lable at ScienceDirect

Animal Behaviour 77 (2009) 803–812

Contents lists avai

Animal Behaviour

journal homepage: www.elsevier .com/locate/yanbe

Behavioural syndromes and trappability in free-living collared flycatchers,Ficedula albicollis

Laszlo Zsolt Garamszegi a,*, Marcel Eens a, Janos Torok b,1

a Department of Biology, University of Antwerpb Behavioural Ecology Group, Department of Systematic Zoology and Ecology, Eotvos Lorand University

a r t i c l e i n f o

Article history:Received 12 September 2008Initial acceptance 24 October 2008Final acceptance 17 December 2008Published online 1 February 2009MS. number: 08-00585R

Keywords:adaptationcollared flycatchercoping styleFicedula hypoleucapersonalitytemperamenttrappability

* Correspondence: L. Z. Garamszegi, Department owerp, Campus Drie Eiken, Universiteitsplein 1, B-2610

E-mail address: laszlo.garamszegi@ua.ac.be (L.Z. G1 J. Torok is at the Department of Systematic Zoology

University, H-1117 Budapest, Hungary.

0003-3472/$38.00 � 2009 The Association for the Studoi:10.1016/j.anbehav.2008.12.012

The concept of behavioural syndromes hypothesizes that consistent behaviours across various situationsmediate important life history trade-offs, and predicts correlations among behavioural traits. We studiedthe consistency of behavioural responses across three ecological situations (exploration of an environ-ment altered with a novel object, aggression towards conspecifics, risk taking) in male collaredflycatchers. We developed behavioural tests that could be applied in the birds’ natural habitat, thus notrequiring the capture of animals. Across individuals, we found positive covariation between exploration,aggression and risk taking, but the magnitude of these relationships varied. Variation in behaviour wasalso related to capture probability. Exploratory and risk-taking individuals were more likely to entera trap than individuals with averse characteristics. Moreover, with the trapped birds, there was anassociation between the time needed for successful capture and exploration, and we found strongercorrelations between behaviours in comparison with effects calculated from the whole sample of indi-viduals. These patterns were independent of territory quality, male age, condition and breeding expe-rience. Consequently, behavioural responses to different ecosocial challenges are determined byindividual-specific characteristics that are manifested in correlative behaviours. Hence, behavioural typesmay be potential subjects for reproductive and life history adaptations. Our results have importantimplications for field studies of animals, because they suggest that capturing protocols may not randomlysample the observed population.� 2009 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.

Behavioural syndrome (sometimes referred to as personality,temperament or coping style in the human and ecological litera-ture, terms that we avoid using here to circumvent confusingterminologies) is defined as the consistency of behaviouralresponses that individuals display in different situations (Sih et al.2004a, b). Correlated behaviours respond to selection pressuressimultaneously, which may determine how animals generally copewith challenges in their physical and social environment. Recently,the phenomenon has been recognized as ecologically and evolu-tionarily relevant in many animals, because consistent individualvariation in a suite of behavioural traits may drive important lifehistory trade-offs (e.g. Hedrick 2000; Bell 2005; Sinn & Moltscha-niwskyj 2005; Duckworth 2006; Bell & Sih 2007; Duckworth &Badyaev 2007; Wolf et al. 2007; Smith & Blumstein 2008). Forexample, one response may be advantageous in a given context,while its correlated response in another context may involve costs(e.g. aggressive individuals may achieve high social dominance on

f Biology, University of Ant-Antwerp, Belgium.

aramszegi).and Ecology, Eotvos Loraand

dy of Animal Behaviour. Publishe

the one hand, but may be subject to frequent predation on theother; Sih et al. 2004a, b). Behavioural types, at least in somespecies, have been linked to mating and/or reproductive successand/or survival via integrated component traits such as dispersal,parental quality and/or dominance, and may thus have conse-quences for fitness (Reale et al. 2007).

Birds are important models in this line of research, becauseextensive field data on their behavioural ecology are availableThese data can be fruitfully amalgamated with the concepts ofbehavioural syndromes (Groothuis & Carere 2005). In fact, the mostcomprehensive study of the adaptive significance of correlatedbehaviours comes from studies on a single species, the great tit,Parus major (Drent 2006). Initial investigations demonstrated thatsubject animals cope with novel objects and environments in anindividual-specific manner. Such individual variation remainedconsistent across experimental situations, and predicted the degreeof aggression (Verbeek et al. 1996). Subsequently, artificial selectionexperiments in combination with field studies separated differentvariance components for behavioural types, and showed that theseare significantly heritable and genetically correlated (Dingemanseet al. 2002; Drent et al. 2003; Carere et al. 2005; van Oers et al.2005). A recent study revealed a relationship between individual

d by Elsevier Ltd. All rights reserved.

L.Z. Garamszegi et al. / Animal Behaviour 77 (2009) 803–812804

variation in behaviour and genetic polymorphism in a neurotrans-mitter-associated gene (Drd4), which suggests genetic contribu-tions to behavioural types (Fidler et al. 2007). The performance ofanimals in behavioural tests in captivity predicted their realizedfitness in natural conditions, as exploration scores were correlatedwith foraging success, dominance and dispersal, with conse-quences for reproductive success and survival (Dingemanse et al.2003; Dingemanse & de Goede 2004; Both et al. 2005).

Although the success of these studies had a remarkable influ-ence on our understanding of the evolutionary role of behaviouralsyndromes in birds, the focus on a single wild model species limitsgeneralizations. Great tits occupy a given ecological niche, showparticular behaviours and face species-specific evolutionaryconstraints that shape their life history decisions. Hence, selectionfactors in association with correlated behaviours may be relevantunder the conditions that the species experiences in its environ-ment. On the other hand, the adaptive significance of behaviouralsyndromes found in great tits is not necessarily applicable to otherspecies. There is tremendous interspecific variation in how avianspecies explore novel objects (Mettke-Hofmann et al. 2002), reactto human approach (Blumstein 2006; Møller et al. 2008) and showfeeding innovations (Lefebvre et al. 1997; Sol et al. 2002), whichmay all represent behavioural type axes, and also appear to relate tointerspecific differences in several ecological factors. Moreover,species may differ not only in average behavioural responses butalso in their frequency distributions which are suited to the givenenvironment (Dingemanse 2003). This increases the importance ofstudying behavioural syndromes in the wild on other modelspecies with a different ecology.

The collared flycatcher is another hole-nesting passerine fromthe western Palaearctic (Cramp & Perrins 1994), which displayselaborate behavioural patterns. This species has been intensivelystudied for its life history and sexual selection (e.g. Gustafsson1989; Part et al. 1992; Gustafsson et al. 1995; Qvarnstrom et al.2000; Michl et al. 2002; Torok et al. 2003; Garamszegi et al. 2004a).Therefore, in addition to the great tit model, the collared flycatchercan also serve as a good subject for studying the evolution ofbehavioural syndromes in an evolutionary context, becausepatterns of individual variation in behaviour can be related to well-described reproductive or ecological traits. However, the breedingbiology of the flycatchers differs conspicuously from that of themembers of the Paridae family (Lundberg & Alatalo 1992; Gosler1993), and thus roles for behavioural types in mediating trade-offswill not necessarily be the same. For example, flycatchers aremigratory and are confronted with an unpredictable environmentupon arrival at the breeding grounds, which may have conse-quences for exploration (Mettke-Hofmann et al. 2005a, b; Mettke-Hofmann 2007). Moreover, differences in diet can also shapespecies-specific behavioural types that help to gather informationabout the environment (Mettke-Hofmann et al. 2002). Such adap-tations may shift the optimal value of different behaviours and theircorrelates, which can have consequences for the role of individualvariation. In a previous study, we found that individuals may showconsistent behavioural responses over time, as the degree ofaggression was significantly repeatable at the within-individuallevel (Garamszegi et al. 2006).

Our main goal in this study was to test for the existence ofconsistent individual behavioural performance across ecologicalsituations in accordance with the definition of behaviouralsyndromes (Sih et al. 2004a, b). In our long-term study of severalbehavioural traits, such as song (e.g. Garamszegi et al. 2004a),copulation (Michl et al. 2002) and risk taking (Michl et al. 2000), wewere faced, on several occasions in the field, with the fact that thereare strong individual differences between birds. To initiate theintegration of behavioural types within ecological and evolutionary

studies in this model species, we established a simple protocol tocharacterize individual variation in behaviours without the needfor capturing animals (see details below). Our approach permittedus to measure behavioural responses to ecologically relevant situ-ations in natural conditions while causing minimal disturbance.Importantly, our experimental scheme did not require thesuccessful trapping of individuals, showing that our sample was notinflated by heterogeneity in trappability. First, we tested fora consistency in individual behavioural performance in three situ-ations: exploration of a breeding environment altered with a novelobject, territorial aggression and risk taking when a humanapproached. We predicted that these traits would covary positivelyacross individuals, if consistent individual variation persists acrossdifferent behavioural traits according to the behavioural syndromeconcept (Sih et al. 2004a, b). Second, we tested the prediction thatbehavioural types have consequences for trappability, as lessexplorative animals may show consistent trap-averse behaviourand may be more difficult to capture. This link has been prevalent inthe literature (e.g. Wilson et al. 1993; Mills & Faure 2000; Realeet al. 2000; Malmkvist & Hansen 2001), but basically remains anuntested assumption because of the difficulty of comparingcaptured and noncaptured individuals. However, as we attemptedto capture individuals after testing, we were able to assign trapp-ability to all tested individuals and to explore the relationshipsbetween different behavioural traits and capture probability.

METHODS

General Methods

The collared flycatcher is a small migratory, hole-nestingpasserine that is socially monogamous with facultative polygyny(Cramp & Perrins 1994). After arriving at the breeding sites, malesimmediately occupy nestboxes or natural cavities, and establishterritories where they start singing and displaying. Females chooseamong them, build nests alone and lay and incubate six or seveneggs. Both sexes provide parental care at the nestling stage. Afterfledging, birds start to prepare for migration, and leave for the sub-Saharan wintering site in early autumn. We established breedingplots at Pilis Field Station near Budapest (47�430N, 19�010E),Hungary in 1981 for the long-term study of the species (see Torok &Toth 1988). Fieldwork for the current study was carried out during2007, when we recorded the behaviour of unpaired males asdescribed below. We chose males for our purposes because theyshow typical nest presentation behaviour, and their elaboratecourtship and territorial behaviour can be well characterized indifferent situations (see below). In addition, during their displayperiod we were able to design protocols that enabled themeasurement of traits without requiring the capture of individuals.This was necessary for both practical and ethical reasons. In a pilotstudy, we found that individuals become very stressed in captivity,thus making it difficult to assess behavioural types in cage or aviaryconditions by using the protocols that have been developed for thegreat tit.

Behavioural Traits

In accordance with Reale et al. (2007), we established experi-mental test conditions in which we characterized three behaviouraltraits in three ecological situations, with consequences for threedifferent trade-offs. The three ecological circumstances werealtered habitat, social challenge and the presence of a potentialpredator, in which we measured exploration, aggressivenessand risk taking, respectively, as elements of correlated behaviours.

L.Z. Garamszegi et al. / Animal Behaviour 77 (2009) 803–812 805

All behavioural observations were made by a single experimenter(L.Z.G).

Exploration of a modified breeding environmentWe actively monitored the breeding area on a daily basis to find

newly arrived males. Following our standard field protocol, weassumed that males displaying at a previously unoccupied nestboxhad arrived on the date of the experiment. If a male was foundsinging in the vicinity of an unoccupied nestbox, we presenteda caged female in its territory to mimic a natural female visit. Weplaced the decoy on the top of the nestbox, from where femalesusually observe nest-presenting males in a natural situation.Moreover, by doing so we could easily judge whether the observedbox belonged to the male seen in its close proximity. If the male inquestion responded to the female stimulus by commencing thetypical presentation displays (Lundberg & Alatalo 1992), weconsidered that the male was the owner of the nest, and wecontinued the experimental session. The nestbox presentationconsists of frequent approaches to the nestbox, by which the resi-dent male calls the female to enter the nestbox and to evaluate itsquality. During nest presentation, males excitedly fly around thenestbox, while repeatedly landing on the top and entrance hole.They may occasionally enter the box and call the female frominside. We assumed that the activity level of this behaviour reflectsthe male’s investment in the current breeding situation (i.e. givenfemale, given territory, given nestbox), and it can be translated intoan advantage in terms of mating success, as less intense presen-tation may result in the female leaving and the loss of a matingopportunity.

According to our long-term field experience, nonrespondingmales are likely to occupy natural holes in the neighbourhood ormay be polyterritorial males seeking a secondary nest. Altogether,we presented females to 41 potential males, of which 33 responded.

To discriminate between general activity and exploratorybehaviour (see Reale et al. 2007) and to control for the fact thatdifferent males may show different reactions towards differentfemales, or respond on the basis of territory quality, we firstrecorded the nest presentation behaviour of territorial maleswithout changing their breeding environment (control activity). Toavoid habituation, we then removed the female stimulus and leftthe male undisturbed for at least half an hour, during which itnormally stopped the typical nestbox presentation and becameengaged in vocal advertisement (Garamszegi et al. 2008). In thisinterval, we continued the observation of males to verify that nonatural female visits occurred and that they remained in theirterritory. If these criteria were met (26 cases out of 33), wemanipulated the breeding environment and measured the samebehavioural traits under the altered conditions by using the samefemale stimulus (experimental activity). The experimentalbreeding situation was established by attaching a white paper sheeton the front of the nestbox, just below the entrance hole. Weinferred that birds had not previously exploited such an artificialbut potential breeding environment (as they breed in conventionalnestboxes or natural cavities). Therefore, we assumed that theresponses of males to the modification achieved by the white paperwould reflect how they cope with altered breeding opportunities. Itwas evident in the field that individuals responded differently tothe treatment. We expressed exploration as the experimentalactivity relative to the control activity, that is, by using each male asits own control; thus stimulus and territory effects were inherentlycontrolled (see below). Following the recommendations of Realeet al. (2007), we systematically avoided adopting neophilia/neo-phobia terminologies in association with behaviours in response toa new situation, as used by some studies (Corey 1978; Greenberg1990; Mettke-Hofmann et al. 2002). This is because from the

evolutionary perspective, exploration/avoidance is the mainsubject of selection and it includes reactions to any novelty (newhabitat, new food and novel objects) by definition (Reale et al.2007). On the other hand, neophilia or neophobia would deal withthe mechanisms responsible for individual variation in behaviour(Greenberg & Mettke-Hofmann 2001).

We made an effort to use different females for tests of differentmales, as the identity and attributes of stimuli may be important(see Garamszegi et al. 2006). Hence, we used 17 females randomly,as they became available from our parallel capturing sessions. Onefemale was used a mean � SE of 1.52 � 0.25 times. However, giventhe pairwise design, we always used the same female in thesubsequent sessions with the same male to make sure that we onlymeasured responses to the change in the breeding environment.

We measured the following nestbox presentation behaviouraltraits in both the control and experimental situations duringa 10 min observation of each. The measurements were initiatedafter the experimenter had ascertained that the individualresponded to the female stimulus by commencing the typicalnestbox presentation. We counted the landings on the top of thefemale cage, nestbox and entrance hole and measured the time (s)that the male spent on the cage, entrance hole or within the boxand the time (s) between the appearance of the resident male on itsterritory (i.e. the detection of the female as indicated by the typicalnestbox presentation behaviour) and the first landing on theentrance hole of the nestbox (we expected that males that wereaverse to the new situation would avoid landing on the hole thatwas just above the white page). These measures were stronglycorrelated with each other (Pearson correlation: number of land-ings versus duration of landings: control situation: r31 ¼ 0.750,P < 0.001; experimental situation: r24¼ 0.736, P < 0.001; numberof landings versus latency to land: control situation: r31 ¼ �0.551,P < 0.001; experimental situation: r24¼ �0.734, P < 0.001; dura-tion of landings versus latency to land: control situation:r31 ¼ �0.637, P < 0.001; experimental situation: r24 ¼ �0.600,P ¼ 0.001). Therefore, those males that displayed intense nestboxpresentation behaviour can be characterized by immediateapproach and frequent and long visits to the female/nestbox. Thiscan be reflected by a single axis that we computed by a principalcomponents analysis on correlations that explained 76.5–79.4%(control versus experimental situation) of the variance in nestbox-presenting behaviour. The two components reflecting the twosituations had similar factor structure (number of landings: controlsituation: 0.582; experimental situation: 0.600; duration of land-ings: control situation: 0.604; experimental situation: 0.565;latency to land: control situation: �0.544; experimental situation:�0.565).

Males in the experimental situation were generally less activethan in the control situation. In particular, they consistently madefewer visits (paired t test: t25 ¼ �4.077, P < 0.001; Fig. 1a), spentless time at the nestbox (t25 ¼ �4.000, P < 0.001; Fig. 1b), andlanded on the nest hole later (t25 ¼ 3.964, P < 0.001, as they flut-tered several times in front of the hole, clearly avoiding landing onit; Fig. 1c) implying that at least some males were reluctant toexploit the altered environment, while others displayed behav-ioural patterns similar to the control situation. There was a strongcorrelation between the principal components of nest presentationbehaviours expressed separately for the control and experimentalsituations (r24¼ 0.805, P < 0.001). This indicates that ourmeasurements reflected differences in individual activity levels, orthe same females on the same territory elicited similar responsesfrom the same males. Moreover, the strong correlation acrossdifferent trials also suggests that activity can be measured ina repeatable and reliable manner. To remove effects of activity, andto estimate the degree of response to the altered environmental

50

45

40

35

30

Nu

mbe

r of

lan

din

gsD

ura

tion

of

lan

din

gs (

s)T

ime

to l

and

(s)

25

20

15

10

5

0

500

400

300

200

100

0

500

600

700

400

300

200

100

0Control 1 Experiment

Control 1 Experiment

Control 1 Experiment

Control 2

(a)

(b)

(c)

L.Z. Garamszegi et al. / Animal Behaviour 77 (2009) 803–812806

situation, we calculated residuals from a linear regression betweenthe two principal components summarizing nest-presentingactivity in different situations. We then used these residuals toreflect exploration that is independent of activity, female andterritory quality with high values indicating more explorativebehaviours in a breeding opportunity with a novel object thanlower values. This trait may be relevant in the trade-off situation, inwhich birds have to balance the potential costs and benefits asso-ciated with altered environments.

To eliminate fully the possibility that males responded differ-ently to the experimental situation because of a different degree ofhabituation and not exploration, in some cases (N ¼ 19) we estab-lished a second control situation. We removed the white sheet andplaced a dummy mistle thrush, Turdus viscivorus, approximately0.5–1 m from the nestbox and used the same female as a stimulus.We measured the time taken to land on the hole and compared itwith both the first control and experimental situations. The timetaken measured at the second control could not be discriminatedstatistically from what was measured in the first control (paired ttest: t18 ¼ �0.341, P ¼ 0.737; Fig. 1c), while it was consistentlyshorter than during the experimental situation (t18 ¼ �2.159,P ¼ 0.045; Fig. 1c). Therefore, males resumed their original activityafter the experimental situation and should have maintained theirlevel of response to the female throughout the whole session. Thissupports our assumption that the effects we measured during theexperimental situation were caused by exploration and nothabituation.

Male–male aggressionAfter the female tests, we quantified the intrasexual aggressive

behaviour of the resident males towards territory intruders. Weexposed the focal males to a live stimulus male bird in a small wirecage that was placed 2–3 m from the nestbox and ca. 0.5 m abovethe ground (simulating a natural situation), and measured the time(s) between the detection of the stimulus male and the first attack(see Garamszegi et al. 2006 for further details). Detection wasdefined as above, while an attack was when the resident maletouched the cage of the stimulus male, usually showing clearintention to fight. Based on previous observations on the knownvariance in this behavioural trait (Garamszegi et al. 2006), if theresident male did not attack the decoy bird within 5 min, it wastreated as nonaggressive, and we assigned an attack latency scoreof 301 s to them (N ¼ 4 cases). In our previous study (Garamszegiet al. 2006), we showed that the latency to attack is a goodpredictor of the territorial behaviour in general, because it predictsthe frequency and duration of subsequent attacks. Moreover,latency of the first attack appeared to be an individual-specificattribute, as, other things being equal, the same male respondedsimilarly in different experimental situations, that is, the trait wasrepeatable within males. We also changed the identity of thechallenger males to control for its confounding effect on the resi-dent’s behaviour, which appeared to be important (Garamszegiet al. 2006). We used 10 stimulus males, which were useda mean � SE of 2.5 � 0.45 times. When we measured aggression,we assessed how individuals solved the trade-off between nestdefence and other territorial activities and most importantly femaleattraction which males do when no intruder is present.

Figure 1. Elements of nestbox-presenting behaviour in different breeding environ-ments in collared flycatcher males in Hungary. (a) Number of all landings on the nestbox(top or hole) during 10 min of observation, (b) summed duration of landings and (c) thelatency to land on the hole of the nestbox after the detection of a female. Control 1: onlyfemale stimulus; experiment: female stimulus þwhite paper; control 2: female stim-ulus þ dummy mistle thrush.

L.Z. Garamszegi et al. / Animal Behaviour 77 (2009) 803–812 807

Risk taking at human approachThe distance at which an individual flees from a potential

predator represents a measure of risk taking (Blumstein 2006;Møller et al. 2008). To estimate the risk taking of our subjects, weadopted a standard technique developed by Blumstein (2006).After estimating the latency of attack in the above male decoyexperiment, we allowed males to continue the territory-defendingbehaviour, in which they frequently flew onto the top of the pre-sented male’s cage. When the focal individual was on the cagewhile showing clear intention to fight, the observer moved ata normal walking speed towards it (from a distance of ca. 25–30 m).When the individual took flight from the cage because of thehuman disturbance, the distance between the position of theobserver and the experimental cage was measured as the numberof steps (one step m 1 m). This distance was recorded as the flightdistance. Low numbers for this variable mirror high risk taking,because birds with a short flight distance allow potential predatorsto approach them closely. Møller et al. (2008) showed that flightdistance can be assessed with high consistency by differentobservers (r ¼ 0.79). This situation reflects a trade-off betweenterritorial defence and predator avoidance.

Trappability

After making the behavioural observations, we attempted tocapture the resident males for identification and for standardmeasurements. We set up a conventional nestbox trap directly afterthe male challenge experiments, when males were present on theirterritory; thus they witnessed a human observer manipulatingtheir occupied nestbox. The nestbox trap consisted of an aluminiumdoor with a spring device, placed inside the nestbox. The target birdwas caught automatically when it entered the nestbox. The personcapturing did not wait to see the capture event, but checked thenestbox every 15 min. Because focal birds could have seen parts ofthe trap from the outside, capture success could reflect the bird’sreaction to both the human manipulation and the presence ofa novel object inside the nest. However, we predicted that bothreactions can be related to a behavioural syndrome, as they involverisk taking (reaction to human) or exploration (reaction to novelty).Note that we are primarily interested in ecological relevance andnot in the underlying mechanisms. We recorded the time at trapactivation and at successful capture to calculate the time needed forsuccessful capture. Our capture effort was successful in 19 of 26cases. To estimate trappability, we used a bivariate measure (trap-ped/nontrapped) and a continuous measure (time needed forsuccessful capture) to deal with different resolutions.

Captured males were measured for standard morphologicaltraits. We used information on age, wing length and body mass.Based on the typical subadult plumage coloration of yearling males(brown remiges), we classified individuals as yearlings or olderbirds (Svensson 1984). We measured wing length with a ruler(�1 mm) and used a Pesola spring balance for weighing birds(�0.1 g). For the captured birds, we could assess breeding experi-ence based on our long-term survey data by counting the breedingattempts that we had detected in our study plots prior to 2007.Yearlings and males that had not been captured earlier wereconsidered to be inexperienced breeders and to have had noprevious breeding attempt in our nestbox plots.

Statistical Analyses

Variables were checked for normality and homoscedasticity,and statistical transformations were applied if it was necessary tomeet the parametric criteria. The distribution of data on aggression,even after transformation, was slightly skewed (skewness: 1.041).

Moreover, we assigned an arbitrary latency score to some individ-uals that did not attack the stimulus bird. To remove potentialerrors caused by improper distribution of aggression data, weanalysed attack latency by using nonparametric approaches.

To determine the strength and direction of the relationshipbetween the investigated traits, we estimated the correlationcoefficients between them (in terms of Pearson r or Spearman rS).Because interpretations relying on significance levels aremisleading for conceptual and philosophical reasons, this paperfollows interpretations based on the effect size theorem (Nakagawa2004; Garamszegi 2006; Nakagawa & Cuthill 2007; Stephens et al.2007). Accordingly, we focus on the magnitude of the biologicalrelationships as reflected by effect size, and on the precision bywhich these effects could be estimated from the current sample interms of 95% confidence intervals (CI). To describe the strength ofa relationship from nonparametric tests, effect sizes and CIs can beobtained by treating Spearman rS as Pearson r (Cohen 1988). Fordemonstrative purposes, we present P values for each relationship(as given by the corresponding analysis). We avoid emphasizingstatistical significance, but bring attention to the biologicalimportance of effects while anticipating the uncertainty resultingfrom the data.

As variation in behaviour may also depend on factors other thanindividuality, we considered some potentially confounding vari-ables that may affect a relationship. For example, aggression ispartially determined by the attributes of the challenger and the ageof the focal bird (Garamszegi et al. 2006). Moreover, territoryquality can mediate a relationship between behavioural traits,because males may invest more in the defence and exploration ofterritories of superior quality than they may do at lower-qualitynestboxes. Similarly, males in good condition may be more likely tohave the physical abilities to cope with the energetic needs oftaking greater risks or becoming more aggressive. Finally, previousbreeding experience may also confound our results in relation totrappability, because birds that had already witnessed a capture inprevious years might have been more averse to the trap. Therefore,we controlled for male age, territory quality, body condition andbreeding experience. We estimated territory quality from the dateof the experiment, because early arriving males are likely to obtainhigher-quality territories (Lundberg & Alatalo 1992); thus theywould perceive the territory as a more valuable breeding oppor-tunity than later males. Given the standard and intensive moni-toring of our study site, we assumed that the date on which thefocal male was first seen and experimented upon reflects the arrivaldate of the male. Note that a statistical control for male age mayalso eliminate some of the unwanted variation in territory quality, ifolder males are more likely to obtain better territories than youngermales. We assessed body condition based on the residuals from thewing/body mass regression (F1,18 ¼ 12.885, P ¼ 0.002;slope � SE ¼ 2.120 � 0.591). The confounding variables wereadded one by one to the model testing for the covariance of traits ofinterest. We estimated the effect size for the relationship in ques-tion while controlling for the confounding traits in a multivariatetest (partial correlation or generalized linear model, GLM). Theidentity of the decoy bird was controlled in a GLM design, in whichit was entered as a random factor.

Ethical Note

The current study has implications for animal welfare, as itsuggests that studies of behavioural syndromes at the populationlevel can be designed without the need to capture a large number ofindividuals, thus avoiding causing them unnecessary stress incaptivity assays. However, the field tests require the use of testanimals as stimulus birds. The study was conducted under licence

30

(b)

25

20

15

10Figh

t d

ista

nce

(m

)R

anke

d l

aten

cy t

o fi

ght

Ran

ked

lat

ency

to

figh

t

5

0

30

25

20

15

10

5

0

30

25

20

15

10

5

0

-3 -2 -1 0 1 2 3

-3 -2 -1 0 1 2 3

0 5 10Flight distance (m)

Exploration of novel breeding environment

Exploration of novel breeding environment

15 20 25 30

(a)

(c)

L.Z. Garamszegi et al. / Animal Behaviour 77 (2009) 803–812808

from the Middle-Danube-Valley Inspectorate for EnvironmentalProtection, Nature Conservation and Water Management. Wecaught all birds with conventional nestbox traps that are widelyused in long-term population studies. During the trials, stimulusbirds were presented in small wire cages (15 � 20 cm and 15 cmhigh). These experimental cages had narrow gaps between thewires which minimized the possibility that resident males couldharm the decoy birds with their beaks, or that the decoy birds couldhurt themselves while moving in the cage. However, we could noteliminate all potential physical contact between the test birds.Therefore, prior to the experiments, we had decided to interruptthe experimental tests if the decoy bird became hurt or seriouslydistressed, but at no time did we have to intervene in the experi-ment. The general response of the decoy birds to the experimentalsituation was that they tried to avoid physical contact with theresident male by moving to the opposite side of the cage. We choseto run female presentation tests for 10 min because this timewindow allowed us to obtain information on the responses of theresident males over both the short and the long term. The strongcorrelation between latency to the first landing (short-termresponse) and the number and duration of visits (long-termresponse) implies that in future, we can use shorter time windows,because the short-term response reflects long-term responses well.Since such information for assays using caged males was alreadyavailable (Garamszegi et al. 2006), we only used short-termmeasures of aggression. When kept overnight, birds were housed inlarger cages (40 � 24 cm and 40 cm high), and covered with clothto minimize stress. For ethical and practical reasons, each stimulusbird was used only a few times in the experiments (see ranges andmeans above). We supplied caged birds with mealworms and waterad libitum, and released them at the site of capture after use,verifying their good condition prior to release (i.e. they showedactive behaviour in the cage and a visual inspection revealed thatthey accumulated some body fat). None of the stimulus birds diedin captivity. We always minimized time in captivity for stimulusbirds, but in a few cases our schedule was delayed by unfavourableweather. They spent a maximum of 3 days in captivity. Moststimulus birds were released on the day following capture. Themajority of the stimulus birds (19 of 27) were observed breedingafter testing, which is similar to what we observe for birds that arecaptured during courtship and released immediately (38 of 69 fromGaramszegi et al. 2004a). The body mass of stimulus birds duringbreeding was not distinguishable statistically from the populationmean during the same period (t717 ¼ 0.153, P ¼ 0.878). Thirteendecoy individuals returned to breed in 2008, which is comparablewith the recapture probability obtained from a mark–recaptureanalysis of long-term data (37.2–56.8% from Garamszegi et al.2004b).

RESULTS

Correlated Behavioural Responses

We predicted that if individuals can be categorized alonga general behavioural type spectrum, they should display similarbehavioural responses in different situations. We found thatcollared flycatcher males with a high degree of exploration of an

Figure 2. Relationships between behavioural responses measured in three ecologicalsituations: (a) an index of exploration of an altered breeding environment andaggression, (b) exploration and risk taking, and (c) risk taking and aggression. Note thatlatency to attack and flight distance are inverse estimates of aggression and risk taking,respectively. Aggression was analysed by using nonparametric tests (see main text);thus ranked values are shown for illustration.

L.Z. Garamszegi et al. / Animal Behaviour 77 (2009) 803–812 809

altered breeding environment were also more aggressive towardsan intruder male, as there was a negative correlation between thelatency to attack and the relative nestbox-presenting behaviour inan altered environment (Spearman rS ¼ �0.481, CI% ¼ �0.741/�0.096, N ¼ 24, P ¼ 0.017; Fig. 2a). Moreover, explorative individ-uals also appeared to take high risks when a potential predator wasapproaching, which was manifested in a strong negative relation-ship between our estimate of exploration and flight distance(Pearson r20 ¼ �0.769, CI% ¼ �0.899/�0.514, P < 0.001; Fig. 2b).Correlated behavioural responses would also predict an associationbetween risk taking and aggression. We could only confirm thisprediction with a smaller effect size (i.e. medium effect sensuCohen 1988), but it still showed patterns in the expected direction(Spearman rS ¼ 0.366, CI% ¼ �0.066/0.682, N ¼ 22, P ¼ 0.094;Fig. 2c). Individual behaviour towards the female decoy, assummarized by the principal component for control activity, didnot predict either risk taking or aggression (aggression: SpearmanrS ¼ �0.065, CI% ¼ �0.456/0.348, N ¼ 24, P ¼ 0.764; risk taking:Pearson r20 ¼ �0.054, CI% ¼ �0.465/0.376, P ¼ 0.810; see Methodsfor correlations with principal component for experimental activityand note that activity was statistically factored out from ourmeasure of exploration).

We assessed the effect of some potentially confounding factors,such as the identity of females in the nestbox exploration test, theidentity of males in the aggression tests, age of the resident males, thetime and date of experiments (the latter reflecting territory quality)and body condition. However, the appropriate control forthese variables in multivariate statistical designs resulted in effectsizes similar to those detected in the above pairwise models (explo-ration–aggression: �0.651 < r < �0.436; exploration–risk taking:�0.780 < r < �0.455; risk taking–aggression: 0.307 < r < 0.585;note that when we controlled for male age and body condition, thesample size was limited to those males that could be captured). Theseresults show that correlations between behaviours cannot be causedby systematic correlations with the confounding variables.

Trappability and Behavioural Types

If behavioural syndromes have consequences for the likelihoodof being captured, we predicted that exploratory, aggressive or risk-taking individuals would be captured more easily than others withmore averse characteristics. We found strong effect sizes forexploration and risk taking being more expressed in males that wecould capture than in those we could not (exploration: r24¼ 0.623,CI% ¼ 0.310/0.814, P < 0.001; Fig. 3a; flight distance: r20 ¼ 0.666,CI% ¼ 0.340/0.849, P < 0.001; Fig. 3c). On the other hand, sucha strong relationship was not found for aggressiveness (r22 ¼ 0.026,CI% ¼ �0.381/0.425, P ¼ 0.892; Fig. 3b).

When we focused on the variation within the successfullytrapped individuals, we found somewhat similar patterns. Explo-ration was also a strong predictor of the time needed for successfulcapture (Pearson r17 ¼ �0.510, CI% ¼ �0.782/�0.072, P ¼ 0.026;Fig. 3d). On the other hand, we failed to find evidence for aggressionand risk taking being strongly associated with capture time (latencyto attack: Spearman rS ¼ 0.162, CI% ¼ �0.330/0.585, N ¼ 18,P ¼ 0.337; Fig. 3e; flight distance: Pearson r15 ¼ 0.211,CI% ¼ �0.300/0.628, P ¼ 0.416; Fig. 3f).

The outcome of capturing may be sensitive to the time of day, asflycatchers tend to decrease their territorial activity around the nestas the time progresses (Lundberg & Alatalo 1992). Moreover,behaviour and the probability of capture may change as the seasonprogresses, and date effects could mediate our results as well.Finally, we also considered breeding experience in our breeding plot,because familiarity with the capture method from previous yearsmay influence the outcome of our current capture effort. However,

statistical control for the time and date of capture and also forprevious breeding experience did not modify the detected effectsizes for the relationship between trappability and behaviour (effectsizes for success of capture and latency to capture, respectively:exploration: 0.619 < r < 0.626 and �0.527 < r < �0.513; aggres-sion: 0.011 < r < 0.028 and 0.030 < r < 0.124; risk taking:0.658 < r < 0.678 and 0.128 < r < 0.308). Trappability was inde-pendent of general activity as characterized during nestboxpresentation at the control (natural) situation (success of capture:r24 ¼ �0.157, CI% ¼ �0.513/0.245, P ¼ 0.444; latency to capture:r17 ¼ 0.116, CI% ¼ �0.284/0.481, P ¼ 0.636). Note that activity wasalso factored out from our measure of exploration. Therefore, indi-viduals are unlikely to enter the trap readily because they are veryactive and so frequently enter the nestbox.

We investigated the relationships between behavioural traits byusing the subsample of individuals that had been successfullycaptured. From this sample, effect sizes for aggression were gener-ally stronger than before when using the whole sample of birds(exploration–aggression: Spearman rS ¼ �0.650, CI% ¼ �0.857/�0.263, N ¼ 18, P ¼ 0.004; exploration–risk taking: Pearsonr15 ¼ �0.513, CI% ¼ �0.797/�0.043, P ¼ 0.035; aggression–risktaking: Spearman rS ¼ 0.670, CI% ¼ 0.261/0.875, N ¼ 16, P ¼ 0.005).

DISCUSSION

We have provided evidence that behavioural syndromes canpotentially determine trade-off decisions in at least three ecologicalsituations, across which individual differences are preserved. Wefound that exploratory behaviour may be a good predictor of thedegree of aggression displayed in a territorial challenge. Moreover,strong effect sizes supported the hypothesis that males that explorea modified breeding site more intensively are also those that takehigher risks when a potential predator is approaching. Such corre-lations corroborating the concepts of behavioural syndromes havebeen repeatedly shown in the great tit (Groothuis & Carere 2005;Drent 2006), but this is one of the first studies to show similar roles ina wild population of a bird species with a different ecology (Kralj-Fiser et al. 2007; Herborn et al. 2008). On the other hand, malebehaviour during nestbox presentation to a female did not correlatewith how it responded to a male intruder or to an approachinghuman. Therefore, general activity does not interfere with estimatesof aggression, exploration or risk taking (Reale et al. 2007). More-over, correlations between behaviours were independent of severalconfounding factors, such as territory quality, male age or condition.The strength of our analysis is that we could directly characterizeindividual differences in behavioural responses based on an exper-imental design that enabled the ranking of individuals in theirnatural environment. Hence, we were not constrained to scorebehavioural types in captivity, which may have involved the risk ofmeasuring ecologically irrelevant traits under artificial conditions.Moreover, by eliminating the need to capture the subject animals,our sample was not biased by the probability of being caught. This istheoretically important, because we proved that trappability of birdsis associated with their behavioural profile, implying that any studybased on capturing protocols at nestboxes will unavoidably samplerisk-taking or explorative individuals.

As the implications of our results depend on the quality of thedata at hand, certain methodological issues warrant some atten-tion. In this study, we assayed birds on the same territory, duringa single day with different tests, because we were interested inestimating the consistency of behaviours across situations (but seeGaramszegi et al. 2006 for the consistency of aggression across timesuch as days or even years). Therefore, the observed behavioural(co)variations could be caused by differences between the qualityor the perception of the environments (Riechert & Hall 2000), in

L.Z. Garamszegi et al. / Animal Behaviour 77 (2009) 803–812810

which our tests were executed rather than differences betweenindividuals. To avoid this problem, we made an effort to assay ourindividuals in an experimental aviary, in which environmentalfactors can be controlled (see Reale et al. 2007). Unfortunately,captive birds showed stressed behaviour rather than patterns ofexploration; thus it was practically impossible to characterizebehavioural traits in a standard environment. However, we caninfer that environmental effects are unlikely to have shaped ourresults for several reasons. First, we controlled statistically forterritory quality, as assessed by arrival date and male age. Second,the pairwise design of our exploration test also eliminates envi-ronmental effects. Exploration scores were expressed relative toa control situation, which corresponded to the same environmentwith the only difference between consecutive control/experimentalsessions being the presence of a novel object. By focusing on thedifference in behaviours between the two sessions, we inherentlyremoved effects of territory quality or resource-holding potential.Third, our previous study on aggression suggests that consistentbehavioural responses may exist across territories, because thetemporal correlation of aggression was assessed on the basis ofsampling individuals at different nestboxes occupied at differenttimes (Garamszegi et al. 2006).

Different Correlations between Behaviours

Studies on captive bird species generally show significantcorrelations for various behavioural traits, but similarity with the

1

0.5

0

–0.5

–1

–1.5 7 19Unsuccessful Successful

Capture

Unsuccessful

C

Exp

lora

tion

of

nov

el b

reed

ing

envi

ron

men

t

6

50

45

40

35

30

25

20

15

10

5

0

Late

ncy

to

figh

t (s

)

2.5

2

1.5

1

0.5

0

Tim

e n

eed

ed f

or c

aptu

re (

h)

–3 –2 –1 0 1 2 3Exploration of novel breeding

environment

0 5 10Ranked la

Tim

e n

eed

ed f

or c

aptu

re (

h)

3

2.5

2

1.5

1

0.5

3

(a) (b)

(e)(d)

Figure 3. Relationships between trappability and behavioural types (exploration, risk takingof catch effort) and (d–f) on a continuous scale (i.e. the time needed for successful capture). Nshow means, except in relation to aggression (b), which was analysed in nonparametric te

great tit model is less than clear in certain details (Groothuis &Carere 2005). Data from different species suggest that the patternof covariation between traits is species specific (Harvey & Freeberg2007; Kralj-Fiser et al. 2007). Interspecific differences may implythat different lifestyles require different organization of behaviours,and thus different roles for behavioural syndromes accompanyingthe evolutionary ecology of species.

The current study with collared flycatchers in their ownbreeding environment may suggest selection pressures thatmaintain behavioural syndromes in natural populations. Nestboxoccupation and presentation are important components of thespecies’ reproductive behaviour (Lundberg & Alatalo 1992; Cramp& Perrins 1994), and patterns of exploration of a modified nestboxmay thus be relevant for fitness maximization. Individuals appar-ently handled altered breeding environments in a different way andthus had different strategies to achieve mating success. Here, weshowed that such differences can be translated into more generaldifferences in several behaviours, because explorative behaviourcorrelated with risk taking and aggression. We infer that behav-ioural syndromes are evolutionarily linked to several aspects of thespecies’ ecology and reproduction.

However, the strength of relationship between differentbehaviours can vary. For example, exploration showed very strongcorrelation with risk taking, and a smaller but still remarkableassociation with aggression, whereas the relationship betweenaggression and risk taking was weaker (Fig. 2). Aggression isa complex trait, and in the collared flycatcher is determined by

Successful

apture

Unsuccessful Successful

Capture

1825

20

15

10

5

0

Flig

ht

dis

tan

ce (

m)

5 17

15 20 25 30tency to fight Flight distance (m)

0 5 10 15 20 25 30

3

2.5

2

1.5

1

0.5

Tim

e n

eed

ed f

or c

aptu

re (

h)

(c)

(f)

and aggression), when trappability was measured (a–c) on a bivariate scale (i.e. successumbers in bars (a–c) indicate associated sample sizes. Error bars (a, c) indicate SEs. Bars

sts, and thus bars depict the 25% and 75% quantiles, and the lines are medians.

L.Z. Garamszegi et al. / Animal Behaviour 77 (2009) 803–812 811

several extrinsic and intrinsic factors (e.g. the value of the territory,the quality of the challenger, age), one of which is an individual-specific attribute that may be linked to behavioural types(Garamszegi et al. 2006). Hence, the manifestation of consistentbehaviours perhaps varies from trait to trait depending on the roleof additional factors that can shape the observed behaviours andblur the effect of individuality. Moreover, the strength of relation-ship between behavioural traits may also depend on the overlapbetween the context (i.e. trade-off and situation) to which theycorrespond (Wilson & Stevens 2005). For example, exploration mayhave a risk-taking element, as the benefits of exploiting an unfa-miliar environment (e.g. mating success) should be traded againstthe suspected costs (e.g. investing in poor-quality habitat) that mayalso involve the presence of predators. Functionally, explorationmay share a component with risk taking, as both can be mediatedby neophobia, which determines how birds generally treat novelsituations including predators (see Greenberg & Mettke-Hofmann2001; Reale et al. 2007). The correlation between behaviours mayalso be enforced by environmental effects that select for the mostadaptive combinations of traits (the ‘adaptive hypothesis’ sensuBell 2005). Accordingly, in the ecological niche of the collaredflycatcher, individuals that explore a novel environment also takehigher risks in a dangerous situation, but these individuals are notnecessarily selected to show very strong aggression towardsconspecifics, as some context-dependent variation is maintained.

Behavioural Syndromes and Trappability

If animals inherently differ along the shy–bold continuum, thisshould have consequences for the probability of being captured bya trap, because shy animals may be less likely to enter a trap thanbold individuals (Wilson 1998). Accordingly, the pool of success-fully caught animals is more likely to include bold individuals thanthe pool of animals that avoided the trap. A relationship betweenbehavioural syndrome and trappability has been shown to exist infishes (Wilson et al. 1993), mammals (Reale et al. 2000; Malmkvist& Hansen 2001) and birds (Mills & Faure 2000). We also detecteda strong association between the success of capture and the degreeof exploration of a habitat with a novel object and risk taking underthe pressure of a potential predator (Fig. 3). On the other hand,trappability was not influenced by general activity. Therefore, thedetected patterns confirm that trappability is a component ofbehavioural syndromes. These results have implications for fieldstudies on nestbox populations that rely on capturing schemessimilar to those we used in the current study.

If a generally strong correlation exists between the probability ofcapture and behavioural syndromes in birds (Mills & Faure 2000;this study) or in other animals (Wilson et al. 1993; Reale et al. 2000;Malmkvist & Hansen 2001), a sampled population will not behomogeneous, as the capturing protocol will necessarily filter risk-takers or explorer individuals. This is important for designingstudies of behavioural syndromes in captivity, because the samplesof animals used in laboratory tests could miss the lower tail ofnatural distributions of phenotypes, which may decrease thestatistical power of subsequent correlations of the measuredbehavioural traits. In addition, the problem not only raises type IIerrors, but can also generate bias, even in a broader context. Forexample, we showed that effect sizes for the relationship betweenbehavioural traits can be different when using the whole sample ofbirds and when using only a subsample of captured individuals.Ecological and behavioural approaches usually require the identifi-cation, measurement or manipulation of individuals, which demandcapturing protocols. However, if trap avoidance correlates with anentire suite of phenotypic traits, identified, measured or manipu-lated animals will not be representative of the population as a whole.

For example, behavioural syndromes may affect the magnitude ofthe adrenocortical response to social stress (Carere et al. 2003; Kralj-Fiser et al. 2007) or parasite levels via the degree of contact withdifferent environments that favour adaptive immune responses(Wilson 1998; Garamszegi et al. 2007). If the measured or manipu-lated trait also interacts with behavioural types (e.g. physiologicalstress response or immune defence) and thus trappability, thesuccess of capture may influence the outcome of experiments.Moreover, behavioural syndromes and the problems of trappabilitymay be relevant in detecting long-term population trends. Accord-ingly, survival analyses take great care to discriminate between truesurvival events and recapture probabilities (Lebreton et al. 1992)which can be related to factors shaped by behavioural types (e.g.dominance; Crespin et al. 2008). Moreover, as the adaptive value ofa behavioural response should be adjusted to the structure of thesocial environment, the structure of the population along the shy–bold continuum will necessarily be density dependent (Wilson1998; Sih et al. 2004a). If the distribution of individual behaviouralstrategies fluctuates in time, and if we are only able to captureindividuals above a given degree of exploration, each year-samplewill focus on different subsets of the population. Consequently,differences in capture probability caused by behavioural syndromesshould be taken into account when focusing on only a subsample ofanimals that entered the trap.

In conclusion, we have demonstrated in a natural population ofan intensively studied avian model species in ecology and evolutionthat behavioural responses in three biologically relevant situationsmay vary along a common axis. Hence, behavioural syndromes andcorrelated behaviours are likely to be important for a species’reproductive and life history adaptations and trade-offs. We havealso provided evidence that our chance of capturing an animal forscientific investigations may depend on its trappability, which hasa strong component that varies along the shy–bold axis. Individualdifferences in trappability may have important, and as yet not fullyrecognized, implications for the field studies of nestbox populations.

Acknowledgments

During this study L.Z.G. received a postdoc fellowship from theFWO Flanders, Belgium. A. Kobler and two anonymous refereesprovided valuable comments. G. Hegyi and E. Szoll}osi assisted in thecapturing protocols. We are grateful to J. Meaney-Ward for linguisticcorrections. The fieldwork was supported by the National ScientificResearch Fund, Hungary (grants: T049678 and T049650). FWOFlanders and the University of Antwerp are thanked for financialsupport to M.E. We are very grateful to the Pilis Park Forestry.

References

Bell, A. M. 2005. Behavioural differences between individuals and two pop-ulations of stickleback (Gasterosteus aculeatus). Journal of Evolutionary Biology,18, 464–473.

Bell, A. M. & Sih, A. 2007. Exposure to predation generates personality in threes-pined sticklebacks (Gasterosteus aculeatus). Ecology Letters, 10, 828–834.

Blumstein, D. T. 2006. Developing an evolutionary ecology of fear: how life historyand natural history traits affect disturbance tolerance in birds. Animal Behav-iour, 71, 389–399.

Both, C., Dingemanse, N. J., Drent, P. J. & Tinbergen, J. M. 2005. Pairs of extremeavian personalities have highest reproductive success. Journal of Animal Ecology,74, 667–674.

Carere, C., Groothuis, T. G. G., Mostl, E., Daan, S. & Koolhaas, J. M. 2003. Fecalcorticosteroids in a territorial bird selected for different personalities: dailyrhythm and the response to social stress. Hormones and Behavior, 43, 540–548.

Carere, C., Drent, P. J., Koolhaas, J. M. & Groothuis, T. G. G. 2005. Epigenetic effectson personality traits: early food provisioning and sibling competition. Behav-iour, 142, 1329–1355.

Cohen, J. 1988. Statistical Power Analysis for the Behavioral Sciences. Hillsdale, NewJersey: L. Erlbaum.

Corey, D. T. 1978. The determinants of exploration and neophobia. Neuroscience andBiobehavioral Reviews, 2, 235–253.

L.Z. Garamszegi et al. / Animal Behaviour 77 (2009) 803–812812

Cramp, S. & Perrins, C. M. 1994. The Birds of the Western Palearctic. Oxford: OxfordUniversity Press.

Crespin, L., Choquet, R., Lima, M., Merritt, J. & Pradel, R 2008. Is heterogeneity ofcatchability in capture-recapture studies a mere sampling artifact or a biologi-cally relevant feature of the population? Population Ecology, 50, 247–256.

Dingemanse, N. J. 2003. Natural selection and avian personality in a fluctuatingenvironment. Ph.D. thesis, Utrecht University.

Dingemanse, N. J. & de Goede, P. 2004. The relation between dominance andexploratory behavior is context-dependent in wild great tits. Behavioral Ecology,15, 1023–1030.

Dingemanse, N. J., Both, C., Drent, P. J., van Oers, K. & van Noordwijk, A. J. 2002.Repeatability and heritability of exploratory behaviour in great tits from thewild. Animal Behaviour, 64, 929–938.

Dingemanse, N. J., Both, C., van Noordwijk, A. J., Rutten, A. L. & Drent, P. J. 2003.Natal dispersal and personalities in great tits (Parus major). Proceedings of theRoyal Society of London, Series B, 270, 741–747.

Drent, P. J. 2006. Avian personalities: the great tit story. Journal of Ornithology, 147, 6.Drent, P. J., van Oers, K. & van Noordwijk, A. J. 2003. Realized heritability of

personalities in the great tit (Parus major). Proceedings of the Royal Society ofLondon, Series B, 270, 45–51.

Duckworth, R. A. 2006. Behavioral correlations across breeding contexts providea mechanism for a cost of aggression. Behavioral Ecology, 17, 1011–1019.

Duckworth, R. A. & Badyaev, A. V. 2007. Coupling of dispersal and aggressionfacilitates the rapid range expansion of a passerine bird. Proceedings of theNational Academy of Sciences, U.S.A., 104, 15017–15022.

Fidler, A. E., van Oers, K., Drent, P. J., Kuhn, S., Mueller, J. C. & Kempenaers, B.2007. Drd4 gene polymorphisms are associated with personality variation ina passerine bird. Proceedings of the Royal Society of London, Series B, 274,1685–1691.

Garamszegi, L. Z. 2006. Comparing effect sizes across variables: generalizationwithout the need for Bonferroni correction. Behavioral Ecology, 17, 682–687.

Garamszegi, L. Z., Møller, A. P., Torok, J., Michl, G., Peczely, P. & Richard, M.2004a. Immune challenge mediates vocal communication in a passerine bird:an experiment. Behavioral Ecology, 15, 148–157.

Garamszegi, L. Z., Torok, J., Michl, G. & Møller, A. P. 2004b. Female survival,lifetime reproductive success and mating status in a passerine bird. Oecologia,138, 48–56.

Garamszegi, L. Z., Rosivall, B., Hegyi, G., Szollosi, E., Torok, J. & Eens, M. 2006.Determinants of male territorial behavior in a Hungarian collared flycatcherpopulation: plumage traits of residents and challengers. Behavioral Ecology andSociobiology, 60, 663–671.

Garamszegi, L. Z., Erritzøe, J. & Møller, A. P. 2007. Feeding innovations andparasitism in birds. Biological Journal of the Linnean Society, 90, 441–455.

Garamszegi, L. Z., Eens, M. & Torok, J. 2008. Birds reveal their personality whensinging. PLoS ONE, 3, e2647.

Gosler, A. G. 1993. The Great Tit. London: Hamlyn.Greenberg, R. 1990. Feeding neophobia and ecological plasticity: a test of the

hypothesis with captive sparrows. Animal Behaviour, 39, 375–379.Greenberg, R. & Mettke-Hofmann, C. 2001. Ecological aspects of neophobia and

neophilia in birds. Current Ornithology, 16, 119–178.Groothuis, T. G. G. & Carere, C. 2005. Avian personalities: characterization and

epigenesis. Neuroscience and Biobehavioral Reviews, 29, 137–150.Gustafsson, L. 1989. Lifetime reproductive success in the collared flycatcher. In:

Lifetime Reproductive Success in Birds (Ed. by I. Newton), pp. 75–89. New York:Academic Press.

Gustafsson, L., Qvarnstrom, A. & Griffith, S. C. 1995. Trade-offs between life-history traits and a secondary sexual character in male collared flycatchers.Nature, 375, 311–313.

Harvey, E. M. & Freeberg, T. M. 2007. Behavioral consistency in a changed socialcontext in Carolina chickadees. Journal of General Psychology, 134, 229–245.

Hedrick, A. V. 2000. Crickets with extravagant mating songs compensate forpredation risk with extra caution. Proceedings of the Royal Society of London,Series B, 267, 671–675.

Herborn, K., MacLeod, R., Schofield, A., Alexander, L. & Arnold, K. E. 2008. Doespersonality in captivity reflect personality in the wild? A study of the blue tit(Parus caeruleus). In: 12th International Behavioral Ecology Congress (Ed. byS. Vekhrencamp & J. Bradbury), p. 195. Ithaca, New York: Cornell University Press.

Kralj-Fiser, S., Scheiber, I. B. R., Blejec, A., Moestl, E. & Kotrschal, K. 2007. Indi-vidualities in a flock of free-roaming greylag geese: behavioral and physiolog-ical consistency over time and across situations. Hormones and Behavior, 51,239–248.

Lebreton, J.-D., Burnham, K. P., Clobert, J. & Anderson, D. R. 1992. Modelingsurvival and testing biological hypotheses using marked animals: a unifiedapproach with case studies. Ecological Monographs, 62, 67–118.

Lefebvre, L., Whittle, P., Lascaris, E. & Finklestein, A. 1997. Feeding innovationsand forebrain size in birds. Animal Behaviour, 53, 549–560.

Lundberg, A. & Alatalo, R. V. 1992. The Pied Flycatcher. London: T. & A.D. Poyser.

Malmkvist, J. & Hansen, S. W. 2001. The welfare of farmed mink (Mustela vison) inrelation to behavioural selection: a review. Animal Welfare, 10, 41–52.

Mettke-Hofmann, C. 2007. Object exploration of garden and Sardinian warblerspeaks in spring. Ethology, 113, 174–182.

Mettke-Hofmann, C., Winkler, H. & Leisler, B. 2002. The significance of ecologicalfactors for exploration and neophobia in parrots. Ethology, 108, 249–272.

Mettke-Hofmann, C., Ebert, C., Schmidt, T., Steiger, S. & Stieb, S. 2005a.Personality traits in resident and migratory warbler species. Behaviour, 142,1357–1375.

Mettke-Hofmann, C., Wink, M., Winkler, H. & Leisler, B. 2005b. Exploration ofenvironmental changes relates to lifestyle. Behavioral Ecology, 16, 247–254.

Michl, G., Torok, J., Garamszegi, L. Z. & Toth, L. 2000. Sex-dependent risk taking inthe collared flycatcher, Ficedula albicollis, when exposed to a predator at thenestling stage. Animal Behaviour, 59, 623–628.

Michl, G., Torok, J., Griffith, S. & Sheldon, B. C. 2002. Experimental analysis ofsperm competition mechanisms in a wild bird population. Proceedings of theNational Academy of Sciences, U.S.A., 99, 5466–5470.

Mills, A. D. & Faure, J. M. 2000. Ease of capture in lines of Japanese quail (Coturnixjaponica) subjected to contrasting selection for fear or sociability. AppliedAnimal Behaviour Science, 69, 125–134.

Møller, A. P., Nielsen, J. T. & Garamszegi, L. Z. 2008. Risk taking by singing males.Behavioral Ecology, 19, 41–53.

Nakagawa, S. 2004. A farewell to Bonferroni: the problems of low statistical powerand publication bias. Behavioral Ecology, 15, 1044–1045.

Nakagawa, S. & Cuthill, I. C. 2007. Effect size, confidence interval and statisticalsignificance: a practical guide for biologists. Biological Reviews, 82, 591–605.

van Oers, K., de Jong, G., van Noordwijk, A. J., Kempenaers, B. & Drent, P. J. 2005.Contribution of genetics to the study of animal personalities: a review of casestudies. Behaviour, 142, 1185–1206.

Part, T., Gustafsson, L. & Moreno, J. 1992. ‘Terminal investment’ and a sexualconflict in the collared flycatcher (Ficedula albicollis). American Naturalist, 140,868–882.

Qvarnstrom, A., Part, T. & Sheldon, B. C. 2000. Adaptive plasticity in mate pref-erence linked to differences in reproductive effort. Nature, 405, 344–347.

Reale, D., Gallant, B. Y., Leblanc, M. & Festa-Bianchet, M. 2000. Consistency oftemperament in bighorn ewes and correlates with behaviour and life history.Animal Behaviour, 60, 589–597.

Reale, D., Reader, S. M., Sol, D., McDougall, P. T. & Dingemanse, N. J. 2007. Inte-grating animal temperament within ecology and evolution. Biological Reviews,82, 291–318.

Riechert, S. E. & Hall, R. F. 2000. Local population success in heterogeneoushabitats: reciprocal transplant experiments completed on a desert spider.Journal of Evolutionary Biology, 13, 541–550.

Sih, A., Bell, A. & Johnson, J. C. 2004a. Behavioral syndromes: an ecological andevolutionary overview. Trends in Ecology & Evolution, 19, 372–378.

Sih, A., Bell, A. M., Johnson, J. C. & Ziemba, R. E. 2004b. Behavioral syndromes: anintegrative overview. Quarterly Review of Biology, 79, 241–277.

Sinn, D. L. & Moltschaniwskyj, N. A. 2005. Personality traits in dumpling squid(Euprymna tasmanica): context-specific traits and their correlation with bio-logical characteristics. Journal of Comparative Psychology, 119, 99–110.

Smith, B. R. & Blumstein, D. T. 2008. Fitness consequences of personality: a meta-analysis. Behavioral Ecology, 19, 448–455.

Sol, D., Timmermans, S. & Lefebvre, L. 2002. Behavioural flexibility and invasionsuccess in birds. Animal Behaviour, 63, 495–502.

Stephens, P. A., Buskirk, S. W. & del Rio, C. M. 2007. Inference in ecology andevolution. Trends in Ecology & Evolution, 22, 192–197.

Svensson, L. 1984. Identification Guide to European Passerines. Stockholm: BritishTrust for Ornithology.

Torok, J. & Toth, L. 1988. Density dependence in reproduction in the collaredflycatcher (Ficedula albicollis) at high populatinn levels. Journal of AnimalEcology, 57, 251–258.

Torok, J., Hegyi, G. & Garamszegi, L. Z. 2003. Depigmented wing patch size isa condition-dependent indicator of viability in male collared flycatchers.Behavioral Ecology, 14, 382–388.

Verbeek, M. E. M., Boon, A. & Drent, P. J. 1996. Exploration, aggressive behaviorand dominance in pair-wise confrontations of juvenile male great tits. Behav-iour, 133, 945–963.

Wilson, D. S. 1998. Adaptive individual differences within single populations.Philosophical Transactions of the Royal Society of London, Series B, 353, 199–205.

Wilson, A. D. M. & Stevens, E. D. 2005. Consistency in context-specific measures ofshyness and boldness in rainbow trout, Oncorhynchus mykiss. Ethology, 111,849–862.

Wilson, D. S., Coleman, K., Clark, A. B. & Biederman, L. 1993. Shy bold continuumin pumpkinseed sunfish (Lepomis gibbosus): an ecological study of a psycho-logical trait. Journal of Comparative Psychology, 107, 250–260.

Wolf, M., van Doorn, G. S., Leimar, O. & Weissing, F. J. 2007. Life-history trade-offsfavour the evolution of animal personalities. Nature, 447, 581–584.