The contribution of tympanic transmission to fine temporal signal evaluation in an ultrasonic moth
1 Ecology and Evolutionary Biology, University of Kansas, Lawrence, KS
66045, USA
2 Biological Sciences, University of Missouri-Columbia, Columbia, MO 65211,
USA
* Author for correspondence at present address: 223 Tucker Hall, Biological Sciences, University of Missouri-Columbia, Columbia, MO 65211, USA (e-mail: rafa{at}missouri.edu)
Accepted 21 September 2005
![]() |
Summary |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Key words: bioacoustics, mechanisms of mate choice, Lepidoptera, Pyralidae, Achroia grisella
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
We found that transmission of pulse length was linear with responses only slightly longer than the stimuli, because of tympanic frequency tuning. Transmission of asynchrony interval was precise except for the shortest intervals. We discuss implications for possible mechanisms in the receptor cells that may explain the evaluation of these signal characteristics in A. grisella.
![]() |
Materials and methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Laser vibrometry
Achroia grisella have two tympana located ventrally on the first
abdominal segment (Fig. 2A).
Each tympanum consists of a sclerotized ring surrounding a membrane with a
whitish anterior section and a transparent posterior section
(Fig. 2A;
Knopek and Hintze-Podufal,
1986). The attachment point of the scoloparium that contains the
four receptor cells is in the middle of the posterior section
(Fig. 2A, arrow;
Knopek and Hintze-Podufal,
1986
; Scoble,
1992
).
To prepare the moths for observation, we cooled them at 5°C for 15 min, removed their legs, and fixed them with a mixture of resin and beeswax on a 2 cmx3 cm platform, with the thorax slightly pulled back to expose the tympana. We placed this preparation in a 50 cmx50 cmx50 cm box lined with acoustic foam with one open side. The box was on top of a vibration isolation table (Vibraplane, Kinetic Systems, Boston, MA, USA).
We monitored tympanic vibration using a laser vibrometer (Polytec Compact Laser Vibrometer CLV 1000, with a CLV M030 decoder module, at a sensitivity of 25 mm s-1 V-1; flat frequency response up to 250 kHz; Polytec Inc., Auburn, MA, USA). The laser source was 17 cm from the moth, with the beam approximately perpendicular to the tympanum surface. We directed the beam with a micro-adjustable arm, monitoring its position on the tympanum with a dissecting scope. We estimated the accuracy of the positioning of the laser beam by measuring the distance from the edge of the tympanum to the center of the laser beam in repeated positionings of the laser beam on the scoloparium attachment point of the same individual. Accuracy was 4.4% of tympanum width (mean ± S.E.M.; tympanum width=0.51±0.01 mm, N=14; the width of the laser beam was ca. 0.05 mm; see Fig. 2A).
The laser vibrometer output was high-pass filtered (10 kHz, KH 3202, Krohn-Hite, Avon, MA, USA) and recorded using a custom-made analog-digital (A-D) converter system (16-bit, 250 kHz sampling rate, flat frequency response up to 100 kHz). A trigger signal indicating the timing of stimulus presentation was generated by the playback system (see below) and recorded on a second channel. We used this trigger signal to synchronize recordings of the stimuli and of the tympanic response in the analysis of transmission of temporal characteristics (see below).
To describe the vibration of the tympanum, we initially recorded at six points on the membrane (Fig. 2A, minus the points adjacent to that indicated by the arrow) for ten females and five males. We then increased the number of points to eight (Fig. 2A) for nine additional females. To describe frequency tuning and transmission of stimulus temporal characteristics, we recorded at the attachment point of the scoloparium on the posterior section of the tympanum (N=42 females; Fig. 2A, arrow).
Stimulus generation
To describe the tympanic mode of vibration and frequency tuning, we used
41-ms pulses of broad-band ultrasonic noise (20-100 kHz) with a rise time of
0.5 ms. This stimulus permitted measuring the output of the tympanum over most
of the frequency range relevant to the moths for detecting bat echolocation
cries and conspecific signals
(Rodríguez and Greenfield,
2004).
To describe the transmission of temporal characteristics, we used stimuli
modeled after male A. grisella signals
(Jang and Greenfield, 1996):
short pulses with a trapezoid shape, rise and fall times of 8 µs, and
plateaus of length that we varied in the experiments. Recordings of these
stimuli show peak energy at 90-100 kHz with frequencies down to 25 kHz within
6 dB of the peak. To describe pulse length transmission, we used single pulses
and varied their length (100-330 µs). For asynchrony interval transmission,
we used pairs of 100 µs pulses and varied asynchrony interval (120-1120
µs). These values cover the range of pulse lengths and asynchrony intervals
tested in playback experiments (Jang and
Greenfield, 1996
).
Stimulus presentation
The stimuli were presented with a speaker (Technics 10TH400C) placed 28 cm
from the moth. The speaker membrane was approximately parallel to the ventral
surface of the moth. The stimuli were generated using a custom-designed D-A
converter system (16-bit, 250 kHz sampling rate). The trigger signal (see
above) was generated simultaneously by this system. For each moth, we
presented the stimuli 200 times and averaged the recordings to improve the
signal-to-noise ratio.
Stimulus amplitude was calibrated at the preparation site with no moth
present using a Brüel and Kjaer (Naerum, Denmark) 2231 sound level meter
and a -inch free-field microphone (40 BF, GRAS, Vedbaek, Denmark).
Stimulus amplitude was similar to that of a male signaling at 10 cm (mean=86
dB peSPL re: 0 dB=20 µPa, range=73-92 dB peSPL, N=25 males). We
set the RMS-amplitude of the long-pulse (noise) stimulus to 86 dB SPL, using
the fast setting of the sound level meter. For the short pulses, we set peak
amplitude to 84 dB peSPL.
We obtained reference recordings of the loudspeaker output using the
-inch microphone with its protective grid, placed at the position of
the moths, and amplified by the sound level meter and high-pass filtered (10
kHz, KH 3202, Krohn-Hite, Avon, MA, USA). The protective grid caused the
frequency response of the microphone to decrease by 13 dB octave-1
from 40 to 100 kHz; we corrected for this frequency response before performing
the frequency tuning analysis.
Signal analysis
To improve the accuracy of our temporal measurements, we increased the
sampling rate of our recorded files to 1 MHz in CoolEdit 2000 (Syntrillium
Software Corporation, Phoenix, AZ, USA). Analyses of mode of vibration and
frequency tuning were performed with custom computer programs written in
Matlab (Mathworks, Inc., Natick, MA, USA).
For the analysis of the tympanic mode of vibration, we used a program
provided by Quang Su (Watson School of Engineering and Applied Science, SUNY
Binghamton, USA). This program calculated the phase relationships between
points on the tympanum by calculating transfer functions (fft size=8192
points) between a reference file (stimulus recording) and the laser
recordings. The transfer function describes the amplitude ratio and relative
phase between the laser and stimulus recordings at the frequencies evaluated
(Bendat and Piersol, 1986).
Because the amplitude of sound pressure as recorded with a microphone is
proportional to particle displacement, while the laser recordings are
proportional to velocity, we converted the transfer function magnitude to
displacement-relative units by dividing by 2x
xfrequency. (Note
that sound pressure is in phase with particle velocity, but its amplitude is
proportional to displacement.) The program then generated an animation of the
tympanic vibration for each individual. We used only recordings with high
coherence (>0.8 across 20-100 kHz, coherence calculated with fft
size=8192); this reduced our sample of females to five moths whose tympana
were monitored on six points and four moths whose tympana were monitored on
eight points. Coherence is a function in the frequency domain, with values
between 0-1 that indicate how well the input corresponds to the output at each
frequency. Coherence can thus be used to measure signal quality, i.e. the
extent to which the signal is linearly related to the stimulus and free from
unrelated noise (Kates, 1992
).
We used the magnitude-squared coherence function:
![]() | (1) |
where Cxy(f) is the cross-power spectrum between the
signal and the stimulus recording, Cxx(f) is the
auto-power spectrum of the signal, and Cyy(f) is the
auto-power spectrum of the stimulus recording
(Kates, 1992;
Robert et al., 1998
).
|
To evaluate the transmission of stimulus temporal characterisics we measured pulse length and asynchrony interval in CoolEdit. We defined the beginning of pulses at the point in which the amplitude reached 20% of the maximum of the pulse; we defined the end of pulses at the point in which amplitude fell below 30% of the maximum of the pulse and remained low. Because the tympanum did not respond to the lower frequencies present in the stimuli (Fig. 3; recordings of the stimuli showed energy down to 25 kHz, see above), we discarded the low-frequency energy from the stimulus recordings before the analysis of transmission of temporal characteristics. To discard the low-frequency energy we high-pass filtered the stimulus recordings at 70 kHz. This filtering did not alter the laser recordings of the tympanic response (data not shown), because most of their energy was at high frequencies (see Fig. 3). We therefore used the unfiltered laser recordings for this analysis. Finally, we used a custom computer program written in Borland Pascal 7.0 (Borland, CA, USA) to measure the effect of asynchrony interval on the RMS amplitude of the tympanic response to stimulus pulse pairs varying in asynchrony interval. We measured RMS amplitude over a window of 1350 µs from the beginning of the response. This window included the pulse pair in its entirety. For this analysis we used the files resulting from averaging the 200 recordings for each stimulus presentation to each individual, to improve the signal-to-noise ratio (see above). The data showed homogeneity of variance, and we performed one-way analyses of variance (ANOVAs) and Model I regressions.
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
Transmission of temporal characteristics
Pulse length transmission was linear but with responses 20-30 µs longer
than the stimuli (Fig. 4).
Transmission of asynchrony interval was precise for asynchrony intervals
170 µs (Fig. 5A). For
120-µs asynchrony intervals there was overlap between the responses to the
pulses in a pulse pair (according to our criterion of 30% of the peak pulse
amplitude to define the end of a pulse), due to the 20-30 µs longer
response to individual stimulus pulses
(Fig. 5A). Variation in the
mean RMS amplitude of the tympanic response according to stimulus asynchrony
interval was under 2 dB and statistically non-significant
(Fig. 5B).
|
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
The tympanic response in A. grisella consisted of a rotational
mode of vibration, in which the anterior and posterior sections moved out of
phase. The posterior section of the tympanum vibrated with all points in phase
and maximum displacement at the attachment point of the scoloparium that
contains the receptor cells. This relatively simple mode of vibration agrees
with most studies on the vibration of insect tympana
(Larsen and Michelsen, 1978;
Michelsen and Larsen, 1978
;
Schiolten et al., 1981
),
except for locusts, where the tympanic membrane and associated Müller's
organ sclerites vibrate in a complex pattern
(Stephen and Bennet-Clark,
1982
; Breckow and Sippel,
1985
; Windmill et al.,
2005
).
The tympanum of A. grisella was most sensitive at high ultrasonic
frequencies. Peak response occurred at the frequency range of male signals and
there was a pronounced roll-off. However, response amplitude was within 12 dB
of the peak for frequencies as low as 35 kHz. This response reflects the
mechanical tuning of the tympanic membrane and of the scoloparium itself (see
Adams, 1972). In noctuid moths,
the mechanical response of the scoloparium can consist of a resonant vibration
with harmonics (Adams, 1972
).
The two peaks of high sensitivity in our results may thus reflect such
harmonics in the resonant vibration of the scoloparium.
The range of relatively high sensitivity of A. grisella tympana
(peak at 90 kHz and response amplitude within 12 dB down to 35 kHz)
encompasses the frequencies of the echolocation cries of many bat species
(Neuweiler, 1989;
Miller and Surlykke, 2001
).
The tympana of A. grisella are thus most sensitive to the frequencies
of male signals, but not much less sensitive to the frequencies typical of
their natural enemies. This tuning agrees with behavioural evidence of
sensitivity to the frequencies of male signals and bat echolocation cries
(Jang and Greenfield, 1996
;
Greenfield and Baker, 2003
;
Rodríguez and Greenfield,
2004
).
The tympana of A. grisella were also capable of fine temporal
resolution. The transmission of pulse length was linear, but responses were
20-30 µs longer than the stimuli, which led to overlap between pulses
separated by 120 µs asynchrony intervals, but not for longer asynchrony
intervals. This level of resolution is comparable to that of noctuid moths
(Schiolten et al., 1981) and
finer than in katydids and crickets
(Larsen, 1981
).
The combination of frequency tuning and high temporal resolution in A.
grisella tympana may be explained because high temporal resolution may
stem from sensitivity to ultrasound per se
(Schiolten et al., 1981). A
system that vibrates over n cycles will take less time to do so at
higher frequencies than at lower frequencies.
Our results suggest a likely range for the tympanic time constant in A.
grisella. The time constant of a tympanum refers to its ability to follow
brief changes in the temporal pattern of a stimulus, i.e. its acuity
(Green, 1985). This time
constant can be estimated by calculating the impulse response of the tympanum
with a very brief stimulus (much shorter than the time constant one wishes to
estimate), to observe how long the tympanum continues to vibrate after the end
of the stimulus (Schiolten et al.,
1981
). Although we did not use this method, our results give an
indication of the tympanic time constant in A. grisella. According to
our criterion of 30% of peak amplitude to define the end of pulses, tympanic
responses were 20-30 µs longer than the stimuli. The shortest asynchrony
interval faithfully transmitted by the tympanic response was of 170 µs with
120 µs tympanic responses, so the briefest gap resolved was of 50 µs.
Thus, the time constant of A. grisella tympana may be 20-50 µs.
This estimate is shorter, by as much as a third, than the tympanic time
constant of noctuid moths (ca. 60 µs;
Schiolten et al., 1981
), which
is in accord with the tuning of noctuid tympana to lower frequencies
(Schiolten et al., 1981
).
Evaluation of pulse length
Tympanic transmission of pulse length was linear, with responses slightly
longer than the stimuli. The stimulation transmitted by the tympanic response
can probably be encoded by integration of energy over time by the receptor
cells. The requirement for this mechanism to obtain is that male pulses be
shorter than the integration time of the receptor cells. The integration time
is the interval over which stimuli are summated
(Green, 1985; Tougaard,
1998
,
1999
). Summation involves
integration of energy over time, and integration of the probability of
detection over time (Green,
1985
; Tougaard,
1998
,
1999
). Integration of energy
occurs at shorter time scales than integration of probability of detection,
and obtains if the integration time is longer than the stimuli
(Green, 1985
; Tougaard,
1998
,
1999
). We do not have
estimates of the integration time for A. grisella, but estimates for
other moths are in the range of 2-5 ms
(Tougaard, 1998
). If the
integration time of A. grisella receptor cells has similar values, it
is an order of magnitude longer than the male pulses
(Fig. 1). The estimate of a
short time constant for A. grisella tympana (20-50 µs) indicates
that tympanic responses to male pulses will fall within the likely integration
time of their receptor cells. Thus, longer pulses can evoke stronger responses
in the receptor cells and increase the likelihood that an action potential
will occur.
Evaluation of asynchrony interval
The evaluation of asynchrony interval is a more problematic question. Our
results showing precise tympanic transmission of asynchrony intervals 170
µs rule out the possibility that non-linear tympanic transmission
influences the strength of the stimulation delivered to the receptor cells.
Thus, the mechanism responsible for evaluation of asynchrony interval must
reside at the neural level.
Experiments with noctuid moths suggest that summation of receptor
potentials is maximized when the interval separating two short pulses lasts
1-5 ms; single pulses, and pulses separated by intervals longer than the
integration time of the receptor cell (2-5 ms), result in weaker spiking
activity (Tougaard, 1996,
1998
). Extrapolating these
findings to the time scale of A. grisella asynchrony intervals does
not offer an explanation for the evaluation of asynchrony interval. On the one
hand, all the asynchrony intervals discriminated by females are shorter than
the likely integration time of the receptor cells
(Fig. 1C), so that they may
deliver the same amount of energy to the receptor cells. On the other hand, if
the amount of energy delivered varies with asynchrony interval, it would be
the shorter intervals that deliver the stronger stimulation, because summation
of sequential receptor potentials would be higher when the receptor potentials
are separated by smaller intervals. Thus, the potential effect would be for a
preference for shorter asynchrony intervals, instead of the observed pattern
(Fig. 1C). The mechanism
responsible for evaluation of asynchrony interval may therefore involve
non-linear summation of receptor potentials. Features of A. grisella
signals that may influence non-linear summation include the very short length
of their pulses, and the very high amplitude at which they are produced
(Jang and Greenfield,
1996
).
In conclusion, we found no mechanical limit to temporal resolution in the transmission of ultrasonic stimuli by the tympana of A. grisella. Pulse length can be encoded by the receptor cells. Evaluation of asynchrony interval probably occurs at other levels of the nervous system.
![]() |
Acknowledgments |
---|
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Adams, W. B. (1972). Mechanical tuning of the acoustic receptor of Prodenia eridania (Cramer) (Noctuidae). J. Exp. Biol. 57,297 -304.
Bendat, J. S. and Piersol, A. G. (1986). Random Data: Analysis and Measurement Procedures. New York: Wiley Press.
Breckow, J. and Sippel, M. (1985). Mechanics of transduction of sound in the tympanal organ of adults and larvae of locusts. J. Comp. Physiol. A 157,619 -629.[CrossRef][Medline]
Fletcher, N. H. (1992). Acoustic Systems In Biology. New York: Oxford University Press.
Gerhardt, H. C. and Huber, F. (2002). Acoustic Communication in Insects and Anurans. Chicago: University of Chicago Press.
Green, D. M. (1985). Temporal factors in psychoacoustics. In Time Resolution in Auditory Systems (ed. A. Michelsen), pp. 122-140. Berlin, Heidelberg, Germany: Springer-Verlag.
Greenfield, M. D. (2002). Signalers and Receivers: Mechanisms and Evolution of Arthropod Communication. New York: Oxford University Press.
Greenfield, M. D. and Baker, M. (2003). Bat avoidance in non-aerial insects: the silence response of signaling males in an acoustic moth. Ethology 109,427 -442.[CrossRef]
Jang, Y. and Greenfield, M. D. (1996). Ultrasonic communication and sexual selection in waxmoths: female choice based on energy and asynchrony of male signals. Anim. Behav. 51,1095 -1106.[CrossRef]
Kates, J. M. (1992). On using coherence to measure distortion in hearing aids. J. Acoust. Soc. Am. 91,2236 -2244.[CrossRef][Medline]
Knopek, L. and Hintze-Podufal, C. (1986). Über den Bau der abdominalen Tympanalorgane der Kleinen Wachsmotte Achroia grisella (Fbr.). Zool. Jb. Anat. 114, 83-93.
Larsen, O. N. (1981). Mechanical time resolution in some insect ears. II. Impulse sound transmission in acoustic tracheal tubes. J. Comp. Physiol. 143,297 -304.[CrossRef]
Larsen, O. N. and Michelsen, A. (1978). Biophysics of the ensiferan ear. II. The cricket ear as a four-input system. J. Comp. Physiol. 123,217 -227.[CrossRef]
Michelsen, A. and Larsen, O. N. (1978). Biophysics of the ensiferan ear. I. Tympanal vibrations in bushcrickets (Tettigoniidae) studied with laser vibrometry. J. Comp. Physiol. 123,193 -203.[CrossRef]
Miller, L. A. and Surlykke, A. (2001). How some insects detect and avoid being eaten by bats: tactics and countertactics of prey and predator. BioScience 51,570 -581.
Neuweiler, G. (1989). Foraging ecology and audition in echolocating bats. Trends Ecol. Evol. 4, 160-166.[CrossRef]
Robert, D., Miles, R. N. and Hoy, R. R. (1998). Tympanal mechanics in the parasitoid fly Ormia ochracea: intertympanal coupling during mechanical vibration. J. Comp. Physiol. A 183,443 -452.[CrossRef]
Rodríguez, R. L. and Greenfield, M. D. (2004). Behavioral context regulates dual function of ultrasonic hearing in lesser waxmoths: bat avoidance and pair formation. Physiol. Entomol. 29,159 -168.[CrossRef]
Schiolten, P., Larsen, O. N. and Michelsen, A. (1981). Mechanical time resolution in some insect ears. I. Impulse response and time constants. J. Comp. Physiol. 143,289 -295.[CrossRef]
Scoble, M. J. (1992). The Lepidoptera: Form, Function and Diversity. New York: Oxford University Press.
Stephen, R. O. and Bennet-Clark, H. C. (1982). The anatomical and mechanical basis of stimulation and frequency analysis in the locust ear. J. Exp. Biol. 99,279 -314.
Tougaard, J. (1996). Energy detection and temporal integration in the noctuid A1 auditory receptor J. Comp. Physiol. A 178,669 -677.
Tougaard, J. (1998). Detection of short pure-tone stimuli in the noctuid ear: what are temporal integration and integration time all about? J. Comp. Physiol. A 183,563 -572.[CrossRef]
Tougaard, J. (1999). Receiver operating characteristics and temporal integration in an insect auditory receptor cell. J. Acoust. Soc. Am. 106,3711 -3718.[CrossRef]
Windmill, J. F. C., Göpfert, M. C. and Robert, D.
(2005). Tympanal travelling waves in migratory locusts.
J. Exp. Biol. 208,157
-168.