arr043 755..762


 

 

 University of Groningen

Zebra finch females prefer males with redder bills independent of song rate-a meta-analysis
Simons, Mirre J.P.; Verhulst, Simon

Published in:
Behavioral Ecology

DOI:
10.1093/beheco/arr043

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from
it. Please check the document version below.

Document Version
Publisher's PDF, also known as Version of record

Publication date:
2011

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):
Simons, M. J. P., & Verhulst, S. (2011). Zebra finch females prefer males with redder bills independent of
song rate-a meta-analysis. Behavioral Ecology, 22(4), 755-762. https://doi.org/10.1093/beheco/arr043

Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the
author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the
number of authors shown on this cover page is limited to 10 maximum.

Download date: 06-04-2021

https://doi.org/10.1093/beheco/arr043
https://research.rug.nl/en/publications/zebra-finch-females-prefer-males-with-redder-bills-independent-of-song-ratea-metaanalysis(67a53c89-7bce-45bd-9cb4-7344f1cbe248).html
https://doi.org/10.1093/beheco/arr043


Behavioral Ecology
doi:10.1093/beheco/arr043

Advance Access publication 4 May 2011

Original Article

Zebra finch females prefer males with redder
bills independent of song rate—a meta-analysis

Mirre J.P. Simons and Simon Verhulst
Behavioural Biology, Center for Life Sciences, University of Groningen, PO Box 11103, 9700 CC
Groningen, The Netherlands

Male zebra finches display multiple secondary sexual traits such as song and red bill coloration. This color is dependent on
carotenoids, which enhance immune function and are antioxidants. A red bill may thus function as an indicator signal. The zebra
finch is extensively used in the study of carotenoid-dependent signaling. However, studies of female mate preferences for redder
bills show mixed results. Here, we report a meta-analysis of mate-choice studies that reveals that female zebra finches do prefer
males with redder bills (r ¼ 0.61), except when there was reduced opportunity for imprinting or when bill color was experi-
mentally manipulated, which both reduced preference for red bills to approximately zero. The latter may either be due to aspects
of the experimental design or due to bill color being correlated with another trait such as song rate as was previously suggested.
We show, however, in a separate meta-analysis on a different set of studies that the correlation between bill coloration and song
rate (r ¼ 0.14) was significantly lower than the r ¼ 0.61 between bill color and attractiveness. We conclude, therefore, that the role
of bill coloration in mate choice cannot be solely due to an association with song rate. Thus, we conclude that females do prefer
males with redder bills when there was sufficient opportunity for sexual imprinting, but to what extent this is causally related to
the bill color remains to be established. Key words: bill coloration, mate choice, sexual coloration, sexual signaling, song rate,
zebra finch. [Behav Ecol 22:755–762 (2011)]

INTRODUCTION

Zebra finches (Taeniopygia guttata) exhibit brightly red col-ored bills. Sexual traits often signal quality (Grafen 1990;
Kotiaho 2001). In male zebra finches, reproductive success
(Price and Burley 1994) and physiological indicators of qual-
ity such as immune functioning and condition are positively
correlated with bill redness (Birkhead et al. 1998; Birkhead
et al. 2006; Bolund et al. 2010). Experimentally, an immune
challenge (Alonso-Alvarez et al. 2004; Gautier et al. 2008; Cote
et al. 2010) and cold exposure (Eraud et al. 2007) have been
shown to decrease zebra finch bill coloration. Male zebra
finch bill coloration thus exhibits variation that reflects phe-
notypic quality with respect to physiological state.
The redness of sexual ornaments of many species, including

the zebra finch’s bill, is dependent on carotenoids (McGraw
2004; Olson and Owens 2005; Pike, Blount, Bjerkeng, et al.
2007). Carotenoids have multiple physiological functions, act-
ing as antioxidants and supporting the immune system
(Pérez-Rodrı́guez 2009) but cannot be synthesized by the
animal itself and hence carotenoid availability is determined
by the dietary intake. Because carotenoids may be limiting but
physiologically important, carotenoid-dependent sexual
ornaments may be particularly suitable as quality indicators,
signaling immune functioning, and antioxidant capacity
(Olson and Owens 1998). Recently, the role of carotenoids
as antioxidants has been questioned (Costantini and Moller
2008). However, noncarotenoid antioxidants also increase
carotenoid-dependent sexual coloration suggesting that
carotenoid-dependent coloration does signal antioxidant ca-
pacity (Bertrand et al. 2006; Perez et al. 2008; Pike, Blount,

Lindström, and Metcalfe 2007). If carotenoid-dependent sex-
ual traits do reliably signal antioxidant capacity (Hartley and
Kennedy 2004; Pérez-Rodrı́guez 2009), mate choice for these
traits may yield direct and/or indirect fitness benefits, explain-
ing why such traits feature in mate selection.
Given that zebra finch bill color is extensively used in the

study of carotenoid-dependent sexual signaling and that it
reflects phenotypic quality, it is surprising that female prefer-
ences for bill color show little consensus among studies
(Collins and ten Cate 1996). Furthermore, it has been sug-
gested that the importance of bill color in mate choice is
minor and that instead females prefer high song rates (i.e.,
display rate), which covaries with bill color (Collins et al. 1994;
Collins and ten Cate 1996), hence resulting in an apparent
preference for redder bills in some studies.
Here, we combine all female mate-choice studies that we

could find that presented the required information, using
meta-analysis to test the hypothesis that female zebra finches
prefer males with redder bills. Studies that correlate natural
variation in male bill coloration with female choice cannot dem-
onstrate causality because choosing females may rely on traits
that covary with bill color resulting in a choice for redder bills.
We therefore also included studies that manipulated bill color
experimentally in our meta-analysis. We further tested whether
the covariance between song rate and bill color can explain
female choice for bill coloration as hypothesized by Collins
and ten Cate (1996) by summarizing the correlation between
song rate and bill coloration in a different set of studies.

MATERIALS AND METHODS

In a meta-analysis, the individual effect size estimates of differ-
ent studies are weighted by study sample size to combine them
into one average effect size. If this average effect size deviates
significantly from zero, it can be concluded that overall the
null hypothesis can be rejected, and hence this provides an
objective synthesis of studies that tested a specific hypothesis.

Address correspondence to M.J.P. Simons. E-mail: m.j.p.simons
@rug.nl.
Received 4 October 2010; revised 11 March 2011; accepted 12

March 2011.

� The Author 2011. Published by Oxford University Press on behalf of
the International Society for Behavioral Ecology. All rights reserved.
For permissions, please e-mail: journals.permissions@oup.com

 a
t U

n
ive

rsity o
f G

ro
n
in

g
e
n
 o

n
 S

e
p
te

m
b
e
r 2

1
, 2

0
1
1

b
e
h
e
co

.o
xfo

rd
jo

u
rn

a
ls.o

rg
D

o
w

n
lo

a
d
e
d
 fro

m
 

http://beheco.oxfordjournals.org/


To test whether the correlation between attractiveness and
bill color can be attributed to a correlation between bill color
and song rate, our approach was to quantify the association
between bill color and song rate using meta-analysis on a dif-
ferent set of studies and compare the strength of this corre-
lation with the correlation between the color of a male’s bill
and his attractiveness. When the association between bill color
and attractiveness is significantly stronger than the correlation
between bill color and song rate we can infer that the associ-
ation between bill color and attractiveness is not solely depen-
dent on an association with song rate.

Female mate choice for male bill coloration

We searched studies using Google scholar and ‘‘zebra finch,’’
‘‘female choice’’ or ‘‘female mate choice,’’ and ‘‘bill’’ or ‘‘beak’’
color (colour) as search terms and also checked the references
of the retrieved papers for relevant material. Authors were con-
tacted for relevant statistics not reported in papers. When studies
did not report raw proportions of choice or test statistics or the
raw data could not be measured from graphs or we did not
succeed in contacting the authors, they could not be used
(Immelmann 1959; Weisman et al. 1994).
The statistical approach between studies differed, with some

reporting the preference for the reddest male and others
reporting the relationship between the difference in redness
and the resulting female preference. The second approach
includes both the effect of the difference between males in red-
ness together with the overall preference for the reddest male.
We recommend reporting both in future research to ease com-
parison between studies. For the purpose of this review, we
included both approaches because the rejection of either ap-
proach would have resulted in a substantial loss of studies. We
preferred the statistic of the preference for the reddest males
if both approaches were available.
Our own unpublished data were included in the meta-

analysis. A classic 2-way choice trial was conducted in our
outside aviaries (Haren, The Netherlands) for 2 h in between
which males that could hear but not see each other were
switched sides to control for side preferences. Preference
was scored as time spent by the female at the side of the aviary
of the male relative to the total time spent with either of the
2 males. Bill color was measured by the use of digital photog-
raphy under controlled camera settings and lighting condi-
tions (Simons MJP, in preparation). The redness of the bill
was expressed as hue in HSV (Hue, Saturation, Value) color
space.
A key element of meta-analysis is the selection of studies to

include and unfortunately we had to omit the study by Roberts
et al. 2007 on methodological grounds because the authors
used principal component analysis to analyze spectrophotomet-
ric data of bills, which is difficult to compare with the methods
applied by the other studies. The methods employed by the
other studies were the Munsell method, photography, or
photospectrometry, which were all expressed as hue. In the
Munsell method, brightness (as well as saturation) is weighted
but little in comparison with hue (Burley and Coopersmith
1987). Roberts et al. 2007 found that males with brighter bills
(the principal component corresponding to brightness) were
preferred by females; this finding can therefore not be inter-
preted as either positive or negative but rather as an incentive
for further research on which aspects of the light reflected by
zebra finch bills are found to be attractive.

Experimental studies

Some studies manipulated bill color using nail polish, polymer
paint, or marker pen of wild-type or white morph birds

(Burley and Coopersmith 1987; Collins et al. 1994; Sullivan
1994; Vos 1995). To evaluate the effect of such manipulations,
we used a moderator for studies that interfered with the nat-
ural appearance of males.

Sexual imprinting

In birds sexual imprinting, which can shape mating preferen-
ces, seems to be the rule rather than the exception (ten Cate
and Vos 1999). In zebra finches, this process continues at
least up to 46 days of age (which is the median period of
imprinting experimentally shown to be still effective in shap-
ing preference: ten Cate 1987; Vos et al. 1993) and requires
close interaction with adult conspecifics (ten Cate et al.
1984). Bill color specifically has been shown to be a trait
zebra finches imprint strongly on, to the extent that exper-
imental imprinting conditions can reverse mate preferences
with respect to bill color (Weisman et al. 1994; Vos 1995).
The average imprinting opportunity in this set of studies
ranged from short, 30–40 days, to long, 48–100 days, and
we used these imprinting opportunity ranges as a 2-level
categorical moderator in the analysis. This dichotomization
was based on the length of the imprinting process, which was
evaluated experimentally (ten Cate 1987; Vos et al. 1993). We
also report the estimator for a continuous fit, but in further
analyses, we use dichotomization for 3 reasons. First, it is well
known that there are critical phases for imprinting and
hence sexual imprinting is not a linear process. Second, di-
chotomization allowed us to compare the overall effect size
of relatively long-imprinted individuals with the song rate
bill color correlation. Third, it allowed us to include
studies where the information provided on the imprinting
period was not provided in great detail (Table 1). Three
studies isolated chicks from adults at 30–40 days (Balzer
and Williams 1998; Blount et al. 2003; Forstmeier and
Birkhead 2004), 4 other studies kept chicks with adults on
which they could imprint for a period which was on average
over 48 days (see Table 1). Of these 3 studies, only
Forstmeier and Birkhead (2004) reported their imprinting
conditions; housing conditions from Balzer and Williams
(1998) and Blount et al. (2003) were obtained through
correspondence with the authors.

Song rate, bill color correlation

Search terms in Google scholar included ‘‘zebra finch’’ and
‘‘song’’ or ‘‘display’’ and ‘‘bill’’ or ‘‘beak’’ color/colour.
Authors of studies that measured both song rate and bill
coloration, but did not report how they correlated, were
contacted. We restricted our analysis to studies that mea-
sured song rate without other male competitors present be-
cause this confounds song rate with female choice and
behavior of the male competitor (this was the reason to omit
de Kogel and Prijs 1996). This resulted in a total of 6 studies.
In this analysis, we included studies regardless of the age up
to which juveniles could imprint on adults because we are
not aware of indications that this affects either bill color or
song rate.

Statistics

Reported statistics were converted into Pearson’s r using stan-
dard formulas (Rosenthal 1994). Proportions were converted
into a v2 statistic before conversion to r. Pearson’s r’s were
converted into Fisher’s Zr’s before analysis (Nakagawa and
Cuthill 2007). The meta-analyses were performed with the
Metafor package (Viechtbauer 2010) in R (R Development
Core Team 2009) using random effects meta-analysis fitted

756 Behavioral Ecology

 a
t U

n
ive

rsity o
f G

ro
n
in

g
e
n
 o

n
 S

e
p
te

m
b
e
r 2

1
, 2

0
1
1

b
e
h
e
co

.o
xfo

rd
jo

u
rn

a
ls.o

rg
D

o
w

n
lo

a
d
e
d
 fro

m
 

http://beheco.oxfordjournals.org/


Table 1

Summary of studies reporting female choice for male bill coloration

Study Sample size

Independent sample size
(corrected for possible
pseudoreplication) Statistic reported Effect size (r) Moderating variable

Bill color measurement
method and other remarks

Opportunity to imprint on
adults

Burley and Coopersmith
(1987)

14 4 12/14 0.87 Long imprinting Munsell Between 34 and 62 days
24 5 v2 ¼ 13.35 df ¼ 1 0.43 Long imprinting Munsell Between 34 and 62 days

de Kogel and Prijs (1996) 26 9 t ¼ 4.97 df ¼ 24 0.71 Long imprinting Munsell 50 days
Unpublished from our lab 21 21 t ¼ 3.31 df ¼ 19 0.60 Long imprinting Digital photography, hue in

HSV color space
100 days

Houtman (1992) 24 24 F ¼ 10.95 df ¼ 1,22 0.58 Long imprinting Munsell Birds were obtained from
commercial breeders and
the precise rearing
conditions were not
specified, but we have no
reason to assume that
chicks were isolated from
adults before 46 days of age
since to our best
knowledge, commercial
breeding usually takes place
in aviaries.

Roberts et al. (2007) 48 8 v2 ¼ 7.28 df ¼ 1 0.55 Not useable, brightness is
incomparable with Munsell
system data

Principal component of
spectrophotometry
corresponding to
brightness

62 days

Vos (1995) 19 5 t ¼ 20.30 df ¼ 17 20.07 Artificial manipulation Measured from Figure 1a,
t-test of arc sine converted
values against hypothesized
mean

55 days

Sullivan (1994) 11 5 6/11 0.09 Artificial manipulation 49 days
Burley and Coopersmith
(1987)

25 7 18/25 0.44 Artificial manipulation Munsell Between 34 and 62 days

Collins et al. (1994) 8 8 t ¼ 21.64 df ¼ 6 20.56 Artificial manipulation Not available
Blount et al. (2003) 10 10 t ¼ 0.74 df ¼ 8 0.25 Short imprinting Data obtained from author.

t-test of arc sine converted
values against hypothesized
mean, Color measurement
comparable with Munsell
with use of the Dulux Trade
Colour Palette

40 days

Balzer and Williams (1998) 33 33 Z ¼ 0.18df ¼ 31 0.03 Short imprinting Munsell 30 days
Forstmeier and Birkhead
(2004)

77 77 F ¼ 1.465 df ¼ 1,77 20.14 Short imprinting Munsell 35 days

HSV, Hue, Saturation, Value.

S
im

o
n
s
a
n
d
V
e
rh
u
lst

•
F
e
m
a
le

ze
b
ra

fi
n
ch

e
s
p
re
fe
r
re
d
b
ills

7
5
7

 at University of Groningen on September 21, 2011 beheco.oxfordjournals.org Downloaded from 

http://beheco.oxfordjournals.org/


with restricted maximum likelihood. When multiple effect
sizes were extracted from one study, we used the weighted
average. Each study was weighted by independent sample size
n 23 (Cooper et al. 2009). The studies were examined for
within study pseudoreplication, which occurs when stimulus
sets of males or individual females are used repeatedly. The
independent sample size is therefore the sample size that
could be used without risking pseudoreplication. For exam-
ple, if 20 females were tested with 5 stimulus sets of males, we
used 5 as the independent sample size.
Publication bias was investigated using funnel plots. In a fun-

nel plot, publication bias is revealed by an increase in absolute
effect size with decreasing sample size. The significance of
this relationship was tested using a rank test (Viechtbauer
2010).
When studies use different methodology or there are differ-

ences between study populations, this induces variability in
‘‘true’’ effect sizes. The resulting heterogeneity between studies
can be evaluated using the Q test (Viechtbauer 2010). A signif-
icant heterogeneity indicates that there are likely to be moder-
ating variables that explain the variability between studies.
The studies that allowed for short- and long-imprinting

periods and the experimental studies were coded in one mod-
erating variable with 3 levels. These were nonoverlapping cat-
egories because all (except one study that did not report this
information) studies that manipulated bills experimentally
allowed for a long-imprinting period. The estimates reported

are the deviation from the group of studies that allowed for
a long-imprinting period.

RESULTS

Do females prefer males with the reddest bill?

A total of 11 independent mate-choice studies were obtained
(Table 1, Figure 1). Our analysis revealed an average effect
size of r ¼ 0.28, (95% confidence interval [CI] 0.0006 2 0.52)
showing that on average, females preferred males with redder
bills (z ¼ 1.96, P , 0.05). Heterogeneity was significant how-
ever (Q ¼ 26.4, P ¼ 0.003).
The moderator coding for imprinting conditions and exper-

imental bill manipulation explained a significant proportion
of this variance (df ¼ 2, Q ¼ 21.93, P , 0.0001). The average
effect size of studies that allowed for a long-imprinting period
(n ¼ 4) turned out to be significantly higher than the average
effect size of the studies (n ¼ 3) with limited opportunity for
imprinting (difference: 20.65, z ¼ 24.59, P , 0.0001) and
also significantly higher than the average effect size of the
experimental studies (n ¼ 4; difference: 20.67, z ¼ 22.58,
P , 0.01). Less opportunity for imprinting and artificial ma-
nipulation thus reduced female preference for red male bill
coloration to approximately zero (0.61 – 0.65 or 0.67 ¼ 20.04
or 20.06). When imprinting opportunity (in days) was fitted
as a continuous linear moderator (which reduced sample

Figure 1
Effect size of female preference for red bills (r 6 95% CI calculated using independent sample size), ordered within each panel with respect to
sample size, with the lowest sample size at the top. Bottom effect size is the average.

758 Behavioral Ecology

 a
t U

n
ive

rsity o
f G

ro
n
in

g
e
n
 o

n
 S

e
p
te

m
b
e
r 2

1
, 2

0
1
1

b
e
h
e
co

.o
xfo

rd
jo

u
rn

a
ls.o

rg
D

o
w

n
lo

a
d
e
d
 fro

m
 

http://beheco.oxfordjournals.org/


size) this turned out to be also significantly correlated with
preference for red bills (n ¼ 6, z ¼ 21.98, P ¼ 0.048).
In nonexperimental studies where birds did have ample im-

printing opportunity, the average effect size was r ¼ 0.61 (95%
CI 0.41–0.75; z ¼ 5.06, P , 0.001) with nonsignificant hetero-
geneity (Q ¼ 0.26, P ¼ 0.97). The latter effect size, without the
confounding effects of limited imprinting opportunity and bill
color manipulation, we consider the best estimate of the
strength of the relationship between male bill coloration and
female mate choice. We used this effect size in the comparison
with correlation between song rate and bill color.

Song rate and comparison with female mate-choice effect

We found 6 independent studies that reported the correlation
between song rate and bill color (Table 2, Figure 2). The
average effect size was r ¼ 0.14, and the 95% CI included zero
(20.07 to 0.34; z ¼ 1.30, P ¼ 0.19). This suggests that the
relationship between song rate and bill coloration of male
zebra finches is weak on average and not significant. Hetero-
geneity was significant (Q ¼ 12.5, P ¼ 0.03).
When we compared this average effect (r ¼ 0.14) with the

average effect size of female mate choice for males with red-
der bills (r ¼ 0.61), using a Student’s t-test on the average
effect sizes with their corresponding standard errors under
the assumption of unequal variance, this revealed that the
latter was significantly higher (t ¼ 3.23, P ¼ 0.018). Female
choice for red bills can therefore not be solely dependent on
the correlation between song rate and bill coloration (see
MATERIALS AND METHODS).

Publication bias

Funnel plots (Figure 3) did not reveal a publication bias, but
this is difficult to detect with this relatively limited sample of

studies. Rank tests for funnel plot asymmetry were nonsignif-
icant for both analyses (Kendall’s tau . 20.09, P . 0.70).

DISCUSSION

Red is preferred

Previous studies reported mixed results with respect to the role
of male bill color in female mate choice in zebra finches.
Using meta-analysis, we show that female zebra finches on
average prefer males with redder bills, which is in agreement
with the reported link between bill color and phenotypic qual-
ity (see INTRODUCTION). Given that bill coloration signals
phenotypic quality, female choice for redder bills selects for
males with higher phenotypic quality.
It has previously been suggested (Collins et al. 1994; Collins

and ten Cate 1996) that observed mating preference for males
with redder bills might be due to mate selection for song rate,
when this is positively associated with bill redness. However,
our meta-analysis revealed that the association between bill
redness and song rate was not very strong (r ¼ 0.14) and
not significant and significantly lower than the reference ef-
fect size for preference for redder bills (r ¼ 0.61). This implies
that the latter effect cannot be fully explained by the correla-
tion between male bill color and song rate. A large difference
in measurement error between mate choice and song rate
could, however, also be responsible for this difference because
random measurement error reduces effect sizes. However,
measurement error of mate choice is probably higher than
that of song rate, given that mate choice is a behavioral trait
with relatively low repeatability (Bell et al. 2009) whereas song
rate has been reported to be highly repeatable (Birkhead and
Fletcher 1995; Forstmeier and Birkhead 2004). Thus, we con-
sider it unlikely that the difference between the 2 effect sizes is
due to a difference in measurement error.

Figure 2
Effect size of (r 6 95% CI) the
song rate/ bill coloration rela-
tionship ordered from top to
bottom with respect to sample
size, starting with the lowest
sample size. Bottom effect size
is the average.

Table 2

Summary of studies reporting the correlation between male bill coloration and song rate

Study Sample size Statistic reported (effect size, r) Bill color measurement method and other remarks

Birkhead and Fletcher (1995) 10 0.05 Munsell
Houtman (1992) 22 0.45 Munsell
Birkhead et al. (1998) 31 0.13 Munsell, measured and analyzed from their Figure 4
Unpublished from Barbara Tschirren 57 20.25 Hue calculated from spectrophotometry, for methods

see Pryke et al. (2001), Tschirren et al. (2009)
Bolund et al. (2010) 76 0.26 Munsell
Forstmeier and Birkhead (2004) 104 0.20 Munsell

Simons and Verhulst • Female zebra finches prefer red bills 759

 a
t U

n
ive

rsity o
f G

ro
n
in

g
e
n
 o

n
 S

e
p
te

m
b
e
r 2

1
, 2

0
1
1

b
e
h
e
co

.o
xfo

rd
jo

u
rn

a
ls.o

rg
D

o
w

n
lo

a
d
e
d
 fro

m
 

http://beheco.oxfordjournals.org/


The proportion of variance in female choice explained by
male bill coloration is 0.612 ¼ 0.37. The proportion that can-
not be due to the song rate/bill color correlation is 0.37 –
0.142 ¼ 0.35. This considerable amount of explained variance
increases our confidence in the signaling function of the male
zebra finch bill. However, it still leaves a considerable part of
variance to be explained (0.65 if error is ignored). This means
that other (sexual) traits have the potential to be more
important in mate choice as bill coloration.
Due to the correlative nature of the mate choice studies the

effect of bill color does not necessarily imply that females
discriminate between potential mates using bill coloration.
Instead, they may select on traits other than song rate that
do covary with bill color. Possible candidates are song content
(Holveck and Riebel 2007; Riebel 2009), UV reflectanc
(Bennett et al. 1996), and chest plumage symmetry (Swaddle
and Cuthill 1994), which have all been shown to play a role in
zebra finch mate choice. A way to establish to what extent bill
color is causally involved in the strong preference for redder
bills is to study the effect of manipulated bill color on attrac-
tiveness. However, the available experimental studies show no
effect on average (Figure 1). There can be different reasons
for the conspicuous contrast between the experimental and
observational results, including of course, as discussed above,

that females base mate choice on other traits that show, how-
ever, a fairly strong correlation with bill color. Alternatively,
there could be methodological aspects of the experimental
studies that explain the negligible effect size, such as the chal-
lenge to manipulate bill color while maintaining a fully natu-
ral appearance of bill color. Furthermore, even when the
manipulation is successful in maintaining the natural appear-
ance of the bill, the manipulation may create a mismatch
between bill color and other sexual signals, which in itself
may change female perception of the male (e.g., Künzler
and Bakker 2001). Lastly, different control groups are missing
from the experimental studies, which makes these studies dif-
ficult to interpret and compare. When a female is presented
with a male manipulated to be more attractive and an unma-
nipulated male (as in Burley and Coopersmith 1987), the lack
of a sham-manipulated group limits the strict conclusion from
such experiments to: female zebra finches prefer or do not
prefer artificially manipulated males. When both males are
manipulated (as in Collins et al. 1994 and Sullivan 1994),
the resulting artificial signal might cause females to behave
abnormally if it does not adequately mimic the natural signal
(as also argued in Collins and ten Cate 1996). A design that
includes nonmanipulated, sham-manipulated, and manipu-
lated to be attractive or less attractive could increase our un-
derstanding of the causality of female choice for male red bill
coloration. Hence, bill color manipulations can show a causal
effect of bill color, but failing to reject the null hypothesis can
be attributable to the general approach rather than to the
absence of causality.

Imprinting

In zebra finches, mating preferences are at least partly shaped
during imprinting at least up to 46 days of age (ten Cate 1987;
Vos et al. 1993) and bill color specifically has been shown to
be an important trait in this imprinting process (Weisman
et al. 1994; Vos 1995). In agreement with these findings, re-
duced opportunity for imprinting after the age of 30–40 days
resulted in an absence of preference for red male bill color-
ation. This effect may arise because the preference for red bill
ornamentation was not fully imprinted, which could have re-
sulted in reduced kin or sex recognition, which in its turn
affected mate choice. Imprinting could have continued after
30–40 days in the groups in which juveniles were kept (per-
sonal information from authors) among which bill color had
not developed its coloration from black to reddish (de Kogel
1997) resulting in a preference for juvenile coloration. Other
aspects of the imprinting process other than specific imprint-
ing on bill coloration can also be responsible for the differ-
ences we find. Reduced kin or sex recognition, which is the
potential result from a short imprinting period, could affect
mate-choice decisions in general resulting in no preference
for bill coloration. Thus, our analysis confirms that husbandry
practices can critically affect female choice with respect to bill
color, as previously suggested (Forstmeier and Birkhead 2004)
and also reported for song (Riebel 2000). Note, however, that
the imprinting effect we report is based on a limited number
of studies, and in our view, this result should above all be seen
as a good reason to evaluate this pattern with an experiment.
An effect of imprinting will be important in the interpretation
of studies on sexual selection in the zebra finch, and we sug-
gest it would be prudent to control and report imprinting
conditions in detail in future studies.

Male choice

We have focused on female choice, but zebra finches form sta-
ble pair bonds with mutual mate choice (Silcox and Evans

Figure 3
Funnel plots of both meta-analyses. Independent sample size is
plotted against the effect size for each study.

760 Behavioral Ecology

 a
t U

n
ive

rsity o
f G

ro
n
in

g
e
n
 o

n
 S

e
p
te

m
b
e
r 2

1
, 2

0
1
1

b
e
h
e
co

.o
xfo

rd
jo

u
rn

a
ls.o

rg
D

o
w

n
lo

a
d
e
d
 fro

m
 

http://beheco.oxfordjournals.org/


1982; Monaghan et al. 1996). Hence, the evolution of bill
color will also depend on male preferences, but unfortunately
male choice for female bill color has been investigated in only
2 studies (Burley and Coopersmith 1987; de Kogel and Prijs
1996). Burley and Coopersmith (1987) reported a preference
for orange females compared with red, but extreme orange
toward yellowish females were avoided. de Kogel and Prijs
(1996) reported a nonsignificant preference for females with
more orange bills compared with red. Interestingly, females
with more orange bills were found to have increased repro-
ductive output and survival (Price and Burley 1994), suggest-
ing males should prefer females with more orange bills
compared with red. However, females with redder bills deposit
more carotenoids into their eggs and increased yolk carote-
noids are associated with increased hatching and fledging
success (McGraw et al. 2005). Further study is required before
conclusions can be reached regarding the association between
female bill color and her sexual attractiveness, and hence the
role of male mate choice in the evolution of bill color.

CONCLUSIONS

We found a significant overall preference for males with
redder bills and we show that the overall effect is significantly
higher than the correlation between song rate and bill color-
ation. This leads us to reject the hypothesis (Collins et al.
1994; Collins and ten Cate 1996) that the preference for red
bill coloration is a result from this correlation. Additionally,
the significant moderating effect of imprinting on female
choice for bill coloration warrants experimental testing of this
effect.

FUNDING

Netherlands Organization of Scientific Research (NWO) Top-
talent grant to M.J.P.S. and an NWO Vici grant to S.V.

Egbert Koetsier and undergraduate students collected the mate-choice
data in our laboratory. Jonathan Blount and Barbara Tschirren kindly
provided unpublished data, Katharina Riebel alerted us to the impor-
tance of imprinting, and Wolfgang Forstmeier, Tony Williams, Tim
Birkhead, and Jonathan Blount discussed their methodology with
us. We thank Shinichi Nakagawa and Wolfgang Viechtbauer for advice
for the meta-analysis. Carel ten Cate provided valuable comments on
an earlier version of the manuscript.

REFERENCES

Alonso-Alvarez C, Bertrand S, Devevey G, Gaillard M, Prost J, Faivre B,
Sorci G. 2004. An experimental test of the dose-dependent effect of
carotenoids and immune activation on sexual signals and antioxi-
dant activity. Am Nat. 164:651–659.

Balzer AL, Williams TD. 1998. Do female zebra finches vary primary
reproductive effort in relation to mate attractiveness? Behaviour. 135:
297–309.

Bell AM, Hankison SJ, Laskowski KL. 2009. The repeatability of be-
haviour: a meta-analysis. Anim Behav. 77:771–783.

Bennett ATD, Cuthill IC, Partridge JC, Maier EJ. 1996. Ultraviolet
vision and mate choice in zebra finches. Nature. 380:433–435.

Bertrand S, Faivre B, Sorci G. 2006. Do carotenoid-based sexual traits
signal the availability of non-pigmentary antioxidants? J Exp Biol.
209:4414–4419.

Birkhead TR, Fletcher F. 1995. Male phenotype and ejaculate quality
in the zebra finch Taeniopygia guttata. Proc R Soc B Biol Sci. 262:
329–334.

Birkhead TR, Fletcher F, Pellatt EJ. 1998. Sexual selection in the zebra
finch Taeniopygia guttata: condition, sex traits and immune capacity.
Behav Ecol Sociobiol. 44:179–191.

Birkhead TR, Pellatt EJ, Matthews IM, Roddis NJ, Hunter FM,
McPhie F, Castillo-Juarez H. 2006. Genic capture and the genetic

basis of sexually selected traits in the zebra finch. Evolution. 60:
2389–2398.

Blount JD, Metcalfe NB, Birkhead TR, Surai PF. 2003. Carotenoid
modulation of immune function and sexual attractiveness in zebra
finches. Science. 300:125–127.

Bolund E, Martin K, Kempenaers B, Forstmeier W. 2010. Inbreeding
depression of sexually selected traits and attractiveness in the zebra
finch. Anim Behav. 79:947–955.

Burley NT, Coopersmith C. 1987. Bill color preferences of zebra
finches. Ethology. 76:133–151.

ten Cate C. 1987. Sexual preferences in zebra finch males raised by
two species: II. The internal representation resulting from double
imprinting. Anim Behav. 35:321–330.

ten Cate C, Los L, Schilperoord L. 1984. The influence of differences
in social experience on the development of species recognition in
zebra finch males. Anim Behav. 32:852–860.

ten Cate C, Vos D. 1999. Sexual imprinting and evolutionary processes
in birds: a reassessment. Adv Study Behav. 28:1–31.

Collins SA, ten Cate C. 1996. Does beak colour affect female prefer-
ence in zebra finches? Anim Behav. 52:105–112.

Collins SA, Hubbard C, Houtman AM. 1994. Female mate choice in
the zebra finch—the effect of male beak colour and male song.
Behav Ecol Sociobiol. 35:21–25.

Cooper HM, Hedges LV, Valentine JC. 2009. The handbook of re-
search synthesis and meta-analysis. 2nd ed. New York: Russell Sage
Foundation.

Costantini D, Moller A. 2008. Carotenoids are minor antioxidants for
birds. Funct Ecol. 22:367.

Cote J, Arnoux E, Sorci G, Gaillard M, Faivre B. 2010. Age-dependent
allocation of carotenoids to coloration versus antioxidant defences.
J Exp Biol. 213:271–277.

Eraud C, Devevey G, Gaillard M, Prost J, Sorci G, Faivre B. 2007.
Environmental stress affects the expression of a carotenoid-based
sexual trait in male zebra finches. J Exp Biol. 210:3571–3578.

Forstmeier W, Birkhead TR. 2004. Repeatability of mate choice in the
zebra finch: consistency within and between females. Anim Behav.
68:1017–1028.

Gautier P, Barroca M, Bertrand S, Eraud C. 2008. The presence of
females modulates the expression of a carotenoid-based sexual sig-
nal. Behav Ecol Sociobiol. 62:1159–1166.

Grafen A. 1990. Biological signals as handicaps. J Theor Biol. 144:
517–546.

Hartley RC, Kennedy MW. 2004. Are carotenoids a red herring in
sexual display? Trends Ecol Evol. 19:353–354.

Holveck MJ, Riebel K. 2007. Preferred songs predict preferred males:
consistency and repeatability of zebra finch females across three test
contexts. Anim Behav. 74:297–309.

Houtman AM. 1992. Female zebra finches choose extra-pair copula-
tions with genetically attractive males. Proc R Soc B Biol Sci. 249:3–6.

Immelmann K. 1959. Experimentelle Untersuchungen über die biol-
ogische Bedeutung artspezifischer Merkmale beim Zebrafinken
(Taeniopygia castanotis Gould). Zool Jahrb Abt Syst Ökol Geogr
Tiere. 86:437–592.

de Kogel CH. 1997. Long-term effects of brood size manipulation on
morphological development and sex-specific mortality of offspring.
J Anim Ecol. 66:167–178.

de Kogel CH, Prijs HJ. 1996. Effects of brood size manipulations on
sexual attractiveness of offspring in the zebra finch. Anim Behav. 51:
699–708.

Kotiaho JS. 2001. Costs of sexual traits: a mismatch between theoret-
ical considerations and empirical evidence. Biol Rev. 76:365–376.

Künzler R, Bakker T. 2001. Female preferences for single and com-
bined traits in computer animated stickleback males. Behav Ecol.
12:681–685.

McGraw K, Adkins-Regan E, Parker R. 2005. Maternally derived
carotenoid pigments affect offspring survival, sex ratio, and sexual
attractiveness in a colorful songbird. Naturwissenschaften. 92:
375–380.

McGraw KJ. 2004. Colorful songbirds metabolize carotenoids at the
integument. J Avian Biol. 35:471–476.

Monaghan P, Metcalfe NB, Houston DC. 1996. Male finches selectively
pair with fecund females. Proc Biol Sci. 263:1183–1186.

Nakagawa S, Cuthill IC. 2007. Effect size, confidence interval and
statistical significance: a practical guide for biologists. Biol Rev
Camb Philos Soc. 82:591–605.

Simons and Verhulst • Female zebra finches prefer red bills 761

 a
t U

n
ive

rsity o
f G

ro
n
in

g
e
n
 o

n
 S

e
p
te

m
b
e
r 2

1
, 2

0
1
1

b
e
h
e
co

.o
xfo

rd
jo

u
rn

a
ls.o

rg
D

o
w

n
lo

a
d
e
d
 fro

m
 

http://beheco.oxfordjournals.org/


Olson V, Owens I. 1998. Costly sexual signals: are carotenoids rare,
risky or required? Trends Ecol Evol. 13:510–514.

Olson VA, Owens IPF. 2005. Interspecific variation in the use of
carotenoid-based coloration in birds: diet, life history and
phylogeny. J Evol Biol. 18:1534–1546.

Perez C, Lores M, Velando A. 2008. Availability of nonpigmentary
antioxidant affects red coloration in gulls. Behav Ecol. 19:967–973.

Pérez-Rodrı́guez L. 2009. Carotenoids in evolutionary ecology: re-
evaluating the antioxidant role. BioEssays. 31:1116–1126.

Pike TW, Blount JD, Bjerkeng B, Lindström J, Metcalfe NB. 2007.
Carotenoids, oxidative stress and female mating preference for lon-
ger lived males. Proc R Soc B Biol Sci. 274:1591–1596.

Pike TW, Blount JD, Lindström J, Metcalfe NB. 2007. Availability of
non-carotenoid antioxidants affects the expression of a carotenoid-
based sexual ornament. Biol Lett. 3:353–356.

Price DK, Burley NT. 1994. Constraints on the evolution of attractive
traits: selection in male and female zebra finches. Am Nat. 144:
908–934.

Pryke SR, Andersson S, Lawes MJ. 2001. Sexual selection of multiple
handicaps in the red-collared widowbird: female choice of tail
length but not carotenoid display. Evolution. 55:1452–1463.

R Development Core Team. 2009. A language and environment for
statistical computing. Vienna (Austria): R Foundation for Statistical
Computing.

Riebel K. 2000. Early exposure leads to repeatable preferences for
male song in female zebra finches. Proc R Soc B Biol Sci. 267:
2553–2558.

Riebel K. 2009. Song and female mate choice in zebra finches: a re-
view. Adv Study Behav. 40:197–238.

Roberts ML, Buchanan KL, Bennett ATD, Evans MR. 2007. Mate choice in
zebra finches: does corticosterone play a role? Anim Behav. 74:921–929.

Rosenthal R. 1994. Parametric measures of effect size. In: Cooper H,
Hedges L, editors. The handbook of research synthesis. New York:
Russell Sage Foundation Publications. p. 231–244.

Silcox AP, Evans SM. 1982. Factors affecting the formation and main-
tenance of pair bonds in the zebra finch, Taeniopygia guttata. Anim
Behav. 30:1237–1243.

Sullivan M. 1994. Discrimination among males by female zebra finches
based on past as well as well as current phenotype. Ethology. 96:
97–104.

Swaddle J, Cuthill I. 1994. Female zebra finches prefer males with
symmetric chest plumage. Proc R Soc B Biol Sci. 258:267–271.

Tschirren B, Rutstein AN, Postma E, Mariette M, Griffith SC. 2009.
Short- and long-term consequences of early developmental condi-
tions: a case study on wild and domesticated zebra finches. J Evol
Biol. 22:387–395.

Viechtbauer W. 2010. Conducting meta-analyses in R with the metafor
package. J Stat Software. 36:1–48.

Vos DR. 1995. The role of sexual imprinting for sex recognition in
zebra finches: a difference between males and females. Anim Behav.
50:645–653.

Vos DR, Prijs J, ten Cate C. 1993. Sexual imprinting in zebra finch
males: a differential effect of successive and simultaneous experi-
ence with two colour morphs. Behaviour. 126:137–154.

Weisman R, Shackleton S, Ratcliffe L, Weary D, Boag P. 1994. Sexual
preferences of female zebra finches: imprinting on beak colour.
Behaviour. 128:15–24.

762 Behavioral Ecology

 a
t U

n
ive

rsity o
f G

ro
n
in

g
e
n
 o

n
 S

e
p
te

m
b
e
r 2

1
, 2

0
1
1

b
e
h
e
co

.o
xfo

rd
jo

u
rn

a
ls.o

rg
D

o
w

n
lo

a
d
e
d
 fro

m
 

http://beheco.oxfordjournals.org/