Content uploaded by Nancy T. Burley
Author content
All content in this area was uploaded by Nancy T. Burley on Jul 30, 2014
Content may be subject to copyright.
“A Taste for the Beautiful”: Latent Aesthetic Mate Preferences for White Crests in Two
Species of Australian Grassfinches
Author(s): NancyTyler Burley and Richard. Symanski
Reviewed work(s):
Source:
The American Naturalist,
Vol. 152, No. 6 (December 1998), pp. 792-802
Published by: The University of Chicago Press for The American Society of Naturalists
Stable URL: http://www.jstor.org/stable/10.1086/286209 .
Accessed: 11/01/2012 11:48
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
The University of Chicago Press and The American Society of Naturalists are collaborating with JSTOR to
digitize, preserve and extend access to The American Naturalist.
http://www.jstor.org
vol. 152, no. 6 the american n aturalist december 1998
‘‘A Taste for the Beautiful’’: Latent Aesthetic Mate Preferences
for White Crests in Two Species of Australian Grassfinches
Nancy Tyler Burley* and Richard Symanski†
Department of Ecology and Evolutionary Biology, University of
tion of species-specific and secondary sexual traits by
California, Irvine, California 92697-2525
constraining the evolution of unfavored (unattractive)
traits and favoring nonfunctional attractive traits (Burley
Submitted October 9, 1997; Accepted June 17, 1998
1985). These ideas are in keeping with Darwin’s conclu-
sion that many secondary sexual traits are fundamentally
not understandable through viability selection. Instead,
Darwin (1874) ascribed the origin of such traits to unex-
abstract: Darwin first hypothesized that bright colors and elabo-
plained ‘‘aesthetic’’ mate preferences possessed by female
rate ornamentation of male animals evolved in response to the
conspecifics.
‘‘aesthetic’’ mate preferences of females. By this reasoning, poten-
tially costly male secondary sexual traits may evolve not in re-
sponse to selection for demonstration of vigor but, rather, in re-
Historical Perspective
sponse to latent, nonfunctional preferences by females. Recent
comparative evidence for this phenomenon is equivocal. Here we
From Darwin’s time to the present, an interesting intel-
present experimental evidence that two avian species from a lin-
lectual circle has been completed: Fisher (1930) argued
eage devoid of crested species have mate preferences for opposite-
that aesthetic mate preferences evolved from functional
sex conspecifics wearing artificial white crests. Other colors of
ones (see Andersson 1994 for recent variants on this
crests that have been studied are not preferred. Preferences for
white crests did not diminish over the longest experimental inter-
idea); proponents of ‘‘good genes’’ models of sexual se-
val (12 wk). These results are additional powerful evidence for
lection have hypothesized that differential expression of
highly structured aesthetic mate preferences in estrildine finches.
secondary sexual traits typically functions to indicate ge-
Sex differences in the expression of preferences, and the wide-
netic or physiological quality of the bearers of the traits
spread occurrence of facial ornamentation in birds, suggest that
(e.g., Zahavi 1975; Hamilton and Zuk 1982; Kodric-
the preference ‘‘structure’’ is influenced by the central nervous sys-
Brown and Brown 1984), a view that discounts the role
tem. We hypothesize that aesthetic preferences are a potent force
of aesthetic preferences. Most recently, authors have sug-
in the early evolution of sexually selected traits, and that ‘‘indica-
tor’’ traits evolve secondarily from traits initially favored by aes-
gested nonfunctional hypotheses that mirror Darwin’s.
thetic preferences.
These include hypotheses invoking ‘‘sensory bias’’ and
‘‘sensory exploitation.’’
Keywords: sexual selection, aesthetic mate preferences, estrildine
‘‘Sensory bias’’ implies an underlying neurophysiologi-
finches.
cal bias that has an ‘‘accidental’’ influence on mate pref-
erence. Burley (1985, p. 31) hypothesized that sensory
bias was ubiquitous: ‘‘The sensory system of any speciesOver the past decade, several authors have suggested that
organisms possess latent mate preferences for traits that will . . . be preadapted to perceive some not-yet-evolved
stimulus in a particular way.’’ Over the last decade therehave yet to evolve (Burley 1985, 1986b, 1986c; Rowland
1989; Basolo 1990; Ryan and Keddy-Hector 1992). These has been a proliferation of approaches to the concept of
sensory bias, including environmental influences on sen-preferences, which are thought to be caused by properties
of neurophysiological/sensory systems (e.g., Endler 1993; sory structure (e.g., Endler 1992; Marchetti 1993) and
psychological and psychophysical considerations (e.g.,Enquist and Arak 1993), provide structure to the evolu-
Guilford and Dawkins 1991; Arak and Enquist 1993).
‘‘Sensory exploitation’’ implies capitalization by one group
* To whom correspondence should be addressed; E-mail: ntburley@uci.edu.
on the sensory biases of another. Thus, males evolve
† E-mail: rsymansk@uci.edu.
secondary sexual traits that ‘‘exploit’’ the mate prefer-
Am. Nat. 1998. Vol. 152, pp. 792– 802. 1998 by The University of Chicago.
0003-0147/98/5206-0002$03.00. All rights reserved.
ences of females (e.g., Ryan and Keddy-Hector 1992).
Mate Preferences for Artificial Crests 793
Aesthetic versus Functional Preferences guity of evidence for other cases (e.g., Ryan et al. 1990;
Shaw 1995) points out the chicken-versus-egg dilemma
Models in which latent mate preferences produced inherent in this approach: because phylogenetic analysis
through various neurophysiological phenomena have the is subject to revision, and the origin and antiquity of ex-
potential to drive the evolution of sexually selected traits tant traits cannot be determined with precision, compari-
can be parsimoniously classed together as ‘‘aesthetic’’ sons involving closely related species may not provide
(Burley 1985, 1986c) in deference to Darwin’s articula- unambiguous evidence for preexisting preferences.
tion of the idea that a vast array of secondary sexual traits An alternative approach is to demonstrate the occur-
evolved in response to a ‘‘taste for the beautiful’’ (Darwin rence of mate preferences for truly novel traits—those
1874, p. 469), widespread among vertebrates and re- with no expression within a taxon. This approach has
sulting from the organization of the ‘‘nerve cells of the been taken by Burley and colleagues in a series of experi-
brain’’ (Darwin 1874, p. 705). Aesthetic preferences, ments on zebra finches (Taeniopygia guttata), a socially
then, are emergent properties of the central nervous sys- monogamous, sexually dichromatic species. In zebra
tem and sensory systems that originated incidentally, not finches, leg band color affects mating attractiveness of
through active selection on mate preferences. They can both sexes to conspecifics (Burley 1985, 1988b). These
be contrasted with ‘‘functional’’ mate preferences that preferences are not predictable simply on the basis of ex-
have evolved through direct selection on mate prefer- isting species-specific and secondary sexual traits (Burley
ences that enhance offspring numbers and/or viability. 1985). Mate preferences for novel traits are sufficiently
While functional and aesthetic preferences are clear strong to affect social (Burley 1986c) and genetic (Burley
conceptual alternatives, we view them as complementary et al. 1996) parentage. Moreover, individuals whose at-
evolutionary agents. Aesthetic preferences may account tractiveness has been altered by color banding spontane-
for the early evolution of traits, while functional prefer- ously develop mating tactics consistent with their attrac-
ences may shape their subsequent amplification. Func- tiveness (Burley 1986b, 1988a; Burley et al. 1994, 1996).
tional models, notably ‘‘good genes’’ models, typically fail Thus, ‘‘mutant’’ attractive traits could readily increase in
to account for the origin of sexually selected traits. That frequency in response to aesthetic mate preferences. Pref-
is, for example, a random mutation conferring a red spot erences for novel leg band colors have also been found in
on the crown of a bird does not constitute an indicator other estrildines (e.g., Burley 1986a) and in nonestrildine
trait; indicator status has to evolve subsequently. How- finches (Johnson et al. 1993).
ever, neither the red spot nor its indicator status can eas-
ily evolve in the absence of mate preferences or some
other active agent. By contrast, aesthetic mate preferences Crests and Grassfinches
can favor the establishment of novel traits arising Here we report results of experiments performed to de-
through mutation, particularly if the reservoir of latent termine whether estrildine finches have latent mate pref-
aesthetic preferences in a population is large. Thus, docu- erences for novel ornaments in any other body region.
mentation of latent aesthetic preferences will greatly en- We focused on the head because crests, or ‘‘elongated
hance our understanding of sexual selection and specia- feathers originating from the avian forehead, crown, and/
tion processes. or nape,’’ have evolved repeatedly among avian lineages
(Armstrong 1947; Morris 1956; Gill 1990). Crests occur
in at least 20 of the 30 orders of extant birds, and in all
Measuring Aesthetic Preferences major groups of passerines (N. T. Burley, unpublished
data).Contemporary evidence for latent aesthetic preferences
comes from two research approaches. One line of evi- Grassfinches, however, lack crests; indeed, there are no
crested species in this group (Estrildidae [Goodwin 1982]dence involves comparison of preferences of related spe-
cies with differing secondary sexual traits. Basolo (1990), or Estrildinae [Sibley and Ahlquist 1990]), which con-
tains about 120 extant species (Goodwin 1982). The clos-for example, studied poeciliid fishes of the genus Xipho-
phorus, in which an elongated caudal fin or ‘‘swordtail’’ est relatives are the viduines (fewer than 15 species, all
uncrested) and the ploceine weavers (about 110 species).is a common secondary sexual trait. She found that fe-
males of a species lacking swordtails (X. maculatus) pre- Among the weavers, crests occur rarely (Malimbus mali-
mbicus has a well-defined crest; males of several Euplectesferred males with artificial tails and concluded that her
results were evidence for ‘‘sensory bias.’’ Meyer et al. species have elongated nape feathers [Serle et al. 1977]).
Based on DNA-DNA hybridization evidence, the estril-(1994) criticized this interpretation on the basis of a mo-
lecular phylogeny that suggested the swordtail was pres- dine and ploceine lineages diverged 20–25 million years
ago (Sibley and Ahlquist 1990). Thus, if preferences forent in the common ancestor of the genus. Similar ambi-
794 The American Naturalist
Figure 1: Distribution of zebra finches and long-tailed finches in Australia (after Pizzey 1980). Long-tailed finches employed in
experiments 2 and 3 were first-generation, captive-reared offspring of birds captured at two sites in the Northern Territory: Newry,
a cattle ranch, and Larrimah, a small town.
crests occur in extant grassfinches, they did not coevolve (fig. 1) and exported them under permit to the United
States. Birds from Larrimah have bright red beaks, smallwith crests; rather, the preferences are either ‘‘preex-
isting’’ or, if they evolved in some remote, crested ances- bibs and are sexually indistinguishable (Burley 1981a,
1986a). Birds from Newry have yellow-orange beaks,tor, they have somehow survived vast spans of evolution-
ary time without expression. larger bibs, and are mildly sexually dimorphic in plum-
age traits. These differences persist among first and sec-We tested the preferences for crested, opposite-sex
conspecifics of two grassfinch species: zebra finches (Tae- ond laboratory-hatched generations; thus, major popula-
tion differences are not the result of environmentalniopygia guttata castanotis) and long-tailed finches (Poe-
phila acuticauda). These species have typically been con- variables such as diet. In captivity, birds will not readily
form pair-bonds with individuals from the opposite pop-sidered congeners or members of sister genera (Goodwin
1982), but more recent analysis (Christitis 1987) suggests ulation (N. T. Burley, unpublished data).
Three experiments measured heterosexual social pref-they may be somewhat less closely related. Zebra finches
range widely over the continent of Australia (fig. 1), but erences using a design previously established to measure
mate choice (Burley 1986a; Burley et al. 1994, 1996).no regional variation in plumage patterns or soft-part
coloration has been found (Keast 1958; N. T. Burley, un- Birds were given choices of four opposite-sex social part-
ners having randomly assigned experimental phenotypespublished data). Long-tailed finches live in a longitudinal
belt across the Northern Territory and Western Australia. (e.g., crests of various colors). The major goals of the ex-
periments were to determine whether members of an un-The zebra finch occurs, albeit uncommonly, through-
out much of the breeding range of the long-tailed finch crested lineage have preferences for or aversions to crests,
and whether there are species and/or population differ-(fig. 1).
Long-tailed finches are often characterized as having ences in preference. An additional goal of the third ex-
periment was to assess individual variation in preferencetwo distinct subspecies, hecki and acuticauda. Birds of the
acuticauda type have yellow-orange beak, whereas those for crests.
of the hecki type have bright red beaks. In actuality, beak
color shows continuous variation in the Northern Terri- Experiment 1: Zebra Finches
tory. Moreover, populations trapped at intervals 100 km
apart show other distinctive characteristics (N. T. Burley Domesticated, wild-type zebra finches maintained in N. T.
Burley’s laboratory (effective population size greaterand R. Symanski, unpublished data). In 1992, we cap-
tured birds at two locations in the Northern Territory than 100 individuals) were presented with opposite-
Mate Preferences for Artificial Crests 795
sex stimulus sets composed of four individuals. Stimulus
sets were matched for natural phenotypic variation (beak
color, weight, and activity level). Three birds within a set
were assigned crests using a stratified random design; the
fourth bird was not crested.
Crest colors were selected on the basis of color-band
preferences previously measured (Burley 1985) and re-
flect the range of responses to color bands shown by ze-
bra finches. Each male stimulus set contained a red-
crested individual (red leg bands are attractive to fe-
males), a white-crested one (white leg bands are neutral),
and a green-crested bird (green is unattractive). The crest
colors used for female zebra finch stimulus sets were sim-
ilarly selected: black is a leg band color preferred by
males, white is neutral, and light blue is unattractive.
Methods
The general methodology for measuring preferences has
been previously detailed (Burley 1986aand references
therein).
Crests were made by weaving together tiny (‘‘seed’’)
beads into a platform in which a feather, either a rectrix
(from an estrildine or small psittacine) or contour
feather (from a psittacine), was glued in an upright posi-
Figure 2: Artificial white crest worn by male long-tailed finch
tion. Platform color was closely matched to that of the
(left) and male zebra finch (right) during mate choice trials.
feather, except for ‘‘black’’ crests, in which the bead plat-
Photo credit: Kerry Clayman.
form was black, but the feather was dark gray. (Black
feathers were not available.) Within stimulus sets, crests logged 2,000 s or more of social time with stimulus
were closely matched for feather size and shape; trim- type(s) and no technical failures (e.g., burned-out
ming of feathers was sometimes employed to achieve lightbulb) occurred. Test birds that failed to reach the
uniformity. Feathers with uneven coloration were ‘‘tou- 2,000-s criterion were retested up to three times (if nec-
ched up’’ with artist-quality water-based markers. When essary) with the same stimulus set (in no more than one
this was done, marker was lightly and uniformly applied trial per day); likewise, trials with technical failures were
to the entire feather. The procedure was employed for all repeated on another day.
of the red feathers used, and about half of the gray feath- Five stimulus sets were employed in the measurement
ers, but not the blue or green ones. Crests weighed about of female choice, six for male choice, and no individual
0.3 g, or approximately 2.4% of bird weight. was used in more than one stimulus set. Each test bird
Crests were affixed with water-based glue to the crown was tested with two stimulus sets, and its response (per-
feathers of stimulus birds (fig. 2). Most stimulus birds cent of social time spent with each stimulus type) was av-
‘‘accepted’’ (stopped trying to dislodge) their crest within eraged over the two trials before analysis. We attempted
30 min of application; those that persisted in attempting to use each stimulus set the same number of times, but
to remove the crest were removed from the experiment. it was often not possible to do so for reasons such as the
We repaired crests when feathers became bent, broken, necessity to ‘‘retire’’ some stimulus sets early (see above)
or frayed. When a crest was damaged beyond repair or if or the availability of an odd number of test birds.
a stimulus bird lost its crown feathers when its crest fell Birds were maintained on a 14L :10D photoperiod.
off, we discontinued use of the stimulus set. Lighting was provided by four 100-W incandescent
Never-mated adults in excellent phenotypic condition bulbs.
were used as test birds. After test birds were acclimated
to the apparatus (Burley 1986b), test birds received 48 h Results
of exposure to the test phenotypes (in the experimental
apparatus) before testing. Then they were tested in 2-h Female zebra finches (N⫽15) preferred white-crested
males over other stimulus types (fig. 3A; Friedman test,trials. Trials were considered successful when test birds
796 The American Naturalist
Figure 3: Mate preferences of zebra finches for nonornamented (‘‘No crest’’) versus artificially crested conspecifics. Bars represent
the mean proportion (⫹SE) of time spent in social contact with each stimulus type; dots represent mean within-trial rank of social
preferences (1–4, with 4 being most preferred). A, Female preference for males. B, Male preference for females.
χ
2
⫽14.84, P⫽.002) and did not discriminate between caught birds. The effective population size in each popu-
lation was about 50 birds.uncrested males and those wearing other colors of crests.
White-crested males averaged the greatest percentage of Crest colors were selected (1) to determine whether
preferences for white crests existed and (2) to maximizesocial time in four of the five stimulus sets (range of
means: 47.2%–73.2% of all social time); in the fifth set, the chance of finding a preferred crest color other than
white. Red was selected because birds tested from thisthe uncrested male received the most social time
(42.1%), with the white-crested male in second place population are attracted to red-beaked and red-banded
conspecifics (N. T. Burley, unpublished data). Light blue(37.5%). Three sets contained males with rounded crests
(including the one set with the second-place finish for was selected because it accentuates the distinctive blue-
gray cap of the long-tailed finch.white crests), and two sets contained males with vertical
crests. Each stimulus set was used in four to eight suc-
cessful trials (X⫽6.0). Methods
Male zebra finches (N⫽21) preferred uncrested fe-
males to crested ones (Friedman test, χ
2
⫽9.17, P⫽With the following exceptions, methods used for testing
were identical to those of experiment 1. In this experi-.027) and did not discriminate among crest phenotypes
(fig. 3B). Each stimulus set was used in six to nine suc- ment, only white finch rectrices were used: red and blue
feathers were created using water-soluble marking pens.cessful trials (X⫽7.0).
Also, the crests weighed less (about 0.15 g, or 1.4% of
bird weight). Trials were of 1-h duration, and the mini-
Experiment 2: ‘‘Larrimah’’ Long-Tailed Finches mum social time criterion was 1,000 s. Lighting was pro-
vided by fluorescent fixtures containing Vita-light bal-Birds used in this experiment and in experiment 3 were
first-generation, captive-hatched adult offspring of wild- anced-spectrum tubes that emit near ultraviolet as well as
Mate Preferences for Artificial Crests 797
Figure 4: Preferences of Larrimah long-tailed finches for nonornamented (‘‘No crest’’) versus artificially crested conspecifics. Bars
represent the mean proportion (⫹SE) of time spent in social contact with each stimulus type; dots represent mean within-trial
rank of social preferences (1–4, with 4 being most preferred). A, Female preference for males. B, Male preference for females.
wavelengths visible to humans. The fixtures were covered (a) to ascertain whether ‘‘Newry’’ females also prefer
white-crested males; (b) to ascertain whether prolongedwith diffusers, however, which probably blocked UV
transmission. exposure to white-crested males would alter female pref-
erence; (c) to ascertain whether females preferred malesBefore experimentation, the sex of each individual was
ascertained by repeated observation of birds (only males wearing only the beaded base over uncrested males; and
(d) to explore the possibility of individual variation insing). Trials were monitored to verify that both test and
stimulus birds behaved in sex-appropriate ways (Burley preference for white crests. This last goal was facilitated
by presenting test birds with two stimulus individuals of1986a). Four male stimulus sets and five female stimulus
sets were employed. each experimental phenotype. This test design reduces
intraindividual variation in test bird response across
stimulus sets, probably because the effect of a test bird’s
Results preferences for or aversions to specific stimulus individu-
Both sexes of this sexually monomorphic population dis- als is reduced. That is, a test bird is less likely to find
played preferences for white crests. The response of fe- both individuals of a given stimulus type highly aversive
males (N⫽13) was indistinguishable from that of fe- or attractive.
male zebra finches: white crests were strongly preferred
(χ
2
⫽13.41, P⫽.004), and no discrimination was made Methods
among the remaining phenotypes (fig. 4A). Each stimu- In experiment 3.1, each of 13 adult females was tested
lus set was used in six or seven (X⫽6.5) successful tri- with two of four stimulus sets, each consisting of two
als. Males (N⫽15) displayed an aversion to red and white-crested and two uncrested males. For each 1-h
blue crests as well as a preference for white ones vis-a
`-vis trial, percentage of time spent with the two males hav-
uncrested stimuli (χ
2
⫽25.72, P⬍.001; all a posteriori ing the same phenotype (white crested vs. uncrested)
multiple comparisons between no crest and crest types was summed before analysis. Individual preference was
are statistically significant; fig. 4B). Each stimulus set was measured as the average response over a female’s two
used in four to seven successful trials (X⫽6.0). trials.
After completion of experiment 3.1, each female was
Experiment 3: ‘‘Newry’’ Long-Tailed Finches housed in one of two viewing cages placed between two
cages containing males from the stimulus sets usedFemales were allowed to choose between uncrested and
white-crested males. This experiment had several goals: above. The six females in viewing cage A were exposed to
798 The American Naturalist
the males from stimulus sets 1 and 2, with the crested variation in response in the three trials of experiment 3.1
and 3.2 (repeated-measures ANOVA: model F⫽1.803,males in a cage physically abutting one side of the female
cage, and the uncrested males in a cage on the other side df ⫽2, 24, P⫽.19). Nor did individual females display
repeatable preferences (linear polynomial F⫽0.208,of the female cage. Thus, uncrested males could interact
with one another, and crested males could interact with df ⫽1, 12, P⫽.66; quadratic polynomial F⫽2.58, P⫽
.13).one another, but physical interactions between crested
and uncrested males were not possible. At the same time, In experiment 3.3, eight of 13 females spent a majority
of social time with featherless males. This result does notthe seven females in viewing cage B were exposed to the
males from stimulus sets 3 and 4 in an identical fashion. deviate from chance expectation (Wilcoxon z⫽⫺0.455,
P⫽.65; mean preference for featherless males ⫽58.4%).After 3 wk, the male stimulus sets were switched between
female viewing sets, so that females saw a new group of Thus, female preference for crested males cannot be ac-
counted for by a preference for the beaded platformmales (i.e., male sets 3 and 4 were placed with females in
viewing cage A, and male sets 1 and 2 were placed with alone.
females in cage B).
After an additional exposure interval of 3 wk, females Discussion
were tested in experiment 3.2 for their preference of Alternative Explanations for Preferences
white-crested versus uncrested males. Each female was
tested with one of two newly created male stimulus sets The existence of preferences for crests in a taxon devoid
of them is powerful evidence that mate preferences can(5 and 6). None of the males in the original four sets
were in the new sets. have their origin in nonadaptive neurophysiological pro-
cesses. Moreover, the similarity of response by females ofUpon completion of experiment 3.2, the feathers of
crested males in stimulus sets 5 and 6 were cut at the two grassfinch genera suggests that phylogeny provides
structure to these preferences. All together, the series ofpoint of attachment to the beaded platform. Then, in ex-
periment 3.3, each female was tested for her preference experiments involving artificial manipulation of grass-
finch phenotypes (Burley et al. 1982; Burley 1985, 1986a;of ‘‘featherless’’ versus uncrested males. Here each female
was exposed to the stimulus set (5 or 6) that she had not and this article)—and the fitness consequences of bear-
ing preferred versus nonpreferred traits (Burley 1986c,viewed during experiment 3.2.
Before the start of experiment 3, diffusers were re- 1988a; Burley and Price 1991; Burley et al. 1994, 1996)—
provides the most convincing evidence for the power ofmoved from the fluorescent fixtures over the experimen-
tal chambers, making light in the near UV available. In latent aesthetic preferences to date.
There are, nevertheless, several alternative/additionalexperiment 3, feathers used in crests were white finch
primaries, rather than rectrices. Primaries resist breakage hypotheses that deserve attention. Two competing
hypotheses that might predict a preference for crests arebetter than rectrices. Previously, rectrices had been used
preferentially because they appear somewhat more sym- ‘‘rare male advantage’’ (see references in O’Donald 1977)
and ‘‘the handicap principle’’ (Zahavi 1975). Both ofmetrical than primaries.
In all steps of this experiment, each stimulus set was these hypotheses, however, would predict that all crest
types are preferred over the uncrested phenotype. That is,used in either six or seven (X⫽6.5) trials.
since all crest types are uncommon (in fact, novel ‘‘mu-
tants’’), a hypothesis of rare male advantage would pre-
Results dict that all crest types would be preferred. Moreover, a
novelty hypothesis would suggest that continuous expo-In experiment 3.1, all 13 females spent a majority of so-
cial time with white-crested males (Wilcoxon z⫽sure to the crested male phenotype should extinguish or
reduce the preference for it. In experiment 3, however,⫺3.184, P⫽.001; mean percentage of social time spent
with white-crested males [‘‘mean preference’’] ⫽80.4%). the preference for white crests was stable over an interval
(12 wk, including test intervals) that is considerablyIn experiment 3.2, 12 of 13 females spent a majority of
social time with white-crested males (Wilcoxon z⫽greater than that typically taken by grassfinches to dem-
onstrate formation of an enduring pair-bond. The handi-⫺2.909, P⫽.004; mean preference ⫽86.5%). There was
no difference in mean preference for white-crested males cap principle would not necessarily predict that all crest
colors would be equally favored, but if the burden ofbetween experiments 3.1 and 3.2 (Wilcoxon matched-
pairs test, P⫽.53). Averaged over all three trials, 12 of added weight and/or reduced aerodynamic efficiency
caused by upright crests confers some sort of ‘‘indicator’’13 females preferred white-crested males (Wilcoxon z⫽
⫺3.110, P⫽.002; mean preference ⫽82.4%). status to crested males, then all crest phenotypes should
have been preferred over uncrested males.Females did not show significant between-individual
Mate Preferences for Artificial Crests 799
Exaggeration of Courtship Posture or Species-Typical nest lining of white feathers may conceal the white eggs
(Goodwin 1982). Perhaps the emergent mate preferenceTraits. Many birds, including males of some estrildines
(Goodwin 1982), commonly raise crown feathers during ‘‘spills over’’ from an attraction to white feathers in a
functional context. The plausibility of this hypothesis iscourtship. Perhaps the preference for a crest represents a
simple exaggeration of such male display. For both spe- weakened by the fact that while both sexes regularly col-
lect white feathers, male zebra finches are not attracted tocies, however, the preferred crest color is atypical of the
crown color. Since both species form long-term pair- females with white crests. Also, neither species incorpo-
rates feathers in sexual displays, even though such displaybonds and have high parental effort by both sexes, aver-
sions to such species-atypical phenotypes as presented would seem an easy evolutionary ‘‘step’’ from estrildine
stem displays (Goodwin 1982). Birds line nests withhere would be consistent with much of mate choice the-
ory (Trivers 1972). For zebra finches, one might argue feathers only when actively breeding, roosting in unlined
nests at other times. Nevertheless, this hypothesis pre-that the white crest exaggerates the species-typical eye
stripes (fig. 1). The aversion of male zebra finches to sents the testable possibility that a preference for white
crests may be absent in estrildines that do not line theirwhite-crested females, however, is inconsistent with this
explanation. Long-tailed finches have white rumps but nests with white feathers, and especially those species,
such as ground nesters, that line their nests with dark ob-lack other conspicuous white markings.
jects (Goodwin 1982).
The spillover hypothesis suggests that latent aestheticA General Preference for ‘‘White.’’ Two lines of evidence
suggest that results reported here do not reflect a general preferences may diverge rapidly among related species as
ecology and natural history diverge (e.g., Endler 1992).preference for white, or a preference for all-white birds,
at least in zebra finches. (No data are available for long- This hypothesis is more easily tested than is the possibil-
ity that a latent preference is a primitive character in atailed finches.) Zebra finches find white color bands neu-
tral (Burley 1985). There are two unlinked recessive al- tree of a lineage’s phylogeny (e.g., Basolo 1990), in part
because there is always the possibility that the preferenceleles that confer all-white plumage in estrildines, but
wild-type zebra finches reared by wild-type parents do was derived at some earlier time. The preference for
white crests would be ‘‘derived’’ if it had been passednot preferentially mate with white conspecifics (N. T.
Burley, unpublished data). down intact from the time the two species shared a
white-crested ancestor. This possibility is untestable but
is seemingly remote, given the absence of extant crested‘‘Artifacts.’’ The possibility that observed preferences are
an ‘‘artifactual’’ result of unnatural lighting conditions is estrildines and a distinct paucity of crested ploceines.
It is more likely that the preferences observed here re-discounted by the fact that similar results were obtained
under a wide range of lighting conditions. Thus, for ex- flect in part a general avian predilection toward facial or-
namentation (Darwin 1974). This predilection appearsample, while white feathers may reflect UV (S. Anders-
son, personal communication), results indicate the pref- sufficiently broad—given the diverse array of facial orna-
mentation that occurs in birds—that it is probablyerence for white crests is not contingent on UV
illumination. Across the experiments, the white feathers rooted in the central nervous system (CNS). In this con-
text, it is noteworthy that the preference for white crestsemployed came from three different species (one psitta-
cine, two estrildines), so the feathers probably varied is sex specific in the quite dichromatic zebra finch, while
both sexes are attracted to white crests in the sexually in-somewhat in reflectance spectra. One reviewer suggested
that the lack of preference for colored feathers resulted distinguishable (Larrimah) long-tailed finch population.
Similar earlier findings (Burley 1985) led to the hypothe-from the use of markers. If so, we might expect that in
experiment 1 blue and green crests would have been per- sis that responses to novel stimuli are organized by the
CNS in strategic ‘‘decision-making rules.’’ Thus, a centralceived as attractive and that black crests (less than half of
which had marker applied to them) would have been or higher mechanism may have evolved to organize and
even override perceptual biases of sensory systems. Mostfound more attractive than red crests (most of which
were touched up). Instead, we found that no crest color Australian estrildines are relatively monomorphic, the ze-
bra finch being one of the most dimorphic species. Re-other than white was attractive regardless of whether it
was naturally or artificially colored. Thus, the preference sults reported here, and previous results indicating that
male zebra finches are attracted to only a narrow rangefor white crests appears robust.
of novel female phenotypes (Burley 1985), suggest the
possibility that male zebra finches have evolved a mecha-Attraction to White Nest Feathers: The ‘‘Spillover’’ Hypoth-
esis. Both species preferentially use white feathers to line nism to override latent aesthetic mate preferences. By
contrast, in two monomorphic species, the aesthetic matetheir nests. Such a preference may be functional in that a
800 The American Naturalist
preferences of the sexes appear very similar (Taeniopygia and decision-making ‘‘apparatuses’’ is likely. Thus, fe-
males whose preferences for novelty encompass the spec-bichenovii—Burley 1986a; Poephila acuticauda—this arti-
cle). These species and sex differences deserve further at- trum of preferences displayed by conspecifics will benefit
from mating with those unusual males they find attrac-tention as they may provide considerable insight into
mechanisms of evolution of sexual monomorphism ver- tive (i.e., their sons will be attractive to a range of fe-
males); females whose preferences are narrow and idio-sus dimorphism.
syncratic will not so benefit. In turn, as preferences
broaden, selection may favor enhanced mutation rates in
Heritability of Preferences areas of the genome affecting secondary sexual traits, re-
sulting in a coevolutionary race between aesthetic prefer-Some authors have argued that critical evidence for com-
peting models of sexual selection hinges on whether or ences and male traits (Johnson and Burley 1998). If this
scenario is valid, then it should be possible to demon-not mate preferences are invariant and genetically fixed;
by this view, aesthetic preferences must be invariant (e.g., strate that other taxa besides estrildine finches have a
broad range of latent aesthetic preferences. Such demon-Wilkinson and Reillo 1994; Sherman and Wolfenbarger
1995). As Christy and Backwell (1995) point out, how- stration would be powerful evidence for our contention
(‘‘Introduction’’) that aesthetic preferences are an impor-ever, there is no legitimate reason to require that aes-
thetic mate preferences have no heritable component. In- tant force in the early evolution of many secondary sex-
ual and ‘‘species-specific’’ traits.deed, if aesthetic mate preferences have an important
role in generating sexually selected traits, then to explain Latent aesthetic preferences constitute a conceptually
parsimonious explanation for the early evolution of sexu-the radiation of such traits, we must expect that aesthetic
preferences evolve. The existence of local variation for la- ally selected traits. Ironically, once preferred traits begin
to evolve and the preferences are no longer latent, traitstent preferences would provide enormous potential for
speciation via sexual selection, and so it is important to may evolve into indicators and preferences may coevolve,
becoming functional. A novel attractive trait, for exam-determine the extent of within-species variation in latent
preferences. ple, might reach fixation rapidly through the impact of
aesthetic preferences. Assuming that the trait has someUnfortunately, it is difficult to ascertain individual
variation in preferences because mate-choice decisions cost, its expression may be influenced by other loci, thus
propelling the evolution of ‘‘indicator status’’ (e.g., ainvolve multiple traits (e.g., Burley 1981b), not just the
one being experimentally manipulated. Thus, in experi- male with brighter spots that result from his condition
genes could be favored). This process might be especiallyment 3, between-female variation in preference was not
significant, nor did individuals show repeatability of the effective if there is nonrandom assortment of attractive
individuals with mates having superior heritable traits‘‘strength’’ of their preference. While an improved design
might show that individual variation does exist, it does (Burley 1986c, 1988a). These possibilities represent a
sort of ‘‘reverse Fisherian’’ process in which traits evolvenot follow necessarily that such variation is heritable. In-
deed, the expression of latent preferences may be condi- initially through aesthetic preferences but later acquire
functional associations through linkage disequilibriumtion-dependent because females in poor condition may
ill afford the consequences of mating with very attractive and coevolution. Thus, even when traits used in mate
choice are found to be indicators of functional, herit-males (i.e., reduced paternal care that such males provide
[Burley 1988a]); if condition-dependent mate preferences able qualities, researchers should be cautious about in-
ferring a functional origin to the preferences for thosedo occur, ascertaining heritability of latent preference be-
comes even more difficult. traits.
AcknowledgmentsLatent Aesthetic Preferences and Good Genes Models
Although it seems reasonable that latent aesthetic prefer- B. Worden suggested that attraction of finches to white
nest feathers might form the basis of their attraction toences had ‘‘accidental’’ origins, we doubt that this is a
sufficient explanation for the readily observable diversity white crests. We thank C. B. Coopersmith for research
contributions, J. Calkins and D. Enstrom for useful dis-of aesthetic preferences found in estrildines. We hypothe-
size that, over evolutionary time, selection favors the cussion, three anonymous reviewers for comments, and
the Commonwealth of Australia and the Conservationbroadening of mate preferences for novelty. This occurs
because consensus of preference is the driving force fa- Commission of the Northern Territory for permission to
export wild long-tailed finches. We thank J. A. Aulick forvoring the evolution of aesthetic traits (Fisher 1930).
Within a sexual population, some variation in perceptual drafting figures 2–4. This research was supported by the
Mate Preferences for Artificial Crests 801
National Science Foundation (awards DEB-8802956 and Christitis, L. 1987. Biochemical systematics within
palaeotropic finches (Aves: Estrildidae). Emu 89:119–IBN-9507514 to N.T.B.) and the University of California.
123.
Christy, J. H., and P. R. Y. Backwell. 1995. The sensory
Literature Cited exploitation hypothesis. Trends in Ecology & Evolu-
tion 10:417.Andersson, M. 1994. Sexual selection. Princeton Univer-
sity Press, Princeton, N.J. Darwin, C. 1874. The descent of man, and selection in
relation to sex. 2d ed. J. Murray, London.Arak, A., and M. Enquist. 1993. Hidden preferences and
the evolution of signals. Proceedings of the Royal Soci- Endler, J. A. 1992. Signals, signal conditions, and the di-
rection of evolution. American Naturalist 139(suppl.):ety of London, Series B 340:207–213.
Armstrong, E. A. 1947. Bird display and behavior. Ox- S125–S153.
———. 1993. Some general comments on the evolutionford University Press, New York.
Basolo, A. L. 1990. Female preference predates the evolu- and design of animal communication systems. Philo-
sophical Transactions of the Royal Society of Londontion of the sword in swordtail fish. Science (Washing-
ton, D.C.) 250:808–810. B, Biological Sciences 340:215–225.
Enquist, M., and A. Arak. 1993. Selection of exaggeratedBurley, N. 1981a. The evolution of sexual indistinguish-
ability. Pages 121–137 in R. D. Alexander and D. W. male traits by female aesthetic senses. Nature (Lon-
don) 361:446–448.Tinkle, eds. Natural selection and social behavior.
Chiron, New York. Fisher, R. A. 1930. The genetical theory of natural selec-
tion. Clarendon, Oxford.———. 1981b. Mate choice by multiple criteria in a mo-
nogamous species. American Naturalist 117:515–528. Gill, F. B. 1990. Ornithology. W. H. Freeman, New York.
Goodwin, D. 1982. Estrildid finches of the world. Cornell———. 1985. The organization of behavior and the evo-
lution of sexually selected traits. Pages 22–44 in P. A. University Press, Ithaca, N.Y.
Guilford, T., and M. S. Dawkins. 1991. Receiver psychol-Gowaty and D. W. Mock, eds. Avian monogamy. Or-
nithological Monographs no. 37. American Ornitholo- ogy and the evolution of animal signals. Animal Be-
haviour 42:1–14.gists’ Union, Washington, D.C.
———. 1986a. Comparison of the band-colour prefer- Hamilton, W. D., and M. Zuk. 1982. Heritable true fit-
ness and bright birds: a role for parasites? Scienceences of two species of estrildid finches. Animal Behav-
iour 34:1732–1741. (Washington, D.C.) 218:384–387.
Johnson, K., and N. T. Burley. 1998. Mating tactics and———. 1986b. Sex-ratio manipulation in color-banded
populations of zebra finches. Evolution 40:1191–1206. mating systems of birds. Pages 21–60 in P. G. Parker
and N. T. Burley, eds. Avian mating tactics: female and———. 1986c. Sexual selection for aesthetic traits in spe-
cies with biparental care. American Naturalist 127: male perspectives. Ornithological Monographs no. 49.
American Ornithologists’ Union, Washington, D.C.415–445.
———. 1988a. The differential allocation hypothesis: an Johnson, K., R. Dalton, and N. T. Burley. 1993. Prefer-
ences of female American goldfinches (Carduelis tristis)experimental test. American Naturalist 132:611–628.
———. 1988b. Wild zebra finches have band-colour for natural and artificial male traits. Behavioral Ecol-
ogy 4:138–143.preferences. Animal Behaviour 36:1235–1237.
Burley, N. T., and D. K. Price. 1991. Extra-pair copula- Keast, A. 1958. Infraspecific variation in the Australian
finches. Emu 58:219–246.tion and attractiveness in zebra finches. Acta XX
Congressus Internationalis Ornithologici, pp. 1367– Kodric-Brown, A., and J. H. Brown. 1984. Truth in ad-
vertising: the kinds of traits favored by sexual selec-1372.
Burley, N., G. Krantzberg, and P. Radman. 1982. Influ- tion. American Naturalist 124:309–323.
Marchetti, K. 1993. Dark habitats and bright birds illus-ence of colour-banding on the conspecific preferences
of zebra finches. Animal Behaviour 30:444–455. trate the role of the environment in species divergence.
Nature (London) 362:149–152.Burley, N. T., D. A. Enstrom, and L. Chitwood. 1994. Ex-
tra-pair relations in zebra finches: differential male Meyer, A., J. M. Morrissey, and M. Schartl. 1994. Recur-
rent origin of a sexually selected trait in Xiphophorussuccess results from female tactics. Animal Behaviour
48:1031–1041. fishes inferred from molecular phylogeny. Nature
(London) 368:539–542.Burley, N. T., P. G. Parker, and K. Lundy. 1996. Sexual
selection and extra-pair fertilization in a socially mo- Morris, D. 1956. The feather postures of birds and the
problem of the origin of social signals. Behaviour 9:nogamous passerine, the zebra finch (Taeniopygia gut-
tata). Behavioral Ecology 7:218–226. 75–113.
802 The American Naturalist
O’Donald,P. 1977.Mating advantage of raremales in models Sherman, P. W., and L. L. Wolfenbarger. 1995. Genetic
correlations as tests for sensory exploitation? Trends inof sexual selection. Nature (London) 267:151–154.
Pizzey, G. 1980. A field guide to the birds of Australia. Ecology & Evolution 10:246–247.
Sibley, C. G., and J. E. Ahlquist. 1990. Phylogeny andPrinceton University Press, Princeton, N.J.
Rowland, W. J. 1989. Mate choice and the supernormal- classification of birds: a study in molecular evolution.
Yale University Press, New Haven, Conn.ity effect in female sticklebacks (Gasterosteus aculea-
tus). Behavioral Ecology and Sociobiology 24:433–438. Trivers, R. L. 1972. Parental investment and sexual selec-
tion. Pages 136–179 in B. Campbell, ed. Sexual selec-Ryan, M. J., and A. Keddy-Hector. 1992. Directional pat-
terns of female mate choice and the role of sensory bi- tion and the descent of man, 1871–1971. Aldine, Chi-
cago.ases. American Naturalist 139(suppl.):S4–S35.
Ryan, M. J., J. H. Fox, W. Wilczynski, and A. S. Rand. Wilkinson, G. S., and P. R. Reillo. 1994. Female choice
response to artificial selection on an exaggerated male1990. Sexual selection for sensory exploitation in the
frog Physalemus pustulosus. Nature (London) 343:66–67. trait in a stalk-eyed fly. Proceedings of the Royal Soci-
ety of London, Series B 255:1–6.Serle, W., G. J. Morel, and W. Hartwig. 1977. A field
guide to the birds of West Africa. Collins, London. Zahavi, A. 1975. Mate selection—a selection for a handi-
cap. Journal of Theoretical Biology 53:205–214.Shaw, K. 1995. Phylogenetic tests of the sensory exploita-
tion model of sexual selection. Trends in Ecology &
Evolution 10:117–120.
Associate Editor: Marlene Zuk
