FirstCite?
e-publishing
Sex-linked inheritance of
hearing and song in the
Belgian Waterslager canary
Timothy F. Wright1*, Elizabeth F. Brittan-Powell1,
Robert J. Dooling1 and Paul C. Mundinger2
1Department of Psychology, University of Maryland, College Park,
MD 20742, USA
2Department of Biology, Queens College (CUNY),
65?30 Kissena Boulevard, Flushing, NY 11367, USA
* Author and addresses for correspondence: before 1 July 2004:
Genetics Program NMNH, Smithsonian National Zoo,
3001 Connecticut Avenue, Washington, DC 20008, USA
(twright@mvar.nmsu.edu); after 1 July 2004: Department of Biology,
New Mexico State University, Las Cruces, NM 88003-8001, USA
(twright@mvar.nmsu.edu).
Recd 20.03.04; Accptd 26.03.04; Published online
Belgian Waterslager canaries have less sensitive
hearing at high frequencies and produce songs with
more energy at low frequencies than wild-type
canaries. A backcross pedigree between Belgian
Waterslager canaries and a domestic strain with
wild-type song revealed inheritance patterns con-
sistent with a factor of major effect located on the Z
sex chromosome affecting both poor high-frequency
hearing at 4 kHz and the relative energy in the spec-
tra of the learned songs of males. Hearing thresholds
at 4 kHz were significant predictors of the relative
amount of song energy at 4 kHz for individual males.
One hypothesis for the mechanistic basis of this cor-
relation between hearing and song abnormalities is
that a reduction in the ability to hear higher-
frequency songs biases males towards learning
lower-frequency songs.
Keywords: Serinus canarius; hearing; song learning;
sex linkage; Z chromosome; sensory linkage
1. INTRODUCTION
Hearing thresholds in birds are typically correlated with
the frequency spectrum of their song (Dooling et al. 1971;
Wright et al. 2003), but the source of this correlation is
not well understood, particularly for species that learn
their song. Domestic canaries (Serinus canarius) that have
been artificially selected for specific song traits offer an
ideal system in which to examine this correlation. One
domestic strain, the Belgian Waterslager (BWS) canary,
exhibits a shift in the peak energy of its learned song spec-
trum to lower frequencies than typically found in either
wild canaries or other domestic strains (Dooling et al.
1971; Guttinger 1985). This strain also exhibits poor
high-frequency hearing relative to other canaries; while
hearing thresholds below 2 kHz are normal, those above
2 kHz are 20?40 dB higher than those of wild-type
canaries (Okanoya & Dooling 1987). This high-frequency
hearing loss is apparent in chicks as early as 20 days post-
hatching (E. F. Brittan-Powell and R. J. Dooling, unpub-
lished data) and stems from missing and damaged hair
cells on the basilar papilla of the inner ear (Gleich et al.
Proc. R. Soc. Lond. B (Suppl.) ? 2004 The Royal Society
DOI 10.1098/rsbl.2004.0204
1997). The behavioural consequence of this anatomical
abnormality is a shift in the region of best hearing from 2?
4 kHz in wild-type canaries to 1?2 kHz in BWS canaries.
Strain-specific changes in both song and hearing in
canaries may be sex-linked (Okanoya et al. 1990; Mund-
inger 1995, 1999). In birds, sex determination is reverse-
heterogametic with ZZ males and ZW females. Recent
work has suggested that genes controlling sexually selected
traits (such as birdsong) commonly occur on the Z chro-
mosome (Iyengar et al. 2002; Price 2002), despite the fact
that in birds it comprises only 7?10% of the total genome
(Bloom et al. 1993). Here, we report the effect of the Z
chromosome on hearing sensitivities and male song fre-
quency in a backcross pedigree between the BWS strain
and the Border, a domestic canary with wild-type hearing
and song.
2. MATERIAL AND METHODS
(a) Pedigree breeding
The pedigree was founded with reciprocal crosses between a BWS
female and a Border male and between a Border female and a BWS
male (see electronic Appendix A for a pedigree diagram and method-
ological details). Female F1 offspring from these reciprocal crosses
were backcrossed to the BWS male to create backcross offspring with
an average autosomal complement of 25% Border and 75% BWS.
Male backcross offspring had either two BWS Z chromosomes or one
Border Z and one BWS Z depending on the origin of their F1
mothers. If either song or hearing is affected by a dominant Z-linked
factor, then any individual carrying a BWS Z chromosome would
show BWS traits, while a recessive Z-linked factor would require a
full complement of two BWS Z chromosomes for trait expression.
Results from autosomally linked factors would be mixed (Mundinger
1999). Pure strain offspring were also produced by within-strain mat-
ings of the BWS or Border founders.
(b) Song tutoring and analysis
Offspring were raised in acoustic chambers isolated from adult
males. Males were tape tutored in their first autumn with three low-
pitched canary songs and three high-pitched or wild-type songs, each
from different adult males. We recorded 5?10 entire songs from 14
tutored males, created long-term power spectra and measured the
peak spectral amplitude in the one-third octave bands centred at
1 kHz and 4 kHz (figure 1a). We then calculated the difference
between peak amplitudes in the two one-third octave bands to give
a single measure of the relative distribution of energy in songs
between low and high frequencies.
(c) Hearing thresholds
We measured hearing thresholds of the 23 pedigree birds (14 males
and nine females) using the auditory brainstem response (ABR), a
sound-evoked potential from the peripheral auditory system that pro-
vides a measure of hearing threshold. Frequency-specific ABRs were
evoked at 1 kHz and 4 kHz by stimulus trains consisting of nine tone
bursts of increasing intensity presented at a rate of 4 s
1 (Brittan-
Powell et al. 2002). Thresholds were defined as the lowest intensity
at which a response was visible in the ABR waveform (figure 1b).
3. RESULTS
(a) Hearing thresholds
Hearing thresholds in the parental generation (figure
2a,b, black symbols) were higher for pure BWS than for
pure Border canaries at 4 kHz but not at 1 kHz (table 1).
Backcross hybrids (figure 2a,b, grey symbols) with a full
complement of BWS Z chromosomes had higher thresholds
than individuals with one Border Z chromosome at both
frequencies. When all generations were combined, indi-
viduals with a full complement of BWS Z chromosomes
had higher thresholds than those with at least one Border
Z chromosome at both 1 kHz and 4 kHz (table 1).
A nonlinear regression modelling a completely recessive
Z chromosome effect explained 87% of the variation in
hearing threshold at 4 kHz (Y = 23.9Zadd 
 23.9Zdom
04bl0074.S2 T. F. Wright and others Inheritance of hearing and song in canaries
?10
?20
?30
?40
?50
?60
?70
?80
?10
?20
?30
?40
?50
?60
?70
?80
?10
?20
?30
?40
?50
?60
?70
?80
?10
?20
?30
?40
?50
?60
?70
?80
0.5 1.0 2.0 4.0 10.06.0 0.5 1.0 2.0 4.0 10.06.0 0.5 1.0 2.0 4.0 10.06.0 0.5 1.0 2.0 4.0 10.06.0
frequency (kHz) frequency (kHz) frequency (kHz) frequency (kHz)
0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14
re
la
ti
ve
 a
m
pl
it
ud
e 
(d
B
 v
ol
t)
15 ?V
110
105
100
95
90
85
80
75
70
110
105
100
95
90
85
80
75
70
75
70
65
60
55
50
45
40
35
75
70
65
60
55
50
45
40
35
time (ms) time (ms) time (ms) time (ms)
(a) (i) (ii) (iii) (iv)
(b) (i) (ii) (iii) (iv)
Figure 1. (a) Averaged frequency spectra of 10 songs from two parental strain canaries and two backcross hybrids with the
one-third octave bands at 1 kHz and 4 kHz shown by hatched bars. (i) BWS parent; (ii) backcross with a full complement of
BWS Z chromosomes; (iii) Border parent; and (iv) backcross with a partial complement of BWS Z chromosomes. (b) ABR
waveforms evoked by a 4 kHz tone delivered at nine intensities (in dB); hearing thresholds are indicated by arrows. (i)?(iv) as
in (a).
 42.7, corrected r2 = 0.87, where Zadd and Zdom are the
additive and dominance components of the Z chromo-
some effect, respectively). The model allowing additive
and dominant sex chromosome components to vary,
independently explained a slightly higher amount of
the variation (Y = 21.2Zadd 
 30.6Zdom  45, corrected
r2 = 0.91), but the difference between the two models was
not significant (2(log-likelihood ratio) = 3.4, d.f. = 1, 2-
test p 0.05).
(b) Song spectral characteristics
Backcross hybrids (figure 2c, grey symbols) showed no
difference in song spectral amplitude at 1 kHz minus
4 kHz between males with two BWS Z chromosomes ver-
sus individuals with a mixed Z chromosome complement
(table 1). When all three generations were combined,
however, males with two BWS Z chromosomes had sig-
nificantly larger values for song amplitude difference than
males with one or zero BWS Z chromosomes.
A recessive Z chromosome effect explained 79% of the
variation in song peak frequency (Y = 3.7Zadd 
 3.7Zdom

 7.1, corrected r2 = 0.79), whereas allowing additive and
dominant Z chromosome effects to vary independently
explained 86% of the variation in this variable
(Y = 4.5Zadd 
 2.2Zdom 
 8.7, corrected r2 = 0.86). The
Proc. R. Soc. Lond. B (Suppl.)
difference between the two models was not significant
(2(log-likelihood ratio) = 3.2, d.f. = 1, 2-test p 0.05).
(c) Correlation between hearing and song
Linear regressions of the song spectral amplitude differ-
ence on hearing thresholds of males were significant and
positive for thresholds at 4 kHz (figure 2d;
n = 12, y = 0.083x
 9.14, r2 = 0.49, p = 0.01) but not at
1 kHz (n = 12, r2 = 0.17, p = 0.2).
4. DISCUSSION
(a) Inheritance of hearing and song spectrum
Our results demonstrate that the high-frequency hear-
ing deficit in BWS canaries is associated with a genetic
factor of major effect located on the Z chromosome. Hear-
ing thresholds at 4 kHz were significantly higher in birds
with a full complement of BWS Z chromosomes than in
individuals with at least one Border Z chromosome,
regardless of autosomal complement. The BWS Z chro-
mosome is estimated to be responsible for 87% to 91% of
the difference between the two strains in hearing at 4 kHz
depending on the degree of recessive expression assumed.
The non-significant difference between the two models
indicates that the Z-linked factor is largely recessive.
Inheritance of hearing and song in canaries T. F. Wright and others 04bl0074.S3
100
80
60
40
20
100
80
60
40
20
5
0
?5
?10
5
0
?5
?10
0 0.5 1.0
0 0.5 1.0
0 0.5 1.0
20 40 60 80 100
proportion of BWS Z chromosomes
proportion of BWS Z chromosomes
proportion of BWS Z chromosomes
threshold at 4 kHz (dB SPL)
so
ng
 a
m
pl
it
ud
e 
at
 1
 k
H
z 
m
in
us
am
pl
it
ud
e 
at
 4
 k
H
z 
(m
V
)
A
B
R
 th
re
sh
ol
d 
(d
B
 S
P
L
)
(a) (b)
(c) (d )
Figure 2. Hearing thresholds at (a) 1 kHz and (b) 4 kHz for individuals plotted by the proportion of Z chromosomes of BWS
origin (0 equals no BWS Z chromosomes; 0.5 equals hybrid males with one Z chromosome from each strain; 1 equals females
with one or males with two BWS Z chromosomes). (c) The difference in the song spectrum amplitude at 1 kHz and 4 kHz
plotted by Z chromosome complement. (d ) The regression of the difference in song spectrum amplitude against hearing
threshold at 4 kHz for males. Black triangles, parental males; black circles, parental females; white triangles, F1 males; white
circles, F1 females; grey triangles, backcross males; grey circles, backcross females.
Table 1. Comparisons of hearing and song spectra between pedigree birds with a full complement of Belgian Waterslager Z
chromosomes versus those with at least one Border Z chromosome.
higher in birds
with a full
pedigree complement of
generation BWS Z
measure compared n tied Z-valuea tied p-valuea chromosomes
hearing threshold at 1 kHz parental 7 1.1 0.284 no
backcross 9 2.5 0.014 yes
all 23 3.1 0.019 yes
hearing threshold at 4 kHz parental 7 2.1 0.032 yes
backcross 9 2.5 0.014 yes
all 23 4.0  0.001 yes
song amplitude at 1 kHz minus backcross 6 0  1.000 no
amplitude at 4 kHz all 14 2.0 0.048 yes
a Mann?Whitney tests.
The inheritance pattern of song spectral energy was also
consistent with a recessive Z-linked factor; nonlinear
regression models explained 79% to 86% of the variation
in song spectral energy. The songs of pure BWS males
had more energy at 1 kHz than at 4 kHz, while the reverse
was true for the songs of wild-type Border males. Hybrid
Proc. R. Soc. Lond. B (Suppl.)
males were generally intermediate in song spectral energy
when compared with the two parental strains.
(b) Correlations between hearing and song
The hearing threshold at 4 kHz for individual males was
a significant predictor of the frequency spectrum of their
04bl0074.S4 T. F. Wright and others Inheritance of hearing and song in canaries
song, explaining 49% of the variation in song energy. One
potential explanation for this association is that it arises
during song learning and is driven by poor perception of
the high-frequency elements of wild-type song. Biased
perception could affect both the memorization phase,
when young BWS canaries first acquire their auditory
templates, and the motor phase, when they attempt to
match their songs to those of other birds. This bias need
not be absolute; for example, some learning of less-per-
ceptible song elements could have occurred under the
controlled conditions of our tutoring regime where mask-
ing noise was minimized. Another possibility, albeit less
likely, is that the observed association between hearing
and song arises from a genetic correlation through plei-
otropy (one gene affecting multiple traits) or physical link-
age of different genes.
(c) Sensory linkage
The apparent linkage of hearing and song traits through
sensory perception in BWS canaries may have general
implications for the evolution of sexually selected traits.
Learned male song in oscine songbirds is important in
mate choice (Searcy & Yasukawa 1996), and in many
species, females learn their sexual preferences for male
song through early exposure (Nagle & Kreutzer 1997;
Irwin & Price 1999). Thus, changes in traits such as audi-
tory perception that affect song learning could potentially
alter the expression of song in males and the preference
of females simultaneously. Such ?sensory linkage? could
provide a mechanism for rapid trait evolution and repro-
ductive isolation through runaway sexual selection (Fisher
1930; Lande 1981; West-Eberhard 1983), wherein the
learned male trait and learned female preference would be
linked via a common dependence on sensory input rather
than through physical linkage of separate genetic loci.
We have demonstrated a correlation between sensory
perception and male song in hybrids of two canary strains
that probably arises during the learning process. Investi-
gations of variation in male song, female preferences and
hearing sensitivities among wild canary populations could
be used to test the hypothesized sensory linkage.
Acknowledgements
G. Wilkinson, P. Danley, S. Wisely, J. Dumbacher, J. Ortega, K.
Hanley and F. Nottebohm provided valuable comments. Pedigree
breeding was supported by PSC/CUNY grants, nos. 61252 and
62255 (P.C.M.) and approved under ACUC protocols, nos. 07 and
Proc. R. Soc. Lond. B (Suppl.)
88. Auditory work was supported by NIH-DC01372 (R.J.D.) and
NIH-NRSA-MH12111 (T.F.W.) and approved under ACUC proto-
col R-00-74. T.F.W. was supported by a Smithsonian Institution fel-
lowship during manuscript preparation.
Bloom, S. E., Delany, M. E. & Muscarella, D. E. 1993 Constant and
variable features of avian chromosomes. In Manipulation of the
avian genome (ed. R. J. Etches & A. M. V. Gibbins), pp. 39?59.
Boca Raton, FL: CRC Press.
Brittan-Powell, E. F., Dooling, R. J. & Gleich, O. 2002 Auditory
brainstem responses (ABR) in adult budgerigars (Melopsittacus
undulatus). J. Acoust. Soc. Am. 112, 999?1008.
Dooling, R. J., Mulligan, J. A. & Miller, J. D. 1971 Auditory sensi-
tivity and song spectrum of the common canary (Serinus canarius).
J. Acoust. Soc. Am. 50, 700?708.
Fisher, R. A. 1930 The genetical theory of natural selection. Oxford:
Clarendon.
Gleich, O., Dooling, R. J. & Presson, J. C. 1997 Evidence for sup-
porting cell proliferation and hair cell differentiation in the basilar
papilla of adult Belgian waterslager canaries (Serinus canarius). J.
Comp. Neurol. 377, 5?14.
Guttinger, H. R. 1985 Consequences of domestication on the song
structures in the canary. Behaviour 92, 255?278.
Irwin, D. E. & Price, T. 1999 Sexual imprinting, learning and speci-
ation. Heredity 82, 347?354.
Iyengar, V. K., Reeve, H. K. & Eisner, T. 2002 Paternal inheritance
of a female moth?s mating preference. Nature 419, 830?832.
Lande, R. 1981 Models of speciation by selection on polygenic traits.
Proc. Natl Acad. Sci. USA 78, 3721?3725.
Mundinger, P. C. 1995 Behavior?genetic analysis of canary song:
inter-strain differences in sensory learning, and epigenetic rules.
Anim. Behav. 50, 1491?1511.
Mundinger, P. C. 1999 Genetics of canary song learning: innate
mechanisms and other neurobiological considerations. In The
design of animal communication (ed. M. D. Hauser & M. Konishi),
pp. 369?389. Cambridge, MA: MIT Press.
Nagle, L. & Kreutzer, M. L. 1997 Song tutoring influences song pref-
erences in domesticated canaries. Behaviour 134, 89?104.
Okanoya, K. & Dooling, R. J. 1987 Strain differences in auditory
thresholds in the canary (Serinus canarius). J. Comp. Psychol. 101,
213?215.
Okanoya, K., Dooling, R. J. & Downing, R. J. 1990 Hearing and
vocalizations in hybrid Waterslager-roller canaries (Serinus
canarius). Hearing Res. 46, 271?276.
Price, T. D. 2002 Domesticated birds as a model for the genetics of
speciation by sexual selection. Genetica 116, 311?327.
Searcy, W. A. & Yasukawa, K. 1996 Song and female choice. In Ecol-
ogy and evolution of acoustic communication in birds (ed. D. E.
Kroodsma & E. H. Miller), pp. 454?473. Ithaca, NY: Cornell Uni-
versity Press.
West-Eberhard, M. J. 1983 Sexual selection, social competition, and
speciation. Q. Rev. Biol. 58, 155?183.
Wright, T. F., Cortopassi, K. A., Bradbury, J. W. & Dooling, R. J.
2003 Hearing and vocalizations in the orange-fronted conure
(Aratinga canicularis). J. Comp. Psychol. 117, 87?95.
Visit www.journals.royalsoc.ac.uk and navigate to this article through
Biology Letters to see the accompanying electronic appendix.