Epigenetic effects on personality traits: early food provisioning and sibling competition

Text-only Preview

Epigenetic effects on personality traits: early food
provisioning and sibling competition
Claudio Carere1,2), Piet J. Drent3), Jaap M. Koolhaas4)
& Ton G.G. Groothuis1,5)
(1 Department of Animal Behaviour, University of Groningen, The Netherlands; 2 Center for
Cellular and Molecular Neurobiology, Behavioural Neuroendocrinology Research Group,
University of Liege, Belgium; 3 Netherlands Institute of Ecology, Center for Terrestrial
Ecology, Heteren, The Netherlands; 4 Department of Animal Physiology, University of
Groningen, The Netherlands)
(Accepted: 4 March 2005)
The relative contribution of genetic and non-genetic factors in shaping personality traits is
of fundamental relevance to biologists and social scientists. Individual animals vary in the
way they cope with challenges in their environment, comparable with variation in human
personalities. This variation has a substantial genetic basis. Here we describe experiments
showing the strength of environmental factors (food availability and sibling competition) in
shaping personality traits in a passerine bird (Parus major). We manipulated the early rearing
condition in two lines (F4) bidirectionally selected for different personalities (fast line: high
exploration speed and high aggression; slow line: low exploration speed and low aggression)
with a food rationing protocol inducing an impairment in growth rate and an enhancement
in levels of offspring solicitation (begging behaviour). Growth impairment was more marked
in the slow line. In a first experiment each nest contained experimental and control siblings
of the same line (within-nests design). Slow chicks became much faster than their parents
in the exploration tests regardless of the treatment, whereas fast chicks had scores similar
to their parents and showed no treatment effect. As a consequence, the line difference in
exploration behaviour of the offspring was not apparent in the juvenile phase. Six months
later the offspring of the slow line was still relatively fast, but lines differed in exploration,
since the fast line became even more fast. Food-rationed birds of the fast line were more
aggressive than both controls and their fathers, while treatment did not affect the slow line.
In a second experiment, carried out only in the slow line, each nest contained either control
or experimental siblings (between-nests design). Now, only the food-rationed chicks became
faster in exploration. We suggest that the shift in the controls in the within-nests design was
5) Corresponding author’s e-mail address: [email protected]
© Koninklijke Brill NV, Leiden, 2005
Behaviour 142, 1329-1355
Also available online -

Carere, Drent, Koolhaas & Groothuis
due to enhanced sibling competition, forced by the experimental chick. Indeed, the control
chicks in the first experiment begged more persistently and had higher exploration scores
than the control chicks in the between-nests design. Environmental factors during ontogeny
modulate the expression of phenotypic traits against the background of the reaction norm
allowed by the genome even in selected lines of animals resulting in profound and reliable
differences in behaviour.
Keywords: great tit, ontogeny, food availability, personality, exploration, aggression, sibling
Personalities, coping styles or behavioural syndromes in animals are ‘pack-
ages’ of correlated suites of traits that reflect different ways of dealing with
environmental challenges. They are influenced by genetic as well as non-
genetic factors (Wilson et al., 1994; Koolhaas et al., 1999; Gosling, 2001;
Broom, 2001; Sih et al., 2004a, van Oers et al., 2005), but little is known
about their precise origin. The few genetic selection lines that have been suf-
ficiently characterized indicate a strong genetic basis in birds and mammals
(house mouse, Mus musculus: Benus et al., 1991; Koolhaas et al., 1999; farm
mink, Mustela vison: Malkvist & Hansen, 2003; great tit, Parus major: Drent
et al., 2003; van Oers et al., 2005). These lines are thought to represent the
extremes of response patterns that coexist within populations. The general
response patterns can be described as the ‘proactive’ and the ‘reactive’ styles
(Koolhaas et al., 1999). Proactive copers are more guided by internal mech-
anisms than by environmental stimuli and easily develop routines. Reactive
copers are more flexible and respond more to environmental stimuli (Benus
et al., 1991; Koolhaas et al., 1999; Koolhaas et al., 2001; Carere et al., in
Unfortunately, ontogenetic studies rarely consider a sufficiently wide
spectrum of behavioural characteristics to show conclusively that the effects
of rearing conditions are exerted on a coherent set of traits (Koolhaas et al.,
1999). Whether early environmental factors affect a whole profile as a trait
characteristic is unknown. Although descriptive studies in the mouse lines
revealed differences in the maternal environment (higher levels of maternal
care in the more aggressive and proactive line, Mendl & Paul, 1991a; Benus
& Röndigs, 1996), handling and cross-fostering, even by means of embryo
transfer, produced little or no effect, suggesting low behavioural plasticity

Plasticity of personality traits
in these lines (Sluyter et al., 1996; Benus & Röndigs, 1997; Benus, 1999).
However, manipulation of the litter gender composition influenced the de-
velopment of some aspects of the strategies, possibly via sibling competi-
tion for maternal milk (Mendl & Paul, 1991b; Benus & Henkelmann, 1998).
Moreover, it was hypothesized that inadequate nutrition of young pups of the
aggressive line, mediated through the mother, promotes increased intra-litter
competition for maternal milk, predisposing the pups to develop into more
active/competitive individuals (Mendl & Paul, 1991a). It is therefore possi-
ble that in the natural situation food availability is a crucial factor that could
exert long lasting effects on personality traits.
Fluctuation in both quality and quantity of food supply experienced dur-
ing early ontogeny can have important consequences for life history both in
mammals and birds. Curiously, the available information concerns mainly
behavioural effects in mammals (but see Boonstra & Boag, 1987; Ylönen
et al., 2003) and effects on morphology and fitness parameters in birds. In
rodents, effects of undernutrition in early life include increased ‘emotional-
ity’, higher activity rates, and increased social responsiveness and aggres-
sion compared to control animals (Manosevitz & McCanne, 1973; Whatson
et al., 1976; Tonkiss et al., 1987). In birds, poor conditions during early de-
velopment affect growth, body condition and a range of related properties,
such us fledging weight, natal dispersal, clutch size, dominance and qual-
ity of the future breeding habitat (reviews in Gebhardt-Heinrich & Richner,
1998; Lindström, 1999). Unfavourable weather conditions and food short-
age during ontogeny are also an important cause for cohort effects in avian
morphometric traits, which can affect behaviour and population dynamics
(Thessing & Ekman, 1994).
In birds, development is also affected by the nestlings’ ability to compete
for food with its siblings (Mock & Parker, 1997). Therefore, developmental
plasticity is expected to increase the competitive abilities of a food-restricted
nestling (e.g., Bengtsson & Ryden, 1981; Brzek & Konarzewski, 2001). Sand
martin (Riparia riparia) nestlings increased locomotion and sibling compe-
tition escalated in food-rationed broods (Brzek & Konarzewski, 2001). Kitti-
wake (Rissa tridactyla) chicks in broods of two begged more frequently than
singletons when they were treated with corticosterone that enhances begging
behaviour (Kitaysky et al., 2001). Thus, the effect of food availability or hor-
mone treatment may be enhanced via sibling competition, since the behav-
iour of a begging chick depends on its own condition and on the conditions

Carere, Drent, Koolhaas & Groothuis
of its nestmate (Godfray, 1995). Such experiences in the brood environment
may shape personality traits later in life.
In the great tit (Parus major), a small territorial, secondary hole-nesting
songbird, hand-reared individuals originating from wild populations consis-
tently differ in the way they explore a new environment or approach a novel
object (‘fast’ versus ‘slow’) around forty days after hatching (Verbeek et al.,
1994). Bidirectional selection demonstrated a genetic basis for this compos-
ite trait (heritability of 54±5% based on four generations, Drent et al., 2003).
A similar result was found for risk taking behaviour (van Oers et al., 2004).
Such heritabilities were also found in wild populations, although less pro-
nounced (about 30%, Dingemanse et al., 2002). The two types of great tits
also differ in other behavioural domains, such as aggression, foraging be-
haviour, response to social and non-social stress and routine formation (see
Groothuis & Carere, 2005 for a review). A longitudinal study carried out in
the selection lines demonstrated that the line differences in these character-
istics show also temporal consistency over a time span of years (Carere et
al., in press). On the whole, the great tit fast and slow explorers resemble
the rodent proactive and reactive styles respectively (Koolhaas et al., 1999;
Groothuis & Carere, 2005).
Verbeek (1998) indicated the possible occurrence of plasticity in the de-
velopment of the behavioural strategies in a great tit population. In a year
characterized by poor environmental conditions (wet and cold spring), re-
flected in a weight lower than normal at an age of 8-12 days and frequent
starvation episodes, there were about three times more fast than slow birds.
This ratio was significantly different from the circa one to one ratio observed
in ‘normal’ years. Verbeek hypothesized that either fast nestlings survived
better in adverse situations, or that retardation of growth and/or enhanced
sibling competition in the nestling phase stimulated the development of a
fast phenotype.
We studied behavioural plasticity in great tit nestlings from the two se-
lection lines raised under laboratory-controlled conditions. Our study com-
prised two related experiments, both involving manipulation of food avail-
ability (temporal deprivation) during ontogeny. To create different levels of
sibling competition, in a first experiment we manipulated half broods, in a
second experiment we manipulated full broods. We predicted that the manip-
ulation would have an impact on the behavioural profile, changing the scores

Plasticity of personality traits
expected on the basis of those of their parents in the direction of a fast phe-
notype. Birds were tested shortly after independence (exploration tests used
as selection criteria, Drent et al., 2003) and again in adulthood (exploration
tests and aggression). We asked: (i) whether the food rationing influences
heritable personality traits; (ii) whether treatment affects different aspects
of the personality together and in the same way; (iii) whether any influence
is different in the two selection lines (genotype-environment interaction);
(iv) whether a shift in personality, if any, is persistent across age; (v) whether
the effect of food rationing is mediated via sibling competition.
General methods
Subjects and breeding
Pairs of the fast and the slow line that had previous breeding experience were
housed in outdoors aviaries at the end of the winter. The birds belonged to the
3rd and 4th generation of a program of artificial selection that started in 1993
(Drent et al., 2003). The line difference in the exploration score of these
birds was confirmed in adulthood (Carere et al., in press). In spring, eggs
from the same pairs were collected daily and exchanged with dummy eggs
(great tits lay one egg per day). They were stored in a cool and dark room at
constant conditions of humidity (60%) and temperature (19◦C) in artificial
nests covered with moss (great tits use to cover eggs with moss before clutch
is completed) for up to one week after last egg was laid. Clutches from the
same parents were then incubated and reared by foster wild parents breeding
in artificial nest-boxes previously installed in the surrounding woods and
parks. All eggs of the foster nest were removed and allocated to other wild
nests. Lines did not differ in clutch size (Mean ± SD Fast: 6.86 ± 2.7;
Slow: 9.0 ± 1.0, t = −1.6, p = 0.13), hatching success (0.62 ± 0.4; Slow:
0.62 ±0.3, t = 0.004, p = 0.99) and brood size (4.17±2.8; Slow: 5.6±2.7,
t = −0.85, p = 0.42).
At the age of 12 days the chicks were collected and hand-reared in stan-
dard conditions in the laboratory. Chicks were marked at the age of 7 days
with colour and metal rings for individual recognition.

Carere, Drent, Koolhaas & Groothuis
Housing and rearing during the nestling phase
On the day of arrival in the laboratory (age 12 days), the chicks were housed
in standard cages (39.5 × 43 × 44 cm) in sibling dyads from the same clutch.
Inside each cage each dyad was put into open wooden nest boxes (12×13.5×
12 cm). The nest boxes were filled with hay and horsehair, refreshed every
4-5 days. Within each nest one of the two chicks, randomly chosen, was
marked with a white spot with Tipp-ex (toxic free fluid) on the black feathers
of the head for individual recognition. All chicks were housed in a room with
about 70% relative humidity, 25-26◦C, and a 14:10 L:D photoperiod. Chicks
were hand-fed with tweezers every 30 min, from 700 to 2100 hrs, with a
mixture containing beef-heart, a sour milk product, baby cereal multivitamin
solution and calcium carbonate alternated with wasp moth larvae (Galleria
) previously kept in a freezer. Between age 20 and 25 days they
were fed only with the wasp moth larvae and mealworms (Tenebrio molitor).
When the chicks were not begging during feeding, acoustic (a whistle sound)
and tactile (bill touched gently with the tweezers) stimulation was given.
Chicks were always fed until begging stopped, but they were never forced
to eat. Faecal sacs were removed from the nests with different tweezers.
Survival during hand rearing was 95%. At age 18-20 days, approaching the
normal fledging age, the chicks gradually started to leave the nest, hopping
in the cage where the nest was housed. At age 25, a perch and two bowls
containing the beef-heart mixture and water were placed in the cages and
within a few days, after noticing that the food was regularly exploited, hand
feeding was gradually withdrawn (age 25-30).
Housing after independence
Around day 35 the birds were individually housed in another room in stan-
dard cages (80 × 40 × 40 cm) with wooden bottom, top, side and rear walls,
a wire-mesh front and three perches. The cage floor was covered with shell-
sand. Cages were located indoors in a room of 4.6 × 2.8 × 2.6 m with natural
daylight augmented with fluorescent light tubes from 800 to 1700 hrs. Each
bird had auditory and visual contact with other conspecifics. Cages alter-
nated respect to treatment and line. Ad lib water, sunflower seeds and a com-
mercial dry mixture (proteins, trace elements, minerals and vitamins) were
available, supplemented every two days with a fresh mixture of raw heart
and live mealworms.

Plasticity of personality traits
Experiment 1: within-nests design
At 8 days after hatching the chicks were ranked for weight within each brood.
On the basis of this rank they were assigned to the experimental or the con-
trol condition alternatingly. As a consequence, half of each brood was food-
rationed. At age 8, 9, and 10, when the broods were still raised by the wild
foster parents in the field, all chicks were taken out and weighed, but the
control chicks were immediately put back in the nest. The other half of the
chicks of each brood (the experimental subjects) were kept out of the nest-
box for three consecutive hours in wooden nestboxes in a quiet, dark, and
warmed environment. The procedure started at the end of the day between
1700 and 1800 hrs in order to avoid any opportunity for parental compen-
sation after the deprivation period, since parents virtually stop feeding trips
at dusk. In wild great tits of the population used to foster the clutches of the
selection lines three hours of deprivation at the end of the day are equivalent
to approximately 30-35 missed feeds/nest (pers. obs.). At day 11 chicks were
left undisturbed.
In total 30 fast chicks from the fast line (7 clutches of 7 pairs) and 36
chicks of the slow line (7 clutches of 7 pairs) were brought to the laboratory
in the morning of day 12. They were fed in order to habituate to beg towards
humans, a process taking a few hours. They were allocated in siblings dyads
(14 fast and 18 slow dyads, of which one slow dyad became incomplete due
to mortality) consisting of one experimental and one control chick from the
same foster nest and genetic parents. Cages containing fast and slow dyads
were positioned alternatingly.
The food rationing procedure was restarted from day 13 onwards. Again
during the last three hours of the day (from the 1830 feeding session onward)
the experimental birds were not fed and missed six feeding sessions per day
until day 20. Since in a first subset of birds there was an indication that
chicks were compensating the treatment effect (which eventually was not the
case), we changed the food deprivation protocol. From day 20 till 25 chicks
received only mealworms. In this period, during each feeding session the
control chick received ad libitum mealworms, until begging stopped, while
the experimental chick received always one worm less than the control chick.
In this way no compensation by the chicks was possible. This procedure was
carried out during each feeding session over the whole day. To assess the
effectiveness of treatment, chicks were weighed on day 8, 9, 10, 12, 20, 25,
30 and at the age of seven months.

Carere, Drent, Koolhaas & Groothuis
Experiment 2: between-nests design
Due to the poor egg production of the fast line, and since the effect of
treatment in the previous experiment was more pronounced in the slow line,
we applied a between-nests design only to the slow line during the following
breeding season. We used the same parents of the previous experiment and
some parents of the new generation of the slow line avoiding siblings and
first cousins as mates. Clutches were collected from the aviaries and again
fostered in the field. Twenty-two chicks (six clutches) were used. Due to
the small sample size and the poor environmental conditions (long spells
of rainy weather and temperature below average, thereby the female was
brooding in the nest-box very frequently and we decided to avoid disturbance
in the field), we did not food ration the chicks when they were fostered
in the field. Chicks were weighed for the first time at day 10. They were
brought in the lab at day 11 or 12 and allocated in 11 nests of sibling dyads
(five experimental and six control dyads). Each brood contributed to both
control and experimental dyads and half of the dyads contained chicks of
different parents. The treatment started from day 12 or 13, one day after they
were brought in the laboratory. Food-rationed and experimental dyads were
housed in the same room alternatingly. The rearing conditions and the rest
of the treatment were the same as in experiment 1. To assess the effect of
treatment, birds were weighed on day 10, 15, 20 and 25.
Characterization of personality: behavioural tests
Novel environment and novel object tests
The sum of the scores obtained in three tests (0-20) is the trait selected
on, where 0 is the extreme ‘slow’ and 20 is the extreme ‘fast’ exploring
bird (Drent et al., 2003). The juveniles were tested at day 35-40 (novel
environment) and day 45-50 (novel object) after hatching. At 6 months of
age they were performed with the same sequence and an interval of 3-5 days
between tests. In the first test (novel environment) birds were allowed to
explore a room with five artificial wooden trees for 10 min. The time needed
to visit four of the five trees was converted linearly to a 0-10 scale. A score
of 10 means that the bird reached the fourth tree within 1 minute; a score of
0 means that it did not reach the fourth tree within 10 minutes.
In the second test (novel object), two sessions were carried out introducing
a novel object in the home cage on one of the outer perches. A penlight

Plasticity of personality traits
battery was used on the first day and an 8 cm pink rubber toy on the second
day. Latency to approach the object and the shortest distance to it within 120
sec was scored. The results for each session were converted linearly to a 0-
5 scale. A score of 5 was given when the bird pecked the object, a score of
zero when the bird did not land at all on the perch with the object. In-between
scores were based on a combination of latency and distance. Details, as well
as absence of sex difference, can be found elsewhere (Drent et al., 2003).
Aggression test
Fast birds of the parental generation have been previously shown to be more
aggressive than slow birds in a resident-intruder test (Carere et al., in press).
We tested aggression only in male birds of experiment 1 at eight months
of age with a similar test as for their fathers. Each male was individually
confronted for 5 min with one of five adult male ‘intruders’ in his resident
cage. The birds used as intruders had similar rearing conditions, weight and
age. Their genetic background and their behavioural scores were unknown
and they were experimentally naïve. We were prepared to stop the test in
case of physical fight, but they did not occur. Tests were carried out between
0900 and 1400 hrs with a 60-min interval between tests. Each intruder was
used twice each day, with an interval of one hour between confrontations.
All intruders were marked with a white spot on head, tail or wing with Tipp-
ex (toxic free fluid). Tests were recorded on videotape and two independent
persons scored tapes at slow motion. We measured the latency time of the
resident bird to attack or chase the opponent.
Begging behaviour
Begging behaviour was scored at day 14 or 15 (before fledging) and 22 or 23
(after fledging) in most of the sibling dyads of experiment 1 and all of them
in experiment 2. Details of the testing procedures are described elsewhere
(Carere, 2003). Chicks were food deprived for 75 min, and stimulated to
beg by subsequently (i) standing in front of the cage; (ii) holding tweezers
with which the birds were hand reared in front of the bird; (iii) as (ii) but
making the same whistle sound used during hand-rearing to stimulate the
birds to beg. Tests were videotaped and the time spent begging (any offspring
solicitation posture including bill-gaping) was scored with an event recorder.

Carere, Drent, Koolhaas & Groothuis
Data analysis
The data from experiment 1 (within-nests design) were analyzed with re-
peated measures analysis of variance with line (fast vs slow) as the between-
subjects factor and treatment (food-rationed vs control), age (juvenile vs
adult phase) or generation (offspring vs parents) as the within-subjects fac-
tors. Assuming that the between-nests variation has little effect within se-
lection lines breeding in captive standard conditions, we used the sibling
dyads formed in the laboratory as statistical units for the within-generation
comparison (effects of treatment and age). Nest means were used for the in-
tergenerational comparisons with the parental mid-scores. In the data from
experiment 2 (between-nests design, only slow line), treatment formed the
between-subjects factor. Sibling dyads were the statistical units. Where data
were not normally distributed we used a log or a square root transformation
to obtain a normal distribution. Details of the analysis and post-hoc compar-
isons are reported in the result section. All p-values are two-tailed. Scores
for the two exploration tests were combined to reveal the effect of treatment
on the selection criteria used for the establishment of the lines (Drent et al.,
Experiment 1: within-nests design
Body mass
A nested model of repeated measures analysis of variance including all days
until the exploration tests in the juvenile phase revealed that food rationing
induced a impairment in body mass during growth (treatment, F1,29 = 30.8,
p < 0.001, Figure 1a), which started already when chicks were in their fos-
ter wild nests. No main effect of line was evident (F1,29 = 0.26, p = 0.61,
Figure 1a), but the effect of treatment depended both on age and line: treat-
ment was effective in both lines in the beginning, but only in the slow birds in
the second part of the treatment phase while the other birds apparently could
compensate for the food rationing (age × treatment × line F6,174 = 10.7,
p < 0.01, Figure 1a). At seven months of age food deprived birds were
still somewhat lighter than the other birds but this did not reach significance
(F1,19 = 2.7, p = 0.11), nor the interaction between treatment and line