You are on page 1of 9

J Comp Physiol A (2004) 190: 665–673

DOI 10.1007/s00359-004-0525-4

O R I GI N A L P A P E R

Carolyn L. Pytte Æ Millicent S. Ficken Æ Andrew Moiseff

Ultrasonic singing by the blue-throated hummingbird: a comparison


between production and perception

Received: 23 September 2003 / Revised: 18 March 2004 / Accepted: 9 April 2004 / Published online: 26 May 2004
Ó Springer-Verlag 2004

Abstract Blue-throated hummingbirds produce elabo-


rate songs extending into the ultrasonic frequency range,
Introduction
up to 30 kHz. Ultrasonic song elements include har-
Hummingbirds (Trochilidae: Apodiformes) are not
monics and extensions of audible notes, non-harmonic
renowned for the splendor or complexity of their vocal-
components of audible syllables, and sounds produced
izations. Perhaps, however, this is an oversight reflecting
at frequencies above 20 kHz without corresponding
our sensory bias against the shrill and high-pitched
hearing range sound. To determine whether ultrasonic
sounds produced by many hummingbird species. Schu-
song elements function in intraspecific communication,
chmann (1999) noted that high frequencies are common
we tested the hearing range of male and female blue-
among hummingbirds and reported that the song of the
throated hummingbirds. We measured auditory thresh-
black jacobin (Florisuga fusca) may in fact contain
olds for tone pips ranging from 1 kHz to 50 kHz using
ultrasonic components. We have observed a South
auditory brainstem responses. Neither male nor female
American hummingbird perched in a singing posture
blue-throated hummingbirds appear to be able to hear
while fluttering its throat feathers without any audible
above 7 kHz. No auditory brainstem responses could be
sound, indicating the possibility of song frequencies
detected between 8 and 50 kHz at 90 dB. This high-
outside the range of human hearing (M.S. Ficken, per-
frequency cutoff is well within the range reported for
sonal observation). The blue-throated hummingbird
other species of birds. These results suggest that high-
(Lampornis clemenciae) also demonstrates unusual sing-
frequency song elements are not used in intraspecific
ing behaviors which are suggestive of the use of ultrasonic
communication. We propose that the restricted hum-
frequencies, and is notable for its large and complex vocal
mingbird hearing range may exemplify a phylogenetic
repertoire (Ficken et al. 2000). We decided to test whether
constraint.
this species emits ultrasounds, and if so, to determine the
hearing range of blue-throated hummingbirds.
Keywords Hearing Æ Hummingbird Æ Perception Æ
Avian species which echolocate and might be
Song Æ Ultrasound
expected to use ultrasound, such as the nocturnal oilbird
(Steatornis caripensis) and cave-dwelling swiftlets (genus
Collocalia), produce broadband echolocation clicks with
a maximum energy concentration below 8 kHz, well
within the sonic range (Pye 1980; Suthers and Hector
C. L. Pytte (&) 1982, 1985). Ultrasonic harmonics have recently been
Biology Department, Wesleyan University, described in the songs of the oscine rufous-faced warbler
Middletown, CT 06459, USA (Abroscopus albogularis) demonstrating that it is not
E-mail: cpytte@wesleyan.edu outside the capability of the syrinx to generate extremely
Fax: +1-860-6853785
high-frequency sounds (Narins et al. 2004). Humming-
M. S. Ficken birds, along with swifts, compose an independent order
Department of Biological Sciences,
University of Wisconsin-Milwaukee and UWM Field Station,
that is phylogenetically distant from the songbirds (os-
3095 Blue Goose Road, Saukville, WI 53080, USA cine passerines) and parrots (Psittaciformes) which have
largely dominated the study of avian sound production
A. Moiseff
Department of Physiology and Neurobiology,
and perception. Despite the wealth of knowledge of
University of Connecticut, 3107 Horsebarn Hill Road, hearing in other birds, hearing tests have not been pre-
Storrs, CT 06269, USA viously performed on any hummingbird species.
666

Blue-throated hummingbirds exhibit strong niche specific communication. Although high frequencies do
specificity, inhabiting wooded riparian canyons in the not transmit over long distances, ultrasonic singing
extreme southwestern United States and Mexico. In the could enhance signal detection by making use of a fre-
breeding season, males establish and defend territories quency channel that is relatively free from competition
along streams. During this time, males produce two from water noise and vocalizations of other birds,
predominant vocalizations: repetitions of acoustically thereby providing a valuable sound window for broad-
simple notes, ‘‘serial chips,’’ which apparently function in casting song.
long-distance territorial advertisement, and a structurally Here, we present the results of ultrasonic recordings
and syntactically complex song which appears to attract of male song and of hearing tests in the male and female
both males and females. Playback of male song results in blue-throated hummingbird using auditory brainstem
a significant number of approaches by males to the responses (ABRs) to pure tones to reveal their audible
playback speaker (Ficken et al. 2002). Attraction of frequency range.
females to singing males has been observed with regu-
larity in the field, whereas females have never been ob-
served to approach silent males despite the fact that males Materials and methods
spend a much greater portion of time resting than sing-
ing. Female attraction to male song seems to be related to Ultrasonic production
courtship behavior and the attraction of males appears to
occur in aggressive contexts (Ficken et al. 2002). Recordings
There are several aspects of the blue-throated hum-
mingbird’s singing behavior which are puzzling and Recordings of sonic and ultrasonic vocalizations and
which led to our choice of this species for investigation of mechanical (wing and tail) noises were made using a
ultrasound use. Selective pressures for effective mate dual recording system to collect data in the hearing
attraction would presumably result in song transmission range simultaneously with sound in the ultrasonic range.
distance at least to the territory boundary (approxi- We recorded with a Sony Walkman WM-D6C (flat
mately 15 m, based on our observations of territory size). frequency response from 40 Hz to 15 kHz) and an
However, blue-throated song is produced at an extremely Audiotechnica AT877 microphone (flat response from
low amplitude (<30 dB at 10 m), much less than that of 60 Hz to 14 kHz) as well as with another Sony Walk-
agonistic vocalizations or serial chips (Pytte et al. 2003). man WM-D6C connected to a Pettersson Ultrasound
To our ears, the song cannot be heard from more than a Detector D230 (flat response from 10 to 120 kHz, fre-
few meters from the singing bird. In addition, rather than quency division setting). Recordings were made onto
avoiding high background noise, birds generally sing synchronized analog tapes which are archived at the
within several meters of a stream and often perched right University of Wisconsin-Milwaukee Field Station.
above one. In such cases, the background noise entirely Sound detected by the ultrasonic microphone was
obscures our detection of the song. Furthermore, males transposed by a factor of 10 in the frequency domain
have been observed to sing up to 10 min at a stretch in prior to recording on tape while the original amplitude
the absence of other individuals in the vicinity, suggesting envelope and true time domain were maintained. The
that it may not function solely as a short range signal recording input level on both recorders was adjusted to
although the song is also used in dual singing interactions ensure that artificial harmonics and other synthetic
between a male and female perched a few centimeters acoustics were not produced.
apart (Ficken et al. 2000). However, despite the low Eleven songs were used in this study, recorded from
signal-to-noise ratio during song production, we suspect four individually identified males (4 songs from one
that localization of singing males is primarily based on male, 3 from another, and 2 from two others). The songs
auditory cues. This has been supported by numerous were selected from a larger sample of songs within a
observations of females approaching singing males that singing bout. Males were identified by territory occu-
are perched in dense vegetation obscuring visual cues pancy as territories were linear and non-overlapping
(M.S. Ficken, personal observation), as well as playback (Williamson 2000, personal observation). Recordings
studies with a hidden speaker (Ficken et al. 2002). were made in a territory only once to provide additional
The use of ultrasonic frequencies during singing certainty of recording from a specific individual. We
would essentially explain this perplexing singing behav- recorded singing throughout the day during the breeding
ior—the production of a complex, low-amplitude song, season in August 2000 at the Southwestern Research
while visually obscured in a noisy environment. While Station, Portal, Arizona.
the sounds audible to humans are very low in amplitude,
perhaps ultrasonic frequencies carry the majority of the Analysis
sound energy.
High frequencies would increase the signal band- Ultrasonics are defined as sound frequencies above the
width, facilitating sound localization (Park and Dooling upper boundary of human hearing, generally considered
1991; Dooling 1992), and also provide an acoustic to be 20 kHz. Although our ‘‘hearing range’’ recordings
channel, improving the signal to noise ratio for intra- and ‘‘ultrasonic range’’ recordings overlap, we regard
667

ultrasonic frequencies as only those above 20 kHz. Supplemental doses were not necessary. Feathers were
Spectrographs of the hearing range and ultrasonic range removed from the crown of the head and bilaterally
recordings were printed using a Kay Elemetrics Sona- caudal to the external auditory meatus. Electrodes (400-
Graph (7800, 16-kHz scale, 300-Hz bandwidth). Spec- lm steel wire) were inserted subcutaneously at the apex
trographs of the ultrasonic signal display a time axis that (positive), caudal to the right ear (negative), and caudal
is identical to that of the hearing range signal, and a to the left ear (ground), and held in place with tissue
frequency axis that is a multiple of 10 times that of the adhesive. All tests were conducted within a sound-
hearing range spectrographs. Thus, using the 0- to 16- attenuating, anechoic room. The bird was wrapped
kHz setting, the frequency range of ultrasound spec- loosely in a cloth surrounded by a heating pad. Body
trographs is 10–160 kHz. Hearing range and ultrasonic temperature was maintained at 37°C. The bird was
range recordings were aligned temporally and analyzed placed with both ears equidistant to the speaker, with
using Raven (Beta 1, 2002 Cornell Laboratory of the external ear canals 30 mm from the speaker and the
Ornithology), Sound Analysis 3.19 (Tchernichovski beak 90° to the flat surface of the speaker, centered
et al. 2000) and Avisoft (4.2, Specht). Power spectra approximately at the speaker midline.
were produced with Avisoft using the hamming func- Custom designed software (by A.M.) interfaced with
tion, a bandwidth of 2.2 Hz, resolution of 1.3 Hz, and a sine wave generator (AG-7001C Audio Generator,
smoothing over 41.7 Hz. Peak detection of the power frequency range 10 Hz–1 MHz, distortion <0.1% at
spectra were determined using )10 dB from the maxi- 500 Hz–100 kHz; EZ-Digital) controlled the generation
mum peak amplitude and a hysteresis value of 10 dB. of stimuli, timing of presentation, and ABR acquisition.
The frequencies of maximum energy (peak power) Stimuli were output through an electrostatic driver
throughout the song were identified using Sound Anal- (ED1; Tucker Davis Technologies, Gainesville, Fla.,
ysis. USA) connected to an electrostatic speaker (ES-1, flat
In order to examine ultrasonic components of the frequency response between 1 kHz and 110 kHz; Tucker
song that are not simply extensions or harmonics of Davis Technologies, Gainesville, Fla., USA). Responses
hearing range sounds, we identified sounds that met were band-pass filtered (10–3 kHz) and passed through
these criteria: (1) ultrasonic elements which did not also a 60-Hz notch filter before amplifying. Amplifier gain
contain energy in the hearing range and (2) ultrasonic was calibrated using a voltage calibrator (2010 Omnical
components which differed in acoustic structure from Voltage Calibrator, World Precision Instruments). Re-
simultaneous hearing range sound. In addition, we sponses were viewed in real-time on a digital oscillo-
noted song elements in the hearing range that did not scope (Tektronix 2230) and, simultaneously, 40-ms
contain ultrasonic components. epochs were digitized (25-kHz sample rate) for averag-
ing. Electronic delays were employed so that response
sampling began 4.5 ms before the onset of the sound
Perception stimulus as recorded at the bird’s ears. Averages from 50
stimulus presentations were viewed on a monitor and
Subjects stored for offline analysis. At the end of the recording
session, the birds were given a 0.6 lg g)1 intramuscular
Adult male and female blue-throated hummingbirds injection of yohimbine and placed under a heat lamp
(n=4 males, 2 females) were mist-netted in southeastern until they fully recovered from anesthesia.
Arizona and transported to Wesleyan University where
they were housed until completion of the hearing tests.
The birds were housed in individual 0.5-m3 mesh cages Stimuli
in a communal room and kept in visual isolation in The hummingbirds were tested for ABR thresholds
order to minimize stress. Blue-throated males are highly using tone pips between 1 kHz and 50 kHz, 8 ms in
territorial during the breeding season and display duration including 1 ms rise and fall, presented with a
aggressive behaviors to males within view. The birds 750-ms interstimulus interval. Frequencies were pre-
were able to interact vocally. Throughout the period of sented in an arbitrary order at 0.5- or 1-kHz increments
captivity, the birds were maintained on photo period between 1 and 10 kHz, at 12 kHz, and at 5-kHz incre-
and temperature conditions matched to those of their ments from 15 to 50 kHz. All measurements are pre-
capture location. They were provided with commercially sented in dB re. 20 lPa. At each frequency, stimulus
prepared hummingbird food (Nekton, Nektar-plus) intensity was varied by 10-dB steps until near threshold,
made fresh daily. Mean (±SE) bird weights were and by 3-dB steps surrounding threshold until the ABR
8.15±0.14 (males) and 6.05±0.05 (females). threshold was identified. Stimulus intensity ranged from
30 dB to a maximum of 90 dB and was monitored with
ABR recording a calibrated microphone located midway between the
speaker and the bird’s head, positioned so that it did not
The birds were anesthetized with an intramuscular block the sound path to the ears. Stimulus intensity in
injection of 24 lg g)1 xylazine and 12 lg g)1 ketamine. the audio range (<15 kHz) was confirmed in the free
668

field using a sound pressure level meter (Bruel and Kjaer All song units contain syllables that include ultra-
2226) placed at the position of the bird’s ear by sonic components, some of which appear to be har-
recording SPLs directly using the same tones presented monics of sonic fundamentals. Other ultrasonic elements
as stimuli. Ultrasonic sound intensities were determined are high-frequency extensions of broadband transients
by measuring the voltage output of the ultrasonic and clicks notes. As these elements are not acoustically
microphone on an oscilloscope and calculating dB re. differentiated in the ultrasonic range, there is no com-
20 lPa. pelling reason to suggest that these categories of sounds
may contain additional information not provided within
the sonic range. Instead, we were interested in ultrasonic
ABR analysis elements which did not also contain energy in the
Evoked potential peaks were identified visually in the hearing range, as well as ultrasonic components of notes
recording trace by two observers. The lowest intensity at which differed in acoustic structure from simultaneous
which a response was present was determined to be the hearing range sound. Several ultrasonic elements meet
response threshold (as in Boettcher et al. 1993; Brittan- these criteria (Fig. 1).
Powell and Dooling 2002). The ABR was identified as a
peak above baseline followed by a trough below the Perception
baseline waveform. We did not attempt to identify
multiple successive peaks since a single, broad peak is ABRs
characteristic of long duration stimuli such as those
used. The neural generators of the ABR wave are un- ABR waveforms consisted of a single broad peak con-
known for hummingbirds. Peak latencies and ampli- tained within 4 ms after the sound reached the bird’s ear.
tudes were not measured since our objective was to A typical ABR to an 87-dB stimulus exhibits a 1-ms
identify the presence or absence of a response in order to peak beginning approximately 1.5 ms after stimulus
generate a threshold audiogram. arrival at the ears, followed by a 1.5-ms depression be-
fore returning to baseline. As is characteristic of the
ABR across vertebrates, decreased stimulus intensity
Results resulted in decreased peak amplitude and increased
response latency (Fig. 2) (Corwin et al. 1982).
Ultrasonic production
Hearing range
Blue-throated hummingbirds produce ultrasonic sounds
vocally during singing as well as mechanically by shuf- Most importantly for our purposes, we conclude that
fling tail and wing feathers while perched. Sound pro- neither male nor female blue-throated hummingbirds
duced while tail shuffling is composed of transient clicks hear high frequencies between 8 and 50 kHz (at 90 dB),
(15–30 kHz). Interpulse intervals are highly stereotyped as determined by ABR. We tested tones at 1- to 5-kHz
across tail shuffling episodes. Wing noises are similarly intervals throughout this range in search of discontinu-
broadband short duration clicks, although interpulse ous sensitivity or a sensitivity window as has been re-
intervals are not stereotyped. Tail shuffling appears to be ported in some rodent species (review in Sales and Pye
a ritualized behavior and occurs in the context of close 1974). Instead, the highest frequency at which a
range agonistic encounters, usually produced by a ter- threshold response was noted was 7 kHz for both males
ritory resident when confronted by an intruding hum- and females (Fig. 3a). None of the birds exhibited a
mingbird. Wing sounds are produced in this context as response to an 8-kHz tone, and no intervals between 7
well, and were also observed after agonistic encounters and 8 kHz were tested.
while apparently guarding a food source. Females demonstrated a range of best hearing for
A detailed description of the sonic song elements has frequencies between 1 and 3 kHz and a peak in sensi-
previously been reported (Ficken et al. 2000). As in the tivity at 2.0 kHz. The range of best hearing is arbi-
sonic range, ultrasonic frequencies contain acoustically trarily defined here as the frequencies eliciting a
differentiated syllables composed primarily of broad- threshold response within 20 dB of the peak hearing
band transients, clicks, and trills. Song organization is sensitivity. The mean female hearing threshold
based on five different sets of syllable clusters or song increases 16 dB between 2.5 and 3.0 kHz. Males
‘‘units’’ A–E (Fig. 1). Within a unit, syllables are entirely showed best hearing for frequencies between 1.0 and
stereotyped in acoustic structure and order. The 3.5 kHz, with peak sensitivity at 2.5 kHz. The mean
sequence of units comprising a song may vary in number male hearing threshold increases by 9 dB between 2.5
and order; however, the variation follows specific and 3.0 kHz.
sequencing rules (Ficken et al. 2000). As has been The overall response curve of the males is not sig-
reported for sounds in the hearing range (Ficken et al. nificantly different from that of the females (ANCOVA,
2000), ultrasonic elements are stereotyped within indi- P>0.05). Females appear to have a lower hearing
viduals and between individuals in a population. threshold at 2.0 kHz, however, the differences are not
669

Fig. 1 Blue-throated
hummingbird song. Song of a
male blue-throated
hummingbird recorded at the
Southwestern Research Station
during the breeding season
showing the five component
song units A–E. An example
sound element that is produced
solely within the ultrasonic
range is indicated with (1). An
example of a syllable composed
of differing acoustic
morphologies in the ultrasonic
and sonic ranges is noted with
(2). Examples of other song
element types are broadband
transient notes which extend
from the sonic into the
ultrasonic range (3), syllables
produced within the hearing
range with a possible ultrasonic
harmonic (4), and syllables
produced entirely within the
hearing range (5)

significant (Mann Whitney U, P>0.05, Fig. 3a). The just outside the upper boundary of the range of best
mean of males and females combined reveals a range of hearing (Fig. 5). Furthermore, we find that the peak
best hearing between 1.0 and 3.5 kHz, a peak in sensi- sensitivity frequencies of 2.0 and 2.5 kHz do not corre-
tivity at 2.5 kHz, and an increase in threshold of spond to any distinctive acoustic elements of the song
approximately 11 dB between 2.5 and 3.0 kHz (Fig. 3b). (Fig. 5).

Comparison of hearing range and song vocalizations


Discussion
The frequencies of peak hearing sensitivity (2.0 and
2.5 kHz) do not correspond to a peak in the concen- Production
tration of energy in the song (Fig. 4). Amplitude peaks
in the song power spectrum occur at 3.3, 5.8, 6.9 and The blue-throated hummingbird’s song differs from
8.6 kHz. Of the frequencies produced below 10 kHz, typical oscine songs in that it is atonal and composed
only 25% of the sound energy falls between 1.0 and primarily of broadband transients, fricative sound
3.5 kHz, within the range of best hearing. Interestingly, bursts, and click-note trills (Ficken et al. 2000). It is also
the majority of the peak frequency trace (frequency of unusual in its wide frequency range, extending from
maximum power at any given time) occurs at 4.0 kHz, 1.8 kHz to approximately 30 kHz. The ultrasonic
670

Fig. 2 Example of an ABR waveform. The top trace is a typical


ABR, produced by a male in response to a 2-kHz tone presented at
87 dB SPL. Subsequent traces show decreasing peak amplitude and
increased response latency correlated with decreasing stimulus
intensity. Sound reaches the bird’s ear at time 4.5 ms (arrow)

(>20 kHz) components of the song include continuous


extensions of hearing range sound, harmonics of hearing
Fig. 3 a, b Audiograms. a Mean (±SE) threshold ABRs for males
range elements, non-harmonic sounds which are pro- and females at each frequency. b The combined (mean±SE) male
duced in concert with sonic elements, and also elements and female audiogram. All subjects were tested up to 50 kHz. No
that are produced without corresponding sonic sound. responses to pure tone stimuli were obtained above 7 kHz at up to
High frequencies are not a necessary correlate of 90-dB stimulus intensity, the maximum intensity that we could
present
sound production by blue-throated hummingbirds as
territorial advertisement serial chips do not contain
frequencies above 7 kHz (Ficken et al. 2002). It is the ultrasonic sounds that they produce. Second, we
unlikely that the distinctive acoustic structure of the sought to determine the frequency of peak hearing sen-
song is the result of the small size of the sound pro- sitivity and compare this with acoustic features of the
ducing apparatus since smaller hummingbird species song. The results of ABR tests indicate that neither male
produce whistles, frequency modulations, and tonal nor female blue-throated hummingbirds hear ultrasonic
sounds (reviewed by Kroodsma et al. 1996; Ornelas et al. frequencies. Rather, they have a maximum audible fre-
2002). Furthermore, we could not detect any ultrasonic quency of 7 kHz at 90 dB. Blue-throated hummingbirds
sounds produced by hummingbirds smaller than the hear best between 1.0 and 3.0 or 3.5 kHz with a peak in
blue-throated, including the black-chinned (Archilochus sensitivity at 2.0–2.5 kHz (females and males, respec-
alexandri), Anna’s (Calypte anna) and broad-tailed tively).
(Selasphorus platycercus) hummingbirds. The magnifi- A correlation between the frequency of peak hearing
cent hummingbird (Eugenes fulgens) and white-eared sensitivity and specific spectral features of song has been
hummingbird (Hylocharis leucotis) songs have elements demonstrated in passerines and also psittacines (Dooling
that extend above 15 kHz but these components are a et al. 1971; Konishi 1971; Dooling and Saunders 1975).
continuation of sonic elements and thus not as intriguing We find that the frequency range of best hearing over-
as the ultrasonic sounds of the blue-throated hum- laps the largest peak in the power spectrum of the blue-
mingbird’s song. throated song. However, the peaks in hearing sensitivity
do not correspond to frequencies of increased amplitude
(Fig. 4) nor to any distinctive acoustic elements of the
Perception song (Fig. 5). Furthermore, much of the maximum
power lies just outside the upper boundary of best
The primary goal of the hearing tests was to identify the hearing (at 4.0 kHz) rather than within the 1.0–3.5 kHz
high end of the hearing range in order to establish best hearing range (Fig. 5). The significance of this for
whether or not blue-throated hummingbirds can hear signal processing, if any, is unknown. In addition, there
671

Fig. 4 Power spectrum of song.


Power spectrum of a single song
(units A–E). The box encloses
the mean frequency range of
best hearing for males and
females combined (1–3.5 kHz).
The shaded bar indicates the
sound energy between 2 and
2.5 kHz, the frequencies of
peak sensitivity in females and
males, respectively. Arrows
point to peaks in power at 3.3,
5.8, 6.9, and 8.6 kHz

Fig. 5 Peak frequency trace.


The frequency of maximum
energy at any given time
throughout a single song is
outlined in red. The horizontal
line indicates 4 kHz which
traces the majority of the
maximum sound energy

are no known calls in the repertoires of either adults or ture morphologies. The latter explanation is supported
juveniles that contain substantial energy below 3.5 kHz by numerous examples in mammals in which the
(Ficken et al. 2002). In fact, most of the sound energy of peripheral auditory system appears to have evolved
agonistic calls, alarm calls, serial chips, fledgling calls, as largely independent of behavioral factors (Plassman and
well as female song, is concentrated well above 3.5 kHz, Brändle 1992). Similarly, it appears that in the blue-
and in the case of serial chips is entirely contained near throated hummingbird, neither the maximal audible
the outer extent of the hearing range, between 6 and frequency nor the peak sensitivity frequency correspond
7 kHz. Perhaps a more consistent account of the fre- to any obvious behaviorally adaptive correlates.
quency of peak sensitivity and range of best hearing in A comparison of behaviorally determined audio-
the blue-throated hummingbird is that these character- grams among three groups of birds indicates that non-
istics have not been shaped entirely by pressures of passerines (excluding owls) have a lower maximum
communication. audible frequency (7.5 kHz) than passerines (9.7 kHz),
and both groups have a lower high-frequency cutoff
than owls (11.2 kHz). Non-passerines have the lowest
Evolution of hummingbird hearing peak frequency sensitivity (2.1 kHz), followed by owls
(2.7 kHz), then passerines (2.9 kHz). These data are the
Two explanations are commonly proposed to explain means of two species of Strigiformes (barn owl and great
characteristics of hearing: (1) behavioral adaptations, horned owls), eight species of other nonpasserines, and
for example, sound localization, predator detection, or 13 species of passerines (reviewed by Dooling 1992).
intraspecific communication; and (2) phylogenetic and Ignoring the adaptive specializations of the owls for
developmental constraints on head and auditory struc- nocturnal prey capture, characteristics of hummingbird
672

hearing are nearly identical to the mean of the other 30 dB would not alter the steep drop in hearing sensi-
nonpasserines that have been tested. tivity between 5 and 7 kHz nor the peak sensitivities at
Nonpasserines, including Apodiformes, are thought 2.0 and 2.5 kHz, and would not necessarily result in the
to evolutionarily pre-date passerines. Therefore, the inclusion of ultrasonic frequencies. Furthermore, in
curtailed frequency range of hummingbird hearing generating behavioral audibility curves, the high fre-
compared to their vocal production may represent a quency cutoff is defined as the highest frequency a bird
phylogenetic constraint. Such constraints are generally can hear at a sound pressure level of 60 dB SPL
due to limitations of the cochlear and middle ear (Dooling 1980). Because we used a 90-dB stimulus to
structures, peripheral transduction, or central auditory determine audibility thresholds using the ABR, we have
processing (Sachs et al. 1978). High frequency process- somewhat compensated for the differences between the
ing in birds begins in the basal region of the basilar two methods. The ABR is not limited by ultrasonic
papilla. The barn owl, which has the highest audible frequencies per se, as it has been shown to successfully
frequency, also has the longest basilar papilla described identify hearing ranges that extend up to 80 kHz in
in birds, about 11 mm (reviewed by Carr 1992). Perhaps clupeiform fishes (Mann et al. 2001).
the length of the basilar papilla in the hummingbird is
constrained by small head side, thus prohibiting high-
frequency hearing in small birds. Why ultrasound production?
Similarly, peak sensitivity is determined by the
dimensions of a few middle ear structures which, in turn, Thus far, we cannot provide an adaptive explanation for
are stringently coupled to phylogenetic and develop- the production of ultrasonic vocalizations, which
mental factors. A shift in peak sensitivity toward lower apparently do not function in intraspecific communica-
frequencies can be achieved by increasing the middle ear tion. Blue-throated hummingbirds often alternate sing-
volume, however this is largely constrained by overall ing with catching small flying insects. Perhaps the
head size. A shift in the peak sensitivity range to higher ultrasonic clicks produced during singing flushes insects
frequencies can occur by thickening the basilar mem- from vegetation, or triggers erratic insect flight patterns,
brane in the cochlea (Neuweiler et al. 1980). However, thereby increasing visual salience for the hummingbird.
peak sensitivity is not likely to be subject to behavioral However, we have not observed any such effect of
selective pressure; instead, behaviorally adaptive pres- singing on insects. Furthermore, vocal click sounds that
sures more often expand or shift the overall range of are produced by blue-throated hummingbirds while
hearing, without altering peak sensitivity (Plassman and actually chasing flying insects are audible and do not
Brändle 1992). Our data are consistent with this idea in contain ultrasonic components (unpublished data). We
that peak sensitivity in the blue-throated hummingbird also believe it is unlikely that ultrasound in blue-
does not correspond to any notable vocal attributes and is throated song repels small rodents that prey on nestling
the same as that of other non-passerines, excluding owls. hummingbirds. Male blue-throated hummingbirds do
not guard nests and females have not been observed to
sing near nests (Johnsgard 1983). Blue-throated hum-
Methodological constraints mingbirds have been observed to compete with the lesser
long-nosed bat (Leptonycteris curasoae) and the Mexi-
It must also be considered that while frequency-depen- can long-tongued bat (Choeronycteris mexicana) at
dent ABRs correspond well with the shape of behav- agave plants in early mornings and evenings (Summer
iorally generated audiogram curves, ABR audiograms Bennett, personal communication). These are times of
are not necessarily consistent with absolute auditory peak song production and thus ultrasonic sounds could
sensitivity and may underestimate the perceived sound signal to bats a site of high resource competition.
level thresholds (Borg and Engstrom 1983; Mann et al. However, even though ultrasonic elements might in fact
2001; Stapells and Oates 1997; Wenstrup 1984). Specif- be detected by bats, there is no evidence that this
ically in birds, ABR-generated audiograms reflect the interaction may have provided selective pressure to the
bandwidth and shape of the behavioral audiogram, structure of blue-throated song.
including the frequencies of peak sensitivity (mallard The high frequency cutoff in birds is usually related to
duck, Anas platyrhynchos, Dmitrieva and Gottlieb 1992; the highest frequency in the species’ song, and the
Bengalese finch, Lonchura striata domestica, Woolley hearing sensitivity in many species is correlated with
and Rubel 1999; budgerigar, Melopsitt undulatus, Brit- their spectral output (reviewed by Dooling 1982).
tan-Powell et al. 2002). However, Brittan-Powell and However, this is not an absolute association. As dem-
Dooling (2002) report that in budgerigars, the ABR- onstrated by Konishi (1971), the frequency range of
generated audiogram is 30 dB higher than the behav- audition does not always match the frequency range of
iorally generated audiogram. Thus, it is not unlikely that vocalizations. In fact, Woolley and Rubel (1999) dem-
the hearing threshold levels may be lower in blue- onstrate that Bengalese finches may attend only to the
throated hummingbirds than demonstrated by the ABR fundamental or low dominant frequencies in song
and frequencies higher than 7 kHz may in fact be per- (<3 kHz) in order to maintain complete song structure
ceived. However, shifting the absolute sensitivity level by across a frequency range of up to 10 kHz. Although it is
673

more intuitive to dismiss the sound categories of ultra- Esser KH, Daucher A (1996) Hearing in the FM-bat Phyllostomus
sonic harmonics and ultrasonic extensions of sonic discolor: a behavioral audiogram. J Comp Physiol A 178:779–
785
sounds as epiphenomenal, ultrasonic sounds that are Ficken MS, Rusch KM, Taylor SJ, Powers DR (2000) Blue-
structurally independent, non-harmonic elements of throated hummingbird song: a pinacle of nonoscine vocaliza-
hearing range sounds may nonetheless also be epiphe- tions. Auk 117:120–128
nomenal artifacts generated by the syrinx or upper vocal Ficken MS, Rusch KM, Taylor SJ, Powers DR (2002) Reproduc-
tive behavior and communication in blue-throated humming-
structures during sound production. birds. Wilson Bull 114:197–209
It must be considered that ultrasound may indeed be Johnsgard PA (1983) The hummingbirds of North America, 2nd
produced and perceived by other hummingbird species edn. Smithsonian Institute Press, Washington, DC
not yet tested. However, the production of ultrasonic Konishi M (1971) Comparative neurophysiological studies of
frequencies by blue-throated hummingbirds does not hearing and vocalizations in songbirds. Z Vergl Physiol 66:257–
272
appear to function in intraspecific communication and a Kroodsma DE, Vielliard JME, Stiles FG (1996) Study of bird
likely adaptive explanation for this phenomenon has yet sounds in the neotropics: urgency and opportunity. In: Kro-
to be identified. odsma DE, Miller EH (eds) Ecology and evolution of acoustic
communication in birds. Cornell University Press, Ithaca, New
Acknowledgements We thank the Southwestern Research Station York, pp 269–282
for housing the hummingbirds. Elizabeth Sandlin provided the use Mann DA, Higgs DM, Tavolga WN (2001) Ultrasound detection
of mist nets and training in hummingbird capture. All animal care by clupeiform fishes. J Acoust Soc Am 109:3048–3054
and methods complied with the regulations of the Institutional Narins PM, Feng AS, Lin W, Schnitzler H-U, Denzinger A, Su-
Animal Care and Use Committees of the affiliated universities, the thers RA, Xu C (2004) Old world frog and bird vocalizations
Arizona Game and Fish Department, the Connecticut Department contain prominent ultrasonic harmonics. J Acoust Soc Am
of Environmental Protection—Wildlife Division, and the US Fish 115:910–913
and Wildlife Service, as well as the Principles of Animal Care of the Neuweiler G, Bruns V, Shuller G (1980) Ears adapted for the
National Institutes of Health. This research was funded by NSF detection of motion, or how echolocating bats have exploited
SGER 0077980. the capacities of the mammalian auditory system. J Acoust Soc
Am 68:741–753
Ornelas JF, González C, Uribe J (2002) Complex vocalizations and
aerial displays of the amethyst-throated hummingbird (Lam-
pornis amethystinus). Auk 119:1141–1149
Park TJ, Dooling RJ (1991) Sound localization in small birds:
References absolute localization in azimuth. J Comp Psychol 105:125–133
Plassman W, Brändle K (1992) A functional model of the periph-
Boettcher FA, Mills JH, Norton BL (1993) Age-related changes in eral auditory system in mammals and its evolutionary impli-
auditory-evoked potentials of gerbils. 1. Response amplitudes. cations. In: Webster DB, Fay RR, Popper AN (eds) The
Hear Res 71:137–145 evolutionary biology of hearing. Springer, Berlin Heidelberg
Borg E, Engstrom B (1983) Hearing thresholds in the rabbit—a New York, pp 637–654
behavioral and electrophysiological study. Acta Otolaryngol Pye JD (1980) Echolocation signals and echoes in air. In: Busnel
95:19–26 R-G, Fish JF (eds) Animal sonar systems. Plenum Press, New
Brittan-Powell EF, Dooling RJ, Gleich O (2002) Auditory brain- York, pp 309–353
stem responses in adult budgerigars (Melopsittacus undulatus). Pytte CL, Rusch KM, Ficken MS (2003) Regulation of vocal
J Acoust Soc Am 112:999–1008 amplitude by the blue-throated hummingbird, Lampornis
Carr CE (1992) Evolution of the central auditory system in reptiles clemenciae. Anim Behav 66:703–710
and birds. In: Webster DB, Fay RR, Popper AN (eds) The Sachs MB, Sinnott JM, Hienz RD (1978) Behavioral and physio-
evolutionary biology of hearing. Springer, Berlin Heidelberg logical studies of hearing in birds. Fed Proc 37:2329–2335
New York, pp 511–544 Sales GD, Pye JD (1974) Ultrasonic communication by animals.
Corwin JT, Bullock TH, Schweitzer J (1982) The auditory brain Chapman and Hall, London
stem response in five vertebrate classes. Electroencephalogr Clin Schuchmann K-L (1999) Family trochilidae. In: Hoyo J del, Elliott
Neurophysiol 54:629–641 A, Sargatal J (eds) Handbook of the birds of the world, vol 5.
Dmitrieva LP, Gottlieb G (1992) Development of brainstem Lynx Edicions, Barcelona, Spain, pp 468–535
auditory pathway in mallard duck embryos and hatchlings. J Stapells DR, Oates P (1997) Estimation of the pure-tone audio-
Comp Physiol A 171:665–671 gram by the auditory brainstem response: a review. Audiol
Dooling RJ (1980) Behavior and psychophysics of hearing in birds. Neurootol 2:257–280
In: Popper AN, Fay RR (eds) Comparative studies of hearing Suthers RA, Hector DH (1982) Mechanism for the production of
in vertebrates. Springer, Berlin Heidelberg New York, pp 261– echolocating clicks by the grey swiftlet, Collocalia spodiopygia.
288 J Comp Physiol A 148:457–470
Dooling RJ (1982) Auditory perception in birds. In: Kroodsma Suthers RA, Hector DH (1985) The physiology of vocalization by
DE, Miller EH (eds) Ecology and evolution of acoustic com- the echolocating oilbird, Steatornis caripensis. J Comp Physiol
munication in birds. Cornell University Press, Ithaca, New A 156:243–266
York, pp 95–130 Tchernichovski O, Nottebohm F, Ho CE, Pesaran B, Mitra PP
Dooling RJ (1992) Hearing in birds. In: Webster DB, Fay RR, (2000) A procedure for an automated measurement of song
Popper AN (eds) The evolutionary biology of hearing. similarity. Anim Behav 59:1167–1176
Springer, Berlin Heidelberg New York, pp 545–560 Wenstrup JJ (1984) Auditory sensitivity in the fish-catching bat,
Dooling RJ, Saunders JC (1975) Hearing in the parakeet (Melo- Noctilioleporinus. J Comp Physiol 155:91–101
psittacus undulatus): absolute thresholds, critical ratios, fre- Williamson S (2000) Blue-throated hummingbird (Lampornis clem-
quency difference limens, and vocalizations. J Comp Physiol enciae). In: Poole A, Gill F (eds) The birds of North America, no.
Psychol 88:1–20 531. The Birds of North America, Philadelphia, PA
Dooling RJ, Mulligan JA, Miller JD (1971) Auditory sensitivity Woolley SMN, Rubel EW (1999) High-frequency auditory feed-
and song spectrum of the common canary (Serinus canarius). back is not required for adult song maintenance in Bengalese
J Acoust Soc Am 50:700–709 finches. J Neurosci 19:358–371

You might also like