Rhinolophus ferrumequinum at Woodchester
Introduction
Studies of the ultrasonic calls of bats have tended to concentrate on the echolocation signals, while comparatively little detailed work has been done on the non-echolocation or social calls. The distinction between which calls are used for which function has however only been made on an ‘intuitive’ basis. If a bat is found to use a simple frequency sweep with a couple of harmonics while hunting and flying in the open, then this is taken to be the type of echolocation signal used by that bat, and a complex, maybe slightly noisy call, with several linked up and down sweeps will be dismissed as a ‘squawk’.
In many cases this is quite reasonable, and there is a clear demarcation between raucous screams produced by a hand-held bat, and the clean echolocation signals used in flight. In many cases however the distinction between echolocation and ‘social’ calls is less well defined, and there may be a gradual continuum between a clear echolocation signal and a blatant squawk. If functionality is considered then the situation becomes even more complicated, since it is quite possible for echolocation signals to be simultaneously used for a communicative function, and it may be possible for the bat to extract echolocation information from a ‘social’ call. Evidence for this possibility can be seen in the structure of the ambiguity diagram of the multiple sweep signal of Natalus stramineus, which shows that echolocation information is available from a signal which would normally be thought of as a purely social call.
For convenience, it will be assumed that signals with a relatively simple, ultrasonic structure, and which are used by the bat in flight are echolocation signals, and all other signals will be referred to as ‘social’ calls, even though the true function and intention of the bat are not known.
If we wish to classify the various calls produced by a particular type of bat we may do so either on the basis of their function, or on the basis of their structure. Much of the work done on bat communication has been carried out with captive bats, under controlled conditions, so that it is possible to determine the function of the sounds being used. In the wild however it becomes much more difficult to relate the occurrence of a call to a particular type of activity, although this may be possible in certain cases, such as when looking at mother-young communication. The range of calls produced however is very wide, and it is interesting to consider them solely from a structural point of view, in relation to the auditory system of the bat, since this has been extensively modified and specialised for echolocation. This does not imply that echolocation is more important to a bat than communication, simply that it requires a more specialised auditory system and it is presumably easier to adapt communication signals to fit into this background than vice-versa.
The study described here was designed to examine the range and structure of ‘social’ and echolocation calls of Rhinolophus ferrumequinum in the wild. Since the majority of social communication between bats takes place within the roost a microphone was planted in the breeding site of a colony of these bats. The vocal activity of the bats could then be remotely monitored and recorded without their being disturbed in any way. They were also recorded in flight when entering and leaving the area of the roost, and when there were young in the group, these were recorded separately and with their mothers. Details of the infant calls, and the ontogeny of the echolocation calls have been separately published by Long, who assisted in the recording programme. In addition to the study of the sounds used by the bats, radar measurements of flight speeds and wing beat rates were made, both of adults and of young bats flying within the roost after the adults had left for the evening feed.
In order to compare the ‘social’ calls of the bats with the requirements of the echolocation system, we must first consider the way in which the echolocation system of this bat works. This has been reviewed very well by Neuweiler (1979?).
The basic echolocation signal is a constant frequency pulse at a frequency of about 85 kHz, with a terminal frequency sweep descending to about 70 kHz. The peak energy may occur either in the cf portion (sweep-decay pulse) or in the fm portion (sweep-peak pulse). The fundamental frequency of the pulse, at 42 kHz, is suppressed and the transmitted portion consists almost entirely of the second harmonic. The precise frequency of the cf portion of the pulse is somewhat variable between individual bats, and between different colonies of bats. In Germany, where much of the work on R. ferrumequinum has been done the cf frequency is about 82 kHz, and in the case of R. ferrumequinum nippon from Japan it is as low as 71-72 kHz.
The echolocation signals are generated by the vocal chords, but unlike the fm bats, they are emitted through the nostrils. Resonances in the vocal tract are responsible for the suppression of the fundamental, and precise frequency control is regulated by the cricothyroid muscle. The nostrils have a spacing of one half wavelength of the cf frequency of the pulse. In conjunction with the beaming effect of the nose-leaf this has a marked effect on the directionality.
The transmitted signal has a complex beam pattern, with a pronounced lobe pointing forwards and down. If the sensitivity pattern of the ears is taken into account however the combined sensitivity pattern is a symmetrical forward pointing cone.
Doppler measurement necessitates the measurement of the difference in frequency between the transmitted signal and the received signal. To achieve very fine frequency resolution requires rather special techniques. Rhinolophus has a very specialised cochlea with the region corresponding to the area around the cf frequency covering a greater number of turns than would be expected. additionally this area is very highly innervated and has therefore been termed an acoustic fovea.
If the hearing sensitivity is examined, either by looking at cochlear microphonics, or behaviourally, it is found that there is a very sharply tuned notch close to the resting frequency used by the bat. The width of the notch is very much narrower than the range of Doppler shifts that the bat might expect to encounter, but this problem is obviated by compensating for the Doppler shift on each echo. That is to say, that when the bat receives an echo with a Doppler shift it will change the frequency of the next transmitted pulse so that the echoes will always come back at a preferred frequency which is nearly identical to that of the acoustic fovea.
Materials and Methods
Recordings were made of a colony of about 80 female Rhinolophus ferrumequinum in Gloucestershire. This colony has been studied in some detail by Roger Ransome and it is therefore possible to predict much of the behaviour of the bats during the year.
During most of the summer the colony occupies one of the attics in a large derelict house, although during the early part of the spring and occasionally in bad weather they will move down to the basements. The attic roost is L-shaped and has only a single exit that the bats could use. A second entrance was available for people but was normally closed off so that the bats would all use a single exit and their numbers could be estimated by counting them as they left the roost in the evening. A rook’s nest was located at one end of the attic, but provided no exit to the outside, although the noise of the rooks was frequently detected on the recordings of the bats.
An ultrasonic microphone and headstage was attached to one of the roof beams near the centre of the attic at a height of about 1m, angled slightly upwards and towards the exit. The area pointed at by the microphone was one of the main congregating points for the bats as evidenced by the distribution of bat droppings. A 25m lead from the microphone headstage led to the room below the roost, where it was left when no recordings were being made. It was possible to carefully extract this lead from the room and take one end, via a further 50 or 25m lead to a more distant room where the recording apparatus was located. Entrance to the roost was minimised as far as possible (although regular visits are made throughout the season by Roger Ransome or his representatives to collect the bat droppings) and were always made during the period immediately after the bats had left at dusk when the roost would be empty for a period of about half an hour.
Recordings were also made of the bats as they left the roost and flew along the corridor and down the main stairwell on their way out of the building. An observer would sit at the top of the stairs to count the bats as they emerged; this did not appear to disturb the bats. Some recordings were also made of the bats outside the building as they left for their feeding grounds, and attempts were made, unsuccessfully, to record the bats drinking at the local lakes. Pipistrelles (P. pipistrellus)[1], Noctules (Nyctalus noctula) and species of Myotis (probably M. mystacinus) were detected outside the house and a colony of Rhinolophus hipposideros were observed to be roosting in another part of the house. There was no evidence to suggest that any of these bats shared the roost used by R. ferrumequinum.
Recordings were made over one or two evenings every one or two weeks during the summer of 1976 (May to September) and rather less frequently during the summer of 1977.
Recordings were made using a Pemco 110 instrumentation tape recorder, running at 30 i.p.s. ( 76.2 cm/s), with a purpose built variable gain pre-amplifier. The microphone used was a ‘brass’ microphone as described in ‘Signal Collection and Analysis’ and as described in Pye and Flinn (1964). The microphone was replaced a number of times during the summer when the diaphragm had deteriorated.
Signals were monitored at the input and output of the tape recorder on a Philips PM3010 portable oscilloscope, and on a Pye tuned/broadband bat detector, or later on an S100 tuned/broadband bat detector.
At least one hour prior to the expected emergence of the bats the cable would be retrieved from the room below the roost and connected to a further extension lead going to the amplifier and tape recorder, which were situated in a room about 25m away on the floor below the roost, and with two doors between it and the flight path of the bats within the building. The flight path of the bats outside the building went past the (unglazed) window of this room providing further opportunities for observation and recording.
Recordings were made when there was sufficient vocal activity in the roost to be heard on the bat detector and to show up on the oscilloscope. During long periods of continuous or semi-continuous activity sample recordings would be made at intervals of five or ten minutes. Occasional recordings would also be made during periods of low level activity even if the signal was not detectable on the oscilloscope. When monitoring activity over a continuous 24 hour period, recordings were made at intervals of about half an hour, or during any periods of particular interest.
Recordings of baby bats were made by Glenis Long during the period between the bats leaving the roost and their return. Clusters of bats of specific age groups could be identified and recorded separately using a hand-held microphone. Results of this part of the work have been reported independently by Long.
Microwave radar was used to measure the flight speed and wing beat rates of bats. The technique was as described by Halls (1975), using a 13.5 GHz Doppler radar. Adult bats were recorded both inside the house as they left the roost, and outside the house on their way to their feeding grounds. Juvenile bats were also recorded flying within the roost.
Recordings of the bat sounds were analysed using the techniques described in chapter ‘Signal Collection and Analysis’.
Results
The range of sounds which were produced by R. ferrumequinum in the roost was very large. The different types of signal however are very variable, and blend into one another, making a discrete classification difficult. This effect was also noted by Fenton ( ) for Myotis. The categorisation of these signals on the basis of their structure is therefore somewhat arbitrary, but indications will be given where different types of signals merge.
The basic division will be into signals which are modified echolocation signals, or which are derived from echolocation signals, and those whose structure appears to be unrelated to the echolocation signals. In the latter group, further sub-division into tonal, and non-tonal signals, and into simple or complex signals is possible. Tonal signals are those in which there is a single pure tone, seen as a discrete line on the sonagram, which may or may not be accompanied by harmonic components. Non-tonal signals on the other hand consist of broad bands of noisy or unstructured sound, or bands of closely related sidebands. If analysed on the sonagraph with a broad-band filter these may appear as a series of short impulses. A signal is said to have a simple structure if it consists entirely of a single type of component, or a complex structure if it changes in type.
Simple tonal signals.
Only two types of signal were found in this category. One of these (fig. 1) consists of a long tone, gradually increasing from 28 to 30 kHz over a period of 50 or more ms. A trace of second harmonic may be present and the tone may degenerate at the end into a broadband non-tonal portion. Increase in the extent of this non-tonal region provides a continuous variation into a complex type of signal.
Figure 1 Long Tonal Signal
The second simple tonal signal was very uncommon, and consisted of a short tone, less than 10 ms, at about 63 or 68 kHz, with the starting and finishing frequencies lower than the centre frequency (fig. 2).
Figure 2 Short Tonal SIgnal
Non-tonal signals.
Non-tonal signals were far more common. The most frequently encountered form is similar to the longer tonal signal(fig 1). It has a duration of 50-80 ms, starting at about 20 kHz, rising to just over 30 kHz and then dropping slowly back to its starting frequency. The instantaneous bandwidth of the signal is about 5 kHz, consisting of several closely spaced sidebands (fig. RR14). Some second harmonic is often present, especially where the frequency is greatest, and occasionally a trace of third harmonic is also detectable.
A variation of this signal, with a less well defined sideband structure and far greater duration, often several hundred milliseconds, is shown in fig. RR13. The frequency is about 25 kHz and varies less than the signal in fig. RR6. A trace of second harmonic is also present.
Shorter variants of fig. RR6 are also quite common, usually with a second harmonic component and often with a third harmonic as well (fig. RR5 and RR12).
A slightly lower frequency signal at about 20 kHz was also found, (fig. RR11) which was otherwise similar in structure to fig. RR6.
A different type of non-tonal signal was a broad-band ‘rasping’ sound (fig. RR15). In this case there is little structure to the signal which occupies frequencies up to 60 kHz.
Complex signals.
Complex signals are frequently combinations of the simple signals described above. Fig. RR4 shows an atonal signal of 25-30 kHz which then becomes a tonal signal of about 29 kHz, with traces of second harmonic of the early part of the atonal segment and of the early part of the tonal segment.
The tonal part may not be as steady as in that signal, and fig. RR9 shows a similar signal in which the tonal part rises in frequency in a series of ‘warbles’ to about 40 kHz. Again there is a trace of second harmonic present. In fig. RR8 it can be seen that sometimes the second harmonic is the only detectable component, and it is debatable whether this is in fact a second harmonic with a suppressed fundamental, or a higher frequency fundamental signal. In fact the former is the more likely since this type of signal can end with a tonal portion which is very similar to the echolocation signal, albeit lower in frequency. Fig. RR3 shows such a signal in which a trace of the terminal downsweep can be seen.
The lower frequency atonal signals may also develop into complex signals such as fig. RR2, at about 20 kHz.
Echolocation derived.
The most interesting signals are those which are derived from echolocation signals, or are modified echolocation signals. With modified echolocation signals it is difficult to know whether the signals are functionally intended for echolocation or communication. The echolocation derived signals however are usually more clear cut (fig. RR 16). In this example the normal echolocation pulses become shorter and noisier and the frequency of the cf portion decreases until the signals are short broadband pulses with three harmonic components, the fundamental being about 28 kHz.
Modified echolocation pulses may also show this decrease in pulse length and cf frequency (fig. RR19). The pulses may have a duration as short as 2 ms, and the initial frequency upsweep, occasionally seen in normal echolocation pulses, can merge with the downsweep. The fundamental is also more strongly present than is usual for echolocation signals.
A more subtle variation is seen in fig. RR17, where the cf portion of one pulse dips and then rises again. It is hard to know whether a pulse such as this is deliberate, or inadvertant. Frequency variation within the cf portion of echolocation pulses is however not uncommon. The echolocation system of Rhinolophus implies that the best signal would be of very constant frequency, and it is true that the echolocation signals used by these bats in the laboratory when presented with an echolocation task are very constant in frequency. The pulses in fig. RR18 however can be seen to vary by several kHz. This can be seen more clearly in fig. RR20, in which the replayed signal from the tape recorder has been multiplied with a fixed frequency so that the sonagram covers a frequency range from 78 to 90 kHz, for a period of 1s. It can be seen that the frequency of the pulse varies over a wide range several times during the pulse.
Radar Studies
During the summer of 1977 we also used a handheld microwave Doppler radar system to measure flight speeds and wing beat rates. The system is described in detail in Halls (1975). The radar was built for us by Marconi Research Laboratories, and operated at 13.5GHz, with a 2MHz frequency modulation which minimised sensitivity extremely close to the unit, but gave a flat response out to a range of about 50m, instead of the usual 1/R4. The signal from the radar was used to modulate a voltage controlled oscillator (VCO) for recording on high speed tape.
During analysis the tape was replayed at 37.5 ips (recording speed was 30 ips) and analysed on a sonagraph with nominal axes of 8kHz x 2.4s, but with the sonagraph’s frequency expansion enabled to stretch this to a nominal 2kHz x 2.4s. After scaling to allow for the increase in tape speed sonagrams had nominal axes of 1.6kHz x 3.6s, although due to the frequency expansion process the higher frequencies could be lost. A calibration pulse provided calibration marks at 400Hz intervals (equivalent to 4.43 m s-1).
Bat sonar signals, recorded simultaneously on the second track of the tape recorder were analysed by replaying through a mixer with a local oscillator set to 102kHz the output of which was filtered and analysed on the same sonagram as the radar trace. Because of the mixing process the frequency axes of the sonagram are 81.6kHz-89.6kHz.
Figure 3Typical Radar and Bat Sonar Sonagram
Radar recordings were made in two distinct situations. In the attic roos while the adult bats were out hunting and mature youngsters were flying in the roost (August 1977), and in the hallway as adult bats were leaving the roost to go out hunting.
| Tape Ref. | FSR max (Hz) | VRB max (m s-1) | WBR (s -1) |
| B684 | 280 | 3.1 | 10.7 |
| B646 | 340 | 3.8 | 12.5 |
| B352 | 257 | 2.9 | 9.0 |
| B129 | 271 | 3.0 | 10.8 |
| B562 | 229 | 2.5 | 12.5 |
| B527 | 257 | 4.0 | 9.8 |
| Means +/- 1SD | 3.2 +/- 0.6 | 10.9 +/- 1.4 |
Figure 4 Table of flight seppeds and wing beat rates for Baby bats flying in the roost. FSR is the Doppler shift on the 13.5GHz radar; VRB is the relative velocity of the bat; WBR is the wing beat rate.
Figure 4 Shows the measured parameters from 6 recordings of juvenile bats flying in the roost and Figure 3 shows a typical radar analysis sonagram with the radar trace and bat sonar overlayed. It can clearly be seen in the second pass that the bat sonar signals are synchronized with the wing beats. On the radar trace the thicker central line represents the body of the bat while the excursions either side are the returns from the wings.
| Tape Ref. | FSR max (Hz) | VRB max (m s-1) | WBR (s -1) |
| 305 | 643 | 7.1 | |
| 307 | 529 | 5.9 | 9.6 |
| 328 | 486 | 5.4 | 9.1 |
| 483 | 543 | 6.0 | |
| 519 | 486 | 5.4 | 9.3 |
| 667 | 543 | 6.0 | 11.1 |
| 150A | 457 | 5.1 | 9.7 |
| 695 | 485 | 5.4 | 8.0 |
| 695b | 8.6 | ||
| 245c | 674 | 7.5 | |
| 270a | 484 | 5.4 | |
| 399 | 484 | 5.4 | |
| 489b | 610 | 6.8 | |
| 489a | 442 | 4.9 | |
| 245a | 568 | 6.3 | |
| 589 | 6.5 | ||
| Mean (+/- 1SD) | 5.94 +/- 0.77 | 9.3 +/- 0.97 |
Figure 5 Table of flight speeds and wing beat rates for adult bats leaving the roos and flying in the hallway.
It can be seen from Figure 5 that adult flight speeds are higher than those of the juvenile bats although the wing beat rates are marginally lower. This may be due to the better flying ability or due to having a clearer flight space along the hallway.
| Species | Flight Speed |
| Rhinolophus ferrumequinum (juvenile in roost) | 3.2 +/- 0.6 |
| Rhinolophus ferrumequinum (adults in hallway) | 5.9 +/- 0.8 |
| Rhinolophus landeri (o/d) | 4.12 +/- 0.5 (n=4) |
| Rhinolophus landeri (i/d) | 3.12 +/- 0.2 (n=3) |
| Rhinolophus hildebrandti | 2-2.8 |
| Coleura afra (o/d) | 5.7 +/- 1.7 (n=3) |
| Coleura afra (i/d) | 3.5 +/- 0.6 (n=3) |
| Rousettus aegyptiacus | 4.44 |
| Pipistrellus pipistrellus/pygmaeus (o/d) | 6.0 +/- 0.6 (n=7) |
| Pipistrellus pipistrellus/pygmaeus (i/d) | 2.9 +/0.6 (n=2) |
| Hipposideros commersoni (in roost) | 5.4 +/- 1.0 (n=3) |
| Hipposideros commersoni (i/d) | 2.7 +/- 0.4 |
| Epomophorous wahlbergi (o/d) | 5.5 +/- 0.6 |
Figure 6 Table of flight speeds of bats recorded using the radar system in the UK and in East Africa.
[1] In 1976 no distinction was made between P. pipistrellus and P. pygmaeus and both species were classified as P. pipistrellus.
