The Laboratory of Comparative Psychoacoustics





The Effects of Environmental Noise on Communication in Birds
When noise masks the biologically important signals of birds in the wild,
and interferes with their ability to communicate effectively, it surely has
a detrimental effect on their normal behavior and breeding biology. We know
much about the
hearing sensitivity of birds from laboratory tests
with simple stimuli, but almost nothing about how they
perceive the vocalizations of their own species in noise. Laboratory data on
the effect of noise on the masking of tones can be combined with the
scattered information available on other factors influencing the distance
over which biologically meaningful signals can be used under natural
conditions, such as the location and source intensity of a singing bird, the
sound-attenuating and masking characteristics of the environment, and the
location and hearing sensitivity of the bird receiving the signal. From such
data we can obtain a rough estimate of possible communication distance, but
without more information we have only a very crude idea of the effect of
environmental noise on the perception of vocal signals in nature. There are
virtually no data published on the effects of different types of noises
(e.g. traffic or aircraft noises) on the perception of species-specific
vocalizations. Such laboratory data are critical for understanding the
effect of noise on acoustic communication and for developing reasonable
guidelines for noise abatement.
Many of the major factors determining how far away a sound can be heard
have been identified in a preliminary way, including source location and
intensity, inverse square attenuation with distance, excess attenuation,
spectrum level of the background noise, and the receiver's location and
auditory sensitivity. Taken together these factors form the basis for a
crude algorithm that may be used to predict how environmental noise will
influence a particular species' ability to detect a sound in the
environment. An example of how such an algorithm might be used is given in
the accompanying figures for the song sparrow (Melospiza melodia).
These figures are based on a simplified (and somewhat unnatural) measure of
masking of pure tones by broadband (white) noise and serve to illustrate how
such a model may be developed. Ultimately, however, data incorporating bird
thresholds for the detection, discrimination, and identification of natural
vocalizations in traffic or aircraft noise would be needed in order to
provide an accurate representation of noise effects on communication
abilities.
The figure at left illustrates the theoretical maximum distance that one
song sparrow can detect the song of another bird in this species. The
dependent variables represented here are different source intensities,
assuming an excess attenuation of 5 dB/100 m. Background noise level is
given as overall SPL (C-weighted), spectrum level (per cycle energy
distribution over the entire band of noise), and the amount of masking
assuming a critical ratio of 26 dB (as measured in the laboratory). We know
from laboratory studies that the threshold in the quiet for a 2 kHz pure
tone is about 0 dB SPL for song sparrows. Critical ratio data for this
species shows that background noise levels must be at least 26 dB below the
power in a 2 kHz pure tone in order for the tone to be detected by song
sparrows (Okanoya
& Dooling 1988,
1990). Thus, for a noise spectrum level
(energy per Hertz) of -26 dB, masked threshold is the same as absolute
threshold in the quiet, and no masking occurs. If the spectrum level of the
background noise is 0 dB SPL, however, the threshold of the signal is raised
to 26 dB SPL, causing 26 dB of masking. The spectrum level of the background
noise can be calculated from the overall sound pressure level, and depends
on the bandwidth and distribution of the noise. For a typical sound level
meter reading of noise in the frequency band 0 - 10 kHz, an overall SPL of
40 dB is equivalent to a spectrum level of about 0 dB if the noise is flat.
The figure at right is based on the same data as in the figure above and
shows the theoretical maximum distance for song detection as in the previous
figure. Here the dependent variables are different values of excess
attenuation, ranging from 5 dB/100 m to 25 dB/100 m, assuming a source
intensity of 95 dB. Excess attenuation depends almost completely on various
characteristics of the habitat in which a bird is signaling. A 5 dB/100 m
excess attenuation is appropriate for a species that sings 10 or more meters
above the ground in an open field, while an excess attenuation of 20 dB/100
m is more appropriate for birds singing at ground level in a coniferous
forest. Song sparrows have relatively small territories (a diameter of 20
-30 m) and reside in relatively open habitats. Thus, in contrast to many
other songbird species, signal attenuation and masking is therefore expected
to be minimal for this species. Songbirds occupy a wide array of habitat
types and have a broad range of territory sizes, with some songbirds having
territory diameters in excess of 100 m. These figures are for theoretical
conditions and represent something of a best case under ideal circumstances.
Moreover, these figures are based on thresholds using pure tones which are
an ideal, simplified signal. The value of using pure tone data is just to
show how a simple model might predict the effects of environmental noise on
acoustic signal detection. This model is very limited and will depend on
supplementary data (sound level measurements, noise spectrum levels,
behavioral observations of species' singing behavior and activity levels,
and laboratory tests of masking of vocal signals) obtained for any specific
example.
These kinds of masking studies are very relevant to the ecology of
birds. It is important to understand that intense traffic or aircraft noise
probably represents a substantial risk to normal acoustic communication in
birds. It is not practical to think of eliminating the detrimental effects
on communication completely. It is practical, however, to try and quantify
the risks so that intelligent judgments can be made concerning the extent to
which noise may interfere with the normal behavior and breeding biology of
birds in the wild, and reasonable arguments can be made about how much risk
is acceptable. At present, predictions made for detection of vocalizations
in the environment only address the simplest case, the ability of a bird to
tell whether a sound occurred (i.e. detection). Such a measure does not
reflect a bird's ability to communicate effectively in a particular acoustic
environment, and may have little bearing on it. Consider the case of human
speech communication as an example. It is one thing to hear a voice, and it
is quite another to understand what is said. For humans, the ability to
discriminate one speech sound from another requires a higher signal-to-noise
ratio than simply detecting a speech sound. Identification, or the ability
to recognize a specific, biologically relevant signal, may require even
higher signal-to-noise ratios. We need equivalent data for birds in order to
understand communication in noise.
At the present time, these worthwhile environmental projects are not
funded. The ultimate goal of these studies is to generate a predictive model
for evaluating the impact of noise on acoustic communication in birds in
order save endangered species. Future studies will examine not only whether
birds are capable of detecting a signal in noise, but are also whether they
are able to discriminate or identify relevant sounds such as vocalizations.
To do this, several different kinds of tests must be performed in the
laboratory with vocalizations and anthropogenic (human made) noises such as
traffic noises of different levels. These data, combined with measurements
from the field, will allow us to better apply our predictions to specific
cases, and estimate relevant distances on territory maps for individuals of
particular species and in specific habitats. If you are interested in
supporting this environmental research effort, please contact
Dr. Robert Dooling.
Link to Laboratory Members with Related Projects:
Bernie Lohr
Return to the
Lab projects page
Return to the
Laboratory of Comparative Psychoacoustics Home Page