82
Hans Winkler
Sorjonen 1983; Shy and Morton 1986), particularly in cases where at least some
song characteristics match transmission properties of the habitats. Pertinent data
also come from studies that compare the transmission of sounds between speciesspecific and atypical habitats (e.g., Shy and Morton 1986). In most cases,
however, it remains unclear what exactly determines the acoustic climate of the
habitat and the specific adaptive responses to it.
In closed habitats, tall grass, reed, foliage, branches and tree trunks may scatter
sound. From empirical data and theory, the following picture of sound attenuation
in forests close to the ground (1-2 m) emerges (Embleton 1963; Aylor 1971;
Morton 1975; Linskens et al. 1976; Marten and Marler 1977; Marten eta!. 1977;
Wiley and Richards 1982; Price et al. 1988). Attenuation peaks at low frequencies
due to ground effects, it is relatively low at frequencies from 0.5 to 2.5 kHz, and it
steadily increases with frequency (between 3 and 6 dB per doubling of frequency)
above approximately 1.5 kHz due to scattering. At these higher frequencies small
additional peaks due to resonance, probably insignificant for avian communication
in most cases (but see Heuwinkel 1982, 1990 for reed stands), may appear (Price
et al. 1988). Scattering in dense growth, on the other hand disrupts ground
interference (Price et a!. 1988).
Wiley and Richards (1978) presented a model of sound propagation in forest
canopies that considers mainly the effects of refraction and treats the canopy as a
more or less even surface. The model certainly underestimates surface roughness,
and wind induced turbulence generated by canopy and emergent trees (Fitzjarrald
et al. 1990). The canopy also acts as a low-pass filter so that only large-sized
eddies may penetrate into the forest interior (Gao and Li 1993). Waide and Narins
(1988) suggest that a considerable proportion of sound from the canopy does not
reach the ground.
Background noise in forests exhibits clear spectral patterns, diurnal variation and
stratification. It peaks between 4 and 9 kHz and is about 10 dB higher than in
grassland (Ryan and Brenowitz 1985). Noise of biological origin is substantial
particularly in tropical forests. It concentrates in the canopy and peaks at dusk, and
increases markedly above 4 kHz (Riede 1996). Noise generated by wind, on the
other hand, declines with frequency and may constitute the main source for noise
in temperate forests (Klump 1996). Waterfalls and torrents produce strong noise
that decreases in dB level towards higher frequencies (Martens and Geduldig
1990; Riede 1996).
Open fields, such as grasslands exhibit strong ground attenuation effects and
refraction caused by wind or temperature gradients exert powerful influences on
sound propagation (Marten and Marler 1977; Romer, this Vol.). Noise levels seem
generally to be lower than in forests, however. In the reed beds of the Old World,
a habitat for many songbird species, attenuation is negatively correlated with
height above ground, and is strong above 5 kHz. Amplitude modulation is present
above 1 kHz. Resonance effects may contribute to low and even to negative
attenuation below 4kHz (Jilka and Leisler 1974; Heuwinkel 1982, 1990).
Hans Winkler
Sorjonen 1983; Shy and Morton 1986), particularly in cases where at least some
song characteristics match transmission properties of the habitats. Pertinent data
also come from studies that compare the transmission of sounds between speciesspecific and atypical habitats (e.g., Shy and Morton 1986). In most cases,
however, it remains unclear what exactly determines the acoustic climate of the
habitat and the specific adaptive responses to it.
In closed habitats, tall grass, reed, foliage, branches and tree trunks may scatter
sound. From empirical data and theory, the following picture of sound attenuation
in forests close to the ground (1-2 m) emerges (Embleton 1963; Aylor 1971;
Morton 1975; Linskens et al. 1976; Marten and Marler 1977; Marten eta!. 1977;
Wiley and Richards 1982; Price et al. 1988). Attenuation peaks at low frequencies
due to ground effects, it is relatively low at frequencies from 0.5 to 2.5 kHz, and it
steadily increases with frequency (between 3 and 6 dB per doubling of frequency)
above approximately 1.5 kHz due to scattering. At these higher frequencies small
additional peaks due to resonance, probably insignificant for avian communication
in most cases (but see Heuwinkel 1982, 1990 for reed stands), may appear (Price
et al. 1988). Scattering in dense growth, on the other hand disrupts ground
interference (Price et a!. 1988).
Wiley and Richards (1978) presented a model of sound propagation in forest
canopies that considers mainly the effects of refraction and treats the canopy as a
more or less even surface. The model certainly underestimates surface roughness,
and wind induced turbulence generated by canopy and emergent trees (Fitzjarrald
et al. 1990). The canopy also acts as a low-pass filter so that only large-sized
eddies may penetrate into the forest interior (Gao and Li 1993). Waide and Narins
(1988) suggest that a considerable proportion of sound from the canopy does not
reach the ground.
Background noise in forests exhibits clear spectral patterns, diurnal variation and
stratification. It peaks between 4 and 9 kHz and is about 10 dB higher than in
grassland (Ryan and Brenowitz 1985). Noise of biological origin is substantial
particularly in tropical forests. It concentrates in the canopy and peaks at dusk, and
increases markedly above 4 kHz (Riede 1996). Noise generated by wind, on the
other hand, declines with frequency and may constitute the main source for noise
in temperate forests (Klump 1996). Waterfalls and torrents produce strong noise
that decreases in dB level towards higher frequencies (Martens and Geduldig
1990; Riede 1996).
Open fields, such as grasslands exhibit strong ground attenuation effects and
refraction caused by wind or temperature gradients exert powerful influences on
sound propagation (Marten and Marler 1977; Romer, this Vol.). Noise levels seem
generally to be lower than in forests, however. In the reed beds of the Old World,
a habitat for many songbird species, attenuation is negatively correlated with
height above ground, and is strong above 5 kHz. Amplitude modulation is present
above 1 kHz. Resonance effects may contribute to low and even to negative
attenuation below 4kHz (Jilka and Leisler 1974; Heuwinkel 1982, 1990).
