Scale and state dependence of the relationship between personality and dispersal in a great tit population



1. Dispersal is a key process in population biology and ecology. Although the general ecological conditions that lead to dispersal have been well studied, the causes of individual variation in dispersal are less well understood. A number of recent studies suggest that heritable temperament – or personality – traits are correlated with dispersal in the wild but the extent to which these ‘personality-dispersal syndromes’ are general, how they depend on an individual’s state and on spatial scale and whether they are temporally stable, both within and across individuals, remains unclear.

2. Here, we examine the relationship between exploration behaviour – an axis of personality that appears to be important in animals generally – and a variety of dispersal processes over 6 years in a population of the great tit Parus major.

3. Exploration behaviour was higher in immigrant than in locally born juveniles, but the difference was much larger for individuals with a small body mass, though independent of sex, representing one of the first examples of a state-dependent effect in a personality-dispersal syndrome.

4. Despite a temporal trend in exploration behaviour at the population level, the difference between immigrants and locally born birds remained stable over time, both across and within individuals. This suggests that the personality difference between immigrants and locally born birds is established early in development, but that the process of immigration interacts with both personality and state.

5. We found that the number of immigrant parents a locally born bird had did not influence exploration behaviour, suggesting either the difference between immigrants and residents was environmental or that the effect is overridden by local environmental sources of variation.

6. In contrast to previous work, we found no evidence for links between personality and natal dispersal distance within the population, either in terms of an individual’s own exploration behaviour or that of its parents.

7. Our results suggest that there are links between individual differences in personality and dispersal, but that these can be dependent on differences in state among individuals and on the scale over which dispersal is measured. Future work should aim to understand the differences between dispersal within and between populations and the ways in which personality and state interact to determine the outcome of these processes.


Dispersal has a key influence on a wide range of processes in natural populations (Hamilton & May 1977; Comins, Hamilton & May 1980; Greenwood 1980; Greenwood & Harvey 1982; Rannala & Mountain 1997; Clobert et al. 2001; Sakai et al. 2001). Understanding the factors that influence dispersal decisions by individuals, and the spatial scale over which they occur, therefore remain important objectives in ecology and evolution. Individual variation in dispersal has been linked to a variety of factors, including natal conditions caused by parental life-history decisions (Pärt 1990; Verhulst, Perrins & Riddington 1997; van der Jeugd 2001), individual physiological profiles and body condition (reviewed in Dufty & Belthoff 2001). Morphological variation has long been known to influence dispersal tendency (reviewed in Ims & Hjermann 2001), sometimes leading to integrated suites of traits underlying different dispersal strategies (Zera & Denno 1997; Roff 2001). Phenotypic correlations between dispersal and other behavioural traits have also been shown to drive variation in the propensity to disperse (Cote et al. 2010a). In particular, correlations with temperament traits are receiving much attention because they often capture variation in many different behavioural traits simultaneously. These correlations are said to reflect ‘personalities’ at the individual level, behavioural syndromes at the population level (Wilson et al. 1994; Wilson 1998; Sih et al. 2004; Bell 2007; Reale et al. 2007) and dispersal syndromes in the specific instance of behavioural traits being correlated to dispersal. As these temperament traits are limited in plasticity, and have often been shown to be heritable, their linkage with dispersal is potentially of importance, because it suggests that dispersal involves the simultaneous flow of suites of behavioural traits and that any selection on dispersal behaviour will have multiple correlated effects.

Three main types of consistent behavioural trait are known to be involved in dispersal syndromes (Cote et al. 2010a): aggression, sociality and exploration behaviour, the latter term here including boldness and activity level. Several important issues are emerging in dispersal-syndrome research. The first is that while the link between dispersal and ‘exploration behaviour’ seems primarily positive (Myers & Krebs 1971; Wilson et al. 1994; Wilson 1998; Fraser et al. 2001; Dingemanse et al. 2003; Krackow 2003; Sih et al. 2004; Bell 2007; Reale et al. 2007), the direction and strength of the relationship between dispersal and both aggressiveness and sociability is highly system-dependent; as yet, too few studies exist to understand this variation. A second emerging issue concerns the scale over which these effects occur. Most studies to date are from captive or semi-captive populations and demonstrate correlations between consistent behavioural variation and dispersal over very small spatial scales. Three notable exceptions have involved a much greater range of scales. The first showed a dispersal syndrome within a single-stream population of Trinidad killifish Rivulus hartii (Fraser et al. 2001), two others in the great tit Parus major (Dingemanse et al. 2003) and the common lizard Lacerta vivipara (Meylan et al. 2009) occurred at the meta-population level, while the final example was detected along the entire range-expansion front of the western bluebird Sialia mexicana in the USA (Duckworth 2006a, 2008; Duckworth & Badyaev 2007; Duckworth & Kruuk 2009). Another emerging issue concerns the proximate causes of behavioural-dispersal syndromes. Although often assumed to have a genetic basis, this has rarely been tested, either indirectly or directly. Quantitative genetic analysis showed that aggressiveness and dispersal were genetically correlated in western bluebirds (Duckworth & Kruuk 2009), while natal dispersal distance in the great tit was predicted by paternal exploration behaviour alone, which may indicate genetic factors are more likely to play a role than maternal effects at the laying stage (Dingemanse et al. 2003). A fourth issue is whether dispersal syndromes are state-dependent. Dispersal is often strongly state-dependent in animal populations, especially in relation to sex, age, physiology and morphological traits such as body condition (Dufty & Belthoff 2001; Ims & Hjermann 2001; Clobert et al. 2009; Meylan et al. 2009). Complex correlations between state and multiple behaviours may involve a combination of genetic, environmental, G × E or developmental factors (Meylan et al. 2002, 2009; Meylan & Clobert 2005), but even phenotypic evidence for state-dependent personality-dispersal syndromes is scarce. A fifth emerging issue concerns the within-individual permanence of dispersal syndromes. One study has demonstrated that sociability was consistent over 3 weeks spanning the dispersal period in a semi-natural population of mosquitofish (Cote et al. 2010b); consistency was also reported over 1 year in a semi-captive population of common lizards (Cote & Clobert 2007), but the stability of personality differences with respect to dispersal under natural conditions remains untested. Lastly, it is largely unknown to what extent dispersal syndromes are stable with respect to population-level processes such as changes in population density, or population history. In the common lizard, sociability remained consistent from birth to 1-year post-dispersal within experimental populations, and this was independent of the original population’s density (Cote & Clobert 2007). In contrast, the aggression-dispersal syndrome in western bluebirds breaks down over time within specific parts of a geographic range, dependent on year since colonisation (Duckworth & Badyaev 2007; Duckworth 2008). Although personality traits are sometimes strongly influenced by environmental factors (Carere et al. 2005; Groothuis & Carere 2005; Arnold et al. 2007; Costantini et al. 2008; Groothuis et al. 2008; Quinn et al. 2009), they are also heritable and may be the target of strong selection (Dingemanse et al. 2004a; Boon, Reale & Boutin 2008; Smith & Blumstein 2008; Quinn et al. 2009). Hence, we might expect to see changes in the mean characteristics of populations’ dispersal syndromes over small spatial and temporal scales.

Our aim was to explore these emerging issues in the context of a natal-dispersal syndrome involving ‘exploration behaviour in a novel environment’ (exploration behaviour or EB) in a long-term study population of the great tit. Extensive work on dispersal has been conducted in this species, revealing sex-biased natal dispersal to be the major dispersal event shaping population genetic structure (Greenwood, Harvey & Perrins 1978; Verhulst, Perrins & Riddington 1997; Szulkin & Sheldon 2008; Szulkin et al. 2009). Exploration behaviour was one of the first personality traits to be linked to dispersal under natural conditions in an animal population (Fraser et al. 2001; Dingemanse et al. 2003). It is also heritable (Dingemanse et al. 2002; van Oers et al. 2004; Quinn et al. 2009), correlates with variation in a range of other types of behaviours – including dominance (Verbeek, Boon & Drent 1996; Verbeek et al. 1999; Dingemanse & de Goede 2004b), aggression (Carere et al. 2005) and territory defence (Amy et al. 2010) – and potentially represents a cause of gene flow within and between populations. Our main objectives were the following: (1) to test for an association between exploration behaviour and immigration; (2) to examine whether natal dispersal distance within the study population was correlated with the exploration behaviour of the individual or of their parents; (3) to test whether any links between exploration behaviour and dispersal were state (sex, body mass and body size)-dependent or (4) likely to have a genetic basis as indicated by the effect of parental origin (immigrant vs. locally born parents) on individual exploration behaviour; and (5) to test whether any effects were stable within individuals.

Materials and methods

Study population and dispersal

This work was carried out as part of a long-term study on a population of great tits at Wytham Woods, Oxford, UK. The study site consists of a discrete ca. 385 ha area of mixed woodland that is effectively isolated from other woodland by surrounding agricultural and urban landscapes. Great tits are socially monogamous and in Wytham breed almost exclusively in nest boxes distributed throughout the woodland (Perrins 1970). Population density has increased markedly over the last 20 years (Garant et al. 2004) and was relatively high from 2004 to 2009, the period over which the data analysed here were collected. Mean nest box occupancy by great tits was 37·6% (range 28–46%; = 1021 nest boxes). Reproductive success is monitored annually by a team of up to 10 fieldworkers and nest boxes checked at least weekly during spring. When nestlings are c. 7 days old, the parents are caught at the nest and are aged as adults (if more than 1 year old) or as immature (<1 year old) and are either ringed for the first time or identified by an existing metal leg-ring. All young are ringed and weighed when 14 days old (see McCleery et al. 2004 for further details), allowing immigrants to be distinguished from Wytham-born birds (natal origin), establish the natal site of Wytham-born birds, and thus the natal dispersal distance for locally born recruits. All nest boxes had previously been mapped with the aid of a GPS system (Wilkin et al. 2006), and the natal dispersal distance from the natal site to first-breeding or settling site was determined as the straight-line distance. The dispersal distances of birds classed as immigrant are unknown. While there is evidence that a small proportion of immigrants are born to great tits breeding in marginal habitats close to the boundary of the study site (e.g. in hedgerows and gardens), 95% of immigrants are thought to originate elsewhere (Verhulst, Perrins & Riddington 1997) and can thus be classed as genuine dispersers from adjacent populations more than 1 km away from the study site.

Exploration behaviour assays

Assays were conducted over five winters, from February to March 2005 and from September to March in the subsequent four winters. Birds were usually caught with mist-nets at sunflower-seed feeders in the woods, weighed, aged and sexed based on plumage (Svensson 1984), identified and brought into captivity at Wytham field station, ca. 2 km from the centre of the wood, usually for not more than 24 h; a minority of birds were caught during nocturnal checks of nest boxes for roosting birds. Birds were individually housed, and assays were conducted the following morning between 08.00 and 13.00 h. An assay commenced 20 s after a bird was first coaxed from its cage through a trapdoor leading to an adjoining novel environment room and lasted for 8 min, after which the bird was coaxed back into its cage. The assay room design was based on (Verbeek, Drent & Wiepkema 1994). The frequency and location of all movements were recorded using a hand-held computer, generating 12 closely related behavioural measures; the first component (PC1) from a principal components analysis had a positive loading for all measures, reflecting a combined measure of activity levels and propensity to explore novel objects and areas (Quinn et al. 2009). PC1 was right skewed and was therefore transformed (PC1T) by the equation PC1T = √(PC1 + (PC1min × −1)) where PC1min was the smallest (negative) value of PC1 in the data set and was used as the basis of our measure of EB for objectives 1–3.

A total of 2258 EB assays were conducted on 1805 individuals from February 2005 to March 2010, of which 627 (34·7%) were immature (i.e. in their first winter) immigrants and 709 (39·3%) were locally born immature birds. Among birds that were first assayed as adults, 244 had been ringed as nestlings in Wytham, and 226 were immigrants from outside, of which 77·9% and 73·5%, respectively, had been recorded as breeding birds previously. The link between personality and natal dispersal distance could be studied directly for 252 locally born birds that were assayed as immature birds and recruited to the population in their first year, for an additional 34 that recruited when older and for 81 that were first assayed as adults (total N = 367). Of the 367 immature birds that were assayed or recruited at any stage, two or more siblings of known EB recruited from 58 nests, yielding possible within-brood comparisons from 58 nests, for which within-brood comparisons of same-sexed siblings were possible from 17 nests for males and 17 nests for females. Of the 252 locally born birds assayed in their first winter, two or more siblings of known EB recruited from 34 nests, allowing within-brood comparisons from 11 nests for males and 8 nests for females.

The indirect link between parental personality and offspring dispersal distance within populations could be estimated for 139 different fathers (recruiting 232 offspring from 168 breeding attempts) and 134 mothers (recruiting 232 offspring from 167 breeding attempts) that were subject to personality assays and also recruited offspring to the breeding population; the relationship between mid-parent EB values and offspring dispersal distance could be estimated for the 86 male–female pairings that generated dispersal distances for 132 offspring. These sample sizes are several times larger than those used in an earlier study of this question (Dingemanse et al. 2003).


Objectives 1 and 3 were both tested in similar models, where the response variable was the PC1T values from the first assays conducted on individuals, and controlled for days since 31 August, or ‘Day’, because EB increases over time within seasons. Objective 1 tested whether EB differed between immigrant and resident immature birds (natal origin, two-level factor). Objective 3 tested whether any dichotomy in EB between residents and immigrants was dependent on state (sex, body mass, body size) or variation in EB among cohorts at the population level, tested by two-way interactions between natal origin and the relevant variable. Because birds were caught at different times of the day and season, both of which factors have a large influence on body mass, we used residual body mass (henceforth, body mass) when first captured as immature birds in winter based on the equation mass = 0·136*h − 0·004*day + Bi*cohort + 1·154 (males) – 0·079 (immature birds), where Bi denotes a cohort-specific parameter (not shown), and the parameter for females was set to 0. Wald statistics for each of these variables were 271·9, 82·0, 66·3, 969·6 and 3·6, respectively; all had P values <0·001, except for age, where P = 0·057. The model accounted for 38·8% of the variation, while quadratic terms for hour and time of season were not significant. Body size was estimated from the first principal component of wing length (mm) and tarsus length (mm), with latent vectors of 0·14 and 0·99, respectively. This component, henceforth referred to as body size, accounted for 92·3% of the variation in these two linear measures but was only weakly correlated with body mass (= 0·156, < 0·001, = 2007).

Objective 2 asked whether EB and natal dispersal were correlated within the study population. Here, we used a series of models to test whether ‘natal dispersal distance’ (√transformed to improve error normality) was predicted by the individual’s own EB or that of either parent (see Dingemanse et al. 2003). Here, EB was an explanatory variable and corrected beforehand for several fixed effects, including date within the season and the number of times the bird had been assayed; all assays were included in a GLM of PC1T, and the parameter estimate for an individual added to the constant was used as the corrected EB (full details given in Quinn et al. 2009).

Objective 4 asked whether the link between EB and natal origin was likely to have a genetic basis. Many immigrants are thought to originate >2 km from the boundaries of our study area (Verhulst, Perrins & Riddington 1997). Therefore, we were unable to measure the EB of birds from the external population that did not immigrate to our study population and cannot determine whether the difference between immigrants and residents reflects a fixed difference at the population level, a process of selection within the external population, or developmental changes at the individual level because of the process of immigration. However, we can make some inferences based on comparisons of the effect of parental origin on the phenotype of their offspring. Therefore, we asked to what extent having zero, one or both parents born locally had on the EB of their offspring. If the link between natal origin and EB is caused by the immigration process itself, or an environmental effect on EB in the source-immigrant population, we expected no link between parental origin and the EB of their offspring. Alternatively, a link between parental origin and offspring EB would suggest either a genetic basis to the syndrome, or that selection acted on EB in the source-immigrant population.

Finally, objective 5 tested whether EB was consistent over time within individuals. Here, we used the first PC1T values for individuals when they were assayed for the first time as immature birds and again when assayed for the first time as adults, which was available for 61 immigrants and 74 residents. All analyses were conducted using GLMs or LMMs. For the GLMs in objectives 1, 3 and 4, the analyses shown include multiple offspring from an unknown number of immigrant birds and a small number of the same nests within Wytham. We elected not to present analyses for single birds per nest for the local birds, because this could not be done for immigrants, but all analyses gave similar results whichever approach was taken. For objective 2, nesting attempt was also included as a random effect to account for multiple offspring dispersing from the same nest. We did not include either paternal or maternal identify in this analysis because (i) doing so often prevented the models from converging, presumably because most individuals had only one datum and (ii) doing so made no difference to the fixed effects when convergence did occur. The statistical significance of random effects was tested using log-likelihood ratio tests and of fixed effects using Wald statistics. All analyses were conducted in Genstat or SPSS.


EB and natal origin

We found a clear effect of natal origin on exploration behaviour: EB scores among immigrant first years were on average 11% higher than among locally born first years (predicted means ± SE: 1·07 ± 0·018 and 0·968 ± 0·016, respectively), controlling for significant between- and within-season effects (Table 1). The analyses in Table 1 were repeated on immature birds that subsequently bred in the population (= 515), and the natal origin main effect remained unchanged (= 5·923, = 0·015; ± SE = −0·094 ± 0·039 for local birds), showing that the effect detected in winter was not driven by any qualitative difference between immigrants that did and did not breed, and that the effect had a direct impact on the breeding population.

Table 1.  GLM of exploration behaviour among immature great tits with respect to natal origin (resident vs. immigrant) and other fixed effects in Wytham. N = 1278, including multiple individuals born from the same nesting attempt; selecting only the first individual originating from each parental nest attempt had no substantial effect on the model
Fixed terms B SE W P
  1. aFemales set to 0.

  2. bImmigrants set to 0.

Natal originb−0·1010·02517·41<0·001
Residual body mass (g)−0·0050·0120·140·712

Mean EB varied at the population level from year to year (Table 1; Fig. 1), but the difference between immigrants and locally born birds was independent of this variation (cohort × natal origin, = 2·28, d.f. = 5, = 0·809; model otherwise similar to Table 1; Fig. 1), suggesting that whatever caused the annual variation in EB among cohorts had equivalent effects in immigrant and locally born birds.

Figure 1.

 Variation in exploration behaviour among immature great tits with respect to time and natal origin. (• immigrants; ○ residents). The natal origin effect on exploration behaviour was maintained over time.

EB-dispersal syndrome within the main population

As previously shown for this population (Verhulst, Perrins & Riddington 1997), natal dispersal distance among birds born in Wytham was greater for females than for males in this sample (= 41·49, < 0·001; median and, in parentheses, IQ range for males and females were 496 m (607 m) and 737 m (885 m); full LMM is in Table 2a). There was no evidence for a relationship between an individual’s EB score and natal dispersal distance (Table 2a). This was true when the analysis was repeated only on those birds that had already bred and therefore undergone post-natal dispersal before being assayed (= 0·33, = 0·569, B ± SE = 1·267 ± 2·219, = 115); therefore, the assaying procedure itself did not explain the lack of an association between EB and natal dispersal distance. Similarly, there was no main effect of parental EB, either for each parent separately (paternal, Table 2b; maternal, Table 2) or using mid-parent values (Table 2d) on the natal dispersal distance of their offspring. We were also able to make within-family comparisons for 34 families where >1 same-sexed offspring were assayed and recruited to the breeding population at any stage of their life. The differences in EB and natal dispersal distance for sibling dyads were not correlated for males (rs = −0·10, = 0·701, = 17) or for females (rs = −0·201, = 0·439, = 17), agreeing with the previous analyses showing no link between within-population dispersal and EB. Similar results were obtained for the smaller sample of birds that were assayed when immature and recruited in their first summer of life. Finally, to allow direct comparison with a similar study elsewhere (Dingemanse et al. 2003), we also repeated all of the analyses in Table 2 using the numbers of hops and flights in the first 2 min as our measure of EB (corrected for other fixed effects as done for our main estimate of EB); all of the effects remained non-significant, and conclusions as to statistical and biological significance are unchanged (analyses not shown). Collectively, our analyses show that a dispersal-EB syndrome existed at the between-population scale but not within our main study population.

Table 2.  The influence of (a) an individual’s own exploration score, (b) paternal exploration score, (c) maternal exploration score and (d) mid-parental score (mean of paternal and maternal) on (natal dispersal distance (m))½. All analyses are linear mixed models. Sample sizes were 252, 232, 232 and 132 for (a), (b), (c) and (d), respectively; d.f. = 1 for all terms, except for those in which year was involved, in which case d.f. = 4
Random termsLog-likelihood ratio P Component ± SEFixed termsWald P B ± SE
  1. aFemale set to 0.

Nesting attempt0·530·46746·93 ± 20·11Constant  26·02 ± 2·219
Residual  66·17 ± 19·15Sex26·36<0·001−6·615 ± 1·288a
        Individual EB1·090·2971·609 ± 1·539
Nesting attempt0·490·48426·51 ± 37·43Constant  35·12 ± 4·457
Residual  89·58 ± 37·66Sex10·610·001−4·658 ± 1·430a
        Paternal EB1·230·2681·961 ± 1·767
Nesting attempt0·001·001·899 ± 33·200Constant  30·90 ± 6·11
Residual  110·9 ± 34·7Sex9·030·003−3·308 ± 1·382a
        Maternal EB0·020·9000·399 ± 1·578
Nesting attempt2·910·08876·00 ± 31·65Constant  26·47 ± 4·728
Residual  50·59 ± 28·02Sex2·960·090−3·174 ± 1·845a
        Mid-parent EB0·160·692−1·195 ± 3·010

The influence of state

There was a tendency for EB to be marginally higher among males than females (predicted means ± SE: 1·04 ± 0·017 vs. 0·995 ± 0·017), but the difference between immigrants and locally born birds was independent of sex (Natal origin × Sex, = 0·17, = 0·682). Immigrants tended to be lighter than locally born birds (= 2·738, = 0·098: mean residual mass, which corrects for sex as well as other fixed effects = −0·051 ± 0·042 SE vs 0·042 ± 0·038 SE), while the difference in EB between immigrants and residents was more pronounced for light birds (body mass × natal origin, = 6·97, = 0·008, B ± SE = 0·066 ± 0·0·025; Fig. 2; controlling for season, day, sex), an effect that was independent of sex (body mass × natal origin × sex, = 0. 49, = 0·486). The body mass × natal origin effect was driven by a positive correlation between exploration behaviour and body mass among local-born immature birds (= 4·47, = 0·035, B ± SE = 0·036 ± 0·017, = 676; controlling for day, sex, season) and no effect of exploration behaviour on body mass among immigrants (= 2·506, = 0·114, B ± SE = −0·029 ± 0·018, = 602, controlling for day, sex, season).

Figure 2.

 State-dependent association between exploration behaviour (PC1T; see text) and natal origin (• immigrants; ○ residents) in Wytham great tits (N = 1276) from 2004 to 2009. The state variable, body mass, was residualised on time of day, sex and other factors (see text) and categorised based on cuts at mean and ± 1SD points.

There was no difference in body size between immigrants and local-born birds (= 0·01, = 0·919, = 1268; controlling for day, season and sex), and the link between natal origin and EB was not influenced by body size (body size × natal origin, = 0·46, = 0·497, B ± SE = 0·008 ± 0·012; controlling for season, day, sex); this result was similar even when body mass × natal origin was also included, which itself remains significant. Therefore, body mass per se rather than body size seemed to be the source of the state-dependent effect on EB.

Within the population, there was no evidence that state-dependent effects could have masked an underlying link between natal dispersal distance and an individual’s EB. This was true for body mass (EB × body mass, = 0·16, = 0·687) and body size (EB × body size, = 0·46, = 0·499). It was also true for sex (EB × sex, = 0·33, = 0·568) and for interactions between sex and both body mass (EB × body mass × sex, = 0·54, = 0·464) and body size (EB × body size × sex, = 3·03, = 0·083).

Parental origin

Of the 645 locally born offspring, 149, 357 and 139 had zero, one and two locally born parents, respectively (including only birds where both parents were known). There was no difference in EB among offspring between these three groups (= 3·520, d.f. = 2, = 0·173; controlling for cohort, sex and day in a LMM with nesting attempt as a random effect; predicted means ± SE were 1·002 ± 0·037; 0·963 ± 0·026; 1·045 ± 0·040). Similarly, the effect was not parent specific because there was no difference in EB between birds born to an immigrant mother and those born to resident mothers (N = 186 and 149, respectively, = 1·11, = 0·292), or between birds born to an immigrant father and those born to resident fathers (= 167 and 149, respectively, = 0·26, P = 0·613, = 347). All of the preceding effects were uninfluenced by time since the parents had immigrated (i.e. parental age; results not shown). We cannot discriminate between those resident offspring that emigrate and those that die; to test whether the estimates of EB with respect to parental origin might be biased because of this, we tested whether there was any difference in local recruitment to the population with respect to parental origin over the years when this study was conducted. However, we found no significant difference in recruitment rate between offspring whose parents were locally born or immigrants (both parents locally born: −0·602 ± 0·236; back-transformed predicted recruitment rate = 0·548 recruits per breeding attempt), both parents immigrants (−0·569 ± 0·238, 0·566 recruits per breeding attempt); one of each type (−0·636 ± 0·229, 0·529 recruits per breeding attempt; = 0·50, d.f. = 2, = 0·779, = 1495 breeding attempts from 2004 to 2009); GLMM with Poisson error and a log-link function; mother identity, father identity and year were included as random effects; lay date, brood size and average fledgling mass were all significant fixed effects – not shown).

Recruitment data for birds assayed during the winter were available for birds born up to the 2008 breeding season (assayed in the 2008–2009 season), and of these, 235 of 506 immigrant birds (46·4%) and 286 of 577 (49·6%) locally born immature birds subsequently recruited to the Wytham breeding population. There was no difference in recruitment rate between these two groups (= 1·464, = 0·226, controlling for: date of assay (W = 9·146, P = 0·002); sex (= 0·0009, = 0·985); cohort (W = 11·853, d.f. = 4, = 0·018). EB did not predict the likelihood of recruitment (= 2·00, = 0·158, B ± SE = −0·205 ± 0·145, logit scale; other effects same as previous model) and this was equally true for immigrants and residents (EB × natal origin, = 0·387, = 0·534). Taken together, these results and those in the preceding paragraph suggest that, at the between-population level, the dispersal-EB relationship is not influenced by differences in recruitment likelihood or by direct parental (e.g. genetic) effects.

Stability within individuals

Analysis of data from individuals assayed when both immature and adult showed that EB did not change with age and that birds consistently differed from one another over time (Age fixed effect and bird random effect in Table 3). Within-individual changes in EB were independent of natal origin (Age × Natal origin, Table 3; Fig. 3). The preceding tendency for immigrants to be smaller than local-born birds was not detected (= 0·11, = 0·732) among those immature birds that subsequently recruited to the population, while the difference in EB between immigrants and locals was no longer influenced by body mass (body mass × natal origin, = 0·038, = 0·846, B ± SE = −0·001 ± 0·040, = 515). This was not caused by any detectable size-dependent recruitment effect because the probability of assayed immature birds recruiting to the population was not influenced by body mass (= 0·024, = 0·877, B ± SE = −0·010 ± 0·064; controlling for season, sex, time of season, natal origin), by natal origin (= 1·311, = 0·252, B ± SE = 0·144 ± 0·126 for local birds) or the interaction between the two (= 0·250, = 0·617, B ± SE = −0·063 ± 0·127). Together, these analyses suggest that the difference between resident and immigrant birds is consistent over an individual’s lifetime after immigration.

Table 3.  Linear mixed model of exploration behaviour with respect to natal origin and age for 115 individuals (N = 294 assays) that were assayed when immature and again as adults
Random termsLog-likelihood ratio P Component ± SE
  1. n/a is not applicable.

  2. aAdult parameter set to 0.

  3. bImmigrant set to 0.

  4. cFemales set to 0.

  5. dParameters not shown.

  6. eParameter of immature residents shown.

Individual76·63<0·0010·132 ± 0·024
Residual (Bird × Age)n/an/a0·099 ± 0·011
Fixed termsWald P B ± SE
Constantn/an/a1·120 ± 0·129
Age0·050·8240·024 ± 0·067a
Natal origin5·710·019−0·154 ± 0·087b
Sex1·750·1890·103 ± 0·078c
Day10·960·0011·74 × 10−3 ± 5·27 × 10−4
Cohort (year)9·370·100 d
Age × Natal origin0·780·379−0·041 ± 0·083e
Figure 3.

 Exploration behaviour and natal origin among great tits that were assayed when immature (•) and again as adults (○).


Here, we provide evidence for a link between the personality trait ‘exploration behaviour’ and dispersal, and that the link is at least partially state- and scale-dependent. Immigrants to our study population had higher exploration scores than locally born birds; this difference was consistent across years and between the sexes, but was state-dependent because it was primarily present in light individuals. In contrast, despite large sample sizes relative to similar studies (Dingemanse et al. 2003), we found no evidence for any relationship between within-population dispersal distance and EB, whether assessed within individuals or within families.

The exploration behaviour-dispersal syndrome

Exploration behaviour among immigrants was on average 11% or 0·23 standard deviations higher than among residents in our study population. Furthermore, the difference between residents and immigrants occurred in each cohort, despite annual variation in mean exploration behaviour between years. Annual fluctuations in environmental conditions and in selection on EB have both occurred over the course of this study (Quinn et al. 2009; Quinn et al. unpublished), either of which might drive these cohort-specific fluctuations, but the consistency of the syndrome indicates resilience to this population-scale process. The effect may also be robust across studies because the difference between immigrants and residents we detected was similar to that reported for a Dutch study population, where exploration scores were about 14% higher among immigrants (estimated from Fig. 2 in Dingemanse et al. 2003). In other systems, personality-dispersal syndromes have tended to be population specific. For example, correlations between a variety of (non-dispersal related) behavioural traits among three-spined sticklebacks seem to be population specific across Californian (Bell & Stamps 2004) and Welsh (Dingemanse et al. 2007) populations. Similarly, the aggressiveness-dispersal syndrome in the western bluebird in the USA is prominent only in populations far from those of the congeneric mountain bluebird Sialia currocoides with which it competes for breeding sites. However, too few replicate studies exist to generalise about the spatial or temporal robustness of dispersal syndromes involving behavioural traits at the between population level.

In contrast to both a previous study of a Dutch great tit population (Dingemanse et al. 2003), and also several other studies of similar exploration behaviour syndromes in other species (Cote et al. 2010a), we found no evidence of a link between exploration behaviour and natal dispersal distance amongst individuals born within our population. Moreover, there was no evidence of a positive correlation between natal dispersal distance and paternal, maternal or mid-parent exploration behaviour. The sample sizes were quite large for these analyses (N = 132–252), and the standard errors for the effects were thus relatively small; analyses gave similar results when repeated on an alternative 2-min assay used in the Dutch study. Population differences in ecology and in the precise way EB influences dispersal are likely to explain different effects across studies. While the present study was carried out in a single continuous woodland in a landscape dominated by agriculture, the Dutch study site (Westerheide) is almost contiguous with the much larger Hoge Veluwe national park in central Netherlands (Dingemanse et al. 2003). Dispersal over Westerheide may be less constrained than in a more fragmented landscape where the strength of selection against emigration might be relatively strong, perhaps decoupling any correlation between exploration behaviour and dispersal distance within our population, which would not necessarily affect the difference in EB between immigrant and resident birds. Against this constraint hypothesis, in fact great tits in Wytham dispersed over larger distances than those in Westerheide (females, 935 vs. 643 m; males 648 vs. 498 m; from Dingemanse et al. 2003), possibly because Wytham is larger (385 vs. 250 ha). Similarly, females dispersed 44% further than males in Wytham but only 30% further in Westerheide. Other unknown ecological factors undoubtedly are at play, but more generally, EB is likely to influence the dispersal process within populations in different ways across populations. Natal dispersal can be viewed in three distinct phases – departure from the natal site, transience and settlement (Clobert et al. 2001, 2009; Cote et al. 2010a) – and EB has the potential to influence each of these processes through correlations with a variety of different behavioural traits, for example dominance and responsiveness, both of which are also likely to be important determinants of condition-dependent and ‘informed’ dispersal (see Clobert et al. 2009), which influence dispersal decisions at multiple stages.

State dependence of the syndrome

The difference in exploration behaviour between residents and immigrants was independent of sex because both male and female immigrants had higher scores than male and female residents, respectively. However, immigrants with small body mass had higher EB scores than locally born birds with a similar body mass. Previous work on this population suggested state-dependent effects in the immigration process, with larger birds born in external, low-quality habitats being more likely to immigrate successfully into Wytham (Verhulst, Perrins & Riddington 1997). Indeed numerous studies show that dispersal is often dependent on an individual’s state, for example on morphological or physiological phenotype (Clobert et al. 2001, 2009), but very few studies provide evidence for state-dependent links between dispersal and behavioural phenotypes (Holekamp 1986; Ims 1990; O’ Riain, Jarvis & Faulkes 1996; De Fraipont et al. 2000; Meylan et al. 2002). Body size or condition play a role in most of these studies, as it does in ours, and may be generally important in behavioural syndromes.

The link between exploration behaviour and body size may shed light on the behavioural mechanisms causing the EB-dispersal syndrome in our population. These mechanisms generally remain poorly understood but are hypothesised to be driven by differences in the outcome of social interactions (Dingemanse et al. 2002). While fast birds tend to be relatively dominant (Verbeek, Boon & Drent 1996; Verbeek et al. 1999), which may facilitate the immigration process (Chitty 1967; Krebs 1978), slow birds are better able to cope with social defeat (Carere et al. 2001), which may facilitate remaining in high-quality natal habitats where competition is stronger, and social defeat is therefore more frequent (Dingemanse et al. 2003). It is plausible that the outcome of these interactions is modulated by body mass, which can be a determinant of social status (Gosler & Carruthers 1999), and we suggest that the pattern in Fig. 2 may be explained by the dominance/social defeat model outlined in Fig. 4a (see legend for description). Figure 4b shows an alternative exploration behaviour/dispersal model where fast birds are more mobile generally and are therefore likely to both emigrate and immigrate, irrespective of body size. This model could not explain the pattern in Fig. 2 because, on average, immigrants were faster than local birds. In reality, the processes involved are likely to be more complicated than indicated in Fig. 4a. For example, the influence of EB on social interactions could be density or frequency-dependent. Furthermore, in the great tit fast explorers are dominant in some circumstances but slow explorers are dominant in others (Dingemanse & de Goede 2004b) and in the common lizard, dispersers from high population densities tend to be repelled by conspecific odour (asocial), while dispersers from low-density areas tend to be attracted to the odour (sociable; Cote & Clobert 2007). Although we show that the exploration behaviour-dispersal syndrome is body-mass dependent, clearly further research is needed to understand how this effect arises. Furthermore, if dominance plays a role in the body-mass dependent effect, it remains unclear why body size did not give a similar effect within the population, given that this measure can also predict dominance status (Garnett 1981). The answer is likely to be non-trivial because body mass is determined by body size, muscle condition and fat reserves (Gosler & Carruthers 1999), each of which may relate to dominance in different ways, dependent on the relative importance of the risks of foraging and predation (Gosler 1996, 2001; Gosler & Carruthers 1999; Gosler & Harper 2000).

Figure 4.

 Proposed models of the exploration behaviour/dispersal syndrome in our population. Both body mass (heavy = uppercase and light = lowercase) and exploration behaviour (f or F = fast and s or S = slow) are continuous phenotypes but for simplicity are shown here as discrete. (a) The dominance/social defeat model: heavy birds of all types are likely to immigrate to Wytham, but especially heavy, fast birds because they are also aggressive and more dominant; light birds are more likely to emigrate because they are less dominant, but light slow birds less so because they are better able to cope with social defeat. (b) The exploration behaviour/dispersal model: fast birds are more mobile generally and therefore likely to both emigrate and immigrate, irrespective of body mass.

Consistency of the syndrome

We found that the difference in exploration behaviour between immigrants and residents was maintained over the first year of life post-immigration. Elsewhere, we reported exploration behaviour was repeatable in this population (Quinn et al. 2009) but the analysis here further suggests that immigrants did not change their exploration behaviour once they had become established in the population over time, suggesting that the differences in exploration behaviour between immigrants and residents were permanent. One study showed that sociability was correlated with dispersal and that it was consistent within individuals before, and 10 months after, dispersal in a semi-captive population of lizards (Cote & Clobert 2007), though sociability could still have changed differentially with respect to dispersal status over time despite this consistency. In another study, a number of behavioural differences between philopatric and dispersing common lizards were found to be stable over periods of up to 10 months after the dispersal phases, even when both maternal and natal environmental conditions were manipulated (Meylan et al. 2009). This consistency suggests an inherent, possibly genetic, cause of these syndromes, but direct evidence for a genetic basis to dispersal syndrome linked to consistent behavioural variation is rare (Duckworth & Kruuk 2009). Exploration behaviour is heritable in our population (Quinn et al. 2009) but is also influenced by permanent environment effects (Quinn et al. 2009). The lack of an association between parental origin and the difference in exploration behaviour between immigrants and residents in the current study suggests that the syndrome may be largely environmental. We cannot distinguish between emigration and death among individuals that were born locally in our population but that did not recruit, but nevertheless, our recruitment analysis suggested the estimates of EB with respect to parental origin may not be biased because of differential emigration with respect to parental origin, though we acknowledge our approach is indirect. We did not measure EB before dispersal took place and therefore could not test directly whether the difference between immigrants and residents arose as a result of changes within individuals during the immigration process itself, whether it results from selection for a particular part of the external population, or whether the external population differs from the main study population with respect to mean EB phenotype (a difference that might have either environmental or genetic causes). Furthermore, G × E effects could equally explain the difference and we note that identifying the genetic basis of dispersal syndromes clearly remains a major challenge in the field (Duckworth 2006b).


Our data suggest that the exploration behaviour-dispersal syndrome in great tits from Wytham is independent of sex and large-scale population-level processes, albeit under a restricted set of environmental conditions – for example, population density has been consistently high over the course of this study (Quinn, J.L., Patrick, S., Cole, E.F., Bouwhuis, S., & Sheldon, B.C. et al. unpublished). They support the growing view that consistent behavioural differences over and above those associated with commonly studied condition factors may have important implications for our understanding of the causes and consequences of dispersal. Although our measure of personality is heritable in this study population, the dispersal syndrome seems to be largely state-dependent and is not influenced by parental origin. Unless this latter effect can be explained by G × E, given the effect size (0·23 SD) of the difference between immigrants and residents, together these analyses point to a limited impact of the syndrome on gene flow into the population. Our data also showed that the syndrome was dependent on spatial scale in our population because it was detected only in the context of differences between immigrants and residents, and not within-population variation in natal dispersal. Although this questions the generality of the effects found elsewhere, it also demonstrates that explaining differences in the existence of syndromes between populations is a major challenge.


We thank J. Cote, J. Clobert, T. Coulson and an anonymous referee for commenting on the MS. We are grateful to the many people who collected data during the breeding season for the long-term Wytham study. We also thank the following members for assistance with the personality assays: S. Bouwhuis and P. Pisa, and A. Gosler, J. Howe and M. Wood for assistance in catching birds. NERC, BBSRC and the Royal Society contributed to the funding of this work.