| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Institute of Biochemistry and Biology, University of Potsdam, 14471 Potsdam; 2 Zoological Institute, University of Frankfurt, 60323 Frankfurt am Main, Germany; 3 Department of Animal Physiology, University of Habana, Calle 25 #455, Vedado, La Habana, Cuba; 4 School of Biology, University of Sussex, Falmer, Brighton BN19QG, United Kingdom
Submitted 3 February 2003; accepted in final form 18 June 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
; Schnitzler and Kalko 1998) are typically altricial mammals with the young born naked and blind, relatively helpless and incapable of flight (for review, see Brown and Grinnell 1980). They depend on their mothers for several postnatal weeks before they can forage prey on the wing. The ontogeny of sonar has been studied in several bat species (e.g., Eptesicus fuscus: Gould 1975a,b; Myotis lucifugus: Konstantinov 1973; Moss et al. 1997; Rhinolophus ferrumequinum: Matsumura 1979; Rhinolophus rouxi: Rübsamen 1987; Hipposideros speoris: Habersetzer and Marimuthu 1986; for review, see Brown and Grinnell 1980). In all species, the calls of infants differ from those of adults, and as a common pattern, the vocalizations rise in frequency and become increasingly stereotyped with age. A parallel development of the receiver is seen in the widening of the hearing range at high frequencies with age (for review, see Brown and Grinnell 1980; Konstantinov 1973; Rübsamen 1987; Rübsamen and Schäfer 1990a; Rübsamen et al. 1989). Except for a preliminary report on neonatal vocalizations (Brown and Grinnell 1980), no data are available on the development of vocal repertoire in the mustached bat (Pteronotus parnellii), a species that represents a central animal model in echolocation research and general auditory neurobiology.
Although P. parnellii is widespread in Middle America and the Caribbean, data on its breeding ecology are scarce. Silva (1979) reports that in Cuba, female P. parnellii from different cave populations gather in certain hot caves, where they form mixed maternity colonies with large numbers of Phyllonycteris poey, and related mormoopid bats. P. parnellii give birth in the middle of July and the last lactating females are observed at the end of September. The first postnatal weeks are spent in the hot cave. Only when the forearm reaches almost adult length, do the young bats start foraging by themselves with their hunting grounds restricted to regions close to the cave (Silva 1979).
Audiovocal behavior and auditory processing in adult P. parnellii are characterized by several specialized features that enable the bat to hunt fluttering prey insects in dense forest. The stereotyped echolocation calls of adult P. parnellii contain 4–5 harmonics, each consisting of a long constant-frequency (CF) component flanked by brief frequency-modulated (FM) sweeps (Goldman and Henson 1977; Keating et al. 1994; Novick and Vaisnys 1964) (Fig. 1). The long CF portion is used to evaluate relative motion, and the terminal downward FM is used to determine target distance. In addition to a sophisticated sonar system, mustached bats possess a rich repertoire of communication calls (Kanwal et al. 1994).
|
The frequency of the dominant 2nd harmonic CF component (CF2) at about 61 kHz is regulated with high precision (for review, see Gaoini et al. 1990; Keating et al. 1994; Schnitzler and Henson 1980). The CF2 frequency emitted by a nonflying bat (resting frequency) is characteristic for the individual and typically falls in a narrow (150–500 Hz) band (e.g., Keating et al. 1994). During flight, the bats compensate for upward Doppler shifts of echo frequency by proportionally lowering the frequency of the emitted calls, so that returning echoes are stabilized at a reference frequency that is slightly higher (100–400 Hz) than the individual's resting frequency (e.g., Gaioni et al. 1990; Keating et al. 1994). At this frequency, cochlear tuning is strongly enhanced (Pollak et al. 1972; Suga and Jen 1977) due to specializations of the passive mechanical systems (for review, see Kössl and Vater 1995; Kössl et al. 1999), and a foveal cochlear representation of a narrow frequency band around CF2 leads to a vast overrepresentation of this signal range throughout the auditory pathway (for review, see Kössl and Vater 1995). The stabilized echo frequency acts as a carrier for periodic frequency and amplitude modulations caused by the wing beat of prey insects, which form acoustic glints that are clearly separable from the randomly structured background echoes by the exceptionally sharply tuned auditory system (Goldman and Henson 1977; Henson et al. 1987; for review, see Schnitzler 1987).
A further significant feature of the mustached bat's sonar system is the establishment of computational maps in the auditory cortex that contain neurons sensitive to specific combinations of the 1st harmonic of the call with a higher harmonic of the echo. "CF-CF" neurons encode target velocity, whereas "FM-FM" combination-sensitive, delay-tuned neurons encode information on target distance (for review, see O'Neill 1995).
This study describes the postnatal developmental dynamics of the vocal behavior of mustached bats. We characterize sonar call structure systematically in prevolant and volant P. parnellii and investigated the onset and quality of Doppler-shift compensation in young bats. Additionally, we provide preliminary data on the development of communication signals. We show that the features of Doppler-sensitive sonar mature gradually in prevolant bats starting out with brief CF-FM signals emitted at low duty cycles that differ from the adult in frequency content, harmonic composition, and duration of CF2 component.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
; Rübsamen 1987). The exact date of birth cannot be determined with certainty in our sample. At the start of our observation period, on July 17, 1998 we found the first small group of neonates (about 50 individuals). The youngest infants were identified by the presence of the umbilical cord; the smallest of these had a FAL of 20 mm, assumed to represent an age of about 1 day. During the following week, birthing activity peaked, and the colony of young bats contained several thousand individuals. About 4 wk after the start of birth activity in the colony, most infants in the cave population had reached FAL values of 46–50 mm, and individuals with FAL values of 51 mm, slightly below the average adult value of 52 mm (Silva 1979), were found at the end of the 4-wk observation period. Fur started to grow in infants with FAL of 49–50 mm. A growth of about 31 mm/28 days would be comparable to a growth rate of about 1 mm/day as observed in Rhinolophus rouxi (Rübsamen and Schäfer 1990a). On the 1st day of captivity, and in some individuals during several consecutive days of captivity, the calls were recorded with a U30 bat detector, fed into a PCMCIA card of a notebook computer and directly digitized at a sampling frequency of 333 kHz. The signals were analyzed with BatSound 2.1 (Petterson). The system was calibrated with a synthetic CF signal. The CF2 frequency and harmonic content of calls were analyzed from spectra obtained with fast Fourier transforms calculated for 4,096–16,384 data points using Hanning windows. The duration of the CF component and the duration and bandwidth of the downward FM component were measured from spectrograms using optimally adjusted time and frequency windows. For individuals with FAL >30 mm, the mean call parameters and SDs were calculated from 10 to 20 calls. For younger individuals, who typically vocalized only sporadically, the analysis is based on at least 5 calls. After sound recording, the bats were either returned to the cave or used in electrophysiological experiments.
Echolocation calls of young bats were recorded while they were sitting unrestrained on a pillow and facing the microphone in 2 situations. 1) Stationary bat: the pillow was placed at 50 cm in front of the microphone; the bat vocalized spontaneously. 2) Passive body motion: the experimenter moved the pillow up and down at a rate of about 2–4 Hz. This procedure is neither natural nor standardized and introduces noncalibrated and variable Doppler shifts of recorded call frequency. However, out of all methods tested to induce vocalizations in young bats [hissing, finger snapping, or key jingling to get the bats' attention; launching them from a defined height; placing them on a pendulum that moved at a defined speed (see following text); confronting them with air flow], it was the most effective in inducing the production of CF-FM calls in many of the youngest bats that did not vocalize spontaneously. We tested bats of all ages in this situation because sound analysis revealed significant differences to the stationary case not only in CF2 frequency but also in duration and repetition rate of emitted calls.
Exposure to ambient temperature (30–33°C) for the purpose of sound recordings was kept as brief as possible. Echolocation calls of adults were recorded from individuals at rest, hanging unstrained in an open cage. The pulse CF2 frequency in this situation is termed "resting frequency."
Doppler-shift compensation (DSC) was tested in bats swung on a pendulum (method after Gaioni et al. 1990; Henson et al. 1980). The bat was restrained in a cushioned body mold with the head protruding. The bat holder was mounted on a 2-m-long pendulum that was suspended from the ceiling of a half-open veranda. The microphone was attached to the pendulum at a distance of 30 cm in front of the bat. Pendulum position was monitored by a potentiometer positioned in the pendulum hinge and adding the DC signal to the microphone output. The pendulum was released manually at a constant height of 152 cm and reached a peak velocity of 5.08 m/s, which is slightly faster than the naturally occurring maximal flight velocity of 4.66 m/s (Lancaster et al. 1992). A large sheet of plywood served as a target at 30 cm beyond the most forward point of the swing. Additionally, the bat was confronted with echoes from multiple targets along the line of motion (house wall, ceiling, floor, a column). Rather than causing a deterioration of the behavioral response, multiple targets parallel to the line of movement are known to improve DSC performance (Keating et al. 1994). Each test session consisted of at least 3 to 6 trials. Each trial included 2 forward and 2 backward swings and recording of echolocation calls for 10 s. The frequency of emitted pulses was determined from spectra; echo Doppler-shift was calculated from the Doppler formula.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Vocal behavior and repertoire in young mustached bats
FAL 22–29 MM. Specimens of FAL <30 mm were rather inactive and quiet for most of the time both in the cave and in captivity. In response to hisses or finger snapping they performed oriented ear and head movements, and thus seemed to be able to passively localize a sound source. The vocal repertoire at this developmental stage consisted of different types of communication calls (Fig. 2, A–C) and echolocation call-like CF-FM signals (Fig. 2D). Figure 2A illustrates a sequence of monosyllable, multiharmonic tonal signals that were produced at long interpulse intervals by a specimen of 22 mm FAL while it was handled. Signals of similar structure were recorded from groups of young bats (FAL >30 mm) while they were interacting in the cage. These communication signals include long (11–52 ms) and short (8–10 ms) upward FM sweeps, long (10–32 ms) and short (4–9 ms) downward sweeps, short (5–9 ms) or long (10–37 ms) CF or quasi-CF signals, long (25–100 ms) arched FM signals, and combinations of quasi-CF and FM signals. Typically, the 1st harmonic around 12–14 kHz is most intense, and, with few exceptions, sound energy was below 50 kHz. A signal of larger bandwidth containing 8 harmonics is shown in Fig. 2B. All of these signals are faint and audible to healthy human ears only at very close range (<50 cm). Loud signals audible to human listeners at distances of several meters occurred in the form of buzzing calls that consisted of stereotyped trains of brief (2–3 ms) downward FM sweeps (Fig. 2C) emitted at high repetition rates and regular interpulse intervals (30–33 ms). The signals contained up to 4 harmonics with the dominant 1st harmonic sweeping from 15 to 10 kHz. Buzzing sounds were emitted over several seconds. They typically occurred before and after feeding time in captivity and were associated with increased motor activity: the buzzing bats approached other group members, and explored their skin with mouth and nose. Buzzing typically occurred during interaction with conspecifics of the same age or older, and in one case (FAL 21 mm) while being hand-fed. Neither at this age nor in older bats did we register calls that would qualify as "isolation calls" known from infants of many other bat species.
|
Figure 2D illustrates 2 calls out of a sequence that was recorded from a bat with 22 mm FAL while its body was moved passively up and down. The first signal consists of 2 syllables. It starts with a long upward sweep followed by a 10 ms CF component. The 1st and 3rd harmonics are dominant, and the frequency of the CF1 is at 24 kHz. A terminal brief FM downward sweep occurs after a 2 ms gap. The next 4 pulses of the sequence (only one pulse shown) are candidates for the precursors of echolocation calls: they have a stereotyped CF-FM time structure, with a dominant 1st harmonic at 24 kHz. They thus resemble CF-FM echolocation calls, except that they contain up to 7 harmonics, CF2 is suppressed, and its frequency is about 12 kHz lower than the adult CF2 frequency. While producing these calls, the bat did not open its mouth. Figure 2E shows a sequence of induced CF-FM signals emitted by an individual of 26-mm FAL, which illustrates a variation in harmonic strength and signal onset. CF2 is suppressed in the first 3 signals, whereas the last signal contains all higher harmonics. The second and last signals start with a brief plosive component. Only half of the bats (n = 10 out of 20) from the youngest age group could be induced to emit CF-FM signals, and only one individual (FAL 29 mm) produced CF-FM signals spontaneously.
FAL 30–40 MM. This age group is characterized by heightened motility and vocal activity. Calls recorded from a group of young bats in the cage are shown in Fig. 3A. These are predominantly faint, multiharmonic tonal calls consisting of CF, QCF, and different types of FM signals that are similar to those emitted by younger bats. Buzzing activity persists. Bats that are the focus of attention of the buzzing bats often reacted with loud broadband protest calls. The broadband calls consist of fast periodic amplitude modulations with rippled spectra often terminated with humped or arched FM (Fig. 3B). Examples of echolocation call-like signals are shown in Fig. 3, C–E. These signals were emitted with widely opened mouth. Most CF-FM signals started with brief plosive components and contained 4 harmonics as illustrated by the 2nd call in Fig. 3C. In many signals, the 1st harmonic is most intense but in a significant proportion of the signals, the 2nd harmonic is increasingly prominent (for quantification see following text). Occasionally, CF-FM signals started with brief downward sweeps (Fig. 3D) or prolonged upward FM components (Fig. 3E). Typical for this age group, the individual with a FAL of 34.5 mm frequently interspersed brief FM calls with CF-FM signals. These FM signals started at about the same frequency as the CF2 portion of bimodal calls.
|
Out of this age group (n = 28 bats), only one individual (PP40, FAL 34 mm) did not produce CF-FM calls. All others could be induced to produce orientation calls and n = 15 bats echolocated spontaneously.
FAL >40 MM. Bats of this age were very active, frequently emitting loud audible broadband calls during interactions with conspecifics, and starting at a FAL of about 46 mm, they attempted to fly. All bats with FAL >40 mm (n = 41) echolocated spontaneously, and reached adultlike CF2 frequencies at a FAL of about 51 mm. They could rapidly switch from production of CF-FM signals to the production of broadband calls within time intervals as brief as 20–50 ms (Fig. 4A), which is similar to behavior seen in adults. The fundamental of CF-FM signals was suppressed as it is in adult calls. There was some variability in the expression of the upper harmonics as shown in Fig. 4B, where the first 2 signals strongly emphasize the CF3 component relative to the CF2 component.
|
Spectral parameters of CF-FM signals
CF2 FREQUENCIES. The CF2 frequency of induced and spontaneous CF-FM calls is significantly correlated with FAL (Fig. 5A). The mean CF2 frequency of induced calls increases from about 47.4 kHz in the youngest specimen (FAL 22 mm) to values of 60.5 kHz in subadults with a FAL of 51 mm (Spearman's rank rs = 0.830; P = 0.01), and the CF2 of spontaneous echolocation signals increases from about 52 kHz (FAL = 28 mm) to 60.5 kHz in the oldest animals (rs = 0.709; P = 0.01). These values are slightly lower than the mean CF2 frequencies measured in adults [61.5 ± 0.35 kHz in females (n = 23); 61.0 ± 0.48 kHz in males (n = 6)].
|
The data points for induced and spontaneous calls were fitted separately with 2nd-order regression lines. The mean CF2 frequency of spontaneous vocalizations was higher than that of induced calls with the difference decreasing with age. This is attributed to the fact that, in individual bats that were tested in both situations (n = 29), the frequency of induced calls was typically lower than that of spontaneous calls (n = 22 bats; paired t-test; P < 0.01) with the largest differences observed in the youngest age group (Fig. 5B). There were differences of up to 3 kHz in the mean CF2 frequency of spontaneous and induced calls at FAL <40 mm, but the differences were typically <300 Hz in larger individuals.
There was a large variation in CF2 frequency across individuals of the same FAL during early postnatal development. Remarkably, 2 bats (FAL 35 and 45 mm) emitted calls of 63 and 64 kHz, well above the CF2 frequencies of adults. Interindividual variation became smaller in the group of bats with FAL >45 mm, but the variation of 1.8 kHz is still above the 1.3 kHz interindividual variation that is seen in the adult population (n = 29 bats).
The SDs of CF2 frequency (Fig. 5C) did not decrease gradually with increasing FAL, but decreased abruptly when the FAL reached 40 mm (i.e., around the end of the 3rd postnatal week). In individuals with FAL <40 mm, the SDs reached maximum values of ±1.5 kHz, whereas bats with FAL >40 mm typically emitted within a frequency band of ±250 Hz. Some individuals of the oldest age group (FAL 46–51 mm, i.e., within the 4th and 5th postnatal weeks) emitted with adultlike precision of about ±100 Hz.
Adult female bats emitted at a slightly higher CF2 frequency than male bats but there were no consistent differences between the CF2 frequencies of male and female bats during postnatal development (Fig. 6A). CF2 frequencies of induced calls were not significantly different between sexes in the group of bats with FAL <30 mm (Mann–Whitney test; P = 0.443) and in bats with FAL >40 mm (Mann–Whitney test; P = 0.138). In the age group with FAL 30–40 mm, male bats emitted at higher frequencies than females (Mann–Whitney; P = 0.001).
|
In hand-raised bats, FAL and CF2 frequencies increased during development (Fig. 6, B and C) but neither of the 2 bats reached the values that are typical for the normal population during the observed time span. In PP8, which was captured at a FAL of 27 mm, the mean CF2 frequency of induced calls increased from 48.1 to 56.8 kHz within the following 18 days. During this time period, FAL reached 35 mm. PP1 was captured at a FAL of 22 mm and did not echolocate during the first days of captivity. At day 5 it could be provoked to produce echolocation calls with the CF2 frequency at 52.2 kHz. The CF2 frequencies of induced vocalizations increased to a value of 56.0 kHz during the following 18 days. It reached a FAL of 33 mm. In both bats, spontaneous echolocation calls occurred later than induced calls. The temporal delay to the onset of induced vocalizations amounted to 14 days in PP1. PP8 started to echolocate spontaneously on day 10 of captivity.
BANDWIDTH OF TERMINAL FM2. The bandwidth of the terminal FM sweep of induced calls increased with FAL (Fig. 6D; rs = 0.2; P = 0.01) from a mean value of 11 ± 2 kHz in the 1st postnatal week (FALs 22–27 mm) to a mean value of 13 ± 2 kHz in the 4th and 5th postnatal weeks (FALs 46–51 mm), whereas the values for spontaneous calls were not significantly correlated with FAL (rs = 0.03). Interindividual variation was large. The bandwidth of the terminal FM of induced vocalizations varies between 7 and 15 kHz in bats <30 mm and between 10 to 17 kHz at later ages. The bandwidth of the terminal FM sweep of spontaneously emitted calls ranges between 11 and 16 kHz in infants with FAL <35 mm and between 8 to 18 kHz in older bats.
HARMONIC COMPOSITION. The relative strength of the 1st, 3rd, and 4th harmonic CF components (CF1, CF3, CF4) with respect to CF2 is plotted versus FAL in Fig. 7. In the youngest group (FAL <30 mm), CF1 was up to 22 dB louder than CF2. There was a pronounced decrease in relative strength of CF1 with age up to a FAL of 40 mm to a mean value of –23 dB and a slightly further decrease to adultlike values of –28 dB at a FAL of 49–40 mm (Fig. 7A). CF3 and CF4 were more pronounced in calls of infants with FAL <30 mm than at a later age (Fig. 7, B and C).
|
In contrast to the stereotyped nature of adult echolocation signals, CF-FM calls of infant bats exhibited a variety of harmonic patterns (Fig. 8). To analyze the relation between harmonic patterns and age groups, the observed harmonic patterns were classified into 9 types according to the relative strength of CF1–CF3 (Figs. 8 and 9). In patterns 1–5, CF1 dominates or is as strong (within ±5 dB) as CF2 (Fig. 8). These patterns differ in relative expression of CF2 and CF3. Pattern 1 was produced only by infants with FAL <30 mm and was characterized by a pronounced attenuation of CF2 relative to both CF1 and CF3. Mean relative levels (dB relative to CF2) of CF1 and CF3 amounted to +22 ± 8 and +11 ± 5 dB, respectively (n = 29). While producing this signal, the mouth appeared closed or was opened just a little. Pattern 2 is dominated by CF1 (+17 ± 6 dB) with CF2 and CF3 of equal intensity (n = 33). Pattern 3 exhibits a staircase formation in level of CF1–CF3, with CF1 being the most intense (+12 ± 5 dB) and CF3 being the least intense component (–11 ± 4 dB; n = 22). In pattern 4, CF1–CF3 are of equal intensity (within ±5 dB; n = 26). In pattern 5, CF1 and CF2 are equally strong and CF3 is attenuated (–17 ± 4 dB; n = 23).
|
|
Patterns 6–8 are characterized by an attenuated CF1. In pattern 6, CF1 is attenuated by –11 ± 4 dB relative to CF2 and CF3 is –15 ± 7 dB weaker than CF2 (n = 81). Pattern 7 differs from pattern 6 by a relative attenuation of CF1 by more than 20 dB (CF1: –28 ± 6 dB; CF3: –19 ± 7 dB; n = 125). Pattern 8 exhibits an attenuated CF1 (–19 ± 8 dB; n = 46) and equally strong CF2 and CF3. In pattern 9, CF3 is the loudest component (+10 ± 3 dB; n = 16).
The distribution of harmonic patterns among age groups is shown in Fig. 9. Patterns 1–5 dominate the calls of infants with FAL <30 mm (Fig. 9, top). At a FAL of 30–40 mm (Fig. 9, middle), bats produce a significant proportion of calls that still contain a strong CF1 but they also begin to produce adultlike signals with an attenuated first harmonic (patterns 6 and 7), with pattern 6 being most common. In bats with FAL >40 mm (Fig. 9, bottom), the variety of harmonic patterns was restricted to patterns 6–9 with pattern 7 representing the prevailing signal type. The development of harmonic structure in CF-FM calls of hand-raised bats followed a similar sequence (not shown).
To display the relationships among harmonic patterns and to separate the age stages by maximizing the total inertia (i.e., a measure for the variance), a correspondence analysis (Legendre and Legendre 1998) was performed. The correspondence analysis is a dimension-reduction technique based on eigenanalysis, and thus similar to a principal-component analysis. The results of the correspondence analysis are given in a distance biplot (Fig. 10), which is a plot of the first 2 principal axes in which the objects (i.e., the age groups T1–T3) and the descriptors (i.e., the harmonic patterns H1–H9) are displayed. The age groups (T1–T3) form the centroids of the patterns (H1–H9), and those patterns that are closest to the centroids largely contribute to the age stages. The youngest group (T1;
|
FAL <30 mm) is predominantly defined by H1–H4; the intermediate group (T2; FAL 30–39 mm) is predominantly defined by H4, H5, H6, and H9; and the oldest group (T3; FAL 40–51 mm) by H7 and H8, respectively. The positions of the age groups are determined by the 
2 distance, which is preserved by the correspondence analysis. Thus the larger the distance between the age groups, the more dissimilar they are in terms of the harmonic patterns. For example, T1 and T3 have the largest separation because of their most dissimilar patterns.
Temporal parameters of CF-FM signals
DURATION. The duration of the CF2 components of induced and spontaneous echolocation calls is plotted against FAL in Fig. 11A. There is a large increase in CF2 duration of spontaneous vocalizations with age (rs = 0.667; P = 0.01). In the youngest bats (FAL about 30 mm) the mean duration amounts to 5–7 ms; in the oldest specimens (FAL 50–51 mm), the mean duration is about 16–18 ms.
|
The CF2 duration of induced calls also increases with FAL but the effect is less pronounced (rs = 0.275; P = 0.01). Interindividual variations in call duration are largest in the age group with FAL <30 mm, with durations ranging between 3 and 11 ms [7 ± 2.6 ms (mean ± SD)]. The scatter of values is smaller in older bats (FAL 45–51 mm) and ranges between 5.8 and 9 ms (mean 8.2 ± 1.5 ms).
In all bats tested in both situations, the duration of the CF2 component of spontaneous calls is significantly longer than that of induced calls. The mean duration of the FM2 component of both spontaneous and induced pulses is 3 ± 0.4 ms (Fig. 11B) and shows no significant change with increasing FAL (induced calls: rs = –0.034; spontaneous calls: rs = 0.042).
REPETITION RATE. As a measure for sustained echolocation activity, we analyzed the number of pulses emitted during 2.5 s of recording time and expressed it in pulses/s (Fig. 12A). There were clear differences in sustained echolocation activity among age groups. Bats with FAL <30 mm emitted pulses only sporadically. Maximum rates of induced vocalizations were below 6 pulses/s (mean 2.5 ± 2 pulses/s). In bats with FAL 30–40 mm, induced vocalizations were emitted at higher rates than spontaneous calls (15/s as opposed to 12/s). All bats with FAL >40 mm showed a high sustained echolocation activity. Induced vocalizations occurred at rates higher than 15/s (mean 17 ± 5 pulses/s) with maximum values of 34/s. Spontaneous vocalizations were also produced at higher rates than at earlier ages (mean 9 ± 4.5 pulses/s compared with 4.4 ± 3 pulses/s).
|
Instantaneous repetition rates (pulses/200 ms) also differed among age groups (Fig. 12B). With one exception, instantaneous repetition rates in young bats (FAL <30 mm) are below 4 pulses/200 ms. At intermediate age (FAL 30–39 mm), maximum rates amounted to 7 pulses/200 ms. The highest instantaneous rates (7–11 pulses/200 ms) occurred in bats with FAL >40 mm. With the exception of 2 bats with FAL about 30 mm, the maximum instantaneous rate of induced vocalizations is typically greater than that of spontaneous vocalizations.
The development of sustained echolocation activity and instantaneous emission rates in both of the hand-raised bats was very similar to that observed in the pooled data set. The first vocalizations were produced sporadically (below 6 pulses/s) and sustained activity increased to values 
15 pulses/s, which is comparable to the performance seen in the 40- to 46-mm age group.
Doppler-shift compensation
Young bats of all developmental stages were submitted to behavioral testing for Doppler-shift compensation in the pendulum task. The performance of one adult mustached bat is shown in Fig. 13A that illustrates the pulse frequency data for 2 consecutive forward and backward swings compared with the calculated frequency of the Doppler-shifted echoes. Adults emitted almost continuous trains of echolocation calls during the pendulum swings. The first few pulses produced directly after pendulum release toward the target (first forward swing) were variable in frequency and 
2.5 kHz lower than the CF2 frequency emitted at rest. Such nonspecific lowering of pulse frequency also occurred when the pendulum was released in the backward direction (not shown). After a fast increase in pulse frequency toward the resting level, the bat started to compensate for positive Doppler shifts of echo frequency during the forward swing. It reduced the pulse frequency during the acceleration phase and increased the pulse frequency during the deceleration phase. During the backward swing, the emitted pulse frequency stayed at values around resting frequency. Doppler-shift compensation during the second forward swing was not accompanied by nonspecific lowering of pulse frequency. The maximal compensatory response was smaller in magnitude than the maximum occurring positive Doppler shift of echo frequency (i.e., the bats undercompensated). All adults (n = 3) undercompensated by a characteristic amount (20% in the best case to 30% in the worst case; Table 1). Doppler-shift compensation in adults showed habituation: after about the 4th trial, calls were emitted only during the acceleration phase of the pendulum forward swing and their frequency was not related to velocity (not shown).
|
|
Young bats reacted to the Doppler-shift compensation task with 3 types of behavioral responses (Table 1). The first type of response was the absence of echolocation. This behavior occurred in bats with FAL <40 mm (n = 12), which either did not vocalize at all (n = 9) or emitted few faint monosyllable communication calls (n = 3).
The second type of response is illustrated by the example in Fig. 13B. Echolocation signals were emitted only during the acceleration phase of the forward swings (n = 5 bats; FAL 40–46 mm). The first pulses that were emitted directly after pendulum release were about 2.5 kHz lower than spontaneous echolocation calls emitted at rest, and lower than those of induced echolocation calls. The following pulses successively increased in frequency toward resting frequency (Rf) without reaching it. A smaller, nonspecific lowering of pulse frequency also occurred during the second forward swing.
The third type of behavioral response included the emission of echolocation calls during all phases of pendulum motion (n = 7 bats) without or with Doppler-shift compensation (Fig. 13, C and D). The performance of the individual PP14 shown in Fig. 13C was characterized by a highly variable pulse frequency during all phases of the pendulum swing during the 1st trial (open circles). In the 2nd trial, however, it responded with a systematic lowering of pulse frequency during the second forward swing. This bat had no previous flight experience, had been kept in captivity for 5 days, and had failed to echolocate on the pendulum on the 1st day of captivity (Table 1). Robust Doppler-shift compensation by a young bat that was capable of short active flights and probably had flight experience in the cave before capture is illustrated in Fig. 13D. Undercompensation, however, was more pronounced than in the adult shown in Fig. 13A. None of the 4 young bats that compensated for Doppler shifts reached a performance comparable to that seen in the best adult (Table 1).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Onset of echolocation behavior and properties of developing sonar
A basic problem when studying the development of echolocation is the definition of the precursor of stereotyped orientation calls and the evaluation of the function of the vocalizations produced by young bats: communication or echolocation? Echolocation calls in adult bats can have a double function and also serve as communication signals (Fenton 1985; Matsumura 1979), and some communication signals of P. parnellii possess echolocation signal-like harmonic design (Kanwal et al. 1994). It has been proposed that in phylogeny and ontogeny, echolocation calls develop from communication calls by a continuous change in signal parameters such as duration, frequency content, SPL, modulation rate, and harmonic content (Fenton 1985; Moss 1988; Moss et al. 1997).
As the most parsimonious approach, we investigated the developmental transitions in acoustic properties of vocalizations that most closely resemble the typical CF-FM echolocation calls of the adult. This signal type was already part of the vocal repertoire at the youngest investigated age but was not produced spontaneously by single stationary bats or during group interactions. Its emission could, however, be induced in 50% of bats with FAL <30 mm by subjecting them to rapid body movements. It is thus likely that, shortly after birth, vestibular inputs or general arousal provides a trigger to a vocalization pathway that may ultimately be involved in echolocation at a later age. Such inputs could also account for the finding that neonates of another bat species, Eptesicus, can be provoked to vocalize by launching them from a platform (Moss et al. 1997).
The rather stereotyped nature of echolocation-like CF-FM calls within the youngest age group contrasts with the diversity in design of communication signals at the same age. This suggests that motor pathways controlling emission of echolocation signals and communication signals are already organized in parallel at the beginning of vocal and acoustic function.
Although P. parnellii are able to emit CF-FM signals in the 1st postnatal week, the signals are faint and typically emitted at a low duty cycle. These findings, together with the insensitive cochlear microphonic and N1 thresholds at the CF2 frequency (see companion paper, Kössl et al. 2003) of the auditory system, indicate that acoustic imaging could be used only at close range and with a poor resolution, if used at all.
The transition from a mammal that localizes sound passively, to an echolocator that additionally probes its environment with self-generated sounds occurs within the 2nd postnatal week in young P. parnellii that are not yet able to fly (FAL 30–35 mm). This corresponds well with observations in other bat species where the onset of spontaneous echolocation was timed to the 2nd postnatal week (Brown and Grinnell 1980; Konstantinov 1973; Matsumura 1979; Rübsamen 1987). The echolocation signals emitted by young P. parnellii differ in several aspects from sonar pulses of the adult. This fact has important consequences for the information-bearing capacities of sonar itself and sensory-neuronal feature extraction.
The first salient difference is in the signal-carrier frequency, which is about 0.4 octave lower in the young bats. Recordings of cochlear potentials show that right at the beginning of function both sender and receiver are matched in frequency (see companion paper, Kössl et al. 2003). Assuming that P. parnellii is similar to horseshoe bats (Rübsamen and Schäfer 1990a; Vater 2000), in that a developmental shift occurs in the frequency tuning of the cochlear fovea, this ensures that the CF signals are analyzed by the same sensory and neuronal substrate throughout development.
The second important difference is the much shorter CF duration in young bats. The CF duration of the spontaneously emitted signals ranges between 4 and 8 ms, which corresponds to those of adults calls during the terminal phase of target-directed flight. The short CF duration and thus wider bandwidth implies that the calls of young bats do not provide the frequency resolution required by adult bats that, from the echo to a single call, can precisely measure relative velocity and wing beats of their insect prey. CF duration of spontaneously emitted signals increases continuously during 4–5 wk of age when individuals have reached adultlike FAL and are able to fly. Consequently, the major adaptation of signal design to detect fluttering insects in clutter, and the prerequisite for classifying insects according to flutter patterns in the echo (for reviews, see Neuweiler 1990; Schnitzler 1987), emerges gradually during postnatal development and before the time of exposure to fluttering prey in hunting grounds outside the cave.
The third important difference concerns the distribution of energy between the harmonics. Up to a FAL of 40 mm, the harmonic composition is highly variable with a dominant or less-attenuated fundamental. Consequently, the input signal to target range and velocity-extracting neuronal circuits (for review, see O'Neill 1995) differs from the adult up to an age of about 3 wk.
The fourth important difference concerns the capability to produce, and maintain, high vocalization rates over extended time periods as a prerequisite for effective acoustic imaging. Only young bats, with FAL >40 mm, attain adultlike performances.
In summary, adultlike acoustics of the sonar calls are achieved just before the time when young bats start foraging for insect prey on the wing. At their first flight outside the cave, they are fully equipped with adultlike sender properties (reported here) and cochlear receiver properties (see companion paper, Kössl et al. 2003).
Mechanisms: development of vocal tract acoustics and audiovocal control
HOW IS SOURCE FREQUENCY INCREASED DURING DEVELOPMENT? In P. parnellii, the source frequency of the laryngeal generator is set by the tension of the vocal folds, as it is in other bats (Hartley and Suthers 1990; Suthers 1988). The tension of the vocal folds is regulated by the action of the cricothyroid muscles that are innervated by the superior laryngeal nerves (Griffith 1978; Suthers 1988). These muscles, to some extent, also influence the harmonic composition of calls (Suthers and Durrant 1988).
The most likely cause for the low source frequency in young bats is a reduced tension of vocal folds as a consequence of smaller force development by the cricothyroid muscles. As in other skeletal muscles, the contraction efficiency of this muscle will depend on properties of the muscle fibers, electromechanical coupling, and neural drive. The increase in CF2 could thus involve maturation at the effector (laryngeal muscles, cartilage stiffness, tendons) and at the level of central neuronal control (auditory system; sensory-motor interface; premotor and motor circuits).
In adult P. parnellii, bilateral denervation of anterior and posterior cricothyroid muscles (Suthers and Durrant 1988) causes the fundamental frequency to drop by more than one octave, from values around 30 kHz to values of 10–12 kHz. There is also an increase in harmonic content of the signal from 4–5 to 9 harmonics with the 2nd and 4th harmonics being suppressed. Interestingly, in young bats with FAL <30 mm, we observed a similar signal structure with a fundamental frequency at 24 kHz (Fig. 3) which is about 0.4 octave lower than the typical frequency in adults. This ontogenetic "drop" in emission frequencies is not as large as the drop in emission frequency created by complete denervation of cricothyroid muscles but much larger than the maximal decreases of emission frequency in Doppler-shift–compensating adults (Gaioni et al. 1990; Keating et al. 1994) or during conditions of lowered body temperature (200 Hz; Huffman and Henson 1993). We suggest that in young bats, either the forces generated by the cricothyroid muscles are less than those in adults, or that there is a level of tonic activity in the superior laryngeal nerve in the adult that is not present in the young. Rübsamen and Schäfer (1990b) concluded from experiments where they either disrupted audition or vocalization in young horseshoe bats, that throughout postnatal development, the frequency of echolocation pulses is under auditory feedback control. The fixed innate developmental increase in foveal frequency by 12–14 kHz is assumed to cause a shift of the frequency setpoint of the audiovocal control, thereby causing a parallel increase in vocalization frequency.
WHAT DETERMINES THE NUMBER AND WEIGHTING OF HARMONIC COMPONENTS IN THE CALL? In contrast to the mouth-emitted echolocation call, the tracheal source spectrum of echolocation calls in adult P. parnellii is multiharmonic with a dominant fundamental and decreasing intensities of the higher harmonics. Consequently, the supralaryngeal vocal tract acts as a notch filter for the fundamental based on antiphasic reflections from the nasal passage combined with pharyngeal resonance not only in nose-emitting bats but also in mouth-emitting bats (Hartley and Suthers 1990). The predominance of H1 in calls of young P. parnellii with FAL <40 mm is thus most likely the result of an immaturity of the supralaryngeal system. Its suppression in adults could be achieved by a change in the dimensions of the nasopharyngeal passage system, which is indicated by an increase in size of the facial skull with age. The missing 2nd harmonic in calls of the youngest individuals (FAL <30 mm) could also be attributed to the fact that these signals were emitted with the mouth almost shut. Thus the buccal cavity may contribute to the typical sonar sound structure in adult P. parnellii.
Apart from passive filtering of the fundamental, the vocal system of adult mustached bats can to some extent actively control the relative intensity of the higher harmonics (Gooler and O'Neill 1987). A postnatal refinement of the circuits controlling vocalization could account for the findings that in P. parnellii 40 mm FAL, there is a striking variation in the relative intensities of CF2 and higher harmonics with respect to each other and the fundamental, not only across individuals but within the same individual.
HOW IS THE ACOUSTIC EFFICIENCY INCREASED DURING DEVELOPMENT? The acoustic efficiency of vocal behavior, as judged by sound pressure level and capability to sustain long periods of echolocation call emissions, clearly improves during postnatal development of P. parnellii. Adult nonflying P. parnellii achieve a more extensive vocal activity than other nonperch hunting bat species (Lancaster et al. 1995) that is attributed to specializations of the sublaryngeal system (Griffith 1978; Hartley and Suthers 1990) and of the respiratory muscles (Lancaster and Henson 1995).
Preliminary observations (Vater, unpublished) indicate a postnatal increase in the diameter of the trachea and tracheal sac in young P. parnellii (<36 mm). It can also be assumed that young bats have a smaller lung volume, and less-efficient respiratory muscles. Thus the increase in size and efficiency of the respiratory apparatus and associated musculature is likely to be responsible for the improvement of acoustic efficiency during development.
WHAT DETERMINES DEVELOPMENTAL CHANGES TO THE FREQUENCY–TIME STRUCTURE OF THE CALL? Pye (1980) proposed that different types of echolocation calls (downward, upward FM, CF, or combinations of FM and CF) are generated by simply opening and closing the glottal gate at different phases of the contraction/relaxation cycle of the cricothyroid muscles that modulate the tension of the vocal folds. P. parnellii additionally may use another technique for gating phonation: glottal resistance is kept constant while expiratory muscles decrease or increase laryngeal airflow (Suthers 1988). Consequently, a fine neuronal control of the relative timing of muscular activity and a high degree of neuronal coordination of different muscle groups is a prerequisite for the production of stereotyped echolocation calls in rapid sequence. Our data on signal design in young P. parnellii suggest that these mechanisms are continuously refined in postnatal development. The diversity in design of onset of the CF-FM signals of young bats (prolonged or wrinkled upward FM, downward FM, or plosive components) probably reflects a difference in timing between the contractions of cricothyroid muscles and the opening of the glottal gate. Such a mechanism could also account for the findings that bats of FAL 30–40 mm frequently interspersed linear downward FM with regular CF-FM pulses (see Fig. 4).
WHAT CAUSES THE CALL DURATION TO INCREASE DURING DEVELOPMENT? The duration of the CF2 component emitted by flying adult P. parnellii ranges from 23 ms during the search phase to 4 ms in the terminal phase (Henson et al. 1987; Lancaster et al. 1992; Novick and Vaisnys 1964). During postnatal development, the duration of CF2 components of spontaneously emitted CF-FM signals increases from about 5 to 18 ms. The CF2 signal length at FAL of 30 mm thus corresponds to the briefest adult pulses. The maturation of CF duration may depend on the capability of keeping the glottal gate open for extended time periods, while holding the cricothyroid muscles at constant tension. Counter to this proposal, the induced CF-FM calls in bats <30 mm FAL can be of long duration, and the duration of many CF and QCF communication calls significantly exceed the length of CF-FM signals at each age tested. A more intriguing possibility is the emergence and refinement of voluntary control of signal length. Perhaps the age-related gain in cochlear sensitivity and tuning at the CF2 frequency (see companion paper, Kössl et al. 2003) serves to focus the bats' attention to this frequency range. Furthermore, the bat may learn to exploit the information-bearing capacity of the CF signal, in conjunction with its sensory specialization, by lengthening the CF duration during spontaneous echolocation.
CF2 duration of induced vocalizations lengthens to a much smaller degree with age. The most likely reason for this phenomenon is the much higher repetition rate of signal emission during passive body motion. Starting at a FAL of about 30–35 mm, young P. parnellii clearly distinguish between a stationary situation and passive body motion and adjust their signal length and repetition rate.
The variation in CF2 frequency is greater in young bats
Interindividual variations in CF2 frequencies are greater in the population of young bats of comparable FAL than in the adult population, and the CF2 of 2 young bats (63, 64 kHz) was even higher than that in adults. There are several possible causes for these findings. Differences in developmental rate among individuals are likely to occur depending on nutritional state; thus individuals of the same FAL may not have the same age and differ in maturation of the audiovocal system. This could account for some of the variation at the same FAL but not for CF2 values above the adult frequency. Another possibility would be a mismatch of auditory tuning and vocal behavior in young bats and thus the necessity of a learning process during sensory-motor development. The companion paper (Kössl et al. 2003) describes how CM resonance can also be elicited by stimuli at frequencies of 64 kHz, which is considerably above the natural frequency of the resonator of 58 kHz at this age.
Standard deviations in CF2 frequency were large in bats <40 mm FAL and decreased to almost adult values at a FAL of 40 mm. This improvement in the precision of CF2 frequency control could correlate with the finding that cochlear sensitivity for CF2 as well as the Q-value of the cochlear resonance increase relatively fast until a stage of 35- to 40-mm FAL is reached and the subsequent further improvement is only small (see companion paper, Kössl et al. 2003). Furthermore, at a FAL of 40 mm, echolocation signals are typically emitted at high duty cycles matching those in the adult. In addition to the maturation of cochlear filter properties, this feature may also be critical for an improved precision in frequency control, given that in adult P. parnellii, production of rapid sequences of calls tends to reduce the SD in call frequency (Gaioni et al. 1990). The increased emission rate and frequency stability of consecutive pulses may represent crucial prerequisites for the onset of Doppler-shift compensation.
Doppler-shift compensation and the influence of nonauditory cues on echolocation performance
Doppler-shift compensation is one of the most striking examples of precise audiovocal control (Schnitzler 1968). This behavior occurs rather late in postnatal development of P. parnellii, emerging only in bats with FAL >40 mm. Doppler-shift compensation was most robust in individuals of FAL 48–50 mm, which were capable of short active flights but had not yet reached the adult CF2 frequency and cochlear tuning properties (see companion paper, Kössl et al. 2003).
Although the best performance in Doppler-shift compensation achieved in our sample of adult P. parnellii agrees with previous reports (e.g., Gaioni et al. 1990), young bats undercompensated by a larger amount and only one individual (50 mm) reached a performance close to that of the best adult. This implies that at the onset of Doppler-shift compensation, the differences between resting frequency and reference frequency can be considerably larger than those in the adult, 800 Hz compared with an average of 146 ± 98 Hz in adults (Gaioni et al. 1990). Our sample, however, is too small to reveal systematic age- or experience-related trends in the quality of Doppler-shift compensation.
The behavioral data obtained from the pendulum experiment in young bats are interesting with respect to the influence of nonauditory clues on echolocation performance. Vestibular input and behavioral states controlled by attention, arousal, and motivation have been implicated in the modulation of DSC behavior on the pendulum in adults (e.g., Gaioni et al. 1990; Keating et al. 1994). Their influence appears to be expressed by a nonspecific lowering of call frequency at swing start, and habituation of DSC behavior after multiple swings.
Our data on spontaneous and induced CF-FM calls show clearly that young bats with FAL <40 mm are competent to produce echolocation call-like signals. Their failure to echolocate in the pendulum task could be simply related to the fact that they were unfamiliar with the situation and restrained in a holder. The performance of the youngest bats that produced echolocation calls on the pendulum resembled the performance of habituated adults. A triggering of vocal activity by the onset of passive motion, combined with a lack of specific auditory attention to the target, could explain this behavior. Interestingly, there were several parallels between pulse parameters during the transient emission of bursts of CF-FM signals at the start of the forward swing seen by us and others (Gaioni et al. 1990) and those of CF-FM signals that were induced by simply swinging young bats up and down. These include high repetition rate, short pulse duration, and a significantly lower frequency than that of spontaneous echolocation calls.
Comparative and evolutionary aspects
The development of echolocation in P. parnellii and old-world CF-FM bats, both of which have evolved Doppler-sensitive sonar in convergent evolution (e.g., Schnitzler and Henson 1980), reveals several interesting parallels. As a common feature, CF2 frequency increases with age by about 1/3 octave in Rhinolophus (Konstantinov 1973; Matsumura 1979; Rübsamen 1987; Rübsamen and Schäfer 1990a), and more than an octave in Hipposideros speoris (52 to 133 kHz: Habersetzer and Marimuthu 1986). Similar to P. parnellii, the echolocation calls of young Rhinolophoidea contain multiple harmonics with the 1st harmonic being dominant until about the end of the 3rd postnatal week. A developmental suppression of the 1st harmonic is thus a common pattern in nose-emitting and mouth-emitting species. The main difference in call ontogeny between nose- and mouth-emitting bats is a stronger suppression of H1 and the suppression of the higher harmonics (H3–H5) in nose-emitting species. The anatomical basis and location of the filtering mechanism for higher harmonics is unclear, although it is likely to reside in specializations of the supralaryngeal tract or nasal cavities (Hartley and Suthers 1988; Suthers 1988).
An increase in sound frequency of echolocation calls and call repetition rate also takes place in postnatal development of FM bats (Myotis oxygnathus: Konstantinov 1973
