ANIMAL BEHAVIOUR, 2003, 65, 6981

doi:10.1006/anbe.2002.2041, available online at http://www.ScienceDirect.com

Mate choice for multiple ornaments in the California quail,

Callipepla californica

JENNIFER DEVLIN CALKINS & NANCY TYLER BURLEY

Department of Ecology and Evolutionary Biology, University of California, Irvine

(Received 8 February 2001; initial acceptance 18 April 2001;

final acceptance 10 May 2002; MS. number: A8987R)

Two processes that may be important in the evolution of multiple ornaments are (1) preference flexibility
based on the distribution of local traits and (2) preferences for composite (versus partitioned) traits. To
examine the occurrence of these processes we conducted three experiments measuring the mate
preferences of California quail. We included analyses of composite traits as well as the partitioned traits
(isolated patches of coloration) typically invoked in ornithological descriptions. In two experiments, in
which the same pool of wild-caught females was presented pairs of males with natural versus experimen-
tally exaggerated traits, the focus of mate choice shifted to the trait specifically exaggerated in that
experiment. In a third experiment, in which females were presented males displaying natural plumage
variation, females chose primarily based on a composite trait (as revealed using stepwise analysis).
Composite traits were also found to be stronger predictors of female behaviour in one of the manipu-
lation experiments. These results indicate that female California quail have active, flexible mate
preferences for a variety of novel and extant traits and illustrate the importance of incorporating
composite traits into the analysis of mate preferences.
2003 The Association for the Study of Animal Behaviour. Published by Elsevier Science Ltd. All rights reserved.

Results of both quantitative models (e.g. Mller &
Pomiankowski 1993; Andersson 1994; Iwasa &
Pomiankowski 1994; Johnstone 1995; Ligon & Zwartjes
1995; ) and numerous empirical studies (e.g. Burley 1981;
Johnson 1988; Moore 1990; Zuk et al. 1990; Brooks and
Caithness 1995; Backwell & Passmore 1996; Omland
1996a, b; Houde 1997; Marchetti 1998; Mller et al. 1998;
Hill et al. 1999; Qvarnstrom et al. 2000) indicate that
mate choice is responsible for generating and maintain-
ing the multiple secondary sexual traits observed in many
species of animals. Despite the widespread acceptance of
the role of mate choice in the evolution of ornamentation
patterns, authors of several recent studies have suggested
that mate preferences for ornamentation are absent,
especially among gallinaceous birds (Ligon & Zwartjes
1995; Hagelin 1999; Hagelin & Ligon 2001). Hypotheses
invoked to account for the maintenance of traits after
preferences for them have been lost range from the idea
that most traits lack current signal value and persist as
ghosts of selection past (e.g. Mller & Pomiankowski
1993; Ligon & Zwartjes 1995; Holland & Rice 1998) to the
Correspondence and present address: J. D. Calkins, Swanson Labora-
tory, Department of Genome Sciences and Medicine, University of
Washington, Box 357730, HSB K-340B, Seattle, WA 98195-7730,
U.S.A. (email: jennic@citrus.ucr.edu). N. T. Burley is at the Depart-
ment of Ecology and Evolutionary Biology, 900 University Avenue,
University of California Irvine, Irvine, CA 92697, U.S.A.
69
00033472/03/$30.00/0 2003 The Association for the Study of Animal Behaviour. Published by Elsevier Science Ltd. All rights reserved.
possibility that discrete traits convey redundant infor-
mation about the bearers mating quality (Mller &
Pomiankowski 1993).
Two explanations for the differing results reported in
empirical studies are that (1) females choose males based
upon combinations of traits and that (2) mate preferences
are flexible rather than fixed. For example, females may
be constrained by the number of mate choice criteria that
they can employ at any one time (N. T. Burley & J. D.
Calkins, unpublished data) and either combine aspects of
many traits into composite trait impressions or prioritize
their choices based on immediate environmental and
social conditions (e.g. Jennions & Petrie 1997 and refer-
ences therein; Marchetti 1998). In order to build a frame-
work for the coherent discussion of these possibilities,
greater attention needs to be focused on the following
two questions. What constitutes a trait used in mate
choice? How do organisms evaluate potential mates with
multiple, variable ornaments?
What Constitutes a Trait Used in Mate Choice?
A survey of the literature on sexual selection suggests
that the standard definition of a trait is any measurable
aspect of an organisms phenotype intuited by a given
researcher that is considered to be independent and
important (Bowler 1996). A weakness of this approach is
70 ANIMAL BEHAVIOUR, 65, 1

that it implicitly assumes that nonhumans perceive trait

divisions in ways similar to humans, whereas they may
actually partition traits differently (Jennions & Petrie
1997). For example, a researcher evaluating adjacent
black and white areas of plumage on a bird might tend to
describe these as separate patches or traits and focus on
some characteristic of these separate units, whereas the
species under study might actually focus on the relation-
ship between these patches (relative size, contrast).
Another problem with a perception-based approach to
trait definition is that the types and range of stimuli
perceived vary greatly among taxa (e.g. avian perception
of UV wavelengths: Bennett et al. 1994; perception of
high-frequency sounds: Bradbury & Vehrencamp 1998,
and references therein). Finally, this approach does not
account for individual traits (e.g. carotenoid-based
patches) that are comprised of developmentally indepen-
dent units (Badyeav et al. 2001).

How Do Organisms Evaluate Potential Mates with

Multiple Ornaments?
Theoreticians have suggested some simple decision-
making rules that organisms might use for making
sequential (e.g. Janetos 1980; Wittenberger 1983; Real
1990) or pooled (Janetos 1980; Real 1990) mate choices.
These rules, however, largely fail to address the complex- Figure 1. Measured plumage characteristics of male California quail.
ity of decisions faced by multiply ornamented organisms For all patches, areas were calculated as the product of the length
residing in moderate-to-large social groups. We hypoth- and the width. Measurement of the length of each patch was taken
esize, first, that such organisms are able to assess the along the vertical midline. The width of the black patch was
overall composite-trait impression left by several traits measured across the bottom of the beak. The width of the tan patch
as perceived by humans rather than make an indepen- was measured between the top of the wings. The width of the
dent assessment of each trait, thus simplifying and brown patch was measured from the top of the legs.
streamlining the assessment phase of mate choice.
Second, we hypothesize that organisms can alter priori-
& Bradbury 1985; Petrie et al. 1991; Zuk et al. 1992) the
ties facultatively, changing mate choice criteria as the
socially monogamous Odontophorinae have received
distribution of ornamental variation changes.
comparatively little attention (but see Brown & Gutierrez
We report here on a series of three mate choice exper-
1980; Brown et al. 1998; Hagelin 1999; Hagelin & Ligon
iments performed on wild-caught California quail
2001). Odontophorine species show a range of ornamen-
designed to address these questions. For two experiments
tation from full sexual dimorphism (with males more
we phenotypically altered males. In a third experiment
ornamented) to sexual indistinguishability. Despite being
we explored female mate preferences for males with
size monomorphic (Calkins et al. 1999), California quail
naturally varying traits. In analysing the data we ask the
are quite dichromatic, with males possessing several orna-
following questions. (1) Do female California quail
ments that females lack (Fig. 1). Recent fieldwork has
respond to qualities of particular male ornaments? (2)
established that these ornaments influence mating suc-
Does a population of females change the focus of their
cess in nature and that ornamentation varies temporally
preference when presented with different distributions of
and spatially (Calkins 2000).
male phenotypes created through experimental manipu-
Female California quail have ample time to assess male
lations? (3) Is male mating attractiveness better predicted
attractiveness, probably at minimal cost (Wittenberger
by a series of traits partitioned into relatively discrete
1983), as the sexes remain together in stable flocks (cov-
units as traditionally characterized by humans (Fig. 1) or
eys) during the nonbreeding season, and females tend to
by composite traits grouped on the basis of proximity and
mate with males from the same covey (J. D. Calkins,
similarity of coloration?
unpublished data). Since the species is nonterritorial,
females are expected to choose a mate only on the basis
Study System of male characteristics. Quail communicate interest in
mating via maintaining proximity and through court-
The California quail is a member of the galliform family ship signals such as courtship feeding (tidbitting;
Odontophorinae. Although both promiscuous and Stokes & Williams 1971). Most wild-caught adults ad-
socially polygynous galliforms have been frequent sub- just well to being held in aviaries, for an interval of
jects of sexual selection studies (e.g. Darwin 1871; Gibson 35 months.

METHODS
We conducted three mate choice experiments in which
we presented females with a choice between two males.
For two of the experiments we asked whether females
responded to a single trait (cap colour and crest length)
manipulated outside of the species-typical range. These
experiments were conducted in the same year, using a
common pool of females, and therefore, they also
allowed us to explore the potential for population-level
flexibility in mate choice. In a third experiment, we
presented females with males possessing naturally vary-
ing traits. The data from the three experiments were then
used to test female response to multiple male traits
through stepwise regression analyses.
Trapping and Captivity
All test females were wild-caught as adults to ensure
that they had been exposed to the phenotypes of parents
and other conspecifics during ontogeny. Males used were
also wild-caught adults, with the exception of two
captive-reared birds that formed one stimulus set in one
experiment (crest length experiment). Experimental birds
were caught prior to the beginning of the breeding season
in baited walk-in traps (on the Ecke Ranch in 1996, and in
Rancho Santa Fe and Manchester Open Space in 1997,
in California, U.S.A.). We banded birds with numbered
metal bands at the time of capture. Birds were held in
groups of 216 individuals from the same covey/
geographical location at an approximate density of 0.33
birds per m3 in single-sex aviaries located on the Ecke
Ranch. Birds were visually isolated from the opposite sex.
Food (turkey gamebird breeder crumble, 3-way chick
scratch, lettuce and mealworms) and water were available
ad libitum. There was adequate cover for all birds as well
as dirt for dustbathing, and aviaries were cleaned every
second day. Individual birds were held for 34 months
and released at their capture sites. Released birds received
supplemental food and water and were monitored to
ascertain reaclimation to natural conditions.
Plumage and Size Measurements
The mass of each bird was measured using a 300-g
Pesola scale both at the start and end of the holding
period and was averaged for the interval under consider-
ation. Tarsus length (in mm) was measured using a dial
caliper drawn across the upper side of the leg from the top
of the tibiotarsal joint to the knuckle just above the
middle toe. Plumage measurements (size and colour) were
taken as described below. To reduce measurement error,
all measurements except mass were scored three times on
the same date, and mean values were used in analyses.
J.D.C. conducted all measurements.
We delineated markings as separate partitioned traits
if they are typically considered independent for species
and subspecies identification (Fig. 1; for summaries of
identifying characteristics, see Johnsgard 1973; Leopold
1977; Calkins et al. 1999). We measured the length (mm)
of the crest, the area (mm2) of the black, tan, brown and
cap patches and the colour of the brown and cap patches.
CALKINS & BURLEY: MATE CHOICE IN CALIFORNIA QUAIL 71
Crest length was measured (in mm) by flattening the crest
against a metal ruler. The areas (mm2) of the black, tan
and brown patches were estimated from the product of
the width times the length of the patch taken at standard
locations on the birds bodies. The area of the cap in
(mm2) was estimated using the formula
((t+b)/2)m),
where t is the width across the top of the cap behind the
eye, b is the width across the bottom edge, and m is the
length along the midsection of the brown patch. Colour
hue, value (lightness versus darkness) and chroma (inten-
sity) of the brown and cap patches were measured using
Munsell chips (Munsell Book of Color, Glossy Finish)
in bright daylight. Hue scores were slightly modified
(Burley et al. 1992), and then all measurements were
entered into the calculation of the colour score using the
following formula
3.0(20
hue)+1.5(5
value)+0.5(chroma
7).
The formula converts the mixed numeral Munsell no-
tation into a standard decimal notation in which colours
perceived by humans as redder, darker and brighter have
higher scores than yellower, lighter and duller colours.
Munsell chips were developed for human perception
and may not accurately reflect avian perception,
especially UV reflectance (Bennett et al. 1994). Although
it is possible that quail perceive near-UV, it is unlikely
that the feathers measured for colour reflect UV since (1)
the feathers of the closely related Gambels quail, C.
gambelii, do not demonstrate strong UV reflectance (J.
Hagelin, personal communication) and (2) the feathers
we measured are not iridescent and are pigmented, red-
brown to yellow-brown. Feathers that are likely to reflect
UV are black, blue and/or are iridescent. Ultimately, our
manipulation altered the colour of the cap dramatically,
and therefore, if preference were conservative for species-
specific cap colours, we would expect females to show an
aversion to the manipulated cap colours.
Experimental Design
Mate choice trials
Dyadic mate choice trials began in May of 1996 and
1997, the time of peak gonadal maturity (Lewin 1963).
During each trial the test female was presented with a pair
of males matched for weight (a stimulus set). Stimulus
males were tethered to prevent physical contact between
them, but males could interact visually and acoustically,
and could be in contact with females, since physical
contact appeared necessary for females to make a mate
choice. Males did not demonstrate aggressive behaviour
towards females. Food and water were provided ad
libitum and all individuals had access to cover (Fig. 2).
At the start of a trial, a single female was released into a
neutral area and was able to travel throughout the cham-
ber. Trials ran for 180 or 185 min, the shortest trial length
necessary for the majority of females to make a mate
72 ANIMAL BEHAVIOUR, 65, 1
(a)
Neutral area
x a potential mate (Calkins et al. 1999). We generated an
x birds using two activities recorded during the exper-
y iments: (1) the number of times a female approached a
In
In residence y residence
Perch 0.3 m
Cover Tether
(b)
Figure 2. Mate choice chamber. (a) A floor plan of the mate choice
chamber showing perches, tethers, cover areas, neutral areas and
areas in which females were in residence with a male. The area in
which a female was considered in residence with a particular male
depended upon that males location in the chamber. Residence fell
behind two lines, a horizontal line 0.3 m above the male (x; towards
the females area) and a vertical line 0.3 m to the side of the male (y;
towards the other males area). These lines are dotted in the figure.
The area in which a female was in residence extended from these
lines to the edges of the chamber. (b) Mate choice chamber from
the observers point of view. The female is in the neutral area.
choice (J. D. Calkins, unpublished data). During each trial
interval, two trials were conducted synchronously, in
which each female was observed for half of the total trial
duration. One female was observed during the first half of
each 3-h trial, and then for 5 min after the last half-hour.
This female was observed for 95 min and had a total of
185 min to interact with males. The other female was
observed during the second half of each 3-h trial. This
female was observed for 90 min and had a total of
180 min to interact with males.
An observer (J.D.C.) recorded all occurrences and dur-
ations of affiliative and aggressive behaviours for both
stimulus and test birds. Affiliative behaviours included
courtship feeding (tidbitting), backrolling, copulation
crouches, and follow invitations (Williams 1969; Stokes &
Williams 1971; Calkins et al. 1999). The intensity of male
courtship behaviour, as represented by tidbitting rate
(TB), was entered into stepwise models of female prefer-
ence for the three experiments (see Regression analysis
below).
Affiliation has been shown to be a strong predictor of
actual mate choice in Old World quail, Coturnix japonica
(White & Galef 1999), and free-ranging California quail
demonstrate affiliation through following or sitting near
affiliation score based upon this behaviour of free-ranging
male (courtship visits) and (2) the total time she spent in
physical proximity with a male (social residence time;
see Fig. 2). These behaviours were included in an affili-
ation score (VT) as follows: for male 1, VT1 =(V1 +T1)/2
with T1 equal to the proportion of the total social time
that the test female spent with male 1 and V1 equal to the
proportion of the total social visits she made to male 1.
Total social time and total social visits were defined as the
total amount of time spent, and the total number of visits
made, to either stimulus male respectively, thus:
VT1 +VT2 =1. The overall trends revealed in an analysis of
this affiliation score (VT) are identical to those that result
if only the proportion of social time is used as the
measure of preference (J. D. Calkins, unpublished data).
Trials were considered successful (revealing a female
mate choice) if the VT score of one stimulus male was
equal to or greater than 55%. This percentage, although
close to 50%, was selected to maximize the range of VT
scores over the analysed data. This percentage also takes
into account that the VT score is a conservative estimate
of female preference since our data include female be-
haviour during both assessment stage and preference
stage (Gibson & Langen 1996).
Additionally, for a trial to be considered successful, the
female must have accumulated at least 150 s of social
time (observed range of times: 1504644 s, XSD=
1415.381196.78; observed range of visits 1110,
XSD=25.03  22.45; Fig. 3). Although a social time of
150 s is a small fraction (3%) of the trial interval, the
two females displaying this duration showed un-
ambiguous preferences: both had visit numbers equal to
or higher than the mean (80 and 25) and both spent more
time with one male (60 and 97% social time), resulting
in VT indices of the favoured males of 78 and 96%,
respectively.
Trials were dropped if a females social time was less
than 150 s (29/95 trials; in such cases females were rerun
at a later date whenever possible); if the VT of the
winning male was less than 55% and/or if the male with
the most social visits was not the male with the most
social time (10/95 trials); if the female showed a side
preference (running up and down a side and not interact-
ing with the male; 2/95 trials); or if there was a technical
problem (loss of artificial crests: 2/95 trials).
The number of trials run per experiment depended
on bird availability. We attempted to test each female
twice with different stimulus sets in the natural variation

1 Table 1. Cap patch colour scores (XSD) for C. callipepla, C. gambelii
0.8
0.6
VT score
0.4
0.2 Dyed males
Normal males
0
0 5 10 15 20 25
Cap colour score
Figure 3. Relationship between cap colour score and mean VT index
for males in experiment 2. As described in the methods, stimulus sets
were comprised of one male from one group and one male from the
other group. Some males were included in two or more stimulus
sets. Males with mean VT scores 0.5 were considered winners in the
analysis.
experiment, but due to time constraints, some females
could only be tested once. Some females were tested twice
in the cap colour experiment as well. When females were
tested twice, resulting VT scores were averaged across the
stimulus sets for paired univariate analysis of female
preference (see below). This procedure eliminates pseudo-
replication in the univariate analysis.
Experiments were designed to use each stimulus set the
same number of times, but some trials were discarded due
to lack of female mate choice. In two of the experiments,
each male was used in only one stimulus set, but in the
cap manipulation experiment some males were used in
more than one stimulus set. This was done in order to
present females with a continuous range of cap scores. For
the stepwise regression analysis of the data, a males status
(winner or loser) was determined based upon his mean VT
over all trials in which he was used (whether he was in
one or two stimulus sets; see below). This procedure elimi-
nates pseudoreplication in the multivariate analysis.
We attempted to use individual stimulus males in only
one experiment but, due to a shortage of birds, two
males were used in both the cap colour and the crest
length manipulation experiments. Stimulus sets were
composed of males from the same geographical location
(Manchester Open Space or Rancho Santa Fe) or covey
(Ecke Ranch) and each test female was exposed to a
set of unfamiliar males from a different geographical
location or covey. (Coveys generally show low intermix-
ing; Emlen 1939; J. D. Calkins, unpublished data.)
To compare the results between the manipulation
experiments, we used the same pool of females in both.
Testing for Female Response to Single Male Traits
Cap colour experiment
Cap colour differs between California quail and its
sister species, the Gambels quail (Table 1). We used an
CALKINS & BURLEY: MATE CHOICE IN CALIFORNIA QUAIL 73
and experimental C. callipepla
Population Cap colour score N
C. callipepla 10.071.50* 17
C. gambelii 15.744.63 11
Experimental
Normal 9.933.39 6
Dyed 18.622.20 6
*Measurements are from individuals trapped in the field.
Measurements are from museum specimens gathered in San Diego
County, California, U.S.A.
experimental manipulation to explore this traits role in
mate choice as well as its possible role in maintaining
species barriers. For this experiment, we dyed the caps of
six males to approximate the cap colour score of Gambels
quail using a colour treatment (Clairol Hydrience coral;
5 min) followed by bleach (Eugene Fair Blonde; 2 min).
Dye was applied using Q-tips and birds were held while
the dye set. The dye was then rinsed off and birds were
hand-held for approximately 15 more minutes to allow
their feathers to dry. They were then released back into
their aviaries. Mate choice trials were initiated 1 week
after manipulation.
Dyed males were paired with unmanipulated males in
stimulus sets matched for weight and crest length (12
total males). Each stimulus set was used in one to four
trials, with a different female for each trial. Each male
participated in one or two stimulus sets (total of eight
sets); each males VT score was averaged over all of the
trials in which he was involved. Males were included in
two stimulus sets in order to achieve a distribution of
stimulus sets with a continuous range of relative cap
scores. Altogether, 16 successful trials were run with 15
females; one female was tested twice and 14 females were
tested once.
Crest length experiment
Gambels quail tend to have longer crests than
California quail (J. D. Calkins, unpublished data; J. Gee,
personal communication). For this experiment, we elon-
gated the crests of six males. (Due to a shortage of
available birds, one male was used in two stimulus sets;
his crest was cut and reglued twice and his VT score was
estimated over all trials in which he participated). Elon-
gation was accomplished by (1) cutting the entire crest
(six feathers) off near the scalp of donor males, (2)
cutting the crests of recipient males approximately
17 mm above the scalp, and (3) attaching the full-length
feathers of the donor males to the cut stubs of the
recipients. The six donor feathers were glued together at
their base to achieve natural conformation prior to
attachment (via Superglue) to the crest of the recipient.
The effect of this manipulation was to produce crest
lengths of 3850 mm in the recipients, which is beyond
the natural range of crest length observed in the Ecke
population (2038 mm, with mean of 33.46 mm; N=
135), but overlap the normal range for Gambels quail.
74 ANIMAL BEHAVIOUR, 65, 1
Donor males were then released from captivity, whereas
recipient males were assigned to stimulus sets and paired
with a male of similar weight that had a crest of normal
length. As a control for the manipulation, the crest
feathers of unmanipulated stimulus males were cut at
approximately 17 mm and reglued to their own crest
stubs, creating crest lengths of 2934 mm.
In total, six stimulus sets were created (11 males total,
one used twice), each consisting of one male with an
elongated crest and one male with a normal crest. We
tested each of 18 females once; each stimulus set was used
one to five times. One stimulus set was used in only one
trial because a males crest broke and could not be
repaired.
Testing for Female Response to Multiple Male
Traits
Natural variation experiment
To explore preference for natural plumage variation, we
tested females with seven stimulus sets, each matched for
mass, in 1996. Each stimulus set was used in three or four
trials. In all, 22 trials were run with 13 females; nine
females were tested twice (with different stimulus sets)
and four females were tested once.
Regression analysis
To analyse female preferences for multiple traits,
including both partitioned and composite traits (see
below), we employed a multivariate approach. In this
procedure, we identified winning males as those males
that had achieved average VT scores greater than 0.5 over
all successful trials in which they were involved. Losing
males were those males with average VT scores below 0.5.
Some winning males, therefore, had VT scores below 0.5
in individual trials, and some losing males had VT scores
above 0.5 in individual trials. When each male was used
only in one stimulus set (most males in the crest length
experiment and all males in the natural variation exper-
iment), that stimulus set contributed one winning and
one losing male. When males were used in more than one
stimulus set (most males in the cap colour experiment,
one male in the crest length experiment), each males VT
score was averaged over all trials in which he participated;
thus, winning and losing males did not represent individ-
ual stimulus sets. We entered the winning males
attributes as independent variables in a series of stepwise
forward regression analyses (F to enter=4.00; F to
remove=3.99) with the winning males VT scores as the
dependent variable.
This procedure differs from that used to measure pref-
erences for single traits as it asks whether winning males
with the most extreme traits (manipulated or not) had
the most extreme VT score, and then measures prefer-
ences for absolute values of male traits. By comparison,
the t and sign tests ask whether females preferred individ-
ual males that were manipulated over those that were not
(i.e. whether or not winners had manipulated pheno-
types), thus testing preferences for relative values of male
traits. In the stepwise regression model we chose to focus
on the trait values of the winning males rather than the
difference in trait values between winning and losing
males in part because females had unlimited opportunity
to remain in a large neutral area. Therefore, a female
choosing to associate with a given male did so because of
something she found attractive in his traits rather than
because of something repulsive in the traits of the losing
male. These two approaches are complementary rather
than redundant.
Partitioned and composite traits
We used two types of trait divisions in the stepwise
regression analyses: partitioned and composite. We use
the term partitioned traits to refer to traits characterized
as independent traits by natural historians. We attempted
to examine the possibility that females do not view these
partitioned traits as independent when making mating
decisions. To do this we combined the values of parti-
tioned traits with similar characteristics into independent
composite traits. We generated two types of composite
traits based upon shared characteristics of partitioned
traits: (1) multipatches: combined scores of multiple par-
titioned traits with similar colours and locations on the
body and (2) total impact patches: combined scores of
the colour and size of a patch.
The two multipatch composite traits were the total
nonblack area and the total black area. The nonblack area
was the sum of the tan, brown and brown cap patches
and was most influenced by the size of the brown patch
(Fig. 1). The areas of the tan and brown patches are
adjacent on the body and were negatively correlated in
the males used in the experiment testing response to
natural variation (Spearman rank correlation: rS =
0.622,
N=14, P=0.03). The cap patch is held at the elevation of
the other two patches during tidbitting. The total black
area was the sum of the crest area and the area of the
black patch and was most influenced by the size of the
black patch. The black patch and the crest area frame
the males eyes and beak. The two total impact composite
traits included the impact of the brown patch and the
impact of the cap patch. These were generated by multi-
plying the colour score and the area of a patch, which
tend not to be correlated (Calkins 2000).
We created composite scores using physical attributes
of the patches rather than through principal component
analysis (PCA) because we hoped to interpret the relation-
ship between patches in a biologically meaningful way.
This is not always possible when using PCA, and our
preliminary exploration of this technique did not yield
components that we could interpret biologically. Further-
more, PCA builds traits based upon their covariance; thus,
the use of PCA would not necessarily allow us to generate
trait groups that females view as composites, whereas
developing rules based upon trait characteristics might
allow us to approach female perception more readily.
Although tidbitting rate (the number of bouts of
tidbitting/s) is a partitioned trait, we included this in both
the analysis of the partitioned traits and the analysis of
the composite traits, since it is the single clear courtship
behaviour demonstrated by California quail (Calkins
et al. 1999). Tidbitting may indicate both willingness and
 CALKINS & BURLEY: MATE CHOICE IN CALIFORNIA QUAIL 75

1 dyed to match caps of Gambels quail (sign test: P=0.22;

N=9 required for Pc0.05; Fig. 3).
0.9
0.8
Crest length experiment
0.7 Males with elongated crests had higher VT scores
0.6 (0.640.26) than did unmanipulated males (0.360.26;
VT score

paired t test: t16 =2.224, P=0.04). Winning males in all

0.5
stimulus sets had elongated crests (sign test: P=0.03;
0.4 Fig. 4).
0.3
0.2 Female Response to Multiple Traits
Control
0.1 Elongated
Naturally varying male traits
0
0 10 20 30 40 50 60 We conducted two stepwise regression analyses with
Crest length (mm) the winning males traits entered as independent vari-
Figure 4. Relationship between crest length (in mm) and mean VT
ables and the winning males VT score as the dependent
index for males included in experiment 3. Stimulus sets were variable. The first stepwise regression included parti-
comprised of one male from one group and one male from the other tioned traits as independent variables (Table 2), and did
group. Some males were included in two or more stimulus sets. not yield a significant model. The second stepwise regres-
Males with mean VT scores 0.5 were considered winners in the sion included the winning males composite traits as the
analysis. independent variables (Table 2), and yielded a significant
model in which VT score was predicted by the multipatch
ability to invest in a mate and young and has been shown trait brown patch+tan patch+cap patch areas (stepwise
to be important in mate choice of other species of quail regression: F1,6 =8.826, N=7 winning males, P=0.03,
(Hagelin 1999; Hagelin & Ligon 2001). R2 =0.57; mean square regression=0.068, error=0.008,
Statistics were calculated using Statview for Macintosh standardized coefficient=0.80).
(Haycock et al. 1992).
Cap colour experiment
In the first stepwise regression analysis using par-
RESULTS
titioned male traits (Table 2), VT score was predicted by
Female Response to Single Manipulated Male negative cap colour score and positive black patch area
Traits (N=6 winning males; Table 3). In the second stepwise
regression analysis using composite male traits (Table 2),
Cap colour score experiment a model was generated in which VT score was predicted
There was no difference between the VT scores of by the total impact of the cap patch and black patch+
dyed males (XSD=0.5450.384) versus unmanipu- crest areas (N=6 winning males, Table 3).
lated males (0.4550.384; paired t test: t14 =0.464, In this experiment there was a positive correlation
P=0.65; N>1000 required for P=0.05). However, winning between cap patch area and cap patch colour for stimulus
males in five of the six stimulus sets were those with caps males (Spearman rank correlation: rS =0.565, N=12 males,

Table 2. Traits entered into stepwise analyses of experimental populations

Partitioned traits
Black patch: black patch just below the eyes and the beak, covers the throat
Tan patch: tan patch, scaled with black, high on the abdomen
Brown patch: beneath the tan patch, scaled with black, low on the abdomen
Crest length: black crest comprised of six feathers
Cap patch: brown patch covering the top of the skull, just behind the crest
Cap patch colour score: see above
Brown patch colour score: see above
Tidbitting rate: rate of bouts of courtship feeding
Composite traits
Multipatch
Brown patch+tan patch+cap patch areas
Black patch+crest areas
Total impact
Brown patchbrown patch colour score
Cap patchcap patch colour score
Tidbitting rate

Although tidbitting rate is not a composite trait, it was included in the regression models employing composite
traits.
76 ANIMAL BEHAVIOUR, 65, 1
Table 3. VT score as a function of winning male traits in the 1997 colour manipulation experiment
Trait Standard coefficient
VT score as a function of partitioned traits
Cap patch colour score
Black patch area
F2,5 =8.895, P=0.05, R2 =0.759, residual=0.002
VT score as a function of composite traits
Total impact of cap patch
Black patch+crest area
F2,5 =17.187, P=0.02, R2 =0.866, residual=0.001
Forward stepwise regression models. Partitioned traits with similar characteristics were increasingly aggregated into
dependent composite traits (see Table 2). Numbers in parentheses indicate the order of entry into the forward
stepwise regression model. N=6 winning males.
P=0.04). This correlation was the inadvertent result of the
dyeing procedure, which tended to increase patch size.
Crest length experiment
None of the winning males traits predicted VT score in
either the first stepwise regression using partitioned traits
or the second regression using composite traits (Table 2).
Trait Correlations and Distribution
Table 4 shows results of correlation analyses for stimu-
lus set males within each year (1996 and 1997). Following
Bonferroni correction, no traits were significantly corre-
lated. The two manipulated traits, crest length and cap
colour score, differed significantly among the three
experiments after Bonferroni correction (Table 5).
DISCUSSION
This study establishes that female California quail use
male plumage characteristics when making mating
decisions. Females chose on the basis of natural trait
variation, but also responded to artificially manipulated
traits. Females in different experiments focused on differ-
ent traits, even when they were from the same test pool.
Thus, females respond to novel values of plumage
traits and populations of females use different traits, or
combinations thereof, in different contexts.
Evidence suggests, then, that female choice has driven
the evolution of male ornamentation in this species. Two
factors that we hypothesize contribute to the evolution
and maintenance of multiple ornaments in this species
are flexible mate choice behaviour and mate choice for
composite traits. This study illustrates that the analysis of
composite traits has a capacity to reveal patterns hidden
from the analysis of partitioned traits.
Novel Trait Values, Species Boundaries and
Flexible Mate Choice
Results of the cap colour manipulation experiment
suggest that cap colour plays a role in mate choice, and
that females do not show an aversion to novel trait states.
The univariate analysis did not result in a significant
Partial r t P
0.658 0.809 2.843 0.07
0.474 0.683 2.048 0.13
0.617 0.838 3.409 0.04
0.516 0.78 2.85 0.07
model although winning males tended to be those with
cap colours resembling Gambels quail. The multivariate
analysis of the winning males traits revealed that, for
winning males, cap colour was the most important trait
in mate choice but that females had stronger preferences
for males possessing lower cap colour scores (more like
California quail). Thus, although higher, more Gambels
quail-like, cap colour scores tended to be important for a
male to be a winner, and among the winning males,
those with more California quail-like cap colour scores
tended to have the highest VT scores. Together these
results suggest that females do not prefer extreme values
of this trait but rather have intermediate or incremental
preferences. Also, since the results of the stepwise regres-
sion model indicate that the area of the black patch
contributed to a females behaviour, the strength of
female preference for a male with a particular cap colour
may be tempered by her response to the size of his black
patch.
Results of the crest length experiment also suggest that
females respond to novel male trait values. The univariate
analysis showed that females preferred males with experi-
mentally elongated crests and all of the winning males
had elongated crests. However, the stepwise regression
procedure on the winning males traits did not result in
any significant model. There was a nonsignificant ten-
dency for males with more modest elongations to have
higher VT scores. Thus, similar to the results of the cap
colour experiment, females did not show the strongest
preferences for the most extreme trait values tested. At
the same time, females did not show aversions to novel
values of the crest length and in fact prefered males with
novel trait values.
We manipulated the cap colour score and the crest
length into the range of the California quails sister
species, the Gambels quail, in order to test the hypoth-
esis that secondary sexual traits function as species
barriers (Dobzhansky 1937; Huxley 1942; Mayr 1963).
Results presented here do not support this function for
the manipulated traits. Females appeared to prefer, rather
than be averse to, the exaggerated values of traits (large
cap colour scores; longer crests) characteristic of Gambels
quail. Thus, the results presented here correspond to
some other studies in which secondary sexual character-
istics do not represent barriers to hybridization but rather
Table 4. Pearson correlation coefficients for male traits of stimulus males
Black patch Tan patch Brown patch Cap patch Brown colour
Male trait Mass Tarsus Crest length area area area area score

Experiment 1
Tarsus (0.003) 0.99
Crest length (0.31) 0.29 (0.29) 0.32
Black patch area (0.001) 0.998 (0.07) 0.82 (0.14) 0.63
Tan patch area (0.11) 0.73 (0.28) 0.33 (0.26) 0.37 (0.004) 0.99
Brown patch area (0.50) 0.07 (0.05) 0.87 (0.30) 0.30 (0.17) 0.58 (0.62) 0.02
Cap patch area (0.16) 0.60 (0.26) 0.37 (0.06) 0.85 (0.49) 0.08 (0.11) 0.72 (0.08) 0.79
Brown colour score (0.11) 0.71 (0.35) 0.23 (0.01) 0.98 (0.17) 0.58 (0.44) 0.12 (0.51) 0.07 (0.17) 0.56
Cap colour score (0.23) 0.46 (0.20) 0.53 (0.28) 0.37 (0.63) 0.020 (0.05) 0.87 (0.04) 0.90 (0.03) 0.92 (0.18) 0.57
Experiments 2 and 3
Tarsus (0.57) 0.002
Crest length (0.14) 0.53 (0.05) 0.82
Black patch area (0.06) 0.76 (0.14) 0.48 (0.06) 0.78
Tan patch area (0.21) 0.29 (0.40) 0.03 (0.14) 0.52 (0.14) 0.48
Brown patch area (0.48) 0.01 (0.32) 0.10 (0.05) 0.82 (0.01) 0.95 (0.01) 0.95
Cap patch area (0.51) 0.006 (0.20) 0.31 (0.20) 0.36 (0.16) 0.41 (0.13) 0.52 (0.01) 0.94
Brown colour score (0.09) 0.68 (0.07) 0.77 (0.46) 0.02 (0.12) 0.57 (0.03) 0.90 (0.02) 0.43 (0.43) 0.03
Cap colour score (0.10) 0.66 (0.07) 0.74 (0.03) 0.86 (0.07) 0.74 (0.26) 0.22 (0.11) 0.60 (0.09) 0.69 (0.36) 0.08

N=14 males included in experiment 1, N =24 males included in experiments 2 and 3. (Coefficient)P value. Corrected significance level Pcrit0.0014.
CALKINS & BURLEY: MATE CHOICE IN CALIFORNIA QUAIL
77
 78
ANIMAL BEHAVIOUR, 65, 1

Table 5. MeanSD of measured male traits included in the stepwise analysis for each experiment
Brown patch Cap patch
Experiment Tidbitting Crest length* Area black patch Area tan patch Area brown patch Area cap colour score colour score*

Natural variation 0.0030.002 33.502.44 1267.64122.57 1424.00269.69 1506.71367.46 310.3352.48 16.091.53 9.911.44
Cap colour 0.0010.001 34.000.99 1284.0171.17 1701.35418.83 1664.69513.07 394.6563.62 15.732.90 14.285.29
Crest length 0.0130.014 36.677.06 1288.50102.45 1498.08278.60 1375.99661.19 384.3883.55 17.163.04 8.624.49

Sample sizes were 14, 12 and 11 males for experiments 1, 2 and 3, respectively.
*Trait values that differed significantly between years according to an ANOVA (P corrected for multiple tests: Pcorrected0.006). Crest length: F2,52 =7.42, P<0.0023. Cap patch colour score:
F2,52 =6.08, P=0.0062.

act to increase the potential for hybridization and gene
flow between species (e.g. Ryan et al. 1990; Christy &
Salmon 1991; Strecker & Kodric-Brown 1999).
Our conclusion that these traits do not function to
maintain species boundaries is tempered by two consid-
erations. The first is that the populations studied here are
not sympatric with Gambels quail and therefore may not
be subject to strong selective pressures against hybridiz-
ation. However, the finding that birds from an isolated
population are not indifferent, but, as in the case with
crest length, actually prefer the trait states of a sister
species, suggests that some factor, such as sampling drift
(Kaneshiro 1980; Nei et al. 1983; Barton 1989; Otte 1989)
might constrain the evolution of long crests in California
quail.
A second consideration is the possibility that quail see
cap colour differently than we do and that the human-
based Munsell scale does not allow us to detect subtle
differences in cap colours that would be apparent to
female quail. This would make a manipulation of cap
colour into the Gambels quail range difficult to achieve,
as we might not be able to detect the most important
characteristic of the colour that differs between the two
species in the eyes of the female. We did manipulate the
colour component (i.e. hue) that differs most between
Gambels and California quail in human perception.
Thus it seems likely that we did manipulate the colour in
one dimension relevant to quail, although there may be
other important dimensions as well. Our results indicate
that the manipulation did influence female quail percep-
tion of males. We conclude from the lack of any aversion
to novel cap colour scores that this trait is not used as a
basis of species-specific choice by female quail.
The two manipulations were conducted in a single year
using the same pool of females. For each experiment,
females focused primarily upon the manipulated trait,
thus changing their focus of attraction between manipu-
lations. We observed a similar trend when data for only
the nine females used in both experiments were analysed
(J. D. Calkins, unpublished data). This change in the
focus of female choice reveals a potential for flexible mate
choice in this species at the population level. This is
suggestive of flexible mate choice at the individual level
as well. Further study of individual preference functions
in this species would be useful for elucidating the degree
of preference flexibility at the individual level, as well as
the factors that drive flexibility (Basolo 1998; Wagner
1998).
One possible factor that might influence flexible female
preference is a changing distribution of male trait values.
The distribution of several male traits in California quail
in the field varies among years (Calkins 2000). Female
quail could either respond to changes in the ecological
and demographic environment by changing the basis of
their mating decisions and/or they could use the traits
themselves as a source of information about a changing
environment (e.g. Reid & Weatherhead 1990; Jennions &
Petrie 1997; Qvarnstro m et al. 2000). They could also
choose males based on more highly variable traits that
might increase the efficiency of mate choice by making
discrimination easier (Jennions & Petrie 1997). In this
CALKINS & BURLEY: MATE CHOICE IN CALIFORNIA QUAIL 79
study, the only two traits that differed significantly
among experiments were the manipulated traits (cap
colour score and crest length); both had the highest
means and standard deviations in the experiment in
which they were the focus of mate choice. This supports
the possibility that females are choosing the most
variable and easily discriminated traits.
Composite Trait Preferences and Mate Choice for
Multiple Ornamentation
For two of the experiments, the cap colour manipu-
lation and the natural variation experiment, the use of
composite traits generally resulted in regression models
that included larger groups of traits and explained a
greater proportion of the variance in female preference
behaviour than did the use of partitioned models. In the
experiment in which we examined female response to
naturally varying male traits, composite traits yielded
successful models even when no partitioned trait was
found to contribute to a significant regression model.
Composites that entered the best regression models
included both those based on overall size of several
patches with the same ground colour, multipatch com-
posites (combined areas of brown, cap and tan patches;
and total black patches), and composites reflecting over-
all patch impact, the total impact composites (cap patch
area and cap patch colour score). No traits correlated
significantly with any other trait (after Bonferroni correc-
tion), and thus correlations among traits do not explain
the power of composite traits to explain mating behav-
iour. The general lack of correlation is similar to that
observed in field populations of California quail (Calkins
2000).
The results presented above, therefore, suggest that
composites may be the focus of female choice and indi-
cate future directions of research. Our conclusions regard-
ing choice for partitioned versus composite traits are
tempered by the limits of the methodology and small
sample sizes. In particular, clearer data on perception and
decision-making processes of female quail are necessary
to differentiate between choice via partitioned versus
composite traits. Similarly, studies with larger sample
sizes would be useful.
The results also point to the usefulness of creating
composite traits as a statistical tool when sample sizes are
too small to reveal preferences for partitioned traits. This
is because mate choice response probably varies among
females; for example, some females may weigh one com-
ponent of the composite trait more heavily than the
others, while other females may weigh other components
more heavily. In this way, several partitioned traits would
be under pressure from mate choice because they com-
bine to yield the composite trait impression important to
females as a group.
Multiple Ornamentation and Ghosts of Selection
Past
These results differ from those found for other galli-
formes, including other species of Callipepla, where male
80 ANIMAL BEHAVIOUR, 65, 1
behaviour was found to be more important in determin-
ing female behaviour than male plumage traits (Ligon &
Zwartjes 1995; Hagelin & Ligon 2001). There are three
possible reasons for the difference in experimental results.
The first is that the difference in preference patterns is
simply a species difference. It is possible that the impor-
tance of plumage characteristics in mate choice differs for
different species or even for different populations of the
same species (e.g. guppies, Poecilla reticulata, Houde 1997
and references therein). A second explanation is the
increased statistical power from our use of composite
traits. This methodology might have allowed us to
observe patterns in a system where mate choice is rela-
tively weak for any one partitioned trait alone. A final
possibility is that our use of adult females that had been
exposed to a rich social environment during develop-
ment allowed us to detect preferences arising from early
exposure to adults and other individuals (Calkins et al.
1999).
This contrast between results reported here and those
of other studies of galliforms has implications for one
hypothesis that attempts to explain the evolution of
multiple ornaments: the ghosts of selection past hypoth-
esis (e.g. Holland & Rice 1998). This hypothesis would
suggest that most secondary sexual plumage traits are no
longer subject to female mate choice in extant species,
but persist because the traits were fixed via mate choices
expressed in the past. The data presented here imply that
rather than most traits not being subject to female mate
choice in this species, most traits are actually, at one time
or another, under pressure from female mate choice
behaviour.
These data also suggest that inconsistencies of mate
preferences results, which are sometimes used as evidence
that observed preferences are too weak to be important in
the evolution of traits (e.g. Ligon & Zwartjes 1995), may
represent flexible responses. Whereas inconsistent or flex-
ible female preferences within a species have occasionally
been assumed to lack selective power to affect the evol-
ution of male traits (e.g. Ligon & Zwartjes 1995), theor-
etical and conceptual studies do suggest that active
selection on male traits can occur through flexible mate
choice (e.g. Jennions & Petrie 1997; Brooks & Endler
2001). Clearly, this is an area that needs further study.
Future Directions
These results indicate that future study of mate choice
should focus on questions of more complex mate choice
behaviour using larger sample sizes and novel analysis
techniques. These investigations should focus on a study
of flexible preferences across a range of species, both
between and within females, as well as analyses of female
preference for both composite and partitioned traits,
especially in the context of mate choice flexibility. Along
with these empirical investigations, a theoretical exami-
nation of the benefit of flexible mate choice in the light
of both partitioned and composite traits would be useful.
Theoreticians should focus on the following issue: what
qualities of secondary sexual traits, and trait composites
(e.g. perceptual or genetic), make them most useful for
mate choice in different contexts? Finally, to determine
the full impact of the evolution of female choice on the
distribution of male traits, we need to understand how
environmental (e.g. physical, demographic) change
affects both male and female reproductive success via
active mate choice on trait distributions.
Acknowledgments
This research was funded by a fellowship from the
Pointsettia Foundation to J.D.C. We would like to thank
Drs D. Campbell, D. Enstrom, J. Gee, A. Hedrick, K.
Johnson, R. Levin, K. McGraw, L. Oring and four anony-
mous referees for comments on the manuscript. K. Beale,
J. Chow, L. Potyondy, N. Sayano and D. Vien helped with
field research. J. Chow, A. Coburn and D. Vien helped
with maintenence of captive birds. Dr J. Swanson aided in
the statistical analysis. We would like to thank the Ecke
family for permission to conduct field research and mate
choice experiments on their property. J.D.C. benefited
from discussions with J. Gee and Dr J. Hagelin. C. Calkins,
D. Calkins and Dr W. J. Swanson helped in all aspects of
the research. Research was covered by the California
Department of Fish and Game permit No. 6942 and
conducted under supervision of the Institutional Animal
Care and Use Committee of the University of California,
Irvine (Permit No. 951664).
References
Andersson, M. 1994. Sexual Selection. Princeton, New Jersey:
Princeton University Press.
Backwell, P. R. Y. & Passmore, N. I. 1996. Time constraints and
multiple choice criteria in the sampling behaviour and mate
choice of the fiddler crab, Uca annulipes. Behavioral Ecology and
Sociobiology, 38, 407416.
Badyaev, A. V., Hill, G. E., Dunn, P. O. & Glen, J. C. 2001. Plumage
color as a composite trait: developmental and functional integra-
tion of sexual ornamentation. American Naturalist, 158, 221
235.
Barton, N. H. 1989. Founder effect speciation. In: Speciation and Its
Consequences (Ed. by D. Otte & J. A. Endler), pp. 229256.
Sunderland, Massachusetts: Sinauer.
Basolo, A. L. 1998. Evolutionary change in a receiver bias: a
comparison of female preference functions. Proceedings of the
Royal Society of London, Series B, 265, 22232228.
Bennett, A. D., Cuthill, I. C. & Norris, K. J. 1994. Sexual selection
and the mismeasure of color. American Naturalist, 144, 848860.
Bowler, P. J. 1996. Lifes Splendid Drama. Chicago: University of
Chicago Press.
Bradbury, J. W. & Vehrencamp, S. L.. 1998. Principles of Animal
Communication. Sunderland, Massachusetts: Sinauer.
Brooks, R. & Caithness, N. 1995. Manipulating a seemingly non-
preferred male ornament reveals a role in female choice. Proceed-
ings of the Royal Society of London, Series B, 261, 710.
Brooks, R. & Endler, J. A. 2001. Female guppies agree to differ:
phenotypic and genetic variation in mate-choice behavior and the
consequences for sexual selection. Evolution, 55, 16441655.
Brown, D. E. & Gutierrez, R. J. 1980. Sex ratios, sexual selection
and sexual dimorphism in quails. Journal of Wildlife Management,
44, 198202.
Brown, D. E., Hagelin, J. C., Taylor, M. & Galloway, J. 1998.
Gambels quail (Callipepla gambelii). In: The Birds of North America.

No. 321 (Ed. by A. Poole & F. Gill). Philadelphia: Birds of North
America.
Burley, N. 1981. Mate choice by multiple criteria in a monogamous
species. American Naturalist, 117, 515528.
Burley, N. T., Price, D. K. & Zann., R. A. 1992. Bill color, reproduc-
tion and condition effects in wild and domesticated zebra finches.
Auk, 109, 1323.
Calkins, J. D. 2000. Dynamic mate choice and mating behavior in
California quail. Ph.D. thesis, University of California, Irvine.
Calkins, J. D., Hagelin, J. C. & Lott, D. F. 1999. California quail
(Callipepla californica). In: The Birds of North America. No. 473 (Ed.
by A. Poole & F. Gill). Philadelphia: Birds of North America.
Christy, J. H. & Salmon, M. 1991. Comparative studies of repro-
ductive behaviors in mantis shrimp and fiddler crabs. American
Zoologist, 31, 329337.
Darwin, C. 1871. The Descent of Man, and Selection in Relation to
Sex. London: J. Murray.
Dobzhansky, T. 1937. Genetics and the Origin of Species. New York:
Columbia University Press.
Emlen, J. T., Jr. 1939. Seasonal movements of a low-density valley-
quail population. Journal of Wildlife Management, 3, 118130.
Gibson, R. M. & Bradbury, J. W. 1985. Sexual selection in lekking
sage grouse: phenotypic correlates of male mating success. Behav-
ioral Ecology and Sociobiology, 18, 117123.
Gibson, R. M. & Langen, T. A. 1996. How do animals choose their
mates? Trends in Ecology and Evolution, 11, 468470.
Hagelin, J. C. 1999. Sexual selection, plumage ornamentation of
Gambels and scaled quail. Ph.D. thesis, University of New
Mexico.
Hagelin, J. C. & Ligon, J. D. 2001. Female quail prefer testosterone-
mediated traits, rather than the ornate plumage of males. Animal
Behaviour, 61, 465476.
Haycock, K., Roth, J., Gagnon, J., Finzer, W. F., Soper, C. &
Spector, P. 1992. Statview. Berkeley, California: Abacus Concepts.
Hill, J. A., Enstrom, D. A., Ketterson, E. D., Nolan, V., Jr &
Zigenfus, C. 1999. Mate choice based on static versus dynamic
secondary sexual traits in the dark-eyed junco. Behavioral Ecology,
10, 9196.
Holland, B. & Rice, W. R. 1998. Perspective: chase-away sexual
selection: antagonistic seduction versus resistance. Evolution, 521,
17.
Houde, A. E. 1997. Sex, Color and Mate Choice in Guppies. Princeton,
New Jersey: Princeton University Press.
Huxley, J. S. 1942. Evolution, the Modern Synthesis. London: Allen &
Unwin.
Iwasa, Y. & Pomiankowksi, A. 1994. The evolution of mate
preferences for multiple sexual ornaments. Evolution, 45, 1431
1442.
Janetos, A. C. 1980. Strategies of female mate choice: a theoretical
analysis. Behavioral Ecology and Sociobiology, 7, 107112.
Jennions, M. D. & Petrie, M. 1997. Variation in mate choice
and mating preferences: a review of causes and consequences.
Cambridge Biological Review, 72, 283327.
Johnsgard, P. A. 1973. Grouse and quails of North America. Lincoln:
University of Nebraska Press.
Johnson, K. 1988. Sexual selection in pinyon jays. I. Female choice
and malemale competition. Animal Behaviour, 36, 10381047.
Johnstone, R. A. 1995. Honest advertisement of multiple qualities
using multiple signals. Journal of Theoretical Biology, 177, 8794.
Kaneshiro, K. Y. 1980. Sexual isolation, speciation and the direction
of evolution. Evolution, 34, 437444.
Leopold, A. S. 1977. The California Quail. Berkeley: University of
California Press.
Lewin, V. 1963. Reproduction and development of young in a
population of California quail. Condor, 65, 249275.
CALKINS & BURLEY: MATE CHOICE IN CALIFORNIA QUAIL 81
Ligon, J. D. & Zwartjes, P. W. 1995. Ornate plumage of male red
junglefowl does not influence mate choice by females. Animal
Behaviour, 49, 117125.
Marchetti, K. 1998. The evolution of multiple male traits in the
yellow-browed leafwarbler. Animal Behaviour, 55, 361376.
Mayr, E. 1963. Animal Species and Evolution. Cambridge,
Massachusetts: Harvard University Press.
Mller, A. P. & Pomiankowski, A. 1993. Why have birds got
multiple sexual ornaments? Behavioral Ecology and Sociobiology,
32, 167176.
Mller, A. P., Saino, N., Taramino, G., Galeotti, P. & Ferrario, S.
1998. Paternity and multiple signaling: effects of a secondary
sexual character and song on paternity in the barn swallow.
American Naturalist, 151, 236242.
Moore, A. J. 1990. The inheritance of social dominance, mating
behaviour and attractiveness to mates in male Nauphoeta cinerea.
Animal Behaviour, 39, 388397.
Nei, M., Maruyama, T. & Wu, C-I. 1983. Models of evolution of
reproductive isolation. Genetics, 103, 557579.
Omland, K. E. 1996a. Female mallard mating preferences for
multiple male ornaments 1. Natural variation. Behavioral Ecology
and Sociobiology, 39, 353360.
Omland, K. E. 1996b. Female mallard mating preferences for
multiple male ornaments 2. Experimental variation. Behavioral
Ecology and Sociobiology, 39, 361366.
Otte, D. 1989. Speciation in Hawaiian crickets. In: Speciation and Its
Consequences (Ed. by D. Otte & J. A. Endler), pp. 482526.
Sunderland, Massachusetts: Sinauer.
Petrie, M., Halliday, T. & Sanders, C. 1991. Peahens prefer pea-
cocks with elaborate trains. Animal Behaviour, 41, 323331.
Qvarnstro m, A., Part, T. & Sheldon, B. C. 2000. Adaptive plasticity
in mate preference linked to differences in reproductive effort.
Nature, 405, 344347.
Real, L. A. 1990. Behavioral Mechanisms in Evolutionary Ecology.
Chicago: University of Chicago Press.
Reid, M. L. & Weatherhead, P. J. 1990. Mate-choice criteria of
Ipswitch sparrows, the importance of variability. Animal Behaviour,
40, 538544.
Ryan, M. J., Fox, J. H., Wilczynski, W. & Rand, A. S. 1990. Sexual
selection for sensory exploitation in the frog Physalaemus pustulo-
sus. Nature, 343, 6667.
Stokes, A. W. & Williams, H. W. 1971. Courtship feeding in
gallinaceous birds. Auk, 88, 543599.
Strecker, U. & Kodric-Brown, A. 1999. Mate recognition systems in
a species flock of Mexican pupfish. Journal of Evolutionary Biology,
12, 927935.
Wagner, W. E., Jr. 1998. Measuring female mating preferences.
Animal Behaviour, 55, 10291042.
White, D. J. & Galef, B. G., Jr. 1999. Affiliative preferences are stable
and predict mate choices in both sexes of Japanese quail, Coturnix
japonica. Animal Behaviour, 58, 865871.
Williams, H. W. 1969. Vocal behavior of adult California quail. Auk,
86, 631659.
Wittenberger, K. F. 1983. Tactics of mate choice. In: Mate Choice
(Ed by P. Bateson) pp. 435447. Cambridge: Cambridge
University Press.
Zuk, M., Thornhill, R., Ligon, J. D. & Johnson, K. 1990. Parasites
and mate choice in red jungle fowl. American Zoologist, 30,
235244.
Zuk, M., Ligon, J. D. & Thornhill, R. 1992. Effects of experimental
manipulation of male secondary sex characters on female mate
preference in red jungle fowl. Animal Behaviour, 44, 9991006.