Content uploaded by Konstans Wells
Author content
All content in this area was uploaded by Konstans Wells
Content may be subject to copyright.
Behavioral Ecology
doi:10.1093/beheco/arp024
Advance Access publication 24 February 2009
Superb fairy-wren males aggregate into hidden
leks to solicit extragroup fertilizations before
dawn
Andrew Cockburn,
a
Anastasia H. Dalziell,
a
Caroline J. Blackmore,
a
Michael C. Double,
a,b
Hanna Kokko,
a,c
Helen L. Osmond,
a
Nadeena R. Beck,
a
Megan L. Head,
a,d
and Konstans Wells
a,e
a
Evolutionary Ecology Group, School of Botany and Zoology, Australian National University, Canberra
ACT 0200, Australia,
b
Department of Biological and Environmental Science, Division of Ecology and
Evolution, P.O. Box 65, University of Helsinki, 00014 Helsinki, Finland,
c
Australian Antarctic Division,
Kingston, Tasmania 7050, Australia,
d
School of Biological and Chemical Sciences, Queen Mary,
University of London, Mile End Road, London E1 4NS, UK, and
e
Institute of Experimental Ecology,
University of Ulm, 89081 Ulm, Germany
Female superb fairy-wrens Malurus cyaneus initiate extragroup fertilizations by forays to the territory of preferred males, just before
sunrise, 2–4 days before egg laying. Over a prolonged breeding season, males advertise their availability to foraying females by
singing during the dawn chorus. Here, we show that 1) males commence dawn advertisement at the same time of the year
regardless of their quality or status; 2) subordinate males advertise by singing in close proximity to the dominant, or by using
the dominant’s song perch, despite inevitable punishment; 3) low-quality dominants and their helpers sing from the boundary of
their own territory, which increases their proximity to attractive neighboring dominants; 4) each spatial cluster of males use
a common dialect of a song that is implicated in extragroup choice, despite the ability of individual males to sing several dialects;
and 5) there is leakage of paternity to lower-quality helpers and neighbors as a result of their ‘‘satellite’’ behavior. Collectively, these
data suggest that Wagner’s hidden lek hypothesis (Wagner RA, 1998. Hidden leks: sexual selection and the clustering of avian
territories. In: Parker PG, Burley NT, editors. Avian reproductive tactics: female and male perspectives. Ornithological Mono-
graphs No. 49. Allen Press. p. 123–145) can be extended to birds that defend year-round all-purpose territories and that mating
induced by parasitic behavior of low-quality satellites can be one explanation for polyandry in birds. Key words: cooperative
breeding, dawn chorus, extrapair copulation, hidden lek, Malurus, polyandry, song dialect. [Behav Ecol 20:501–510 (2009)]
Despite considerable effort, there remains little consensus
on the ecological determinants of the extreme differences
between bird species in the importance of extrapair fertiliza-
tions (Griffith et al. 2002; Westneat and Stewart 2003). High
breeding density has been suggested to facilitate access to
extrapair mating, but support for this hypothesis remains
equivocal (Westneat and Sherman 1997). Wagner (1998,
1993) has offered the alternative perspective that rather than
determining the possibility of extrapair fertilization, density
may instead be determined by male aggregation that facili-
tates access to extrapair mates (the hidden lek hypothesis).
He proposes that the desirability of obtaining extrapair mat-
ings from neighbors could cause breeding birds to aggregate
into colonies and that territorial migratory birds could occupy
just a small proportion of suitable habitat in order to increase
proximity to conspecifics. Wagner (1998) describes two paths
by which this could occur. In the first case, females may settle
near high-quality males, because even if not paired to those
males, they can procure their sperm through extrapair mat-
ing. In the second, which is analogous to the hotshot model of
lek formation (Beehler and Foster 1988), males aggregate
because physical proximity to high-quality males gives them
the opportunity to parasitize the attractiveness of nearby
males that visit those males in order to procure extrapair
matings. Although there is growing evidence that migratory
birds use the presence of conspecifics in choosing where to
breed (Safran 2004; Fletcher 2007), there are currently few
explicit tests of all aspects of the hidden lek hypothesis,
though criteria for detection of the hypothesis have been re-
fined (Tarof et al. 2005; Fletcher and Miller 2006), and both
supportive (Hoi and Hoi-Leitner 1997; Ramsay et al. 1999;
Formica et al. 2004), and negative (Safran 2004, 2007) results
are starting to emerge.
The hidden lek hypothesis initially seems ill suited to explain
the spatial distribution of breeding birds that defend year-
round, all-purpose territories that collectively occupy much
of the landscape, as settlement options are constrained to
a far greater extent and are completely different from the case
of species settling on return from migration or winter flocks, or
to colonies that are only occupied to rear young (Møller 1992).
Here, we propose an extension of the hidden lek hypothesis
that is applicable to year-round, all-purpose territories. Our
extension is based on the logic of the hotshot hypothesis and
depends on females visiting male territories to obtain extrap-
air fertilizations. According to this view, if only a few males are
likely to be sufficiently attractive to provoke female visits,
lower-quality males should attempt to gain benefits through
satellite behavior and advertise as close to those males as is
allowed by the constraints of territory, generating conspecific
aggregation (Figure 1).
We illustrate this model with our own observations of the
superb fairy-wren Malurus cyaneus, in which extrapair
Address correspondence to A. Cockburn. E-mail: andrew.cockburn@
anu.edu.au.
Received 16 September 2008; revised 28 December 2008; accepted
14 January 2009.
The Author 2009. Published by Oxford University Press on behalf of
the International Society for Behavioral Ecology. All rights reserved.
For permissions, please e-mail: journals.permissions@oxfordjournals.org
by guest on May 18, 2011beheco.oxfordjournals.orgDownloaded from
fertilization is unusually prevalent (Mulder et al. 1994). In this
cooperatively breeding species, females initiate extragroup
fertilization during predawn forays to the territory of the pre-
ferred male (Double and Cockburn 2000). Females will travel
as far as six territories from their own to encounter attractive
males. Here, we use a number of observational data sets to
demonstrate that males aggregate spatially to court females
during the dawn chorus. We report behavioral and parentage
data that suggest that these aggregations are competitive
rather than cooperative, and interpret these data against the
classification by Ho
¨glund and Alatalo (1995) of hypotheses
proposed to explain aggregations of displaying males in lek-
king species.
METHODS
Study species and area
We have studied a color-ringed population of M. cyaneus at
the Australian National Botanic Gardens continuously since
1988, monitoring social organization (Cockburn et al. 2003;
Cockburn, Osmond, et al. 2008), reproductive attempts
(Cockburn, Sims, et al. 2008), and resolving parentage
through microsatellite-based parentage analysis (Double,
Cockburn, et al. 1997; Double, Dawson, et al. 1997). The
breeding season lasts on average from the construction of
the first nests in late August until the last young are fledged
in February. Females lay a succession of clutches that usually
comprise two to four eggs, and although they are only capable
of rearing three successful broods per season, they can initiate
as many as eight clutches in response to frequent nest pre-
dation. Fairy-wrens are facultative cooperative breeders, and
although just one female breeds on the territory, there may be
as many as five adult males resident on each territory during
the breeding season (Cockburn, Osmond, et al. 2008), all
of which contribute to the rearing of young (Dunn and
Cockburn 1996). The oldest male is socially dominant, dis-
placing subordinate males from access to food and the breed-
ing female (Cockburn, Osmond, et al. 2008).
Despite the presence of many males on the territory of the
female, males from outside the social group dominate pater-
nity (Mulder et al. 1994; Dunn and Cockburn 1999). Extra-
group fertilizations are initiated by the female on
extraterritorial forays that take place in the half hour before
sunrise, 2–4 days before the first egg of the clutch is laid
(Double and Cockburn 2000). Males respond to the opportu-
nity for extragroup fertilization by courtship displays during
daytime on the territory of the female (Mulder 1997) and
singing on their own territory in the dawn chorus (Dalziell
and Cockburn 2008). Females prefer to mate with the small
proportion of males that acquire nuptial plumage months
before the start of the breeding season (Dunn and Cockburn
1999; Green et al. 2000; Double and Cockburn 2003; Cockburn,
Osmond, and Double 2008), perhaps because males begin
courtship displays as soon as they have full nuptial plumage
(Peters et al. 2001), and only old, high-quality males can sus-
tain such display (Peters 2000; Peters et al. 2000, 2001; Cock-
burn, Osmond, and Double 2008).
Fairy-wrens produce two sorts of songs (Langmore and
Mulder 1992; Dalziell and Cockburn 2008). The most com-
mon, a variable chatter song, is sung by both males and
females and is implicated in territory defense (Cooney and
Cockburn 1995). The alternative trill song is sung only by
males and comprises a long low frequency trill component
that ultimately degenerates into a higher frequency conclu-
sion that resembles the chatter song. During the day, trill
songs are prompted by the calls and wing beats of large
predatory birds (Langmore and Mulder 1992). Males also
incorporate unprompted trill songs in their dawn recital
(Dalziell and Cockburn 2008). Regardless of time of day,
older males sing longer trill components, implicating this
aspect of singing in extragroup mate choice (Dalziell and
Cockburn 2008).
Focused observations
Here, we synthesize data from several data sets of the behavior
of males during the dawn chorus. As each study was designed
to focus on different aspects of behavior, each employs differ-
ent methods and sampling criteria. The studies are described
below:
a) Breeding season behavior. Between 1999 and 2004, we un-
dertook regular surveys of the behavior of males sing-
ing in the dawn chorus during the breeding season. We
conducted observations from 1 h before sunrise, ini-
tially monitoring the bush in which the group had
roosted overnight. We maintained a continuous narra-
tive of the behavior of males after they moved to the
song perches from which they sang in the dawn chorus.
In territories with more than one male, a separate ob-
server monitored the song post of each male. Because
an additional focus of these surveys was to observe the
interactions between males and their own fertile mate
Figure 1
The formation of hidden leks via spatial aggregation of males in year-
round all-purpose territories (a) advertisement from the center of
territories, (b) advertisement in close proximity to attractive
neighbors.
502 Behavioral Ecology
by guest on May 18, 2011beheco.oxfordjournals.orgDownloaded from
after the chorus concluded, we selected males whose mate
was initiating a new nest. Hence, our sample of predawn
behavior is biased toward males whose mate was fertile.
b) Winter song. In 1999, we studied singing behavior of
males during the winter. We chose 18 focal males ac-
cording to the criteria that they lived on territories that
afforded easy visibility and that a sample of them (N¼
8) had molted early (before 2 July) according to the
criteria of Cockburn, Osmond, and Double (2008).
Seven had retained nuptial plumage in the postbreed-
ing molt, and one acquired nuptial plumage before 2
July. The remaining 10 late molters either fell into the
prebreeding (N¼7) or intermediate (N¼3) classes of
Cockburn, Osmond, and Double (2008), but all were
in brown eclipse plumage for a considerable propor-
tion of the sampling period. We sampled the birds at
approximately 14-day intervals from early June until
late September, by which time some females were start-
ing to initiate their first nest. We arrived on the terri-
tory an hour before sunrise and measured the intensity
of dawn singing by counting the number of songs
males sang during their dawn recital, from the moment
they moved to their song perch until they abandoned
singing, usually just before sunrise.
c) Trill song during the day. Between August and December
2000, we conducted surveys of the variation in the trill
component of trill songs sung by males during the day,
using predator calls to prompt singing. We played
taped calls of a pied currawong Strepera graculina to
targeted males using a Sony Walkman WM-FX195 and
a Sony SRS-A60 speaker. Responses were recorded with
a Sennheiser MZW60 directional microphone and a So-
ny TCD-D10 ProII Digital Audio Tape Recorder. We
attempted to obtain a minimum of 10 trill songs from
each male. Variants (dialects) of the trill component
were distinguished according to the type of elements
and their organization into one of more distinctive
phrases (Figure 5). For example, songs were consid-
ered to belong to the same dialect if they shared the
same element/s, though varied in the extent of repeti-
tion of the element/s. For example, six males classified
as using Dialect G show variation only in the repetition
of the single element (Figure 5a–f). By contrast, their
neighbors singing Dialect H can be distinguished as
singing a distinctive dialect, this time comprising three
elements arranged in a characteristic order (Figure 5g–h).
d) Trill song during the dawn chorus. During the 2004/2005
breeding season, we recorded the full dawn recital of
most of the males in the study area. Our methods were
described in detail in Dalziell and Cockburn (2008).
We used the recordings to determine the dialect/s of
the trill component sung.
e) Spatial distribution of males during the dawn chorus. As part
of our measurement of dawn recitals in 2004/2005, we
mapped the distribution of song posts used by males
during the dawn chorus. Because males sometimes
used several posts, we divided the study area into 15-
m315-m cells. Each cell was also assigned as belong-
ing to one of the territories in the study area or as
unoccupied space (an area of open lawn and a planta-
tion of rainforest). We then assigned the song posts of
each male to a particular cell. Although this meant that
males could have theoretically assigned their time
equally between two cells, in practice, all males in our
date spent at least 90% of the dawn chorus in a single
cell, so there were no problems in assignment.
In order to model whether the resulting distribution of
males was random, we first calculated the number of 15-m units
from the cell of each male and the cell of the male that sang in
closest proximity to him. We then conducted a randomization
test in ‘‘Matlab’’ that involved assigning each male randomly to
one of the cells within its territory, allowing us to calculate the
average nearest male distances for this randomized distribu-
tion. This randomization was repeated 10 000 times, and the
resulting set of simulated values for the mean distance,
under a null hypothesis of random male location, was com-
pared with the observed distribution.
Paternity analysis
Exploratory analysis revealed that because the clutch size of
fairy-wrens is small (2–4), and some paternity can be allocated
to members of the female’s social group, individual broods
provided a poor sample of the males with which the female
may have copulated. However, repeated nesting allowed
a greater sample to be obtained for individual females. Be-
cause the extragroup males available to females changes each
year, we analyzed female choice on an annual basis. We con-
fined our analysis to females that had produced at least six
extragroup young during the course of the breeding season,
as exploratory analysis suggested that we would underesti-
mate the range of possible sires until that many young had
been sampled. We discuss possible biases resulting from this
criterion below.
For each female, we asked: 1) whether there was an a priori
attractive male, defined by early molting, that sired offspring in
all or most (.75%) of the broods the female produced that
season. We called this the preferred male. It is important to
make this distinction on a priori criteria in order to estimate
the paternity preferred males lose to subordinates and other
satellites. We then classified any additional extragroup sires
as: 2) a subordinate of the preferred male; 3) a dominant from
an immediately neighboring territory of the preferred male; 4)
a subordinate from an immediately neighboring territory of
the preferred male; or 5) a more distant male. Some females
have offspring that could not be sired by any male in the pop-
ulation. This is most likely near the edge of our study area, in-
dicating that the female has mated with a male from outside
the study area. Because the status of these males could not
be determined and it is impossible to be certain which sire
was preferred, we excluded females from consideration unless
we knew the sires of all their young.
In 2004/2005, when we knew the song post of each male, we
estimated seasonal variation in the opportunity for extrapair
mating by assigning a female as fertile when she was 2–3 days
before laying the first egg in each clutch, because although
females occasionally undertake forays 4 days before egg laying,
they are only common on day 22 and day 23 (Double and
Cockburn 2000), and within-pair matings are also confined to
this period (Cockburn A, unpublished data). Because, our
analyses indicated that males aggregated into clusters, and
that females only visited one cluster of males, we also esti-
mated the number of visits each cluster was likely to receive,
and whether the absolute and per capita visitation rates were
predicted by the number of males in the cluster.
Statistical analysis
All statistical tests were carried out in ‘‘Genstat for Windows v. 9.’’
For simple models based on least squares or goodness-of-fit,
we report the test used in the Results. Where repeated sam-
pling of individuals occurred, we used the REML procedure
to fit mixed models with the identity of the male as a random
term. We discarded nonsignificant terms (P.0.05) in a step-
wise way to obtain parsimonious models.
Cockburn et al. •Hidden leks in superb fairy-wrens 503
by guest on May 18, 2011beheco.oxfordjournals.orgDownloaded from
RESULTS
Winter song
During winter, males sang very little in the dawn chorus until
late August (Figure 2, approximately day 235). Thereafter, all
males increased singing until late September (day 270), when
most females were commencing their first nests, and males
were producing a continuous recital. We explored a variety of
methods of statistical modeling, but the only technique that
produced a Normal distribution of residuals was to analyze
data from before and after day 235 separately. Before day
235, the number of songs increased gradually, but there was
no difference between birds that were in full nuptial plumage
(early) or had not yet molted (late) (REML test with identity
of the male as a random term: interaction: Wald v
2
¼3.1, df ¼1;
P¼0.08; early versus late; v
2
¼1.2, df ¼1; P¼0.3; Julian
day v
2
¼32.7, df ¼1; P0.001). The same was found
once singing started to accelerate after day 235. All males
increased at a similar rate, though some of the late class
were molting into nuptial plumage (interaction: v
2
¼0.0,
df ¼1; P¼0.98; early vs. late; v
2
¼2.8, df ¼1; P¼0.09;
Julian day v
2
¼41.6, df ¼1; P0.001).
During the breeding season, males generally contributed to
the dawn chorus every morning, regardless of social status
(183/187 mornings; 98%; sample derived from 37 males).
The four exceptions involved three dominant males that did
not sing when their own female had departed on a foray that
morning, though they always sang when their mate had not de-
parted. This is a small proportion of males whose female had
departed on a foray (3/37; 8%), though the males that did sing
when their mate was fertile often delayed the onset of singing.
The actual incidence of singing by males during the season will
be even higher than 98%, as our sampling was biased toward
males whose own mates were fertile.
Singing males are spatially aggregated and adopt a distinctive
dialect
Males either sang exclusively from one song post or moved be-
tween two or three adjacent song posts during the dawn chorus
and used the same sites on each morning they were observed.
Most dominant males sang from a prominent, exposed branch
of a tall shrub, generally between 1 and 4 m in height, though
subordinates sometimes sang from more concealed locations.
Although some males sang from central locations in their ter-
ritory, many did not, and sang instead on the boundary of their
territory, allowing them to sing in very close proximity to their
neighbors (Figure 3). As a consequence, distances between
nearest song posts were considerably shorter (observed mean
¼2.8360.20 SE; N¼67; units ¼15 m cells) than predicted by
the simulations (mean ¼3.42; Figure 4; observed mean lower
than 9980/10 000 of the simulated outcomes; P¼0.002).
Clustering of males becomes even more obvious when the
form of predawn song is considered. Following predator play-
back during daytime, most males sang more than one trill dialect
(95/136 [70%] sang more than one dialect; mean ¼2.161.0 SD
dialects. However, all but one male sang just one dialect in the
dawnchorus,and the males in physicalclustersadopted the same
distinctive dialect (Figure 5). The exception was a subordinate
that sang both the dialect adopted by the other males on his
territory and the dialect used by the neighbors with a territory
boundary in close proximity to his song post.
Dominant males attack subordinates
Although males devoted most of the dawn chorus to continu-
ous singing, we also detected direct physical interactions be-
tween males in two contexts.
First, dominant males flew to the song post of their subordi-
nate/s at least once during the dawn chorus and attacked them
(N¼50/52 mornings [96%] when viewing conditions were
adequate; sample involves 13 different social groups). In the
47 cases where the timing was recorded, the dominant initi-
ated the attack 14 min 66 SD before sunrise. Attacks by
the dominant led the subordinate to stop singing briefly,
but subordinates inevitably recommenced shortly thereafter
(N¼50). The two cases where no attacks took place coincided
with the only two instances where a dominant with a helper
refrained from singing in response to the departure of his
mate on an extragroup foray.
Second, occasionally dominants left their song post for other
reasons, most often in response to movements by their own
mate. When these departures occurred, the subordinate left
his own perch and commenced singing from the song perch
of the dominant (N¼6). In all but one case, when the dom-
inant returned, the helper was physically attacked (N¼3) or
fled and was chased (N¼2). The exception occurred on one
of the cases where the dominant failed to sing in the dawn
chorus at all, because he was waiting on the territory boundary
for the return of his mate from her extragroup foray. In this
case, the subordinate commenced singing from the song
perch of the dominant and did so for the entire dawn chorus.
Preferred males lose substantial paternity to parasitic
satellites
We obtained data for 105 breeding seasons of 85 females that
had produced more than six or more young from extragroup
sires, but were forced to discard 39 of these because a malefrom
outside the study population had sired at least one of the young
they reared. Hence, the sample we report here involves 66
breeding seasons from 54 females, who produced 447 extra-
group young (range from 6 to 11 per season). There is some
bias in our sampling, as females rarely reach the threshold of
six young in their first year of breeding and in drought years.
We previously reported the impact of drought on female choice
of sires (Cockburn, Osmond, and Double 2008) and will report
an analysis of choice by first-year females elsewhere.
There was multiple extragroup parentage in 48 of the 66
samples (73%). Multiple extragroup paternity was almost inev-
itably shared between birds living on the same territory or
Figure 2
The intensity of dawn singing measured as the number of songs
produced by advertising males from their song perches during the
dawn chorus. Males are distinguished by whether they gained nuptial
plumage early (before 2 July; circles) or after that date (triangles).
The line depicts the best fit of separate REML models for the data
collected before and after day 235 (22 August).
504 Behavioral Ecology
by guest on May 18, 2011beheco.oxfordjournals.orgDownloaded from
immediate neighbors, all of which are likely to be members of
the same song cluster (443/447 young; 99%). The four young
that did not come from neighboring territories were from
three cases where females also sampled a male at 180 degrees
from the main cluster. We have discussed these exceptions else-
where (Double and Cockburn 2000). Despite these very rare
exceptions, our data overwhelmingly suggest that females es-
tablish an area from which they will seek extragroup fertiliza-
tions by the first brood of the season, and continue to visit the
same area throughout the season (66/66 seasons).
In 52/66 seasons (79%), the female apparently targeted an
early-molting sire, on three occasions producing young from
two sires that both molted early. In these latter cases, the
two early-molting males were immediate neighbors. Attractive
male/s gained 100% of extragroup parentage in just 17/52 sea-
sons (33%). In 32 cases, the attractive dominant had one or
more subordinates, which also gained parentage in 12 seasons
(38%). However, the attractive dominants also lost paternity to
immediate neighbors in 25 cases (48%). Collectively, the early-
molting male obtained 262/391 (67%) of extragroup young
produced by the female.
The failure of some females to choose an early-molting extrap-
air sire in the remaining 14 cases occurred in a number of con-
texts. In 7/14 cases, the high-quality male visited by the female in
the previous breeding season had once again acquired nuptial
plumage early and displayed to the female throughout the win-
ter, but was no longer present on the territory when she was ready
to mate, because he died close to the start or during the breeding
season (N¼6 males, as two females were apparently affected by
the same death). In the remaining cases involving five females
in seven seasons, we could not contrast preference between
seasons, because the female had not produced extragroup
young in the previous year. However, all five females initially
chose a poor male in circumstances that we have demonstrated
are unfavorable to mate assessment (Cockburn, Osmond, and
Double 2008), either because few males acquired early blue
plumage (n¼3), because there was heavy mortality of the
early-molting males just prior to the season (1), or because
the female was mating for the first time (n¼1). Nonetheless,
two of these females subsequently mated with the same male in
the following year, suggestive of consistent preference, so occa-
sional choice of low-quality males apparently occurs.
Figure 3
The location of the main song
post used by males in the dawn
chorus in relation to the
boundaries of their territories.
The lines represent territory
boundaries. Upper case letters
denote dominants, and lower
case letters denote subordi-
nates. Each letter represents
a distinct dialect characterized
by variation in the trill compo-
nent of the trill song used in
the dawn chorus. One subordi-
nate (indicated by an arrow)
sang two dialects—that of his
dominant and the neighbor
nearest his song post. The two
shaded areas represent habitat
not part of the territory of
breeding fairy-wrens (a rainfor-
est gully and an open lawn),
though it can be traversed by
foraying females.
Cockburn et al. •Hidden leks in superb fairy-wrens 505
by guest on May 18, 2011beheco.oxfordjournals.orgDownloaded from
Visits by females are rare
The number of females we knew to be fertile, and hence we
estimated to be undertaking forays on any particular morning,
never exceeded eight (from a sample of 66 breeding on the
study area) and was well described by a quadratic function
(Figure 6; estimate for quadratic term in multiple regression:
F
1,122
¼41.2; P0.001).
Our data for such visits to particular clusters were less com-
plete, as from the 66 females, two produced only a single brood
of legitimate young, three failed to rear any young to the age
when we sample young for DNA to establish parentage (day 6
of nestling life), and eight apparently preferred males whose
dawn song we had not recorded. In addition, females from out-
side the study area also visit some of our clusters. Our analysis is
therefore based on 53 females visiting the 12 main clusters as
determined by the males that sired their young (Figure 3). The
probability that any cluster group receives a visit on any morn-
ing is low (14.5% of mornings; N¼1488). Although as many
as three females may visit a cluster on a morning, visits by
multiple females are rare (1.5% of mornings), and such mul-
tiple visits are close to that predicted by a random Poisson
process (Figure 7), suggesting that females act independently
in seeking male clusters.
Larger clusters attract more visits (Figure 8a; linear regres-
sion; F
1,10
¼5.9, P¼0.04, R
2
¼37%), but this does not trans-
late into an increased visiting rate per male in the cluster
(Figure 8b; linear regression; F
1,10
¼1.0, P¼0.70, R
2
¼2%;
the significance in this case is for a negative estimate of the
relationship between size of cluster and per capita visit rate).
Copulations are initiated by the female
Although we have previously characterized female forays using
radiotelemetry (Double and Cockburn 2000), it has been
Figure 5
Variation in the trill component between males singing the G dialect, which is defined by repetition of a single element in the introductory
component (a–f), together with the trill components of two of their neighbors that sang the H dialect, which is defined by a single low-frequency
introductory element, followed by two phrases defined by repetition of distinctive elements (g–h). The songs are truncated at 2 s, removing all or
part of the concluding chatter.
Figure 4
The average nearest distance between song posts of males observed
in our study contrasted with the results of 10 000 simulations where
the position of males was allocated randomly within their territory.
506 Behavioral Ecology
by guest on May 18, 2011beheco.oxfordjournals.orgDownloaded from
pointed out that we have not previously described the actual
process of extrapair copulation (Westneat and Stewart 2003).
During the collection of our dawn recital data in the 2004/
2005 breeding season, we witnessed the arrival of four foray-
ing females in 47 mornings of recording (8.5%), which is
close to our overall estimate of a 14.5% visitation rate, par-
ticularly as some of the males in each cluster would not be
visited by females. In each case, the female flew to a perch on
or near the branch from which the male was displaying. The
male then started the characteristic blue–black display used
in day-time courtship. The female moved slowly toward the
male, and in at least one case approached him to the point
their bills touched. This precipitated immediate copulation,
or the birds dropped into denser shrubbery, where copula-
tion presumably took place. Three of the copulations were
with a dominant (two early molters, one not) and one with
a subordinate.
DISCUSSION
Despite being confined to year-round territories, male superb
fairy-wrens aggregate physically during the dawn chorus, which
is when female fairy-wrens seek the extragroup fertilizations that
dominate paternity. Each cluster shares a distinctive dialect of
a song used in intersexual advertisement, although at an indi-
vidual level, males know more than one dialect. Therefore, sev-
eral males singing similar recitals in close proximity confront
females visiting their preferred male. The clusters she encoun-
ters appear to be competitive rather than cooperative, as dom-
inant males impose physical punishment on their subordinates.
Despite this resistance from dominants, both subordinates and
neighbors appear able to parasitize the attractiveness of high-
quality birds in the cluster, and gain one-third of the fertiliza-
tions from females whose preferred sire can be identified. This
supports the proposition that aggregation represents satellite
behavior on the part of lower-quality males and that year-round
territoriality provides no impediment to the applicability of
Wagner’s (1998) hypothesis of the hidden lek.
Participation in the dawn chorus by males is independent of
social status or male quality. Although males differ by many
Figure 7
The probability that a given number of females will visit a dialect
cluster of males singing during the dawn chorus on any morning.
The bars represent the observed probability. The line represents the
expectations from a Poisson distribution given the observed mean of
0.21 visits per cluster per morning.
Figure 8
The effect of the size of a dialect cluster on visits by foraying females
to each cluster during the 2004/2005 breeding season: (a) total visits
received, with a significant regression depicted by the line; (b) per-
capita visits received. One male sang two dialects, and both he and
the visits he attracted have been assigned 50:50 to the two dialect
clusters involved. The estimates are conservative as the choice of
some females could not be assessed.
Figure 6
The number of female fairy-wrens (from a sample of 66 females)
estimated to be making an extragroup mating foray on each morning
of the 2004/2005 breeding season. The line represents the
prediction of a quadratic regression function.
Cockburn et al. •Hidden leks in superb fairy-wrens 507
by guest on May 18, 2011beheco.oxfordjournals.orgDownloaded from
months in the time that they acquire nuptial plumage, and
such differences are a primary criterion of mate choice (Dunn
and Cockburn 1999; Green et al. 2000; Double and Cockburn
2003; Cockburn, Osmond, and Double 2008), all fairy-wrens
increase dawn singing at about the same time, which is also
close to the time when they are developing cloacal swelling in
readiness for the breeding season (Mulder and Cockburn
1993). The only context we found where dawn singing was
reduced was the tendency for a small proportion of domi-
nant males to refrain from singing when their own mate had
left on an extragroup foray, and for those males that sang to
start the chorus later when their mate was fertile. These ob-
servations reject the hypothesis that males should sing more
vigorously when their own mate is fertile as a form of guard-
ing (Møller 1991).
Despite living on year-round all-purpose territories, male
fairy-wrens aggregated to sing, with some males singing as close
as physically possible to the boundary of their territory. This
creates a number of male clusters within the study site, the dis-
tinctiveness of which was reinforced by the use of a common
trill dialect by all members of the cluster. Here, we evaluate
these observations in the context of the framework proposed
by Ho
¨glund and Alatalo (1995) for the formation of conven-
tional leks, because a primary aim of their synthesis was to
explain why breeding males should aggregate.
Two of the main hypotheses for formation of leks are inap-
plicable in the current case. The hotspot hypothesis proposes
that males aggregate where female home ranges or foraging
routes coincide (Bradbury 1981; Bradbury et al. 1986) and
is supported in some lekking frugivorous birds (Westcott
1994). In contrast, female fairy-wrens have nonoverlapping
territories and do not make extraterritorial forays except to
seek extragroup fertilizations. Alternatively, the black hole
hypothesis suggests that in species where sexually receptive
females are subject to considerable harassment, females gain
an advantage in visiting exclusive male territories, and males
are advantaged because they quickly regain any mates
they have lost due to disruption (Clutton-Brock et al. 1992;
Stillman et al. 1993). To the contrary, in fairy-wrens, females
initiate copulation, visit male territories only briefly and ex-
clusively for the purpose of copulation, and make those visits
so rarely that the appropriate dynamics are unlikely to
emerge in any case.
Males could also be forced to aggregate because suitable dis-
play sites are limited (Parker 1978). In the case of our study
site, there is no shortage of song posts. Territories in both the
natural and plantation vegetation in our study site usually
contain large numbers of shrubs of the appropriate growth
form and size, and this is particularly true for the plantation
areas, where gardening practices increase the suitability of
song posts. Aggregating males could also enjoy protection
from predators, through heightened corporate vigilance, di-
lution of risk, or selfish herd effects (Oring 1982). Several
studies have suggested that choice of song post by birds is
sensitive to the risk of predation (Krams 2001; Duncan and
Bednekoff 2006; Møller et al. 2006; Parker and Tillin 2006).
Clustering of birds that sing loudly from exposed locations
may expose the group collectively to greater risk of predation,
but may allow individuals both better monitoring and dilution
of their own predation risk. We doubt the general relevance of
predation for clustering in fairy-wrens for two reasons. First,
the degree of clustering is probably insufficient to provide
a reliable source of information or dilution. Second, neither
of the preceding hypotheses helps explain why each cluster
adopts a distinctive dialect.
By contrast, a suite of hypotheses concerning the attraction
of females predicts that members of clusters should share a di-
alect. The idea that aggregation into conventional leks occurs
because more males provide a unified stimulus has been dis-
credited on theoretical grounds (Bradbury 1981; Ho
¨glund
and Alatalo 1995). However, it is possible that females may
prefer to mate in aggregations of males in order to facilitate
comparison between them (Alexander 1975; Bradbury 1981).
Males may advertise the size of the aggregation by joint dis-
play. For these ideas to promote male aggregation, we would
expect per capita success to be higher on larger clusters, as is
true for some lekking species. Although larger clusters attract
more female fairy-wrens, males enjoy no per capita advantage
(Figure 8), undermining the applicability of this hypothesis.
Our data strongly suggest that males aggregate and converge
vocally to parasitize the attractiveness of high-quality males.
Dominant males respond to participation of their subordinates
in the dawn chorus by physical attack, and by increasing the
performance rate of their song (Dalziell and Cockburn
2008), suggesting that their interaction is competitive rather
than cooperative. Despite the interference from dominants,
low-quality individuals gain two advantages from clustering.
First, because attractive early-molting males are vulnerable
to mortality just before and at the start of the breeding season
(Cockburn, Osmond, and Double 2008), males can ‘‘inherit’’
their attractiveness, and be the beneficiary of forays by females
unaware that their preferred male has died. Such ‘‘temporal
spillover’’ has been previously noted in lekking species such as
sage grouse (Gibson et al. 1991; Gibson 1992).
Second, low-quality males gain parasitic success by mimick-
ing the behavior of attractive males. The deceptive mimicry hy-
pothesis of dialect formation suggests that low-quality males
gain advantages by singing songs similar to their neighbors
in order to minimize aggression from high-quality birds, or
to obtain access to mates (Payne 1981; Podos and Warren
2007). Microgeographic structuring and rapid temporal
change of dialects has been reported for two groups that dis-
play in leks: hummingbirds (Snow 1968; Wiley 1971; Gonzalez
and Ornelas 2005) and cotingas (Kroodsma 2005; Saranathan
et al. 2007), but data on parentage are not available to dissect
the significance of variation in these species. Our data from
fairy-wrens support the mate access version of the deceptive
mimicry hypothesis. We have shown that the best predictor of
extragroup mating in subordinate males was the attractiveness
of their own dominant (Double and Cockburn 2003). Here
we present new data that support the hypothesis that early-
molting males also lose parentage to neighbors. Indeed,
though all our evidence suggests that females are seeking
early-molting males in the dawn chorus, and that early molt
enjoys strong selection as a consequence (Dunn and
Cockburn 1999; Green et al. 2000; Double and Cockburn
2003; Cockburn, Osmond, and Double 2008), parasitic satel-
lite males appear to garner on average 33% of extragroup
fertilizations. Dominant males do obtain two-thirds of pater-
nity from a cluster that on average contains 5.25 males (62.4
SD, N¼12), which represents substantial mating skew, and
unlike conventional lekking species, this skew is unlikely to be
distorted by processes like mate-choice copying (Gibson and
Ho
¨glund 1992; Westneat et al. 2000), as the probability that
females will be in a position to observe the behavior of other
females is extremely low. Despite this advantage obtained by
attractive males, success through spatial and temporal spill-
over represents an important alternative mating strategy,
and is consistent with the predictions of the hotshot hypoth-
esis (Beehler and Foster 1988).
Collectively, these observations lead us to conclude that de-
spite year-round territoriality, fairy-wrens conform to the hid-
den lek concept espoused by Wagner (1998) and suggest
that comparable effects should be examined in other birds.
Finally, our data raise the possibility that at least some cases
of polyandry arise when females fail to mate with their
508 Behavioral Ecology
by guest on May 18, 2011beheco.oxfordjournals.orgDownloaded from
preferred sire and that such failure is a source of selection on
male behavior. Although lekking systems in birds provide the
most extreme example of pure female preference for male
phenotypes (Bradbury et al. 1985; Wiley 1991), in some cases
multiple parentage is common (Lank et al. 2002). Recent
attempts to explain reproductive success among low-quality
males have contrasted the ability of males to force copulations
or fertilizations with advantages females obtain from mating
with diverse or compatible mates (Lank et al. 2002; Reichard
et al. 2007). However, there is growing evidence for strong
selection on reproductive parasitism by deception in lekking
birds (Jukema and Piersma 2006). Deceptive strategies in turn
make female choice more difficult, increasing the likelihood
of error. The role of perceptual error has not attracted atten-
tion in recent discussions of variability in female mating be-
havior (Jennions and Petrie 2000; Simmons 2005; Kokko et al.
2006), though Johnstone and Earn (1999) modeled the effect
of perceptual error and suggested that it would be most po-
tent in affecting skew when lek size was large. Our data suggest
a more general effect.
FUNDING
Australian Research Council (to A.C.), the Academy of Finland
(to H.K.), an Ecological Society of Australia student research
award and an Anjeli Nathan Memorial Scholarship (to
A.H.D.), and the Special Topics undergraduate program in
the School of Botany and Zoology (to C.J.B., N.R.B., M.L.H.,
and K.W.).
We are grateful to the Australian National Botanic Gardens and the
Animal Experimentation Ethics Committee at ANU for permits, and
to Sarah Stuart-Smith who assisted with recordings.
REFERENCES
Alexander RD. 1975. Natural selection and specialized chorusing be-
havior in acoustical insects. In: Pimentel D, editor. Insects, science
and society: proceedings of a symposium on insects, science and
society. New York: Academic Press. p. 35–77.
Beehler BM, Foster MS. 1988. Hotshots, hotspots and female prefer-
ences in the organization of lek mating systems. Am Nat.
131:229–238.
Bradbury JH. 1981. The evolution of leks. In: Alexander RD, Tinkle
DW, editors. Natural selection and social behaviour. New York:
Chiron. p. 138–169.
Bradbury JW, Gibson RM, Tsai IM. 1986. Hotspots and the dispersion
of leks. Anim Behav. 34:1694–1709.
Bradbury JW, Vehrencamp SL, Gibson RM. 1985. Leks and the una-
nimity of female choice. In: Greewood PJ, Harvey PH, Slatkin M,
editors. Evolution: essays in honour of John Maynard Smith.
Cambridge: Cambridge University Press. p. 301–314.
Clutton-Brock TH, Price OF, MacColl ADC. 1992. Mate retention,
harassment, and the evolution of ungulate leks. Behav Ecol.
3:234–242.
Cockburn A, Osmond HL, Double MC. 2008. Swingin’ in the rain:
condition dependence and sexual selection in a capricious world.
Proc Roy Soc B Biol Sci. 275:605–612.
Cockburn A, Osmond HL, Mulder RA, Double MC, Green DJ. 2008.
Demography of male reproductive queues in cooperatively breed-
ing superb fairy-wrens Malurus cyaneus. J Anim Ecol. 77:297–304.
Cockburn A, Osmond HL, Mulder RA, Green DJ, Double MC. 2003.
Divorce, dispersal and incest avoidance in the cooperatively breed-
ing superb fairy-wren Malurus cyaneus. J Anim Ecol. 72:189–202.
Cockburn A, Sims RA, Osmond HL, Green DJ, Double MC, Mulder
RA. 2008. Can we measure the benefits of help in cooperatively
breeding birds: the case of the superb fairy-wren Malurus cyaneus?
J Anim Ecol. 77:430–438.
Cooney R, Cockburn A. 1995. Territorial defense is the major function
of female song in the superb fairy-wren, Malurus cyaneus. Anim
Behav. 49:1635–1647.
Dalziell AH, Cockburn A. 2008. Dawn song in superb fairy-wrens:
a bird that seeks extra-pair copulations during the dawn chorus.
Anim Behav. 75:489–500.
Double M, Cockburn A. 2000. Pre-dawn infidelity: females control
extra-pair mating in superb fairy-wrens. Proc Roy Soc Lond Ser B
Biol Sci. 267:465–470.
Double MC, Cockburn A. 2003. Subordinate superb fairy-
wrens (Malurus cyaneus) parasitize the reproductive success of
attractive dominant males. Proc Roy Soc Lond Ser B Biol Sci. 270:
379–384.
Double MC, Cockburn A, Barry SC, Smouse PE. 1997. Exclusion prob-
abilities for single-locus paternity analysis when related males com-
pete for matings. Mol Ecol. 6:1155–1166.
Double MC, Dawson D, Burke T, Cockburn A. 1997. Finding the
fathers in the least faithful bird: a microsatellite-based genotyping
system for the superb fairy-wren Malurus cyaneus. Mol Ecol.
6:691–693.
Duncan WJ, Bednekoff PA. 2006. Singing in the shade: song and song
posts of northern cardinals near nesting Cooper’s hawks. Can J
Zool. 84:916–919.
Dunn PO, Cockburn A. 1996. Evolution of male parental care in a bird
with almost complete cuckoldry. Evolution. 50:2542–2548.
Dunn PO, Cockburn A. 1999. Extrapair mate choice and honest sig-
naling in cooperatively breeding superb fairy-wrens. Evolution.
53:938–946.
Fletcher RJ. 2007. Species interactions and population density medi-
ate the use of social cues for habitat selection. J Anim Ecol.
76:598–606.
Fletcher RJ, Miller CW. 2006. On the evolution of hidden leks and the
implications for reproductive and habitat selection behaviours.
Anim Behav. 71:1247–1251.
Formica VA, Gonser RA, Ramsay S, Tuttle EM. 2004. Spatial dynamics
of alternative reproductive strategies: the role of neighbors. Ecol-
ogy. 85:1125–1136.
Gibson RM. 1992. Lek formation in sage grouse: the effect of female
choice on male territory settlement. Anim Behav. 43:443–450.
Gibson RM, Bradbury JW, Vehrencamp SL. 1991. Mate choice in lek-
king sage grouse revisited: the roles of vocal display, female site
fidelity, and copying. Behav Ecol. 2:165–180.
Gibson RM, Ho
¨glund J. 1992. Copying and sexual selection. Trends
Ecol Evol. 7:229–232.
Gonzalez C, Ornelas JF. 2005. Song structure and microgeographic
song variation in wedge-tailed sabrewings (Campylopterus curvipen-
nis) in Veracruz, Mexico. Auk. 122:593–607.
Green DJ, Osmond HL, Double MC, Cockburn A. 2000. Display rate
by male fairy-wrens (Malurus cyaneus) during the fertile period of
females has little influence on extra-pair mate choice. Behav Ecol
Sociobiol. 48:438–446.
Griffith SC, Owens IPF, Thuman KA. 2002. Extra pair paternity in
birds: a review of interspecific variation and adaptive function.
Mol Ecol. 11:2195–2212.
Ho
¨glund J, Alatalo RV. 1995. Leks. Princeton (NJ): Princeton Univer-
sity Press.
Hoi H, Hoi-Leitner M. 1997. An alternative route to coloniality in the
bearded tit: females pursue extra-pair fertilizations. Behav Ecol.
8:113–119.
Jennions MD, Petrie M. 2000. Why do females mate multiply? A review
of the genetic benefits. Biol Rev Cambridge Phil Soc. 75:21–64.
Johnstone RA, Earn DJD. 1999. Imperfect female choice and male
mating skew on leks of different sizes. Behav Ecol Sociobiol.
45:277–281.
Jukema J, Piersma T. 2006. Permanent female mimics in a lekking
shorebird. Biol Lett. 2:161–164.
Kokko H, Jennions MD, Brooks R. 2006. Unifying and testing models
of sexual selection. Annu Rev Ecol Evol Syst. 37:43–66.
Krams I. 2001. Perch selection by singing chaffinches: a better view of
surroundings and the risk of predation. Behav Ecol. 12:295–300.
Kroodsma DE. 2005. The singing life of birds: the art and science of
listening to birdsong. Boston (MA): Houghton Mifflin.
Langmore NE, Mulder RA. 1992. A novel context for bird song: pred-
ator calls prompt male singing in the kleptogamous superb fairy-
wren, Malurus cyaneus. Ethology. 90:143–153.
Lank DB, Smith CM, Hanotte O, Ohtonen A, Bailey S, Burke T. 2002.
High frequency of polyandry in a lek mating system. Behav Ecol.
13:209–215.
Cockburn et al. •Hidden leks in superb fairy-wrens 509
by guest on May 18, 2011beheco.oxfordjournals.orgDownloaded from
Møller AP. 1991. Why mated songbirds sing so much: mate guarding
and male announcement of mate fertility status. Am Nat.
138:994–1014.
Møller AP. 1992. Frequency of female copulations with multiple males
and sexual selection. Am Nat. 139:1089–1101.
Møller AP, Nielsen JT, Garamszegi LZ. 2006. Song post exposure, song
features, and predation risk. Behav Ecol. 17:155–163.
Mulder RA. 1997. Extra-group courtship displays and other reproduc-
tive tactics of superb fairy-wrens. Aust J Zool. 45:131–143.
Mulder RA, Cockburn A. 1993. Sperm competition and the reproduc-
tive anatomy of male superb fairy-Wrens. Auk. 110:588–593.
Mulder RA, Dunn PO, Cockburn A, Lazenby-Cohen KA, Howell MJ.
1994. Helpers liberate female fairy-wrens from constraints on extra-
pair mate choice. Proc Roy Soc Lond Ser B Biol Sci. 255:223–229.
Oring LW. 1982. Avian mating systems. In: Farner DS, King JR, Parkes
KC, editors. Avian biology. New York: Academic Press. p. 1–92.
Parker GA. 1978. Searching for mates. In: Krebs JR, Davies NB, edi-
tors. Behavioral ecology: an evolutionary approach. Cambridge Uni-
versity Press. p. 214–244.
Parker TH, Tillin H. 2006. Male blue tits Cyanistes caeruleus choose
early-leafing tree species during spring dawn chorus. Bird Stud.
53:253–257.
Payne RB. 1981. Population structure and social behaviour: models
for testing the ecological significance of song dialects in birds. In:
Alexander RD, Tinkle DW, editors. Natural selection and social
behavior: recent research and new theory. New York: Chiron.
p. 108–120.
Peters A. 2000. Testosterone treatment is immunosuppressive in su-
perb fairy-wrens, yet free-living males with high testosterone are
more immunocompetent. Proc Roy Soc Lond Ser B Biol Sci.
267:883–889.
Peters A, Astheimer LB, Boland CRJ, Cockburn A. 2000. Testosterone
is involved in acquisition and maintenance of sexually selected male
plumage in superb fairy-wrens, Malurus cyaneus. Behav Ecol Socio-
biol. 47:438–445.
Peters A, Astheimer LB, Cockburn A. 2001. The annual testosterone
profile in cooperatively breeding superb fairy-wrens, Malurus cya-
neus, reflects their extreme infidelity. Behav Ecol Sociobiol.
50:519–527.
Podos J, Warren PS. 2007. The evolution of geographic variation in
birdsong. Adv Stud Behav. 37:403–458.
Ramsay SM, Otter K, Ratcliffe LM. 1999. Nest-site selection by female
Black-capped Chickadees: settlement based on conspecific attrac-
tion? Auk. 116:604–617.
Reichard M, Le Comber SC, Smith C. 2007. Sneaking from a female
perspective. Anim Behav. 74:679–688.
Safran RJ. 2004. Adaptive site selection rules and variation in group
size of barn swallows: individual decisions predict population pat-
terns. Am Nat. 164:121–131.
Safran RJ. 2007. Settlement patterns of female barn swallows Hirundo
rustica across different group sizes: access to colorful males or fa-
vored nests? Behav Ecol Sociobiol. 61:1359–1368.
Saranathan V, Hamilton D, Powell GVN, Kroodsma DE, Prum RO.
2007. Genetic evidence supports song learning in the three-wattled
bellbird Procnias tricarunculata (Cotingidae). Mol Ecol. 16:
3689–3702.
Simmons LW. 2005. The evolution of polyandry: sperm competition,
sperm selection, and offspring viability. Annu Rev Ecol Evol Syst.
36:125–146.
Snow DW. 1968. The singing assemblies of little hermits. Living Bird.
7:47–55.
Stillman RA, Clutton-Brock TH, Sutherland WJ. 1993. Black holes,
mate retention, and the evolution of ungulate leks. Behav Ecol.
4:1–6.
Tarof SA, Ratcliffe LM, Kasumovic MM, Boag PT. 2005. Are least
flycatcher (Empidonax minimus) clusters hidden leks? Behav Ecol.
16:207–217.
Wagner RA. 1998. Hidden leks: sexual selection and the clustering of
avian territories. In: Parker PG, Burley NT, editors. Avian reproduc-
tive tactics: female and male perspectives Ornithological Mono-
graphs No 49. Lawrence (KS): Allen Press. p. 123–145.
Wagner RH. 1993. The pursuit of extra-pair copulations by female
birds: a new hypothesis of colony formation. J Theor Biol. 163:
333–346.
Westcott DA. 1994. Leks of leks: a role for hotspots in lek evolution?
Proc R Soc Lond B. 258:281–286.
Westneat DF, Sherman PW. 1997. Density and extra-pair fertiliza-
tions in birds: a comparative analysis. Behav Ecol Sociobiol.
41:205–215.
Westneat DF, Stewart IRK. 2003. Extra-pair paternity in birds: causes,
correlates, and conflict. Annu Rev Ecol Evol Syst. 34:365–396.
Westneat DF, Walters A, McCarthy TM, Hatch MI, Hein WK. 2000.
Alternative mechanisms of nonindependent mate choice. Anim
Behav. 59:467–476.
Wiley JH. 1991. Lekking in birds and mammals: behavioral and evo-
lutionary issues. Adv Stud Behav. 20:201–291.
Wiley RH. 1971. Song groups in a singing assembly of little hermits.
Condor. 73:28–35.
510 Behavioral Ecology
by guest on May 18, 2011beheco.oxfordjournals.orgDownloaded from
