus nilssonii, Vespertilio superans and Nyctalus aviator). Sonograms of the calls of R. ferrumequinum could easily be distinguished from those of all other species by eye. For the remaining calls, seven parameters (measures of frequency, duration and inter-call interval) were examined using discriminant function analysis, and 92% of calls were correctly classified to species. For each species, at least 80% of calls were correctly classified. We conclude that analysis of echolocation calls is a viable method for distinguishing between species of bats in the Tomakomai Experimental Forest, and that this approach could be applied to examine species differences in patterns of habitat-use within the forest.


Disturbances of foraging habitats may have serious effects on populations of insectivorous bats (e.g. Vaughan et al., 1996; Law et al., 1999; Racey and Entwistle, 2003). Bats use a variety of habitats for foraging (e.g. open space, forest edges, cluttered space, above water, Schnitzler and Kalko, 2001; Schnitzler et al., 2003), and their requirements may vary, not only between species (Swift and Racey, 1983; Arlettaz, 1999; Vaughan et al., 1997a; Russo and Jones, 2003), but also between seasons, and between geographical areas (Racey, 1998). For conservation planning to be effective, it is essential to clarify the type, or types, of habitat that each bat species typically forages in over the course of a year, so that provision can be made to protect them.

Methods that have been used to study the distribution and patterns of habitat use by bats include mist-netting (e.g. Kuenzi and Morrison, 2003; Ciechanowski, 2002), radio-tracking (e.g. Arlettaz, 1999) and observation of individuals fitted with reflective tape (e.g. Swift and Racey, 1983). All of these methods involve capturing bats, however, which will inevitably disturb their activity. Moreover, the height at which bats typically forage varies between species (Aldridge and Rautenbach, 1987; Norberg and Rayner, 1987; Neuweiler, 1989; Fenton, 1990; Schnitzler and Kalko, 1998; Schnitzler and Kalko, 2001; Schnitzler et al., 2003), and capture of bats that typically fly at high altitude is very difficult. Consequently, there have been very few examples of studies in which bats were captured at heights of more than a few metres above ground level, such as in or above the forest canopy (Kalko and Handley, 2001).

Recently, the use of ultrasonic bat detectors has become increasingly common in studies of bat activity (e.g. Krusic et al., 1996; Ciechanowski, 2002). Using bat detectors it is possible to monitor the activity of species that normally fly above the canopy, as well as those below it, provided that the calls they give are loud enough. If we can distinguish between species using ultrasonic detectors, it should be possible to assess species differences in patterns of habitat use. Subjective identification of species from their calls can be extremely difficult and its accuracy is highly dependent on the experience and ability of the observer (see Parsons et al., 2000). An alternative way of identifying calls of free-flying bats is to record them using time-expansion bat detectors, measure various call parameters from sonograms, and then categorize the calls on the basis of those parameters (e.g. Vaughan et al., 1997b; Parsons, 2001; Rydell et al., 2002). In order to do this, it is necessary to have a reference collection of calls of individuals of known species. These can then be analysed statistically to establish whether combinations of call parameters follow patterns that are species-specific. Methods that have been used to do this include discriminant function analysis (DFA), which has been used to identify bats in several studies in Europe and North America (Zingg, 1990; Obrist, 1995; Krusic and Neefus, 1996; Vaughan et al., 1997b; Murray et al., 1999; Parsons and Jones, 2000; Russo and Jones, 2002; Rydell et al., 2002), and synergetic pattern recognition algorithms performed by artificial neural networks (Parsons, 2001; Parsons and Jones, 2000).

The structure of echolocation calls varies, not only between species, but also within species. A variety of factors have been shown to influence call structure, including geographical variation, foraging habitat and foraging mode (Aldridge and Rautenbach, 1987; Norberg and Rayner, 1987; Thomas et al., 1987; Fenton, 1990; Barclay and Brigham, 1991; Obrist, 1995; Barclay et al., 1999). This means that reference calls recorded in a particular region, or in a particular habitat type, may not be applicable to other regions, or to other habitat types. Therefore, as far as possible, species identification methods should be developed in the region and habitat type where they are to be used.

The aim of this study was to develop a method for assessing interspecific variation in patterns of habitat-use by bats in the Tomakomai Experimental Forest, Hokkaido, Japan. In this paper, we examine characteristics of echolocation call structure for eight species that are found in forests in the region. We then examine the feasibility of identifying these species from their echolocation calls using discriminant function analysis. This is the first study to present an analysis of interspecific differences in echolocation calls in a community of Japanese bats.


Study area

The main study area was Tomakomai Experimental Forest (TOEF: 42°43′N, 141°36′E), a research facility of Hokkaido University located near Tomakomai City in Hokkaido Prefecture, Japan. TOEF covers 2715 ha and consists of mature deciduous forest, secondary deciduous forest, and coniferous plantations, each of which occupies about 1/3 of the total area. Dominant tree species in the natural forests are Quercus crispula, Kalopanax pictus, Fraxinus mandshurica, Acer spp., and Betula spp. The coniferous plantations consist mainly of Picea jezoensis and Abies sachalinensis.

Five species of bat have been caught in previous surveys in the forest: Rhinolophus ferrumequinum, Myotis macrodactylus, Myotis ikonnikovi, Murina leucogaster and Murina ussuriensis (Fukui et al., 2001). In addition, Vespertilio superans, Nyctalus aviator and Eptesicus nilssonii are suspected be present, although they have not been caught. These species tend to fly above the forest canopy (Abe et al., 1994), and so are difficult to catch in nets set from ground level. However, all 3 species are common in Hokkaido, and one carcass of each of Vespertilio superans and Nyctalus aviator has been collected in the study area (Fukui, unpublished data). Thus, we assumed that these three species are also likely to be part of the bat community at TOEF. To obtain calls for these species, we visited roosts at other sites in the same region. Calls of V. superans and N. aviator were recorded at Centennial Woods Park in Kutchan Town (42°54′N, 140°46′E), where several roosts of each species are known (Bat Research Group of Centennial Woods Fan Club, Kutchan, 2002). The park is approximately 70 km from TOEF and consists of 27 ha of mixed woodland, largely Populus maximowiczii and Alnus japonica, surrounded by agricultural land. The calls of E. nilssonii were recorded at a roost in Kyogoku Town (42°49′N, 140°53′E), about 60 km from TOEF, where the roost is in the wall of a potato storehouse surrounded by fields.

Recording methods

We captured Rhinolophus ferrumequinum, Myotis macrodactylus, Myotis ikonnikovi, Murina leucogaster and Murina ussuriensis in TOEF using mist nets and harp traps from 2001 to 2003 (Capture permit number from Ministry of Environment of Japan, 2001: 2-3∼5, 2002: 2-3∼6, 2003: 2-39∼42). We checked their sex, maturity and reproductive condition, and measured their forearm length using calipers and body weight using a balance. To avoid recording the call of the same individual twice, we tagged them with numbered aluminium bands (2.9 mm, Lambournes Ltd., Leominster, England) on their forearms.

Calls were recorded using a Pettersson D240 bat detector (Pettersson Elektronik AB, Uppsala, Sweden) linked to a digital audio tape recorder (TCD-D100, SONY, Tokyo, Japan). The D240 was set to time-expansion mode, in which it records 1.75 s of ultra-sound, slows the signal down ten times to bring it into the audible range, and plays it back. This technique preserves all characteristics of the original sound and allows accurate measurement of acoustic parameters. Captured bats were initially released inside a mosquito net (8 m3) set on the forest road, and their echolocation calls were recorded as they flew inside the net. For Rhinolophus ferrumequinum, calls were also recorded when the bats hung from the roof of the mosquito net. There were two reasons for recording bats in the mosquito net. First, calls of both Murina species are very weak, which sometimes meant that we were unable to record any calls when they were released from the hand. Second, for some species the structure of echolocation calls given in open spaces is quite different from those given in cluttered areas. The confined space of the mosquito net was intended to simulate proximity to clutter. Once bats had been recorded in the mosquito net, they were released by hand, and the calls produced on release were also recorded. All bats were released on the forest road (width > 5 m) near the point of capture. Recordings were made with the detector held at the same height, and about 2 m from the released bat.

Calls of V. superans, N. aviator and E. nilssonii were recorded from June to August 2003. Recordings were made as the bats left their roosts using the same equipment as in TOEF. The bats usually emitted loud, broadband calls immediately after emergence, which could be recorded from bats that were 20–50 m away from the roost exit.

Analysis of echolocation calls

The structure of recorded calls was analyzed using Bat Sound 3.1 software (Pettersson Elektronik AB, Uppsala, Sweden) with a sampling rate of 44.1 kHz and a Hanning window. FFT size was 512 for sonograms and 4096 for power spectra. One pulse of an echolocation call was selected at random from each bat individual in each recording situation (in the mosquito net, on hand released and during roost emergence) for analysis. We measured the following parameters from each pulse: duration (D), start frequency (SF, frequency at the start of the pulse), end frequency (EF, frequency at the end of the pulse), peak frequency (PF, frequency of maximum energy of the pulse) and center frequency (MF, frequency of highest energy taken at half pulse duration). Interpulse interval was also measured from the sequence of pulses for each individual, in each recording situation. Because the bats did not usually emit pulses at even intervals, we measured the longest interpulse interval (IPI-L) and shortest interpulse interval (IPI-S) in the sequence of echolocation calls. D, IPI-L and IPI-S were measured from oscillograms, PF from power spectra, and all other parameters from spectrograms.

Statistical procedures

For the five species for which calls were recorded from captured bats, multivariate analysis of variance (MANOVA) was used to test for differences between calls recorded in the mosquito net and on hand release. Peak frequency, duration and bandwidth (SF minus EF) were compared using Scheffe's test when significant difference had been confirmed by one-way analysis of variance (ANOVA) on calls of all species, except R. ferrumequinum.

R. ferrumequinum emits FM/CF/FM calls typical of rhinolophids (e.g. Vaughan et al., 1997b), while all other species produced FM or FM/QCF calls. As sonograms of calls of R. ferrumequinum could be easily distinguished from others by eye, they were omitted from the main statistical analyses. Multivariate discriminant function analysis (DFA) was applied to call parameters of the remaining seven species. Quadratic analyses were used because Box's M test showed that covariance matrices were not homogeneous (p<0.001). Quadratic discriminant analyses were performed using all seven parameters. Subsequently, the analyses were performed using only the five parameters (SF, EF, PF, MF and D) that could be taken from a single pulse, to assess the degree to which species discrimination would be possible in cases where only one pulse was recorded. Wilk's lambda values were obtained with a MANOVA to test for statistical significance of DFA models. MANOVA was performed with StatView 5.0 for Macintosh, and other tests were performed with SYSTAT 10 for Windows. In all tests, values of p<0.05 were considered statistically significant.


In total, 171 echolocation calls were recorded from 129 individuals of eight species. Of these, 40.4% were recorded on hand-release, 33.9% in the mosquito net, and 25.1% during roost emergence (Table 1).

Table 1

Species recorded and number of bats recorded at each situation


Characteristics of echolocation call

Calls of each species were categorized into FM/CF/FM, FM and FM/QCF type, as defined by Schnitzler and Kalko (2001).

FM/CF/FM type

Rhinolophus ferrumequinum produced typical FM/CF/FM echolocation calls, i.e. calls with a long, strictly constant-frequency component (CF) preceded and followed by a brief, frequency-modulated (FM) sweep (Fig. 1). Peak frequency in the calls of this species was 65.0 kHz (Table 2). R. ferrumequinum was the only species with FM/CF/FM type calls in the TOEF.

Fig. 1

Sonogram of echolocation calls of R. ferrumequinum.


Table 2

Descriptive statistics for time and frequency parameters in echolocation calls of 8 species of bats. Table shows mean and minimum–maximum of parameters. SF, start frequency; EF, end frequency; PF, frequency of maximum energy; MF, middle frequency; D, duration; IPIS, shorter inter pulse interval; IPI-L, longer inter pulse interval.


FM type

All four species of the genera Myotis and Murina emitted FM calls (Fig. 2). The peak frequency of Murina ussuriensis was the highest of the four species (86.3 kHz; Table 2, Scheffe's test, p<0.05). There was little difference between the peak frequencies of other three species. Duration of calls of the two species of Myotis were longer than those of the two species of Murina (Scheffe's test, p<0.05). Both species of Murina had broader bandwidth than the two Myotis (Scheffe's test, p<0.05). In all species, interpulse interval was very variable (see Table 2). MANOVA showed that the only significant differences between hand-release calls and mosquito net calls were for My. macrodactylus, (Wilk's lambda=0.20, F7,11=6.3, p<0.05). MF for this species in the mosquito net was higher than that of hand released calls (ANOVA, F1,17=6.2, p<0.05), and D, IPI-S and IPI-L in mosquito net were shorter than those of hand-released bats, respectively (D: F1,17=4.5, p<0.05; IPI-S: F1,17=21.2, p<0.05; IPI-L: F1,17=9.1, p<0.05).

Fig. 2

Sonograms of echolocation calls of Myotis macrodactylus, Myotis ikonnikovi, Murina leucogaster and Murina ussuriensis.


FM/QCF type

Calls of Eptesicus nilssonii, Vespertilio superans and Nyctalus aviator were all defined as FM/QCF type. These pulses had two components: they began with steep frequency modulation (FM), and moved to shallow frequency modulation (quasi-constant frequency, QCF) in latter part of the pulse (Fig. 3). Calls of E. nilssonii, V. superans and N. aviator had lower peak frequencies (30.5, 24.2 and 21.1 kHz respectively, p<0.05 by Scheffe's test), longer durations (6.3, 6.2 and 12.0 ms respectively, p<0.05) and shorter bandwidth (31.8, 26.4 and 18.6 kHz respectively, p<0.05) than the other four FM type species (Table 2). Among FM/QCF species, for duration and bandwidth, there were no differences between E. nilssonii and V. superans (Scheffe's test, p=0.09, 0.99). Peak frequencies between each species were significantly differ (Scheffe's test, p<0.05). Calls of N. aviator had longer duration and broader bandwidth than other two FM/QCF type species (Scheffe's test, p<0.05). These calls showed considerable variation in duration (D), interpulse interval (IPI), MF and SF (Table 2).

Fig. 3

Sonograms of echolocation calls of Vespertilio superans, Nyctalus aviator and Eptesicus nilssonii. Two calls of each species are included, in order to show some intraspecific variation.


Discriminant function analysis

Quadratic discriminant function analysis using all 7 parameters resulted in 92% of 171 records of echolocation calls being correctly classified to one of the seven species (Table 3). MANOVA showed that the model was significant (Wilk's lambda=0.001, F42,688=62.0, p<0.05). The first 3 discriminant functions explained 99.9% of total variation. For four species classification was 100%. Ten of the 13 cases of misclassification involved two species, My. ikonnikovi and Mu. leucogaster (Table 3). For the preceding analysis, 45 bats contributed more than one data point, as they were recorded both in the mosquito net and on hand release. When the analysis was re-run with only one call taken for each individual bat (n=126), 85% of calls were correctly classified.

Table 3

Summary of classification of 7 species of bats by discriminant function analysis. Model relied on 7 parameters (SF, EF, PF, MF, D, IPI-S and IPI-L). Overall correct classification rate was 92%.


In the canonical score plot, the cluster of Mu. ussuriensis was distinct from clusters of other species (Fig. 4). The other three FM type species (My. macrodactylus, My. ikonnikovi and Mu. leucogaster) showed some overlap. Clusters of 3 FM/QCF type species were contiguous, but with very little overlap (Fig. 4).

Fig. 4

Canonical score plots of each individual by DFA using 7 parameters. My. i.; Myotis ikonnikovi, My. m.; Myotis macrodactylus, Mu. u.; Murina ussuriensis, Mu. l.; Murina leucogaster, E. n.; Eptesicus nilssonii, N. a.; Nyctalus aviator, V. s.; Vespertilio superans.


Quadratic discriminant function analysis using 5 parameters indicated that 87% of 171 calls were classified into correct species (Table 4). MANOVA showed that the model was significant (Wilk's lambda=0.001, F30,594=98.8, p<0.001). The first three discriminant functions explained 99.7% of total variation. Classification rates in each species ranged from 72% (My. ikonnikovi) to 100% (E. nilssonii and Mu. ussuriensis). Twenty-eight percent of My. ikonnikovi calls were misclassified as Mu. leucogaster and My. macro-dactylus, and 27% of N. aviator calls as V. superans (Table 4). Clusters of N. aviator and V. superans overlapped extensively in the DFA score plot when 5 parameters were used (Fig. 5). The degree of overlap between clusters of other species was similar.

Table 4

Summary of classification of 7 species of bats by discriminant function analysis. Model relied on 5 parameters (SF, EF, PF, MF and D). Overall correct classification rate was 87%.


Fig. 5

Canonical score plots of each individual by DFA using 5 parameters. My. i.; Myotis ikonnikovi, My. m.; Myotis macrodactylus, Mu. u.; Murina ussuriensis, Mu. l.; Murina leucogaster, E. n.; Eptesicus nilssonii, N. a.; Nyctalus aviator, V. s.; Vespertilio superans.



Characteristics of echolocation calls

Of the eight species found in the broad-leaved boreal forests of southern Hokkaido, R. ferrumequinum and E. nilssonii have been studied in other parts of their distribution, and parameters of their echolocation calls have been published (Vaughan et al., 1997b; Parsons and Jones, 2000; Russo and Jones, 2002; Zingg, 1990). For the remaining six species (Mu. ussuriensis, Mu. leucogaster, My. ikonnikovi, My. macrodactylus, V. superans and N. aviator) this study provides the first description of their call structures.

Mu. ussuriensis is one of the smallest of the Murininae (forearm length 28.3–34.0 mm) and is distributed in eastern Russia, Korea and Japan. Mu. leucogaster is the largest species of the genus (forearm length 40–44 mm) and is distributed in eastern Asia (Koopman, 1993). Both Murina species emitted FM calls of very short duration, broad bandwidth and low intensity (Fig. 2, Table 2), which were similar to those described for three Murina species in Malaysia (Kingston et al., 1999). Short, broad-band and low intensity FM calls are thought to be used for the detection of arthropod prey in clutter (Simmons et al., 1979; Neuweiler, 1989; Schnitzler et al., 2003). Kingston et al. (1999) suggested that calls of Murina species in Malaysia facilitate highly accurate target localization in terms of both range and angle estimation and can thus be interpreted as an adaptation to foraging in the highly cluttered environment of the forest understory. Therefore, echolocation calls of Murina in this study area would be suitable for highly cluttered forest understory.

The two Myotis species in this study emitted typical FM calls. The interpulse interval for these species was longer than for the Murina ussuriensis, and the bandwidth was narrower than Murina species. Although there is no information about habitat use by either My. ikonnikovi or My. macrodactylus, studies of other species suggest that this call type is characteristic of “narrow or edge space aerial/trawling forager” (Schnitzler et al., 2003). For My. macrodactylus, temporal parameters (D, IPI-S and IPI-L) of hand-released calls and calls recorded in the mosquito net were significantly different. This suggests that My. macrodactylus varies the characteristics of its calls according to whether it is in cluttered space or uncluttered (or backcluttered) space. For DFA, it is preferable to analyze both types of call structure together, as under natural conditions, it is possible that My. macrodactylus flies in both uncluttered and cluttered space.

The call structures of the three FM/QCF species presented here are similar to those of related European species described in the literature (N. leisleri, N. noctula and E. serotinus; Vaughan et al., 1997b). These species are expected to forage mainly in open spaces (Vaughan et al., 1997a), because FM/QCF calls are suitable for use in open environments with some obstacles (Simmons et al., 1979). In our results, the degree of frequency modulation of the QCF part was very variable (Fig. 3). Such variation has also been reported in previous studies (Vaughan et al., 1997b). It is known that FM/QCF bats emit relatively long, narrow-band signals in uncluttered space, but emit shorter and broader-band signals in background cluttered space such as gaps or edges (Schnitzler and Kalko, 1998, 2001). When we recorded the calls of these species the bats had just left their roost and were dispersing in various directions. Some bats were recorded flying in fairly open space, whereas others were flying near to clutter, and this may have been the source of variation in the structure of calls recorded.

The call structures of R. ferrumequinum (Vaughan et al., 1997b; Parsons and Jones, 2000; Russo and Jones, 2002) and E. nilssonii (Zingg, 1990) have been described previously for other regions. R. ferrumequinum has a very wide distribution across temperate Europe and Eurasia (Koopman, 1993). This species is typical of narrow-space CF bats, which hunt from perches for passing insects (Schnitzler and Kalko, 1998). In Europe, the peak frequency of this species is quite different from that in Japan (England: 82.08 kHz, Parsons and Jones, 2000; Italy: 81.3 kHz, Russo and Jones, 2002; Japan: 65.5 kHz, Taniguchi, 1985 and 65.0 kHz, this study). Heller and Helversen (1989) noted that for R. ferrumequinum in Europe there seems to be a cline along which call frequencies decrease continuously from NW to SE. This trend may possibly extend as far as the Japan, as the frequency of calls of R. ferrumequinum in Japan was much lower than in Europe. The distribution of E. nilssonii also includes parts of temperate Europe and Eurasia (Koopman, 1993). The peak frequency of this species is 28.0 kHz in Switzerland (Zingg, 1990), which is lower than in Japanese individuals (30.5 kHz, this study), although the difference is smaller than that found for R. ferrumequinum. The occurrence of such geographic variation in call parameters means that studies that use characteristics of echolocation calls to attempt to identify species must be sure to compare them to calls of known individuals from the same region.

Discriminant function analysis

The results show that Discriminant Function Analysis of parameters of echolocation calls is a feasible method for identifying bats in the assemblage of species found in broad-leaved boreal forests in southern Hokkaido. The overall level of accuracy of the DFA was high (92%; Table 3), and remained high even when the analysis was based on only 5 parameters obtained from a single pulse (87%; Table 4). The rates of correct classification were comparable to, or better than, those achieved by other studies that have used DFA to classify calls from individual bats of known species (Obrist, 1995; Zingg, 1990; Vaughan et al., 1997b; Krusic and Neefus, 1996; Murray et al., 1999). One reason why this study obtained higher accuracy of classification than some others may have been that it was based on relatively few species. In particular, only two species of Myotis inhabit the study region. In many cases the echolocation calls of sympatric Myotis species are similar in structure and show extensive large overlap in spectral and temporal parameters making them difficult to distinguish from one another (Krusic and Neefus, 1996; Vaughan et al., 1997b).

It is important to note, however, that certain species, such as My. macrodactylus, E. nilssonii, N. aviator and V. superans, changed the structure of their echolocation calls in response to changing environmental conditions of the flight space. It is therefore essential to develop a reference collection of calls that includes recordings of individuals of each species flying in a variety of environments. As far as possible, the collection should include all environmental conditions in which the bats may be encountered in the field.

Although classification by DFA was more accurate for some species than for others, accuracy was high for all species. This suggests that it should be possible to study species-specific patterns of habitat use by bats in forests in the study area using acoustic monitoring alone. In practice this may be less straightforward than it appears to be, however. Levels of activity cannot be compared directly across species, because different species give calls of different intensity (Waters and Jones, 1995), and this influences the distance that they can be detected from. It should be possible to examine the habitat requirements of each species independently by assessing relative levels of foraging activity in various environmental conditions. However, this may also be problematic for certain species. The calls of the three FM/QCF type species were very loud, and could be detected at distances of over 50 m (Fukui, personal observation), which means that they may be recorded from habitat types adjacent to those being surveyed. This problem could be overcome to some extent by ensuring that sampling of a particular habitat type was always done as far as possible from borders with other habitat types. By contrast, the calls of the two Murina species and Rhinolophus ferrumequinum were very quiet, and could only be recorded in close proximity. Consequently, in a survey of habitat use, the calls of these species may be detected very rarely, even in areas where they are frequently active. In view of this limitation, acoustic monitoring should probably not be used in isolation, but should be supplemented with survey techniques involving capture.

Assessing species differences in habitat use

Patterns of habitat use and foraging behavior of bats have been interpreted in terms of the structure of echolocation calls (Neuweiler, 1989; Schnitzler et al., 2003), as mentioned above, and aspects of wing morphology (Norberg and Rayner, 1987). In theory, it should be possible to predict the kinds of habitat that each species in a bat community is likely to be associated with on the basis of these characteristics. Such predictions could then be used to develop plans for conservation management that would enhance habitats for bats. However, it is first necessary to establish whether the predictions are valid for a broad range species by assessing patterns of habitat use in the field.

This study has shown that acoustic identification of species is a promising method for establishing patterns of habitat use by the bat community in boreal broad-leaved forests in southern Hokkaido. When calls of unknown bats are recorded in the field, they can be classified to species using the methods described here. This approach is less invasive, and can provide much more quantitative data about activity patterns, than surveys based on capturing bats. Such data can then be used both to test the accuracy of general predictions based on call structure and wing morphology, and to provide profiles of the habitat requirements of individual bat species which is essential information for promoting their conservation (see Vaughan et al., 1997; Russo and Jones 2003).


We thank Dr. Kuniko Kawai of Hokkaido University and members of the Centennial Woods Park Fan Club in Kutchan for their encouragement, valuable advice and help with the fieldwork, and staff and students of TOEF for much advice and good humour. This work was partly supported by the Japanese Ministry of Education, Science, Sports and Culture (No. 15207008).



H. Abe, N. Ishii, Y. Kaneko, K. Maeda, S. Miura, and M. Yoneta . 1994. A Pictorial Guide to the Mammals of Japan. Tokai University Press. Tokyo. (in Japanese). Google Scholar


H. D. J. N. Aldridge and I. L. Rautenbach . 1987. Morphology, echolocation and resource partitioning in insectivorous bats. J Anim Ecol 56:763–778. Google Scholar


R. Arlettaz 1999. Habitat selection as a major resource partitioning mechanism between the two sympatric sibling bat species Myotis myotis and Myotis blythii. J Anim Ecol 68:460–471. Google Scholar


R. M. R. Barclay and R. M. Brigham . 1991. Prey detection, dietary niche breadth, and body size in bats: why are aerial insectivorous bats so small. Am Nat 137:693–703. Google Scholar


R. M. R. Barclay, J. H. Fullard, and D. S. Jacobs . 1999. Variation in the echolocation calls of the hoary bat (Lasiurus cinereus): influence of body size, habitat structure, and geographic location. Can J Zool 77:530–534. Google Scholar


Bat Research Group of Centennial Woods Fan Club, Kutchan 2001. Bats in Mt. Yotei and Niseko Range, Hokkaido, Japan, No. 1. –Report on 1997∼2000 Faunal Survey–. Bull Otaru Mus 14:127–132. Google Scholar


M. Ciechanowski 2002. Community structure and activity of bats (Chiroptera) over different water bodies. Mammal Biol 67:276–285. Google Scholar


M. B. Fenton 1990. The foraging behaviour and ecology of animaleating bats. Can J Zool 68:411–422. Google Scholar


D. Fukui, T. Ishii, N. Agetsuma, and T. Aoi . 2001. Efficiency of harp trap for capturing bats in boreal broad-leaved forest in Japan. Eurasian J For Res 3:23–26. Google Scholar


K-G. Heller and O. von Helversen . 1989. Resource partitioning of sonar frequency bands in rhinolophoid bats. Oecologia 80:178–186. Google Scholar


E. K. V. Kalko and C. O. Handley Jr . 2001. Neotropical bats in the canopy: diversity, community structure, and implications for conservation. Plant Ecol 153:319–333. Google Scholar


T. Kingston, G. Jones, Z. Akbar, and T. H. Kunz . 1999. Echolocation signal design in Kerivoulinae and Murininae (Chiroptera: Vespertilionidae) from Malaysia. J Zool 249:359–374. Google Scholar


K. F. Koopman 1993. Order Chiroptera. In “Mammal Species of the World: a Taxonomic and Geographic Reference”. Ed by D. E. Wilson and D. M. Reeder . Smithonian Institution Press. Washington DC. pp. 137–241. Google Scholar


R. A. Krusic and C. D. Neefus . 1996. Habitat associations of bat species in the White Mountain National Forest. In “Bats and Forests Symposium”Ed by R. M. R. Barclay and R. M. Brigham . Ministry of Forests. Victoria. pp. 185–198. Google Scholar


R. A. Krusic, M. Yamasaki, C. D. Neefus, and P. Pekins . 1996. Bat habitat use in White Mountain National Forest. J Wildl Manage 60:625–631. Google Scholar


A. L. Kuenzi and M. L. Morrison . 2003. Temporal patterns of bat activity in southern Arizona. J Wildl Manage 67:52–64. Google Scholar


B. S. Law, J. Anderson, and M. Chidel . 1999. Bat communities in a fragmented forest landscape on the south-west slopes of New South Wales, Australia. Biol Conserv 88:333–345. Google Scholar


Ministry of the Environment 2002. Threatened Wildlife of Japan: Red Data Book 2nd ed – Vol. 1, Mammalia. Japan Wildlife Research Center. Tokyo. Google Scholar


K. L. Murray, E. R. Britzke, B. M. Hadley, and L. W. Robbins . 1999. Surveying bat communities: a comparison between mist nets and the Anabat II bat detector system. Acta Chiropt 1:105–112. Google Scholar


G. Neuweiler 1989. Foraging ecology and audition in echolocating bats. Trends Ecol Evol 4:160–166. Google Scholar


U. M. Norberg and J. M. Rayner . 1987. Ecological morphology and flight in bats (Mammalia; Chiroptera): wing adaptations, flight performance, foraging strategy and echolocation. Phil Trans R Soc Lond B 316:335–427. Google Scholar


M. K. Obrist 1995. Flexible bat echolocation – the influence of individual, habitat and conspecifics on sonar signal design. Behav Ecol Sociobiol 36:207–219. Google Scholar


S. Parsons 2001. Identification of New Zealand bats (Chalinolobus tuberculatus and Mystacina tuberculata) in flight from analysis of echolocation calls by artificial neural networks. J Zool 253:447–456. Google Scholar


S. Parsons, A. M. Boonman, and M. K. Obrist . 2000. Advantages and disadvantages of techniques for transforming and analyzing Chiropteran echolocation calls. J Mammal 81:927–938. Google Scholar


S. Parsons and G. Jones . 2000. Acoustic identification of twelve species of echolocating bat by discriminant function analysis and artificial neural networks. J Exp Biol 203:2641–2656. Google Scholar


P. A. Racey 1998. The importance of the riparian environment as a habitat for British bats. In “Behaviour and Ecology of Riparian Mammals” Ed by N. Dunstone and M. L. Gorman . Symp Zool Soc Lond71:69–91. Google Scholar


P. A. Racey and A. C. Entwistle . 2003. Conservation ecology of bats. In “Bat Ecology”. Ed by T. H. Kunz and M. B. Fenton . The University of Chicago Press. Chicago. pp. 680–743. Google Scholar


D. Russo and G. Jones . 2002. Identification of twenty-two bat species (Mammalia: Chiroptera) from Italy by analysis of time-expanded recordings of echolocation calls. J Zool 258:91–103. Google Scholar


D. Russo and G. Jones . 2003. Use of foraging habitats by bats in a Mediterranean area determined by acoustic surveys: conservation implications. Ecography 26:197–209. Google Scholar


J. Rydell, H. T. Arita, M. Santos, and J. Granados . 2002. Acoustic identification of insectivorous bats (order Chiroptera) of Yucatan, Mexico. J Zool 257:27–36. Google Scholar


H-U. Schnitzler and E. K. V. Kalko . 1998. How echolocating bats search and find food. In “Bat Biology and Conservation”. Ed by T. H. Kunz and P. A. Racey . Smithsonian Institution Press. Washington DC. pp. 183–196. Google Scholar


H-U. Schnitzler and E. K. V. Kalko . 2001. Echolocation by insect-eating bats. BioScience 51:557–569. Google Scholar


H-U. Schnitzler, C. F. Moss, and A. Denzinger . 2003. From spatial orientation to food acquisition in echolocating bats. Trends Ecol Evol 18:386–394. Google Scholar


J. A. Simmons, M. B. Fenton, and M. J. O'Farrell . 1979. Echolocation and pursuit of prey by bats. Science 203:16–21. Google Scholar


S. M. Swift and P. A. Racey . 1983. Resource partitioning in two species of vespertilionid bats (Chiroptera) occupying the same roost. J Zool 200:249–259. Google Scholar


I. Taniguchi 1985. Echolocation sounds and hearing of the Greater Japanese Horseshoe Bat (Rhinolophus ferrumequinum nippon). J Comp Physiol A 156:185–188. Google Scholar


D. W. Thomas, G. P. Bell, and M. B. Fenton . 1987. Variation in echolocation call frequencies recorded from North American vespertilionid bats: a cautionary note. J Mammal 68:842–847. Google Scholar


N. Vaughan, G. Jones, and S. Harris . 1996. Effects of sewage effluent on the activity of bats (Chiroptera: Vespertilionidae) foraging along rivers. Biol Conserv 78:337–343. Google Scholar


N. Vaughan, G. Jones, and S. Harris . 1997a. Habitat use by bats (Chiroptera) assessed by means of a broad-band acoustic method. J Appl Ecol 34:716–730. Google Scholar


N. Vaughan, G. Jones, and S. Harris . 1997b. Identification of British bat species by multivariate analysis of echolocation call parameters. Bioacoustics 7:189–207. Google Scholar


D. A. Waters and G. Jones . 1995. Echolocation call structure and intensity in five species of insectivorous bats. J Exp Biol 198:475–489. Google Scholar


P. E. Zingg 1990. Acoustic species identification of bats (Mammalia: Chiroptera) in Switzerland. Rev Suisse Zool 97:263–294. Google Scholar
Dai Fukui, Naoki Agetsuma, and David A. Hill "Acoustic Identification of Eight Species of Bat (Mammalia: Chiroptera) Inhabiting Forests of Southern Hokkaido, Japan: Potential for Conservation Monitoring," Zoological Science 21(9), 947-955, (1 September 2004).
Received: 20 April 2004; Accepted: 1 June 2004; Published: 1 September 2004
discriminant function analysis
echolocation call
Back to Top