Recommended methods: Capture (mist nets, harp traps) and detection (ultrasonic bat detectors and listening for E. maculatum) should be employed simultaneously to determine presence/not detected and relative abundance of bats (Table 3). Note that relative abundance can only be estimated using detection sampling and detection indices.
Anyone involved in the capture and handling of live bats should be familiar with the manual, Live animal capture and handling guidelines for wild mammals, birds, amphibians, and reptiles (No. 3).
Capture of bats allows positive species identification (see Nagorsen and Brigham, 1993, for identification key), age and sex determination, the collection of mass and other mensural data, and an assessment of reproductive condition (Anthony, 1988; Racey, 1988). However, this obviously requires some handling of and disturbance to the animal and not all species or sexes are equally catchable, if catchable at all.
The two most common methods of capture involve the use of mist nets or harp traps, although several other methods (e.g., hand nets, funnels) have been used in the past (e.g., LaVal and LaVal, 1977; Youngson and Mckenzie, 1977; Fenton and Bell, 1979; Kunz and Kurta, 1988). Many of these other techniques require sampling at or in roost sites, and are not recommended because they tend to be disruptive to the bats and may cause them to abandon the roost. Conservation of bats and critical habitats, as well as minimization of disturbance must be considered for all potential sampling protocols.
Mist netting is the most common method used to capture bats (Kunz and Kurta, 1988). Catching bats in mist nets depends on careful selection of netting sites (Fig. 4). Productive netting sites (i.e. areas of high bat activity) can be determined by direct observation of bats or by using bat detectors (see below).
The major advantages of using mist nets to sample bats are that they are relatively inexpensive, highly portable and easy to use and set up. The disadvantages are that they have certain biases associated with them, in terms of which species can be caught, and they require constant monitoring to ensure that bats do not chew their way out, become badly entangled or cause injury to themselves. A further disadvantage is the recent difficulty in obtaining suitable mist nets from suppliers. The success of mist nets at a location decreases if a net is set up at the same location more than once (Kunz and Brock, 1975). In addition, certain species are adept at avoiding mist nets or fly at heights that make their capture difficult, even though they may be present in a study area. For example, Lasiurus blossevilli, L. cinereus and Eptesicus fuscus tend to fly higher than the location of most mist nets and gleaning species such as Myotis evotis, Plecotus townsendii, and Antrozous pallidus seem better able to detect and avoid mist nets, particularly now that monofilament nets are unavailable. Setting nets higher in the canopy can increase the success of capturing these high flying species, and numerous designs for canopy netting is described in Kunz (1996). Also, juveniles may be more susceptible to capture than older age classes creating a biased interpretation of population composition. In addition, environmental factors may influence the effectiveness of mist netting. The presence of wind may decrease capture success by causing the mist net to billow and thus become more detectable (Nyholm, 1965). Rain also adheres to mist nets, rendering them more "visible" to bats.
Mist nets used for capturing bats are usually black, 6 to 36 m in length, 2 m high, have four shelves, a mesh size of 36 mm and are constructed from 50 or 70 denier/2 ply nylon (Fig. 5; Kunz and Kurta, 1988). Unfortunately, recent restrictions by the Japanese manufacturers and a government trade ban by Japan have made mist nets very difficult to obtain and monofilament nets, the most effective ones for capturing bats, are no longer available. Nets less than 12 m in length tend to be easier to handle, especially for one person. Poles made of 3 m lengths of aluminum tubing are often used to support the nets. The tubing should have a wall thickness of about 1.6 mm and should be at least 2.5 cm in diameter. Thin-walled electrical conduit is inexpensive and readily available and makes excellent mist net poles. Connectors (e.g., 20-30 cm long solid aluminum shafts that fit the inside diameters of poles) can be made to join lengths of pole to make sections of the necessary length. To keep mist nets in place, guy lines can be attached to the poles and anchored to vegetation or rocks.
Figure 4. Example of mist net placement. Note that the net is placed in the vegetation such that a potential flight corridor is covered by the net. (From Kunz and Kurta, 1988).
Figure 5. Mist net components and dimensions.
Harp traps, specifically designed for capturing bats, were first described by Constantine (1958) and later modified by Tuttle (1974). Unlike mist nets, harp traps may be set up and left unattended. Similar considerations as those for setting mist nets are used for the placement of harp traps (Fig. 6). Harp traps may be hoisted off the ground by ropes or positioned outside at entrances to buildings, caves, or mines. As for mist nets, trapping success tends to decrease with each successive night in the same location (Kunz and Anthony, 1977).
The major advantages of using harp traps to sample bats are that they are less labour intensive, they do not require constant supervision (thus several can be set up per night) and they can be used to catch species that tend to avoid mist nets (such as Myotis ciliolabrum and Myotis evotis, Holroyd et al., 1994). Disadvantages include the small area sampled by the trap (only about 2 m2 as opposed to several times that for each mist net used ), its limited portability, which may limit its use to areas accessible by roads, and its greater cost (approximately $500 CAN). A collapsible, 7 kg harp trap described by Tidemann and Woodside (1978) which takes 30 minutes to set up or dismantle at least partly solves the portability problem.
Harp traps (Fig. 7) consist of two 2 m by 1.8 m frames of aluminium tubing. Vertically strung across each frame is a bank of 6 - 8 pound (3 - 3.5 kg) monofilament fishing line. Lines are strung 2.5 cm apart. The two frames are spaced 7 to 10 cm apart. Attached to the bottom of the frame is a canvas bag, lined with polyethylene. The trap works on the principle that a flying bat can not easily detect or avoid the bank of lines and will become trapped between the monofilament lines and fall into the holding bag below. The bats drop into this bag and are unable to crawl out over the slippery polyethylene. If a bat manages to fly straight through the first set of lines, it is impeded by the second set. The degree of tension on the lines may have to be increased if bats are able to fly straight through without becoming trapped, or decreased if they simply 'bounce off'.
Figure 6. Examples of harp trap placement, a) along a forest trail, (b) at the entrance to a cave (From Kunz and Kurta, 1988).
Figure 7. Harp trap design and detail. (design from Tuttle 1974, drawn by Tom Swearingen).
Detection involves sampling bats either by visual or acoustic means. Unlike netting and trapping, no handling is involved and therefore disturbance is minimized. However, positive species identification is not always possible nor is an assessment of age, sex, or reproductive condition. Therefore the question being asked and the type of information required will generally dictate whether this sampling method is useful.
Visual detection has been used at sites where bats are known to roost, in order to count the number of bats exiting the roost at or shortly after dusk (e.g., Swift, 1980). This provides a useful and accurate census of the total number of bats using a roost site, provided that all exits from the roost are identified and monitored and that any bats that re-enter the roost are accounted for (Thomas and LaVal, 1988).
It may still be necessary to trap bats at the roost to obtain a positive species identification and ensure that only one species is using the roost. One obvious drawback to this method is that it can only be applied at known roost sites and usually only one roost exit per night per observer can be monitored. In order to accurately extrapolate the results of censuses at a roost to larger geographic areas or populations, it is necessary to locate and census every roost in the area and information regarding home ranges of individuals must be known and taken into account (Thomas and LaVal, 1988). In practice, this is very difficult, if not impossible.
Electronic counting devices such as photo-electric beam splitters, which record each flying bat that interrupts the light beam, have also been used to census bat roosts (e.g., Voute et al., 1974). Although this method does not require an observer to be present, its use in sampling British Columbia bats is limited given the relatively small roost sizes compared to much larger colonies located outside B.C. (Nagorsen and Brigham, 1993), for which the method is usually used.
Only one study has attempted visual censusing of bats away from roost sites. Gaisler (1979) used visual counts along transects in a city environment to census bat populations. However, this approach would be of limited use given the higher species diversity and dense vegetation and canopy that often occurs in B. C. This method may be applicable to urban sites with streetlights or in northern regions where twilight never ends. However, positive species identification is not always possible and at minimum, a highly experienced observer would be required.
Bats in B.C. typically rely on vocalizations for communication (Fenton, 1985) and orientation when commuting or foraging (Griffin, 1958). It is possible to eavesdrop on these vocalizations (i.e., echolocation calls) to detect the presence of bats, assess whether a bat is foraging or commuting, and potentially identify the species emitting the call. Such vocalizations can be used in much the same way that bird song is used to census bird populations, the major difference being that the majority of bat sounds are beyond the range of human hearing and thus require specialized equipment to monitor them. Most humans can only detect sounds with frequencies less than 20 kHz. Sounds above this limit are termed ultrasonic. The calls of all but one species of bat in B.C. are restricted to the ultrasonic range.
Bats emit ultrasonic signals in order to echolocate. By emitting a series of discrete calls and listening for returning echoes, bats are able to locate objects, including prey items (Griffin, 1958). Echolocation signals have a frequency, a duration, and an intensity associated with them (Simmons et al., 1979). The signal may consist mainly of a constant frequency or it may sweep over a range of frequencies. The signal may also include harmonics, in addition to the fundamental (lowest) frequency. Differences in these features allows for a limited degree of species recognition (Fenton and Bell, 1981), although there is considerable geographic and individual variation in call design (Thomas et al., 1987; Brigham et al., 1989; Hayes 1997). In addition, some bats have the ability to change their echolocation call characteristics, depending on the habitat type (e.g., open versus interior forest; Kalko and Schnitzler 1993), which can further complicate species identification.
The repetition rate at which calls are given varies with the activity of the bat and provides a means for discriminating between different behaviours in the field (Thomas and West, 1989). Commuting bats or bats searching for prey emit approximately 10 calls per second. This rate increases to 100 or more pulses per second when a potential prey item has been detected and the bat closes in to attack. This results in a characteristic 'feeding buzz' (Griffin, 1958) and gives a positive indication that the bat is foraging in an area. Thus, it is possible to determine what habitats are important as foraging areas, by detecting the presence of feeding buzzes.
When using detectors to eavesdrop on bats, two pieces of information should be recorded (on a per unit time basis): (1) the number of bat passes and (2) the number of feeding buzzes. A bat pass is defined as a sequence of two or more echolocation calls registered as a bat flies within range of an observer or the detecting equipment (Fenton, 1970; Thomas and West, 1989).
Knowledge about the number of bat passes detected does not allow for an estimate of the number of bats present in a study area because there is not a one to one relationship between the number of bat passes and the number of bats responsible for those passes (Fenton, 1970). That is, it is not possible to discriminate between several bat passes made by a single bat flying repeatedly through the study area versus several bats each making a single pass. Therefore, bat passes do not allow a direct estimate of population densities. However, the technique does allow a relative measure of bat activity in an area and allows for comparisons between areas or over time to be made.
One species found in British Columbia, the spotted bat (Euderma maculatum), uses echolocation calls that sweep in frequency from 15 to 9 kHz, and are thus readily audible to the unaided human ear and require no specialized equipment to detect (Leonard and Fenton, 1984; Fenton et al., 1987). Although, young individuals and females tend to have better high frequency hearing, and are better able to detect E. maculatum. There is evidence that E. maculatum forage in circuits along specific, well-defined routes and thus repeatedly fly through the same area while foraging (Woodsworth et al., 1981; Navo et al., 1992). Therefore, it seems likely that several feeding buzzes or bat passes detected at a sampling location represents the same bat and not several individuals.
To detect the other 15 species of bats found in British Columbia, some type of commercially available ultrasonic bat detector is required. Two types of detectors are available; tunable narrow band detectors and divide-by-n broad band detectors. Both detector types can be operated either remotely or manually, as described below. The ability to discriminate and identify individual species depends to some extent on the sophistication of the detecting equipment. The simplest and least expensive detectors are tunable narrow band (heterodyne) detectors, whereas the divide-by-n broad band detectors generally provide more information, yet at a greater expense.
The audio output from the detector will depend on the structure and energy of the incoming ultrasonic signal. Figure 8 shows the frequency-time displays (sonograms) of some hypothetical signals and describes the corresponding output as heard on a tunable narrow band detector. It is possible using an identification key (e.g., Table 5) to identify some species based on the output from the detector (but see "Precautions and Limitations" below). A tunable detector is particularly useful for identifying the presence of red and hoary bats (Lasiurus spp.), which are rarely captured in mist nets or harp traps. However, it is not possible to discriminate between the different Myotis species, based on the output of a tunable detector, due to the similarity of their calls.
By coupling the tunable narrow band detector with a micro-cassette recorder it is possible to leave the detector unattended in the field and thus sample a number of study areas on any given night. The amount of data that can be collected is limited by the length of tape, from which the data must later be transcribed. It is also possible to sample at different heights in the canopy by using a microphone with a long lead suspended at different heights (Thomas and West, 1989). The major disadvantage of a tunable narrow band detector is that they must be set at one and only one frequency and therefore not all bat species can be sampled, unless several detectors are set at different frequencies and left in the study area.
An advantage of divide-by-n detectors over tunable, narrow band detectors is that they are broad-band and are able to monitor all frequencies (and thus detect most bat species) simultaneously. Therefore, sampling effort can be increased, because it is unnecessary to constantly tune to different frequencies to detect different species. Also, information regarding the time and frequency characteristics of the fundamental frequency are retained, as well as call harmonics when using some detectors (e.g., Petterssen detector). This allows a greater degree of species resolution, although some of the Myotis species still cannot be distinguished from one another (Fenton et al., 1983; Thomas and West, 1989).
The output of a divide-by-n detector can be analyzed by using a zero-crossing period meter coupled with an oscilloscope (Simmons et al., 1979). The period meter displays a frequency-time display (a sonogram) of the fundamental frequency on the oscilloscope screen, which can be used to identify species or species groups (e.g., Table 6; see below for cautionary note regarding the identification of species from echolocation calls). The use of a period meter to identify calls requires many hours of training and experience with free-flying individuals of a known species (Thomas and West, 1989).
Some divide-by-n detectors (e.g., ANABAT or Petterssen systems) can be operated remotely as well as manually. Such a set-up allows automatic monitoring of bat calls, thus freeing the worker for other tasks, and will detect all species unlike a tunable narrow-band detector which can only be set at (and therefore detect) a single frequency at a time. Sampling effort can be greatly increased by using several automated detectors. However, time must still be spent analyzing the recorded signals, although this can be done subsequent to field work. The analyses of data from the ANABAT or Petterssen systems are based on visual representations of bat calls, similar to those in Figure 8. In addition, characteristics such as maximum frequency, minimum frequency, average frequency, duration, and time between calls are available, which generally permits a more accurate assessment of the species or species group of the recording. However, there are limitations (see "Precautions and Limitations" below) which must be considered when interpreting data.
Figure 8. Sonogram of echolocation calls (frequency versus time).
A tunable narrow band detector tuned to frequency 1 would register call "a" as a sharp "tick", call "b" as a "putt" sound, and call "c" as a tonal "chirp"; call "d" would not be detected. At frequency 2 calls "a", "b", and "c" would all be registered as sharp "ticks". At frequency 3 calls "a" and "b" would produce sharp "tick", call "c" would not be detected, and call "d" would result in a long tonal output (Modified from Fenton, 1988).
One commonly used ultrasonic bat detector is the QMC mini bat detector2 (about $300 CAN, UltraSound Advice, 23 Aberdeen Road, London, N5 2UG, UK). These detectors superimpose an internally generated pure tone on the inaudible ultrasonic signal, thus rendering it audible when the detector is tuned to a frequency near that of the incoming signal (Miller and Andersen, 1984). They can detect frequencies between 10 and 180 kHz, but can only scan a single narrow frequency band (about 3-5 kHz) at a time. If the detector is tuned to 35 kHz it can detect any bat within range that is using an echolocation signal with a 35 kHz component to it. Tunable detectors only sample a small portion of the total frequency range of any call. They do not preserve the time and frequency characteristics of a call's fundamental frequency. By changing the tuning of the detector it is possible to sample for several bat species, which employ calls with different frequency components. The detector has a directional range of about 120o (Downes, 1982). It is crucial to calibrate tunable bat detector to a pure tone before using them in the field (Thomas and West, 1984).
Another class of bat detectors are known as divide-by-n (or countdown) detectors. These detectors contain a broad-band microphone coupled with a countdown circuit that produces one cycle for every n cycles (where n is usually 10) of the input frequency, thus giving an audible output (Miller and Andersen, 1984). For example, a call sweeping from 100 kHz to 40 kHz becomes audible as a sweep from 10 kHz to 4 kHz with a divide-by-10 detector (Thomas and West, 1989). This divided output can potentially be recorded for later analysis, or it can be analyzed in the field.
Commercially available divide-by-n detectors are available such as the QMC S200 (about $1500 CAN, QMC Instruments Ltd.) and the ANABAT II detector system (about $2025 CAN, Titley Electronics, P.O. Box 19, Ballina N.S.W. 2478 Australia). The ANABAT II detector systems includes a broad band detector ($ 625), delay switch ($550), timer ($250), and zero-crossing analysis interface module (ZCAIM; $600). The delay switch automatically turns on a tape recorder whenever a call is detected, accompanied by a time cue and calibration tone (40 kHz). The timer can be used to turn the detector system on or off for set durations of time at certain periods of the night. The ZCAIM is used to link the information from the recorded tapes through the delay switch, or data directly from the detector, to a computer. The system comes with software for use on a PC computer (a laptop is convenient in the field) for analyzing calls (rather than using a period meter).
Table 5. Identification key for use with a tunable bat detector for identifying selected species of bat found in British Columbia
(Modified from Fenton et al., 1983).
|
1 |
Audible to unaided ear |
Euderma maculatum |
|
1' |
Not audible |
2 |
|
2 |
20 kHz, calls detectable |
Lasiurus cinereus |
|
2' |
20 kHz, calls not detectable |
3 |
|
3 |
25-35 kHz, output a tonal chirp |
Lasionycteris noctivagans |
|
3' |
25-35 kHz, output a "put" sound |
Eptesicus fuscus |
|
3" |
25-35 kHz, calls not detectable |
4 |
|
4 |
40 kHz, output a tonal chirp |
Lasiurus blossevilli |
|
4' |
40 kHz, output a sharp "tick" |
Myotis species |
Table 6. Characteristics of the ultrasonic calls of selected bat species3 as viewed with a period meter/oscilloscope (Modified from Fenton et al., 1983; Thomas and West, 1989).
|
Species |
Call Characteristics |
|
Lasionycteris noctivagans |
Call starts with 1-2 ms sweep from 30-25 kHz. Call lasts about 10 ms. |
|
Lasiurus cinereus |
Similar to L. noctivagans, but call initially sweeps in 25-30 kHz range. |
|
Lasiurus blossevilli |
Similar to L. noctivagans, but call initially sweeps in 45-40 kHz range. |
|
Eptesicus fuscus |
Call sweeps from 33-28 kHz in first 3 ms. Ends with a 2-7 ms constant |
|
Myotis thysanodes |
frequency (CF) tail around 28 kHz. |
|
M. volans |
Call sweeps from >60-35 kHz in 5 ms. Call has no inflection point. |
|
M. californicus |
2-4 ms call sweeping from >60-40 kHz no inflection point |
|
M. ciliolabrum |
|
|
M. lucifugus |
2-4 ms call sweeping from >60-40 kHz Inflection point near middle of |
|
M. yumanensis |
call. |
|
M. evotis |
1-2 ms call, steep sweep from >100-40 kHz. |
|
M. septentrionalis |
1-2 ms call, sweep from 80-40 kHz. |
|
Plecotus townsendii |
6-7 ms call, straight sweep from 40-28 kHz. |
To make ultrasonic reference recordings for particular bat species, it is necessary to first capture and visually mark bats. These reference recording will assist in species identification for a particular study area. A small chemiluminescent light tag, such as a miniature light stick (2.9 mm X 24 mm), can be glued to the fur on a bat (Buchler, 1975; Barclay and Bell, 1988; Horvorka et al. 1996). This light tag will act as a short term visual mark to keep track of the bat as it's echolocation calls are being recorded. Light tags are also a useful means for determining foraging ranges, habitat use, the vertical distribution of activity and for making behavioural observations (see Barclay and Bell, 1988 for details).
Once the light tag is affixed, the bat can be released and reference calls recorded. Fenton (1988) describes how to record bat calls. Ideally a high speed tape recorder (76 cm/sec) should be used, to preserve the sound characteristics exactly as they were produced, but the high cost ($20,000) of such tape recorders may be prohibitive. Alternately the output from a divide-by-n detector (i.e. a divided call) could be recorded. This should provide enough information about the time and frequency characteristics of a call to make a usable reference recording. Such a recording, along with a corresponding sonogram will help facilitate species identification by experienced field workers. At the very least, the output from a tunable detector set at various frequencies should be recorded to provide verification of species identity in different geographic regions. If possible, several individuals of each species from several areas in the province should be recorded, to allow an assessment of variation in calls between individuals or over geographic regions.
Caution should be exercised when identifying species on the basis of echolocation calls, because it has been shown that calls can vary quite widely between individuals or over geographic areas and this may lead to misidentification (Thomas et al., 1987; Brigham et al., 1989). The extent of this variation has yet to be fully determined. The keys presented in this manual were developed by Fenton et al. (1983) for identifying bats in the Kootenay, Glacier and Mount Revelstoke National Parks of B.C. Their applicability to other regions of B.C. is unknown and requires verification. Until this is determined, we strongly recommend the production of reference tapes of each species' call for different regions of B.C. to provide a comparison for field recordings (Thomas and West, 1989), and to assess the prevalence of variation in calls. A library of reference calls is currently being complied for bats in North America, and can be found on the web site at: http://sevilleta.unm.edu/~wgannon/batcall.
The exact distance over which a bat detector can detect an echolocating bat depends mainly on the intensity of the echolocation call (Downes, 1981; Fenton, 1988). Maximum detection distances are relatively constant for a species, but differ among species (Thomas and West, 1989). Some species such as P. townsendii, M. evotis and M. septentrionalis use low-intensity calls or high frequency calls which attenuate rapidly (Faure et al., 1990) and are only detectable at a distance of a few metres. These species will thus be under-represented in sampling compared to other species. In contrast M. lucifugus is detectable at a range of over 10 m with a QMC mini bat detector (Downes, 1982), and L. cinereus is detectable up to 30 or 40 m away. Because detection distances vary between species (i.e. species have different degrees of detectability), reliable comparisons of relative abundance between species can not easily be made (Thomas and West, 1989). Further, caution must be exercised when using broad band detectors (e.g., ANABAT or Petterssen) to discriminate between species based on the audible output of the detector. Because these detectors are generally more sensitive than tunable arrow band detectors, and detect the bats over a range of frequencies, differences in the audible output may not be as distinct for different species or species groups. Experience is needed to accurately distinguish species or species groups based on the audible output when using sensitive broad band detectors.
The methods discussed so far are designed towards sampling bats at sites away from roosts. It is also possible to net, trap, or use bat detectors near known roosts to estimate the number of bats using that roost. However it may first be necessary to locate such roosts and as noted below (section 5.1), disturbance at roosts can lead to bats abandoning these sites.
3 Because of the existence of individual and geographic variation in calls, this table provides only a general outline of call characteristics and may not be applicable to all areas of B.C. (see below). This table is presented for illustrative purposes and represents a composite of characteristics described by Fenton et al. (1983) and Thomas and West (1989). There are considerable differences in call characteristics for the same species in the two studies.
