Ossicular differentiation of airborne and seismic stimuli in the Cape golden mole (Chrysochloris asiatica)

Original Paper

DOI: 10.1007/s00359-005-0070-9

Cite this article as:
Willi, U.B., Bronner, G.N. & Narins, P.M. J Comp Physiol A (2006) 192: 267. doi:10.1007/s00359-005-0070-9


Comparison between the middle ear anatomy of the Cape golden mole (Chrysochloris asiatica), which exhibits a club-shaped malleus head, and the Desert golden mole (Eremitalpa granti), with a ball-shaped malleus head, suggests differences in sensitivity to airborne sound. Scanning laser Doppler vibrometric measurements of the ossicular behavior in response to both vibration and airborne sound were made in C. asiatica. Two distinct vibrational modes were observed. In response to low-frequency vibration (70–200 Hz), the malleus oscillates about the ligament of the short process of the incus, whereas in response to high-frequency airborne sound (1–6 kHz) the ossicular chain rotates about the long axis of malleus. It is proposed that the club-shaped malleus head in C. asiatica constitutes an adaptation towards bimodal hearing—sensitivity to substrate vibrations and airborne sound. Possible functional differences between these two middle ear types are discussed.


Middle ear Bimodal Sound Vibration Chysochloridae 



Anterior process of the malleus


Lenticular process of the incus


Scanning laser Doppler vibrometer


Short process of the incus


Tympanic membrane


The members of the golden mole family (Chrysochloridae) are small fossorial mammals inhabiting sub-Saharan Africa. They exhibit a set of specializations for life underground, are image blind (Van der Vyver Nolte 1968; Fielden at al. 1990; Skinner and Smithers 1990; Nowak 1999) and they are “thought to rely heavily on the sense of smell and hearing” (Von Meyer et al. 1995) like other fossorial species (Burda et al. 1990; Francescoli 2000). The lack of light and the poor propagation of airborne sound in the underground environment (Heth et al. 1986) limit the importance and reliability of the corresponding senses. The presence of conspecifics, prey and most predators is conveyed via substrate vibrations, and long-distance communication via seismic signaling has been considered as the principal communication modality in subterranean mammals (Nevo et al. 1991).

The inertial bone conduction is the proposed means by which substrate vibrations are detected in some golden mole species (Henson 1974; Kuyper 1984; Fielden et al. 1990; Hickman 1990; Lombard and Hetherington 1993; Narins et al. 1997; Mason 1998, 1999; Mason and Narins 2002). This mechanism implies the presence of a mass that impedes ossicular motion while the skull vibrates, and thereby induces relative motion between stapes and skull—an effective stimulus for the inner ear.

The Cape golden mole (Chrysochloris asiatica) is a member of the family Chrysochloridae, within which a variety of middle ear adaptations are found. An overview of golden mole middle ear morphology is given by Mason (2003a), and Fig. 1 illustrates five examples from his study. The most obvious variations concern the relative size and shape of the malleus. The ‘freely mobile’ middle ear (see Fleischer 1978) exhibited by Amblysomushottentotus (Fig. 1a) is considered primitive for golden moles (Mason 2003a; Von Meyer et al. 1995), and the hypertrophied mallei in C. asiatica (Fig. 1b), C. stuhlmanni (Fig. 1c), Eremitalpa (Fig. 1d) and Chrysospalax (Fig. 1e) represent middle ear specializations for the detection of seismic vibrations (Narins et al. 1997; Mason and Narins 2002; Mason 2003b). Among the hypertrophied middle ear types, the malleus head shows two distinct forms: spherical or ball-shaped in Eremitalpa and Chrysospalax and club-shaped in Chrysochloris and Cryptochloris. The need for additional mass in order to detect substrate vibrations by means of inertial bone conduction seems straightforward, but what are the functional differences between the two malleus forms? Although the unusual morphology of hypertrophied mallei in some golden moles has been known for a long time (Hyrtl 1845; Doran 1878; Broom 1927; Cooper 1928), possible functional differences between the two malleus forms have not been discussed.
Fig. 1

Middle ear morphologies exhibited by five species of the golden mole family (Chrysochloridae). Only the malleus (M) and the incus (I) of aAmblysomus hottentotus, bChrysochloris asiatica, cChrysochloris stuhlmanni, dEremitalpa granti and eChrysospalax villosus are shown. Illustrations from Morphology of middle ear of golden moles (Chrysochloridae) by Mason (2003a). (Copyright 2003 by The Zoological Society of London. Reprinted with permission)

Hearing in golden moles has not been studied, but a subterranean lifestyle and poor audition generally seem to go hand in hand (Bronchti et al. 1989; Heffner and Heffner 1990, 1992, 1993; Brückmann and Burda 1997). Nevertheless, acoustic communication has been reported in many fossorial mammals (Capranica et al. 1974; Pepper et al. 1991; Credner et al. 1997; Bennett and Faulkes 2000; Francescoli 2000). Since these animals usually use acoustic communication at close range, relatively low thresholds for airborne sound are not under intense selective pressure. Moreover, the degree to which certain species of golden moles rely on audition is unclear.

The goal of the present study was to describe the functionality of one middle ear type, the club-shaped malleus-type ear of C. asiatica (Fig. 1b). The dynamics of this middle ear was tested in response to both substrate-borne vibration and airborne sound.

Materials and methods


Cape golden moles (C. asiatica) were caught in the field of the Cape region (South Africa); after at least 2 weeks of captivity, they were shipped to the University of California, Los Angeles (UCLA). Prior to the initiation of the experiments, the animals remained in captivity for at least 5 days. The present study is based on five specimens: three females and two males. Owing to the highly invasive approach, the animals were euthanized immediately before the experiments with an overdose of isoflurane (Abbott Labs, N. Chicago, IL, USA). Each animal was exposed to the anesthetic gas for 3–4 min in an airtight container (diameter: 0.14 m, height: 0.15 m), 1/3 filled with topsoil. If a heartbeat was absent after an additional lapse of 5 min, the experiment was started after removing the animal from the container.

Vocalization recordings and analysis

Between the arrival of the animals at UCLA and the initiation of the experiments, the animals were handled only twice: once when they were transferred from the shipping cases into the vivarium, and again immediately before euthanization. For about 30 s, a microphone (Brüel & Kjaer, 4134, Naerum, Denmark) was held close (0.1–0.2 m) to each of ten animals in order to record any emitted calls. In a single observation we recorded intra-specific calls between two males sharing a small, shallow tank (0.8×0.7×0.3 m), 1/3 filled with topsoil. The calls were recorded within 0.2 m of the animals and the microphone was moved manually to follow the movements of the animal in view. All recordings were made at room temperature (20–25°C). In order to estimate the distances between the animals when calls were detected, the recording session was video-taped (GL1, Canon, Irvine, CA, USA). All recordings were made with a sampling rate of 48 kHz (16 bit) and later processed with sound analysis software (Avisoft-ASALab Pro, Avisoft Bioacoustics, Berlin, Germany). Spectrograms were generated by using an analyzing filter bandwidth of 162 Hz (overlap: 98%).


The skull was separated from the body and the skin and most soft tissue were removed from the skull. A frontal section parallel to the plane, described by the long axes (Fig. 2a) of the two mallei, divided the skull into two parts (Fig. 2b); the posterior section was mounted on a stainless steel disc (diameter: 38 mm, thickness: 3.2 mm) with acrylic resin (Duralay, Reliance Dental Mfg. Co., Worth, IL, USA) (Fig. 2c). After curing (~5 min), the middle ear was approached laterally by first removing the temporal muscle. The malleus becomes visible through the thin and translucent bony shell that terminates the middle ear cavity laterally. The careful removal of the bony shell led to the opening of the distal portion of the enormous malleus head.
Fig. 2

a Lateral view of skull (C. asiatica). Thin line indicates the long axis of the malleus. b In order to attain optimal coupling between the skull and the vibration exciter, the former was divided into two sections by a frontal plane cut and c the posterior half was mounted on a stainless steel disk with acrylic resin. The disk was secured to a mounting block which could be driven by a vibration exciter. d The long axes of the mallei are horizontally aligned and laterally excited by a vibration exciter. The scanning laser Doppler vibrometer (SLDV) measurement is in line with the driving direction. e The specimen is in the same relative orientation to the SLDV. The middle ear is excited acoustically with a miniature loudspeaker (MS, Knowles, CI-2960) and the sound pressure level is recorded with a probe tube microphone (PTM, Etymotic Research, ER-7C) close to the tympanic membrane

A dome-shaped bony cap constitutes the most proximal portion of the external auditory meatus (Fig. 3). Distally, at the antero–superior quarter of this dome, a round opening (diameter: ca. 1.5 mm) connects the bony portion (dome) with the cartilaginous portion of the external auditory meatus, which is 1.0–1.5 mm in diameter and about 4–5 mm long. The dome covers the “kidney-shaped” (von Meyer et al. 1995) tympanic membrane (TM), with a mean area of 6.83 mm2 (n=5) (Mason 2003a), creating an air space between the TM and the dome of approximately 6 mm3. All soft tissue distal to the TM including the cartilaginous portion of the external auditory meatus was removed. A small opening (diameter: 1 mm) was drilled in the posterior half of the dome. The natural and the drilled openings in the dome were used as entry points for a probe tube microphone and a speaker, respectively, in the acoustic experiment (see Acoustic stimulation).
Fig. 3

Schematic illustration of the external auditory meatus of C. asiatica. The tympanic membrane is facing laterally and is slightly tilted inferiorly. The proximal portion of the external auditory meatus constitutes a bony cap (dome) and forms an extratympanic space. Superiorly an opening in the dome gives way to the cartilaginous, distal portion of the external auditory meatus. BR brain; DO dome; EAM external auditory meatus; MEC middle ear cavity; SK skin; ST soft tissue; TM tympanic membrane

Seismic stimulation

After the surgical preparation, the stainless steel disc was inserted into the circular impression in an aluminum block (44×44×13 mm) (Fig. 2c). In order to align the skull and the laser beam within the horizontal plane (Fig. 2d), the disk could be rotated within the circular impression, and affixed to the block by tightening the screws at each corner. The aluminum block was attached to a vibration exciter (Brüel & Kjaer, 4809, Naerum, Denmark) for lateral seismic stimulation (Fig. 2d). Lateral seismic stimulation was chosen since the SPI and the LPI reside in a plane perpendicular to the measurement axis (laser beam), which allows calculation of the medio–lateral LPI-velocity based on all three motion components. Furthermore, it was shown that the ossicular response is about 6 dB greater for lateral compared to vertical seismic stimulation (Willi et al. 2005).

In order to get rough vertical alignment of the specimen relative to the scanning laser Doppler vibrometer (SLDV), the vibration exciter was placed on a jack (Big Jack, Precision Scientific Co., Chicago, IL, USA). Fine vertical alignment was facilitated by a telescopic vertical lifting column (Fig. 4). The applied signal (periodic chirp) swept upward from 10 to 600 Hz and vibrated the skull laterally (Fig. 2d) at a constant velocity amplitude of (1±0.4)×10−4 m/s. Besides the malleus motion, the skull motion was measured at 3–5 points on the measurement grid for reference purposes. The signal driving the vibration exciter served as a phase reference for all measurement points.
Fig. 4

Setup for both acoustic and seismic stimulation. Except for the attachment of the loudspeaker and the probe tube microphone to the osseous external meatus, the two setups are identical. 1 SLDV; 2 telescopic vertical lifting column; 3X-Y-translation stage; 4 specimen holder; 5 vibration exciter; 6Z-translation stage; 7 rubber mat (6 mm); 8 open cell foam mat (25 mm); 9 laminar-flow isolator

Acoustic stimulation

For the acoustic stimulation the skull remained in the identical position as for seismic stimulation. Prior to the experiments, the probe tube microphone (Etymotic Research, ER-7C) was tested against a calibrated microphone (Brüel & Kjaer, 4134, Naerum, Denmark) in an anechoic booth. Over the frequency range of 0.5–10 kHz, its frequency response deviated by no more than ±1.8 dB from the calibrated microphone. The slope of the phase response was considered in the analysis of the results (see Relative ossicular response). The probe tube microphone was introduced into the posterior (artificial) opening of the dome and was fixed to it using topical tissue adhesive (Nexaband, Abbott Labs, N. Chicago, IL, USA). The tip of a micropipette (Narishige, MM3, Tokyo, Japan) was attached to the outlet of a miniature loudspeaker (Knowles, CI-2960, Itasca, IL, USA) and the tapered end was inserted into the natural opening of the dome, the connection between the dome and the cartilaginous external auditory meatus (Fig. 2e). The acoustic signal (periodic chirp) swept from 0.1 to 10 kHz at a constant sound pressure level of 90±3 dB SPL. A compensation file was applied in order to obtain this flat frequency response. The phase of the sound pressure at the TM served as the phase reference for the velocity measurements.

Velocity measurements

The dynamic response of the malleus was measured by means of scanning laser Doppler vibrometry (SLDV). The system (PSV-300, Polytec, Waldbronn, Germany) allows automated velocity measurements at multiple points. The accessible distal portion of the malleus head and parts of the skull were covered with a measurement grid comprising 20–40 measurement points. The system includes a built-in video camera (FCB-IX47P, Sony, Minokamo, Japan), which is optically in-line with the laser beam, and displays the selected measurement area on the computer monitor enabling grid selection, positioning and focusing of the laser beam as well as facilitating the positioning of the specimen relative to the SLDV. Such alignment is achieved by both the telescopic vertical lifting column (TGC 8AWD3 Magnetic, Liestal, Switzerland) and an X-Y-translation stage (Newport, Irvine, CA, USA) within the horizontal plane. During velocity measurements in response to both acoustic and seismic stimulation, the laser beam was in line with the medio–lateral axis of the skull, and except for the placement of the microphone and the speaker for acoustic stimulation, the two setups are identical (Fig. 4). The jack that supports the vibration exciter was placed on a rubber mat (6 mm) and an open-cell foam mat (25 mm) in order to attenuate any vibrations reaching the SLDV during seismic stimulation. The efficiency of this isolation was demonstrated in a control experiment (Willi et al. 2005). Ambient seismic disturbances were attenuated further by mounting the entire setup on a high-performance laminar-flow isolator (Newport, RS-4000, Irvine, CA, USA).

A scan comprised of 20–40 points covering the distal portion of the malleus head and parts of the skull for reference purposes. Data acquisition occurred at a frequency resolution of 1 Hz for both acoustic and seismic stimulation over the corresponding frequency band. Both amplitude and phase of the measured velocities–and in the case of acoustic stimulation, the amplitude of the sound pressure–were averaged five times for each measurement point. The SLDV system allows animation of the dynamic response of all data points on the measurement grid simultaneously at discrete frequencies.

Relative ossicular response

Since the effective stimulus for the inner ear is the motion of the stapes relative to the skull (oval window), this relative motion was calculated at each measurement point by subtracting the magnitude and phase of the skull velocity from those of the velocity at each point on the malleus. For the acoustic experiment, the resulting phase information now comprised the phase of the ossicular response with respect to the sound pressure level at the TM and the intrinsic phase of the microphone response. The latter phase was compensated based on the microphone calibration as follows: prior to the experiments, the microphone used in the experiment (Etymotic Research, ER-7C) was tested against a calibrated microphone (Brüel & Kjaer, 4134, Naerum, Denmark) in an anechoic booth. The sound pressure level recorded during the experiments was then compensated for the phase as well as amplitude deviations. Then the relative ossicular response was compensated in order to reflect the ossicular response at a constant sound pressure level of 90 dB SPL. This compensation also included the variations in sound pressure level (±3 dB) monitored during the experiment.

In the analysis of the seismic experiment, the skull motion was also subtracted from all measurement points on the malleus, but since the driving signal of the vibration exciter was used as a phase reference, the phase of the skull motion was subtracted from all measurement points. In this way, the phase of the skull motion was set to 0° and became the phase reference for all other measurement points. Finally, the relative ossicular motion was compensated in order to represent the response at a constant skull velocity of precisely 1×10−4 m/s.

Alignment of coordinate system

Following both the acoustic and the seismic measurements, the specimen holder (aluminum block) with the skull tightly affixed to it was removed from the vibration exciter. The structures that masked the proximal portion (SPI, LPI) of the ossicular chain during the experiment were carefully removed. All middle ear structures were preserved in order to ensure that the ossicular chain maintained its position relative to the skull. The skull was then repositioned in the measurement setup and aligned to its previous position. The built-in video camera captured and stored an image of the entire ossicular chain.

During a laser scan, the SLDV-system assigns x- and y-coordinates to each point on the measurement grid. The position and alignment of this coordinate system was scaled and adjusted to anatomical landmarks. The origin of the coordinate system was set at the ligament of the short process of the incus (SPI), whereas the x-axis was coaxial with the long axis of the malleus and the y-axis pointed towards the lenticular process of the incus (LPI). Superimposing the pre- and postexperimental images (Fig. 5a–c) allowed us to refer the coordinates of the measurement grid and the coordinates of the SPI and the LPI to the same coordinate system (Fig. 5d).
Fig. 5

Superimposition of a pre- and b postexperimental images. By superimposing the two images, all structures of the ossicular chain can be referred to the coordinate system of the measurement grid (c). d The coordinate system is aligned to anatomical landmarks, e.g., the short process of the incus (SPI), and the coordinates of the lenticular process of the incus (LPI) were determined

Motion pattern

Velocity measurements on a rigid body measured from a single side can be decomposed into three motion components: a translation in line with the measurement axis (vzt), and two rotations with their axes within the plane perpendicular to the measurement axis (ωx, ωy). The three motion components were calculated by applying the rigid body motion equation to the coordinates and the motion measured at 10–15 points on the malleus. For a more detailed description of this analysis procedure, we refer to Willi (2003) and Willi et al. (2005).

Once the three motion components were calculated, the velocity at any point in the coordinate system could be determined with the rigid body motion equation by applying the coordinates of any point and the three motion components. This method was used to display the motion pattern at discrete frequencies and to calculate the motion of the LPI in the medio–lateral direction over the frequency range tested. In order to justify the reconstruction of the ossicular motion outside of the measured area based on the dynamic response of the measured area, the ossicular chain must dynamically function as one unit over the frequency range tested. One way to test the rigidity of the ossicular chain is to display the motion of several points that reside on a straight line of the measurement grid in the Gaussian plane (real–imaginary) at a discrete frequency (Schön and Müller 1999; Ferrazzini 2003). If all points on that line are represented in a straight line in the Gaussian plane, the structures involved are considered to function as a rigid body at that frequency. Since velocity measurements were to be made on the malleus and the LPI (the structure for which the motion was to be reconstructed), resides on the incus, the described test had to be applied to these two ossicles. These were exposed by opening the middle ear cavity on its posterior side.


Control experiment

Sixteen points, four on the incus and 12 on the malleus, that were located on a line extending from the SPI to the distal end of the malleus head were measured with the SLDV in response to the acoustic stimulus (periodic chirp: 0.1–10 kHz at 90 dB SPL). Figure 6 shows the real–imaginary diagram for these points for five frequencies (1, 2, 4, 6 and 8 kHz). For all five frequencies tested, the points on the incus and the points of the malleus fall on reasonably straight lines. This control experiment was performed in three specimens and the results were consistent, indicating that motion reconstruction within the incudo-mallear complex is justified.
Fig. 6

Rigidity test for the incudo-mallear complex. The velocity measurements of 16 points (four on the incus, one on the malleus) on a straight line extending from the SPI to the distal end of the malleus head fall on a straight line in the real–imaginary space at five discrete frequencies

Ossicular response

The medio–lateral motion of the LPI was calculated for the same five right ears for seismic and acoustic stimulation. The dataset was reduced to these five ears since the LPI motion was successfully reconstructed for them in response to both modalities. Their amplitude- and phase responses are shown in Fig. 7a and b, respectively. All responses were shifted along the logarithmic frequency axis to align the resonant frequencies at 150 Hz for seismic stimulation and at 2 kHz for acoustic stimulation, respectively. In order to significantly reduce computational processing time, the original frequency resolution of 1 Hz was reduced to 20 Hz by averaging the velocities at 20 consecutive frequencies for each measurement point. Both decomposition of the motion into three components and motion reconstruction at the LPI were calculated for each frequency.
Fig. 7

Reconstructed LPI-motion in medio–lateral direction in response to a seismic and b acoustic stimulation. For both stimuli the LPI response is shown for the same five right ears. All responses were shifted along the logarithmic frequency scale to align all resonant frequencies at 150 Hz (seismic stimulation) and 2000 Hz (acoustic stimulation), respectively. c For each of the five original non-frequency-shifted LPI responses, the bandwidth at 6, 12 and 18 dB below the peak response was determined. The horizontal bars in this figure represent the frequency range at which these bandwidths overlap for all five LPI responses

For seismic stimulation the resonant frequencies at the LPI were between 71 and 200 Hz (mean: 135.7±49 Hz). Below resonance, relative motion between the skull and the incus is very small so that the two structures essentially move in phase whereas above resonance, they move out of phase by 180°. In response to the acoustic stimulus, the LPI shows a resonance between 1.3 and 2.2 kHz (mean: 1.76±0.43 kHz). At resonance, amplitudes reach 1.4±0.6×10−4 m/s. In four out of five specimens, a secondary resonance was observed between 100 and 200 Hz (mean: 145±48 Hz), which coincides well with the resonant frequency in response to the seismic stimulation in the corresponding specimen. Owing to the frequency shift imposed on all responses for resonant frequency alignment (see above), some of the secondary resonances were shifted to frequencies below 100 Hz, and thus do not appear in Fig. 7b. Each resonance is characterized by a local maximum and an abrupt phase shift.

From each original (non-frequency-shifted) LPI response curve (n=5), the bandwidths at 6, 12 and 18 dB below the peak response were determined. Figure 7c shows the frequency ranges over which all five LPIs responded (i.e., the frequencies common to all five LPI responses) for the three relative levels. They are 1.8–2.5 kHz (−6 dB), 1.07–4.0 kHz (−12 dB) and 0.98–5.3 kHz (−18 dB).

In order to visualize the ossicular motion pattern in response to both stimulus modalities, we use the reconstructed iso-velocity amplitude maps generously covering the ossicular chain (Fig. 8). These maps represent the results obtained from the left ear of specimen #14. At low frequencies (115 and 120 Hz, respectively), the motion patterns for seismic and acoustic stimulation resemble each other (Fig. 8a, b). The axis of rotation passes through the SPI and is nearly perpendicular to the long axis of the malleus. For seismic stimulation, this rotation axis is maintained up to 600 Hz. However, with increasing frequency of the acoustic signal, this axis of rotation turns clockwise (for lateral view of the left ear), and at the resonant frequency (1.4 kHz) this axis is in line with the long axis of the malleus (Fig. 8c–e). The orientation of the rotation axis does not change uniformly with frequency: between 120 and 200 Hz, it rotates by about 12°, whereas between 200 and 300 Hz the axis performs a clockwise rotation of about 75°. After this rapid change it takes another 1.1 kHz in order to reach the mode at which the ossicular chain rotates precisely through the long axis of the malleus head. This mode is maintained up to about 6 kHz, above which there is a transition to the next higher mode (data not shown).
Fig. 8

Motion patterns of the ossicular chain in response to a seismic and be acoustic stimulation at different frequencies indicated by iso-velocity lines. The rotation axis of each mode is defined by the iso-velocity lines of smallest amplitudes (center of blue lines). At low frequencies the rotation axis is almost perpendicular to the long axis of the malleus (ac), whereas with increasing frequency it approaches the long axis of the malleus (d, e). The transition between modes occurs between 200 and 300 Hz (f)


Vocalizations were recorded to determine the relationship between the dominant frequency components of the calls and the frequency band to which the middle ear is most responsive. Solitary animals do not vocalize spontaneously. However, when handled, they may emit distress calls and when two animals were in close proximity to one another, acoustical interactions were observed.

Distress calls during handling were only emitted by four out of ten animals (five females, five males) and all four animals were males. Vocalizations of females were never recorded nor observed. The sound spectrograms of the distress calls of one male are displayed in Fig. 9a and b. Two distinct types of calls were recorded: chirps (Fig. 9a, call-component 1 and call 3; Fig. 9b, calls 4, 5) and noise bursts (Fig. 9a, call-component 2). Noise bursts varied in duration between 0.1 and 0.2 s and covered frequencies from 1 to 20+ kHz with energy peaks at ca. 1.5 kHz, between 4 and 5 kHz and between 10 and 20 kHz, corresponding to the fundamental frequency, third harmonic and higher harmonics of call-component 1, respectively. The first vocalization in Fig. 9a shows a transition between a (1) chirp and (2) a noise burst, but often these two types of calls are separated in time. Chirps appear as short pulses of 0.02 s (Fig. 9b, call 5) with durations up to 0.5 s (Fig. 9b, call 4). They often contain 3–5 harmonics with most energy contained in the fundamental. High harmonic numbers (<10) as seen in call-component 1 in Fig. 9a were only observed when the chirp ended with a noise burst.
Fig. 9

Spectrogram of calls from Chrysochloris asiatica. During handling, the animals emitted two distinct types of sounds: chirps (call component 1 and call 3 in (a), calls 4, 5 in (b)) and noise bursts (call-component 2 in (a)). The two vocalizations clearly differ in the frequency band covered by them and in their harmonic structure. c Chirps emitted by two vocally interacting males in close proximity to one another. In these recordings, it was not possible to assign a specific vocalization to a specific individual

In one instance, two males were placed in the same tank to observe intra-specific interactions; this procedure appeared stressful to both animals, so it was not repeated. In this single observation period, we were able to record calls that occurred when the animals came within 0.1 m of one another. After the first subsurface encounter, one male remained underground, whereas the other male emerged immediately and remained on the surface during the entire observation period (ca. 30 min). When animals were in close contact, they emitted calls, immediately followed by escape behavior of the male on the surface. This behavior was observed four times. The spectrogram in Fig. 9c shows a representative time segment comprising four consecutive chirps recorded during an encounter. These chirps also varied in duration between 0.05 and 0.15 s and contained several harmonics. For all calls recorded (distress calls during handling and the calls recorded during male–male encounters), the fundamental frequency ranged between 1.0 and 2.2 kHz and, except for noise bursts, contained most of the energy in the calls.


The fact that the hypertrophied malleus most likely comprises an adaptation towards vibration detection has been discussed by several authors (Henson 1974; Kuyper 1984; Hickman 1990; Lombard and Hetherington 1993; Mason 1999 ,2003b; Mason and Narins 2002). Moreover, the foraging behavior of the Namib Desert golden mole (Eremitalpa granti namibensis), which exhibits a similar morphological middle ear specialization, puts this sensory modality in a meaningful ecological context (Fielden 1990; Narins et al. 1997).

The results presented in Fig. 7 show that the middle ear ossicles respond to both modalities, substrate-borne vibration and airborne sound. This does not mean that the animal is capable of perceiving both types of information per se. Most middle ears will respond to high-level airborne sounds independent of the animal’s capability to perceive them. So can we draw any conclusions about the significance of the two modalities for the Cape golden mole (C. asiatica) based on the present study?

Vibration detection

For both airborne sound entering the ear via the tympanic route and inertial bone conduction, the effective stimulus for the inner ear is the relative motion between the skull and the stapes. Such a stimulus is generated if two identical wave forms (skull and stapes) of the same amplitude are out of phase or they are in phase but have different amplitudes. However, as long as the skull and the stapes move in phase with the same amplitude, no such stimulus is generated.

The frequency and phase response shown in Fig. 7a clearly indicate a resonance at low frequencies (71–200 Hz) which represents the first vibrational mode of this middle ear. At resonance, the malleus tip moves at a relative amplitude four times higher than the absolute amplitude of the skull (Willi et al. 2005) which is in turn reduced by a factor of four at the LPI by the lever ratio. The latter depends on the anatomy and the orientation of the rotation axis. As a result, at resonance, the LPI moves with an amplitude very close to that of the skull (1×10−4 m/s). This means that if the skull were perfectly coupled to the substrate, the LPI would move at the same amplitude as the substrate. Although perfect coupling is never achieved between the skull of a living animal and the substrate, this result demonstrates the high responsiveness of the ossicular chain to skull vibrations. Quantitatively, it has to be considered that opening the middle ear cavity reduced the stiffness of the air space behind the TM and thereby its impedance at low frequencies. However, the fact that the middle ear of C. asiatica is highly responsive to vibrations at frequencies that propagate with least attenuation in various soil types (Von Dohlen 1981; Aicher and Tautz 1990; Brownell 2001; Hill and Shadley 2001) further supports the suggestion that the ear of this golden mole is adapted to detect substrate vibrations.

In another study (Willi et al. 2006), the ossicular behavior in response to vertical and lateral seismic stimuli was investigated. In both cases, the ossicular chain oscillates about the SPI showing greatest velocities at the distal end of the malleus head in line with the stimulation direction. The response at the tip of the malleus head for lateral stimulation exceeds that for vertical stimulation by about 6 dB. The study further hypothesizes that the orientation of the ossicular chain at certain head positions enables the detection of both the vertical and longitudinal component of the Rayleigh wave, a surface wave that propagates efficiently in the environment of C. asiatica (Narins et al. 1992).

However, the possibility of a somatosensory route for seismic signal detection as proposed by Nevo et al. (1991) for Nannospalax must be considered. It is unknown whether such somatosensory receptors are responsive to vibration or touch. The responsiveness of somatosensory receptors such as Eimer’s organs and Meissner’s corpuscles to vibration was suggested (Armstrong and Quilliam 1961; Quilliam 1966a, b; Klauer et al. 1997) but convincing evidence for their use in golden moles is still lacking. On the other hand, Catania and Kaas (1996) showed that Eimer’s organs in talpid moles respond to a wide range of tactile cues. Tactile receptors around the nose could potentially deliver important information about the structure of the soil and might aid in prey location and capture in the immediate proximity of the animal.

After Barany’s model, the hypertrophied middle ear ossicles in C. asiatica and E. granti fulfill both conditions to augment inertial bone conduction: an increased ossicular mass and the displacement of the center of mass from the axis of rotation (Barany 1938). In the case of C. asiatica, the latter condition must be specified as follows: The input to the inner ear (stapes) resides between the anchorage point of the ossicular motion (SPI) and the center of mass (malleus head). The present study demonstrates that this system is highly responsive to substrate vibrations.

Detection of substrate vibrations by means of a somatosensory route cannot be excluded, but the compact design of the skull suggests that the space-consuming middle ear in C. asiatica serves a crucial function and the present study demonstrates that it is uniquely specialized for the detection of substrate-borne vibrations.


The mean amplitude at the resonant frequency (mean: 1.76±0.43 kHz, n=5) of the LPI in response to airborne sound was (1.4±0.6)×10−4 m/s. LPI responses measured in humans at the same sound pressure level (90 dB) by applying the same techniques revealed similar values between 1 and 2 kHz (Willi 2003). The response of the golden mole’s middle ear in response to sound might have been boosted by experimentally opening the middle ear cavity. Compared to humans, the middle ear cavity volume in the golden mole is very small and opening this cavity most likely changes the impedance of the system more than that in humans. But still, we suggest the response sensitivity to be in a range appropriate for hearing. This is supported by the fact that the frequency band within which all five tested ears respond to the same relative degree (Fig. 7c) coincides closely with the fundamental frequency of their calls. The fact that the fundamental frequencies in the recorded distress calls and the intra-specific contact calls both fall in a frequency range between ~1.0 and 2.0 kHz and contain the major portion of emitted energy suggests that this frequency band plays an important role in intra-specific communication.

As mentioned in the Results, during their short residence in the lab, we never observed vocalizations in females. However, the absence of female vocalizations may be a consequence of seasonal activity patterns exhibited by C. asiatica (G. N. Bronner, unpublished observations). Despite the absence of female calls, the middle ears of both males and females respond mechanically to acoustic stimuli in a similar manner. It would be interesting to investigate sexual dimorphism in middle ear responses in this species, but a larger sample size would be needed than that was available for the present study.

Modes and bimodality

The velocity patterns shown in Fig. 8 demonstrate that the ossicular chain of C. asiatica exhibits two distinct vibrational modes below 6 kHz. Higher modes appear above 6 kHz, but the relative response amplitude at the LPI rolls off continuously up to 10 kHz (Fig. 7b). Both modes rotate about an axis that passes through the SPI, and the rotation axes of the two modes are almost perpendicular to one another. At low frequencies (~100–200 Hz), be it in response to airborne sound or substrate-borne vibration, the club-shaped malleus head oscillates about the SPI showing maximum amplitudes at its superior end and the rotation axis is nearly perpendicular to the long axis of the malleus (first mode). In response to higher frequency sounds (~1–6 kHz), the axis of rotation runs parallel to the long axis of the malleus head (second mode).

This is not the first report of a middle ear exhibiting such distinct modes. Fleischer (1978) proposed two rotation axes of the micro-type middle ear (Fleischer 1974), one in response to low frequencies and the other one in response to high frequencies, the two axes being perpendicular to one another. Fleischer’s idea and considerations appear to apply to the middle ear of C. asiatica as well. Structurally, the micro-type middle ear and that of C. asiatica exhibit significant differences but they have a crucial common feature that enables a second vibrational mode as hypothesized by Fleischer and as found in C. asiatica. That is, both ear types are endowed with an additional mass–the orbicular apophysis in the micro-type middle ear and the hypertrophied malleus head in the middle ear of C. asiatica–which alters the center of the ossicular mass. In the low-frequency mode, the center of mass resides outside the rotation axis. With increasing frequencies, the additional mass impedes the first mode. In order to partly overcome this impedance, a second mode appears at higher frequencies. In both designs, the micro-type middle ear and the middle ear of C. asiatica, the orientation of the rotation axis changes between the two modes as follows: while one anchorage point, the SPI in C. asiatica and the SPI or the gonial in the micro-type middle ear, is maintained in both modes, the rotation axis passes through the center of mass in the second mode. Since the center of mass is now close to the rotation axis, the impeding effect of the additional mass is reduced. Saunders and Summers (1982), working with the mouse middle ear, confirmed the existence of the first mode described by Fleischer. They identified the second mode as well, but minimized its relevance as an inner ear stimulus.

There are mechanisms other than inertial bone conduction by which substrate vibrations can be transmitted through the inner ear (Rado et al. 1989), but the enormous mallear mass in C. asiatica suggests that substrate vibration detection in this species involves just such a mechanism. Inertial bone conduction can be achieved by various designs (Fig. 1). Whether the malleus is club- or ball-shaped, the additional mass and its displacement relative to the rotation axis will foster this mechanism. However, the shape of the additional mass has an effect on the impedance of the second mode, as seen in C. asiatica. A highly simplified model illustrates this in Fig. 10. The weights of the arrows indicate the angular moment of inertia. For the first mode, the angular moment of inertia, I1 is defined by:
$$ I_{1} = \frac{{\text{1}}} {{\text{3}}}ml^{2} $$
where m is the mass and l is the distance of the point mass from the suspension point (see Fig. 10). Although simplified as a point mass in this example, the impedance in this mode depends on the distance between the center of mass and the anchorage point of the mass itself.
Fig. 10

Simplified cylindrical models of the two hypertrophied middle ear types approximating the club-shaped (left) and ball-shaped (right) malleus for C. asiatica and E. granti, respectively. The weight of the arrows indicates the angular moment of inertia of the ossicular chain in response to low-frequency seismic stimulation (1st mode) and high-frequency acoustic stimulation (2nd mode). SPI tip of the short process of the incus

For the second mode for which the rotation axis passes through the center of mass, the angular moment of inertia, I2 is roughly defined by:
$$ I_{2} = \frac{1} {2}mr^{2} $$
where m is the mass of the cylinder and r its radius (Fig. 10). Assuming the angular moment of inertia for the first mode is identical in both designs, that for the second mode is necessarily different, being larger for the ball-shaped malleus (Fig. 10, right model) compared to the club-shaped malleus (Fig. 10, left model). It is unknown if such a second mode exists in E. granti. And even though Van der Vyver Nolte (1968) reported that E. granti exhibited “little to no response to sounds,” comparative data on the hearing capabilities of the two species is lacking. But these simple physical considerations indicate that the club-shaped malleus facilitates a second mode as seen in C. asiatica.

The middle ear of C. asiatica has been shown to respond to substrate vibrations and airborne sound. Its design is an elegant solution to the problem of how a middle ear can exploit the inertia of an increased ossicular mass for detecting substrate vibrations at little cost to the animal’s airborne sound detection capability. Additional experiments are needed to determine if the two vibrational modes of the middle ear apparatus in C. asiatica transmit differential information to the inner ear.


We thank Richard Klufas for carefully manufacturing many parts of the setup and also thank the staff of the Division of Laboratory Animal Medicine (DLAM/UCLA), particularly Dr. Marcelo Couto and Dr. Greg Lawson for their helpful collaboration. We acknowledge the Zoological Society of London for permission to reprint Fig. 1. We thank Dr. Walter Metzner for the loan of the AVISOFT program for sound analysis. This research was supported by NIH grant no. DC00222 to PMN. All animal procedures were approved by the Animal Research Committee (ARC) at the University of California, Los Angeles, USA.

Copyright information

© Springer-Verlag 2005

Authors and Affiliations

  1. 1.Department of Physiological ScienceUniversity of California at Los AngelesLos AngelesUSA
  2. 2.Department of ZoologyUniversity of Cape TownCape TownRepublic of South Africa
  3. 3.Department of Physiological Science and Department of Ecology & Evolutionary BiologyUniversity of California at Los AngelesLos AngelesUSA

Personalised recommendations