Dominance, pair bonds and boldness determine social-foraging tactics
in rooks, Corvus frugilegus
Jolle W. Jolles
, Ljerka Ostoji
, Nicola S. Clayton
Department of Zoology, University of Cambridge, Cambridge, U.K.
Department of Psychology, University of Cambridge, Cambridge, U.K.
Received 15 October 2012
Initial acceptance 11 January 2013
Final acceptance 20 February 2013
Available online 12 April 2013
MS. number: 12-00790
Socially foraging animals can search for resources themselves (produce) or exploit the discoveries made
by others (scrounge). The extensive literature on producerescrounger dynamics has mainly focused on
scramble competition over readily accessible resources, thereby largely neglecting the variety of
scrounging techniques individuals may use as well as the role of investment in food handling.
Furthermore, although individual differences in boldness and social factors such as dominance have been
described to inﬂuence foraging tactics, their potential interplay and effect in foraging contexts beyond
the conventional producerescrounger game remains unclear. We investigated the relationship between
social-foraging tactic use and dominance, pair bonds and boldness in a foraging experiment focused on
food handling and alternative scrounging tactics. We conducted a producerescrounger experiment in a
captive group of rooks in which individuals could produce by pulling up baited strings, or scrounge by
retrieving fallen food items or joining a producer. There were three key ﬁndings: (1) dominant rooks
adopted the producer tactic more often and more successfully than subordinates; (2) producing and
scrounging by tolerance led to mixed beneﬁts to paired birds; (3) bold birds scrounged by retrieving
more often than shy birds. Importantly, individuals were highly consistent in their tactic use across
conditions differing in food availability. Our study highlights the importance of taking both social factors
and boldness (heterogeneity) into account when studying social-foraging dynamics and offers empirical
data on food handling and alternative scrounging tactics that can be used to extend current models and
experiments on social foraging.
Ó2013 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
Socially foraging animals can either search for resources them-
selves (produce) or exploit the discoveries made by others
(scrounge; Giraldeau & Beauchamp 1999;Giraldeau & Caraco 2000;
Giraldeau 2008). An extensive literature describes theoretical
models and experiments investigating the dynamics behind these
foraging behaviours and the conditions that inﬂuence this pro-
ducerescrounger (PS) game (Barnard & Sibly 1981;Giraldeau et al.
1994;Giraldeau & Caraco 2000;Giraldeau 2008;Broom et al.
2008). Surprisingly, most PS studies have focused only on the
search for food and ignored the time needed to handle food. The
latter is ecologically relevant as many food items are not immedi-
ately available for consumption and require preparation (Giraldeau
& Caraco 2000;Bugnyar & Kotrschal 2002;Broom & Ruxton 2003;
Ha & Ha 2003;Morand-Ferron et al. 2007a;Broom et al. 2008), such
as the breaking of the protective shell of bivalves by many wading
bird species (Norton-Grifﬁths 1967;Hockey et al. 1989;Stillman
et al. 1997). Furthermore, whereas most PS research has focused
on the exploitation of food patches through scramble competition
(Giraldeau & Beauchamp 1999), individuals may use a variety of
other techniques to obtain food from others, such as stealthy
sequestering and food sharing (Brockmann & Barnard 1979;
Giraldeau & Caraco 2000;Bugnyar & Kotrschal 2002). Investigating
the social-foraging dynamics related to food handling and alter-
native scrounging tactics may therefore provide valuable insight
into social foraging and the diversity of tactics that is frequently
observed (Giraldeau 2008).
Recent studies on the conventional PS game, which focuses on
the exploitation of patches through scramble competition, have
shown that phenotypic differences may play a major role in
determining the use of PS tactics (Barta & Giraldeau 1998;
Beauchamp 2000,2006;Liker & Barta 2002;Lendvai et al. 2006;
Giraldeau 2008;Kurvers et al. 2009). For example, the payoffs of
searching for food (producing) and exploiting the discoveries made
by others (scrounging) are often strongly related to an individual’s
competitive ability as well as its explorative and risk-taking ten-
dencies, with bolder individuals mainly producing and dominants
mainly scrounging (Ens & Goss-Custard 1984;Barta & Giraldeau
*Correspondence: J. W. Jolles, Department of Zoology, University of Cambridge,
Downing Street, Cambridge CB2 3EJ, U.K.
E-mail address: firstname.lastname@example.org (J. W. Jolles).
Contents lists available at SciVerse ScienceDirect
journal homepage: www.elsevier.com/locate/anbehav
0003-3472/$38.00 Ó2013 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
Animal Behaviour 85 (2013) 1261e1269
1998;Stillman et al. 2000;Liker & Barta 2002;Lendvai et al. 2006;
Kurvers et al. 2010a). However, in a social-foraging context where
foraging requires extensive handling time and multiple scrounging
tactics are available, different phenotypic effects may be expected
(Brockmann & Barnard 1979;Bugnyar & Kotrschal 2002;Morand-
Ferron et al. 2007b;David et al. 2011). Extensive handling time
increases the investment of producing, making it especially costly
for lower-ranked individuals to produce because of their lower
competitive ability (e.g. Giraldeau 2008;Broom et al. 2008).
Furthermore, paired individuals may defend resources together
(Robichaud et al. 1996;Emery et al. 2007;Seed et al. 2008), tolerate
one another at a food patch (Beauchamp 2000;King et al. 2009)
and share food with each other (Emery et al. 2007;Seed et al.
2008). Therefore, paired and unpaired individuals are expected to
make different decisions as to whether to produce or scrounge. In
addition, consistent individual differences in boldness may be ex-
pected to affect tactic use differently in a context that involves a
large investment in food handling and in which alternative
scrounging tactics are available. Not only are bolder individuals
more likely to enter a feeding area ﬁrst (Dyer et al. 2009), they are
often the ﬁrst to discover food patches (Beauchamp 2001), and rely
less heavily on social information in making foraging decisions
(Kurvers et al. 2010a;Webster & Ward 2011; but see Marchetti &
Drent 2000). Bold individuals may therefore be expected to use
alternative and more risky scrounging tactics especially when
producing involves a long handling time. Thus, although in the
conventional PS context it has been shown that phenotypic dif-
ferences may play a major role on PS tactics, these and additional
phenotypic effects may be expected to play different roles in
foraging contexts in which food requires considerable handling
effort and/or in which alternative scrounging options are available.
Furthermore, investigating social factors in conjunction with
consistent individual differences in behaviour is very relevant as
they may be crucially linked to one another (Réale et al. 2007;
Scheid & Noë 2010;Webster & Ward 2011) and actually drive
consistent individual differences in the use of different foraging
strategies (Beauchamp 2001;Morand-Ferron et al. 2007a;
Bergmüller & Taborsky 2010).
We investigated the relationship between social-foraging tactic
use and three phenotypic factors, namely dominance, pair bonding
and boldness. Importantly, our foraging experiment extends the
conventional PS game by focusing on two often neglected factors:
time needed to handle food and alternative scrounging tactics,
namely joining a tolerant producer or retrieving food items drop-
ped on the ground by producers. To address these questions we
used rooks, Corvus frugilegus, as a model species as they are highly
social, form life-long pair bonds and establish a linear dominance
hierarchy (Clayton & Emery 2007;Emery et al. 2007). We presented
a captive group of 19 birds with bait attached to strings tied to
branches in their aviary, a set-up that has been successfully used in
avian cognitive research (Heinrich 1995;Heinrich & Bugnyar 2005).
Individuals could ‘produce’by successfully pulling up a baited
string, ‘scrounge’by joining or retrieving, or they could choose not
to participate at all. In this particular set-up, producing required a
considerable time investment and the ability to defend the
resource, while scrounging by displacing a producer was not viable
because of the high chance the string would drop. Rooks are rarely
observed foraging alone on the ground in their natural environ-
ment because it entails a higher risk of predation. Therefore,
scrounging by retrieving can be seen as naturally risky for rooks.
Furthermore, this tactic contains an element of uncertainty. A
scrounger’s success will depend not only on its own actions as there
is variability when food items fall after they have been successfully
pulled up by a producer. To determine the consistency of individual
tactic use we presented the foraging experiment in a low (one
baited string) and high (eight baited strings) food availability con-
dition. In addition, we measured individual feeding rates on a
maintenance diet to investigate the effect of hunger on foraging
motivation and behaviour (Lendvai et al. 2004). Prior to the social-
foraging experiment, we determined dominance ranks and afﬁlia-
tion scores for each individual and acquired boldness scores using a
novel object test (Kurvers et al. 2010a).
We predicted that: (1) dominance would be positively related to
the use and success of the producing tactic; (2) paired individuals
would beneﬁt from the high tolerance with their partner as it may
increase their producing success and the ability to scrounge by
joining; and (3) boldness would be positively related to scrounging
by retrieving. We expected individuals to be consistent in their
tactic use independent of food availability.
Subjects and Housing
A group of ﬁve male and 14 female adult rooks served as sub-
jects and were housed in an outdoor aviary (8 20 m and 3 m
high) at the University of Cambridge Sub-Department of Animal
Behaviour, Madingley, U.K. The birds were members of a group that
was collected under English Nature Permit 20030108 from two
colonies in Cambridge on 16 and 17 April 2003 and that were
subsequently hand-raised. Aviaries were constructed of wood and
mesh, had gravel ﬂoors and contained several perches of varying
height, width and length throughout. All individuals could be
identiﬁed by coloured leg rings. Birds had ad libitum access to
water and a maintenance diet of soaked dog biscuits, egg, cheese,
bread, cooked vegetables, seeds, nuts and fruit, except during the
experimental procedures when the group was deprived of their
diet for 4 h. The behavioural observations and boldness tests were
conducted from May to August 2010 and the social-foraging
experiment was conducted from August to October 2010. Birds
were kept for future behavioural experiments after the conclusion
of the study. All experiments were conducted in accordance with
the university’s guidelines under the U.K. Home Ofﬁce project
licence PPL 80/1975.
Prior to the social-foraging experiment we scored agonistic and
afﬁliative interactions to determine the dominance rank and pair
bonds in the group (Table A1 in the Appendix). Observations were
made from 10 m outside the aviary between 1300 and 1430 hours
for 16 days (MayeJuly 2010). Data were collected ad libitum
(Altmann 1974) and interactions were recorded as events with The
Observer 5.0 software (Noldus Information Technology, Wagenin-
gen, The Netherlands) and analysed with MatMan 1.1 (idem). We
scored 345 agonistic interactions, ranging from threats to active
chases with feather pulling.
Pairs were identiﬁed using ‘dyadic afﬁliation scores’(DAS) using
the formula: DAS ¼(afﬁliation directed from x to y þafﬁliation
directed from y to x)/(afﬁliation directed by x to all individuals in
the group þafﬁliation directed from y towards all individuals in the
group), with the calculations based on the ‘index of association’
(Martin & Bateson 2007). We separated afﬁliative behaviours into
active and passive afﬁliations (Table A1 in the Appendix). Active
afﬁliations are important for establishing and maintaining a pair
bond, while passive afﬁliations are related to interindividual
tolerance (Emery et al. 2007). Individuals were only considered
J. W. Jolles et al. / Animal Behaviour 85 (2013) 1261e12691262
paired if they had the highest DAS with each other in the active,
passive and combined afﬁliation categories.
We conducted a novel object test to determine individual
boldness scores. We use the term ‘boldness’for approaching a
baited novel object (Wilson et al. 1993;Greenberg & Mettke-
Hofmann 2001;Frost et al. 2007), although some authors have
proposed using the term ‘exploration’(Réale et al. 2007). For 40
trials each of 5 min (maximum ﬁve trials/day) we presented a novel
object next to a cup baited with eight waxmoth larvae, Achroia
grisella. We presented the novel object to the group rather than to
each individual separately because it has higher ecological rele-
vance (Webster et al. 2007), as in natural conditions it is likely that
rooks would be confronted with such a situation while in the
presence of conspeciﬁcs. The novel object was unique in each trial
(varying in material, colour and size) and its position in the aviary
was randomized across trials. For each trial we recorded which of
the 19 individuals approachedthe novel object to within 1 m. Based
on the average number of trials in which an individual approached
(4.3 1.45 trials) we classiﬁed individuals with approach scores
higher than the average as bold (N¼6) and those with scores lower
than the average as shy (N¼13). To exclude the possibility that
feeding motivation inﬂuenced approach scores in the novel object
test we ran an additional 40 control trials during which only the
baited food cup was presented. The conditions were presented
concurrently with randomized order for condition.
In the social-foraging experiment individuals could scrounge,
either by retrieving food items dropped on the ground by pro-
ducing individuals or by joining a tolerant producing individual,
produce, by pulling up baited strings in the aviary (see Heinrich &
Bugnyar 2005; photo sequence Fig. A1 in the Appendix), or
decide not to participate at all. A week before the start of testing,
the experimenter habituated the birds to the strings and the pro-
cedure of baiting the strings (Heinrich & Bugnyar 2005) by daily
entering the aviary, clearly showing the bait (a peanut or piece of
cheese), and randomly attaching it to the end of one of 10 already
present 50 cm long strings (jute twine) hanging off branches in the
aviary. All strings were positioned at 2 m height and minimally
1 m apart to make each string approachable by multiple birds
simultaneously. We minimized the effect of possible interindi-
vidual differences in string-pulling proﬁciency by starting the
experiment only after all individuals had been observed to pull up a
baited string successfully. In the test, we presented the rooks with
either a single baited string in 42 trials (one-string condition) or
eight baited strings in 35 trials (eight-string condition). The order of
conditions was pseudorandomized across trials such that the same
number of trials was run for each condition. Trials lasted for a
maximum of 5 min. Time between the experimenter entering and
leaving the aviary when baiting the strings was kept constant
across conditions. For each trial, we recorded which individuals
approached the string to within 1 m (‘producing attempt’), suc-
cessfully pulled up the string and obtained the bait (‘producing
success’), joined a producer and thereby obtained food (‘scrounge
by tolerance’) or landed directly under the string to retrieve
dropped food (‘scrounge by retrieving’). The measure of
approaching the string provides information about the extent to
which individuals would decide to bear the cost of approaching the
string. Scrounging by displacing a successful producer was not a
viable tactic because of the high probability that the string would
drop and thus food would again be out of reach. We used an
approach distance of 1 m from the string to indicate attempts to
produce as birds were always observed to land within this distance
when approaching the string to produce.
As a proxy of feeding motivation we measured individual
feeding rates at two adjacent feeding tables (50 50 cm and
140 cm high, approximately 150 cm apart) that are normally used
for the provisioning of the maintenance diet. After the string-
pulling experiment, for 10 consecutive days, we removed the
maintenance diet from the aviary for 4 h and recorded activity at
the tables for 20 min after providing the food again. We determined
the average percentage of time that an individual was on either
feeding table across the 10 trials (‘feeding rate’). The size of the
tables and the distance between them allowed multiple birds to
feed simultaneously. Together with the relatively long time frame
used for recording feeding rates, this set-up ensured that variation
in dominance ranks would not inﬂuence the measure of interest.
We organized all observed dominance interactions between
individuals in a sociometric matrix, which takes into account the
identity of each opponent and all the interactions and minimizes
inconsistencies (de Vries et al. 1993). To test for linearity we
calculated Kendall’s coefﬁcient of linearity K, Landau’s index hand
the index of linearity h
using MatMan 1.1 (Noldus Information
Technology). Statistical signiﬁcance of Kwas calculated using a chi-
square test. Both indices vary from 0 (complete absence of linearity)
to 1 (complete linearity). The index h
is based on hand takes into
account the existence of unknown relationships. Statistical signif-
icance of h
is provided by a resampling procedure using 10 000
randomizations (de Vries 1995). If the dominance is linear, MatMan
calculates a rank order most consistent with a linear hierarchy by
minimizing the number of inconsistencies and then minimizing the
total strength of inconsistencies (de Vries 1998).
Repeatability is a measure of the within-individual variance
compared with the between-individual variance and describes the
phenotypic variance explained by the individual (Kurvers et al.
2009). We determined the repeatability of individuals’perfor-
mance across the novel object trials by calculating the mean
squares from a one-way analysis of variance (ANOVA) with indi-
vidual as the main effect following the method of Lessells & Boag
(1987). We compared the number of birds that approached the
cup between the novel object trials and the control trials using a
ManneWhitney Utest. Spearman correlation tests were run to
determine whether approach scores in the novel object and control
condition were correlated.
We used generalized linear models (GLMs) to test which vari-
ables explained producing attempts, absolute and relative pro-
ducing success, and scrounging by retrieving for each condition.
Data were ﬁtted to a binomial distribution with logit-link function
and separate models were run with the number of trials in which
an individual attempted to produce, had producing success and
scrounged by retrieving as the numerator and the total number of
trials as denominator. For the GLMs on the data of the relative
producing success the number of trials in which an individual
attempted producing was used as the denominator. As ﬁxed effects,
we ﬁtted dominance (rank), if an individual was paired or not (yes/
J. W. Jolles et al. / Animal Behaviour 85 (2013) 1261e1269 1263
no), boldness (bold/shy) and feeding rate (continuous). Models did
not include sex because dominance was strongly related to sex and
additional models including sex revealed it had no signiﬁcant in-
ﬂuence on any of the other numerators. We started with full models
with all explanatory variables. Interactions were not included in the
models as based on our sample size they may be driven by one or
two data points or, conversely, a real interaction may not be
apparent owing to lack of data. Minimal adequate models were
obtained by comparing models with dropped terms using Akaike’s
information criterion (AIC), with the lowest AIC
the best model ﬁt, which is considered the preferable method for
observational data (Burnham et al. 2011). The results of the GLMs
are presented in Table 1. Because of this model selection procedure,
discussion of the GLMs’results are based on the effect sizes and not
on signiﬁcance of Pvalues. For completeness, both effect sizes and P
values are presented in Table 1. Correlations of producing attempts
and scrounging by retrieving between the low food and high food
availability conditions were analysed using the Spearman rank
). Results of all statistics other than the GLMs were
interpreted based on the signiﬁcance of Pvalues, with P<0.10
reported as a trend and P<0.05 as signiﬁcant; means are stated
SE throughout. All data were analysed in R 2.15.2 (R Development
Core Team 2012).
Rooks had a linear dominance hierarchy (Kendall’s coefﬁcient
K¼0.53, P<0.001; Landau’s linearity index h¼0.53, h
P<0.001), such that all birds could be assigned a rank from 1 (most
subordinate) to 19 (most dominant; de Vries 1995). All males were
dominant over all females.
In both the one-string and eight-string conditions, dominance
was positively related to the total number of attempts to produce
and the total number of producing successes (Fig. 1)aswellasto
the proportion of successful trials out of all producing attempts
(Table 1). Dominance did not play a role in scrounging by retrieving.
Based on our observation of 340 active and 516 passive afﬁlia-
tions six pairs were identiﬁed, one of which consisted of two fe-
males. Four of these pairs had built nests in the previous breeding
season. Paired individuals performed fewer producing attempts
than unpaired individuals (one string: paired: 24.0 6.0%; un-
paired: 31.9 5.5%; eight strings: paired: 38.9 5.7%; unpaired:
57.1 12%). Although in the one-string condition the pair bond did
not affect absolute producing success, paired birds had a higher
proportion of successful producing attempts (relative producing
success) than unpaired birds (paired: 33.3 7%; unpaired:
6.6 4%; Table 1). In 33 of the 42 trials, producing success was
achieved by a paired individual, 21 of which when together with its
partner. For 18 of these 21 trials the male was the ﬁrst to arrive and
pull the string; in 17 of these 21 cases the joining partner received
some of the food, and in seven of 15 trials this occurred after both
birds had pulled the string together. Furthermore, in 17 of 21 cases
the primary producer was successful only after its partner had
arrived at the string. In contrast to the one-string condition, in the
eight-string condition not only the absolute success but also the
relative success of paired birds was lower than that of unpaired
birds (Table 1). Scrounging by retrieving was negatively related to
the pair bond, in both the one-string and eight-string conditions
(Table 1,Fig. 2a).
Individuals were highly consistent in their approaching behav-
iour in the novel object condition; the repeatability of individuals’
approaches during the ﬁrst and second half of trials was high (0.93).
Although more birds approached in each trial during the control
than the novel object condition (U¼240.50, N
P<0.001), there was no signiﬁcant correlation between the num-
ber of trials in which an individual approached in the novel object
test and a control condition (40 trials; r
P¼0.876), nor did the number of approaches of shy individuals
across the two conditions counterbalance those of bold individuals,
suggesting foraging motivation did not affect approach scores in
the novel object test.
Consistent individual differences in boldness were negatively
related to both producing attempts and producing successes.
Boldness did not affect relative producing success. These effects
Dominance High >
Ratio of trials produced successfully
Dominance High >< Low
Figure 1. Inﬂuence of dominance rank on producing success in the (a) one-string
(N¼42 trials) and (b) eight-string (N¼35 trials) condition. Points are mean values
from raw data with ratio of trials based on the total number of trials per condition. Line
shows predicted means SE from the minimal model, thus controlling for other sig-
niﬁcant terms in the model.
J. W. Jolles et al. / Animal Behaviour 85 (2013) 1261e1269126 4
were observed in both the one-string and eight-string conditions
(Table 1). In contrast, scrounging by retrieving food items fallen on
the ground was positively affected by boldness, with bold in-
dividuals scrounging considerably more than shy individuals, seen
in both food availability conditions (Table 1,Fig. 2b).
In the one-string condition, individuals with higher feeding
rates made more producing attempts. However, feeding rate did
not affect the number of successful pulls and was negatively related
to relative producing success (Table 1). In contrast, although in the
eight-string condition feeding rate was also positively related to
producing attempts, individuals with higher feeding rates had
more producing success, both in terms of absolute numbers and in
the proportion of pulls that were successful (Table 1). An in-
dividual’s feeding rate did not inﬂuence scrounging by retrieving in
the one-string condition and was negatively related to scrounging
by retrieving in the eight-string condition (Table 1).
Individual Consistency in Tactic Use
In each trial, individuals could either engage in one of the
foraging behaviours or not participate at all. On average, in the one-
string condition, individuals participated in 31.3 3.1% of the trials,
with producing attempts accounting for 27.0 3.9% and
scrounging by retrieving for 7.8 1.7% of trials. In less than 2% of
trials did individuals attempt both to produce and to scrounge,
which was not related to the frequency of producing attempts
¼0.123, P¼0.615) or the relative producing success (r
P¼0.714). The more individuals attempted to produce the less they
scrounged by retrieving (r
¼0.455, P<0.05). In the eight-string
condition, individuals participated on average in 69.0 24.9% of
the trials, with producing attempts accounting for 49.2 24.3% and
scrounging by retrieving for 19.9 18.1% of trials; they did not
attempt to produce in 50% of the trials in which they scrounged.
Individuals were highly consistent in their tactic use across the two
food availability conditions, both in terms of producing attempts
¼0.879, N¼19, P<0.001; Fig. 3a) and in scrounging by
¼0.755, N¼19, P<0.001; Fig. 3b).
In this study we investigated the use of various foraging tactics
by rooks in a PS game that focused on investment in food handling
and alternative scrounging tactics. In particular, we assessed how
two social factors (namely dominance rank and whether a bird was
paired or not) and individual differences in boldness inﬂuenced the
use of producing and scrounging tactics. We tested a group of rooks
in a string-pulling experiment and determined individual consis-
tency in tactic use across conditions that differed in food avail-
ability. The experiment revealed three key ﬁndings. First, dominant
rooks adopted the producer tactic more often, and more success-
fully, than subordinate rooks. Second, paired individuals often
attempted to produce together, which increased the scrounging
success of the joining partner, but resulted in mixed success for the
producing partner. Third, bold individuals scrounged more often by
retrieving than shy individuals. Individuals were highly consistent
in tactic use across conditions with low and high food availability.
Dominance has increasingly been taken into account in studies
on foraging dynamics (Barta & Giraldeau 1998;Stillman et al. 2000;
Liker & Barta 2002;Beauchamp 2006;Lendvai et al. 2006). The
majority of these studies have used the conventional PS game, in
which individuals can either search for food or exploit the discov-
eries made by others, and found that higher-ranked individuals
produce less and scrounge more than subordinates (but see
Beauchamp 2006). These ﬁndings contrast with the result of our
study. Here, dominants made more producing attempts and were
GLMs of approaching, absolute and relative success, and retrieving in the one-string and eight-string conditions of the social-foraging experiment
Fixed effect One-string condition zPEight-string condition zP
(Intercept) 1.410.29 4.80 <0.001 0.540.29 1.88 0.06
Dominance 0.040.02 2.06 0.04 0.050.02 2.63 <0.01
Paired 0.420.21 2.05 0.04 0.520.21 2.50 0.01
Bold 0.360.20 1.79 0.07 0.670.19 3.42 <0.001
Feeding rate 0.070.03 2.75 <0.01 0.110.03 4.12 <0.001
Producing success (absolute)
(Intercept) 4.000.45 8.96 <0.001 1.700.32 5.27 <0.001
Dominance 0.140.03 4.47 <0.001 0.100.02 5.03 <0.001
Paired 0.870.22 3.92 <0.001
Bold 0.680.40 1.72 0.09 0.420.21 1.95 0.05
Feeding rate 0.130.03 4.44 <0.001
Producing success (relative)
(Intercept) 3.290.68 4.87 <0.001 0.640.47 1.35 0.176
Dominance 0.230.05 4.65 <0.001 0.140.03 4.86 <0.001
Paired 0.750.41 1.81 0.07 0.780.33 2.35 0.02
Feeding rate 0.150.08 1.98 0.05 0.100.05 2.08 0.04
Scrounging by retrieving
(Intercept) 2.490.25 10.09 <0.001 1.130.30 3.75 <0.001
Paired 0.770.28 2.89 <0.01 0.700.24 2.94 <0.01
Bold 0.990.27 3.62 <0.001 1.250.21 6.04 <0.001
Feeding rate 0.070.03 2.25 0.02
Minimal models were selected using an information theoretical model selection approach and results are discussed based on coefﬁcient estimates. For completeness, Pvalues
are also presented. Coefﬁcient estimates represent the change in the dependent variable relative to the baseline category and can therefore be interpreted as measures of effect
size. Data were ﬁtted to a binomial distribution with a logit-link function. The analyses were conducted on all 19 individuals.
J. W. Jolles et al. / Animal Behaviour 85 (2013) 1261e1269 1265
more successful at doing so than subordinates. This contrasting
effect is likely to be strongly inﬂuenced by the extensive handling
time needed to pull up a string successfully. This may change the
payoffs and the probability of producing success in such a waythat,
for subordinates, attempting to produce might become too costly.
This effect might have been strengthened further by the ineffec-
tiveness of scrounging by displacing an individual after it had
successfully pulled up the string as this would result in it dropping
the string, thereby putting the bait out of immediate reach. In a
situation with a relatively high availability of food, the probability
of being displaced while attempting to produce is expected to be
low and producing success to be high. If this goes together with a
reduction in clumping of the food, this may minimize the beneﬁts
of being high in dominance status for food intake (see Theimer
1987;Giraldeau 2008). However, when we presented the birds
with eight baited strings (thus high food availability with a
reduction in food clumping), dominance was again positively
related to both the amount individuals attempted to produce and
their success in producing. Although this result might to a certain
extent have been inﬂuenced by the feedback related to the prob-
ability of producing success between the two conditions, the in-
vestment in food handling is expected to play the major role.
Despite food being much more available and less clumped, the
probability of producing success may still be low because of the
probability of being displaced still being high owing to the exten-
sive handling time. Overall, these ﬁndings thus seem to suggest
that a large investment in food handling may have an important
and consistent effect on the payoffs of producing related to
The pair bond (paired/unpaired) had a strong effect on how
often birds attempted to produce and were successful as well as
0.2 0.3 0.4 0.5 0.6 0.70
0.1 0.15 0.2 0.25
Ratio of trials scrounging by retrieving 
rS = 0.879
P < 0.001
rS = 0.755
P < 0.001
Ratio of trials attempting producing 
Ratio of trials attempting producing 
Ratio of trials scroun
Figure 3. Consistency in tactic use across conditions differing in food availability for
attempts of (a) producing and (b) scrounging by retrieving (one string : N¼42;
eight strings : N¼35). Shown are mean SE ratio of trials out of the total number
of trials per condition during which an individual (a) attempted to produce or (b)
scrounged by retrieving.
8 strings1 string
Ratio of trials scrounging by retrieving
Figure 2. Inﬂuence of (a) the pair bond and (b) boldness on scrounging by retrieving in
both the one-string and eight-string condition. Shown are mean SE ratio of trials out
of the total number of trials per condition during which an individual used the
scrounging by retrieving tactic. Paired individuals: N¼12; unpaired individuals:
N¼7; bold individuals: N¼6; shy individuals N¼13.
J. W. Jolles et al. / Animal Behaviour 85 (2013) 1261e1269126 6
how often they scrounged by tolerance and by retrieving fallen food
items. When food availability was low, paired individuals had a
higher chance of producing successfully than unpaired birds. On
average, in one out of two trials, paired individuals joined each
other at a producing attempt and, in more than 80% of these at-
tempts, success was achieved after the primary producer was
joined by its mate. This seems to suggest that paired birds beneﬁt
from their ability to defend resources together (Robichaud et al.
1996;Emery et al. 2007;Seed et al. 2008) and, furthermore, that
unpaired individuals may have to work harder to get their share of
the food. This is indeed supported by the ﬁnding that unpaired
individuals scrounged more by retrieving and performed more
producing attempts than paired individuals. However, looking at
the costs and beneﬁts of the pair bond in more detail, we see that
having a partner did not inﬂuence absolute producing success in
the one-string condition. In addition, paired individuals might join
their producing partners preferentially only at the moment when
producing success seems likely, thus reducing the beneﬁtof
increased defensibility. Furthermore, in more than 80% of cases the
joining partner received (some of) the food (either by food sharing
or taking (some of) the bait); thus producing for paired birds may
actually be costly. This is supported by the ﬁnding that when food
availability was high, both absolute and relative producing success
was lower for paired compared to unpaired birds. Thus, although
for paired individuals scrounging by tolerance was a highly suc-
cessful strategy (see also Bugnyar & Kotrschal 2002), producing
when paired might actually be relatively costly in the short term,
especially when food availability is high. This is interesting as
scrounging by tolerance was performed by the female of a pair 85%
of the time. However, a producing male may still beneﬁt from
tolerating its partner’s scrounging, as this may strengthen their pair
bond (Emery et al. 2007;von Bayern et al. 2007) and ultimately
result in higher ﬁtness beneﬁts (e.g. increased mating opportu-
nities). To our knowledge such a nuance has never been addressed
in PS models. This might also explain the high levels of social
tolerance between the individuals in a pair (Beauchamp 2000;Seed
et al. 2008): during seven trials we observed paired birds to pro-
duce successfully by pulling the string together.
Consistent individual differences in boldness had a very strong
effect on use of the scrounging tactic. As predicted, bold individuals
scrounged more by retrieving fallen food items than shy individuals,
with both low and high food availability. We predicted this effect
because this particular scrounging technique entails both an
element of risk of predation and uncertainty because of dependence
on the producer’s action. In the wild, single rooks are rarely seen
foraging on the ground owing to a high risk of predation. Despite the
absence of predators in the aviary setting, the captive-housed rooks
spend most of their time away from the ground and in most cases
will land on the ground with multiple individuals simultaneously
(J. W. Jolles, unpublished data). Furthermore, retrieving fallen food
items contains an element of uncertainty that is beyond the action of
the scrounger, as food items will not fall during or after every suc-
cessful producing action. Although Kurvers et al. (2010a) reported
that boldness was positively related toproducing in barnacle geese,
Branta leucopsis, which seems to contradict our ﬁnding, both results
may be explained by the commonality of risk and uncertainty in
foraging tactics. In Kurvers et al.’s study, individuals could produce
by discovering new food patches, which is a more risky and uncer-
tain tactic than scrounging on these new discoveries. This would
suggest that the relationship between boldness and the use of pro-
ducing and scrounging tactics might depend on the risk and un-
certainty involved in both tactics. This hypothesis is supported by
the ﬁnding that bold individuals are often the ﬁrst to enter a feeding
area and discover food patches (Beauchamp 2001;Dyer et al. 2009)
and are more ﬂexible in their responsiveness (Marchetti & Drent
2000;Frost et al. 2007). Owing to the small size of dropped food
items and the difﬁculty of discerning them from the ground sub-
strate, we could not investigate the determinants of scrounging
propensity and scrounging success in more detail. Further studies
are needed to provide insights into the relative beneﬁts and costs of
this scrounging tactic and their effect on foraging choices.
In common with previous studies (Beauchamp 2001,2006;
Morand-Ferron et al. 2007a;Kurvers et al. 2010a), we found that in-
dividuals were highly consistent in their use of foraging tactics. In-
dividuals chose the same tactics in both food availability conditions,
even though high food availability could have enabled more in-
dividuals to choose successfully the more proﬁtable producing strat-
egy. Dominant individuals made more producing attempts and were
more successful than subordinates, while, in contrast, scrounging by
retrieving was positively related to boldness and the pair bond. This
effect was observed irrespective of food availability. These ﬁndings
may indicate that dominance, boldness and the pair bond lower the
ﬂexibility of individuals’tactic choices. Still, the current state of in-
dividuals may playa role in their tactic use, as is substantiated by the
ﬁndingthat individual differences in motivationto feed (feeding rates)
had opposite effects on producing success under low and high food
availability. Overall, these results suggest an interplay between social
factors and individual differences in boldness that has an important
effect on individual consistency in tactic use. This would support the
idea of social niche specialization which suggests that individuals
adopt different behavioural strategies to reduce competition with
other group members (Bergmüller & Taborsky 2010).
Our social-foraging experiment conforms to the ﬁrst and main
assumption of PS games, namely tactic incompatibility (Coolen
et al. 2001), and enabled the study of food handling and alterna-
tive scrounging tactics in a social-foraging context. According to the
second assumption of PS games, scrounger payoffs should be
negatively frequency dependent (Mottley & Giraldeau 2000). Our
results show that the investment in handling food may be an
important driving factor for individual foraging tactic choices and
that social factors and boldness may have strong and opposite ef-
fects relative to the conventional PS game. This may have impli-
cations for future studies on PS dynamics. Although foraging
efﬁciency may play a large role in foraging tactic use under certain
conditions (see e.g. Stillman et al. 2000), we do not expect this to be
the case in our study. In our social-foraging experiment food-
handling time was long and competitor density high, factors un-
der which the inﬂuence of foraging efﬁciency is expected to be
minor (Stillman et al. 2000). Furthermore, only after all individuals
had already been observed to pull up a baited string successfully
did we start with the social-foraging task. Given that persistent
individual differences may often be related to dynamics that are
peculiar to the group in which they occur (David et al. 2011;
Morand-Ferron et al. 2011; but see Kurvers et al. 2010b), follow-up
research on our social-foraging experiment should also study howa
change in social composition may affect individual tactic use.
Finally, more research is needed to investigate the payoffs of the
various foraging tactics in relation to food-handling time, and how
these payoffs may differ between different phenotypes.
To summarize, our study shows that individuals were highly
consistent in the use of producing and scrounging tactics in a
social-foraging context that focused on food handling and was
strongly inﬂuenced by dominance, the pair bond and boldness. Our
results add to a growing literature on heterogeneity in group
behaviour stressing the importance of not treating group members
as identical or interchangeable (Dyer et al. 2009;Nagy et al. 2010;
Webster & Ward 2011;Jolles et al. 2013). Moreover, our study
highlights the importance of considering social factors as well as
consistent individual differences (often referred to as ‘animal per-
sonality’) for investigating social-foraging dynamics. The current
J. W. Jolles et al. / Animal Behaviour 85 (2013) 1261e1269 1267
study offers empirical data on food-handling time and alternative
scrounging tactics that could be used to extend both current
models and experiments on social foraging.
We are grateful to the editor and two anonymous referees
whose comments have greatly improved the manuscript. This work
was supported by grants from the Erasmus Programme and the
University of Cambridge. We thank Alex Thornton and Neeltje
Boogert for valuable discussion and feedback, and Ivan Vakrilov and
Charmaine Donovan for avian husbandry.
Altmann, J. 1974. Observational study of behavior: sampling methods. Behaviour,
Barnard, C. J. & Sibly, R. M. 1981. Producers and scroungers: a general model and its
application to captive ﬂocks of house sparrows. Animal Behaviour,29, 543e550.
Barta, Z. & Giraldeau, L.-A. 1998. The effect of dominance hierarchy on the use of
alternative foraging tactics: a phenotype-limited producing-scrounging game.
Behavioral Ecology and Sociobiology,42,217e223.
von Bayern, A. M. P., de Kort, S. R., Clayton, N. S. & Emery, N. J. 2007. The role of
food- and object-sharing in the development of social bonds in juvenile jack-
daws (Corvus monedula). Behaviour,144,711e733.
Beauchamp, G. 2000. The effect of prior residence and pair bond on scrounging
choices in ﬂocks of zebra ﬁnches, Taenopygia guttata.Behavioural Processes,52,
Beauchamp, G. 2001. Consistency and ﬂexibility in the scrounging behaviour of
zebra ﬁnches. Canadian Journal of Zoology,79, 540e544.
Beauchamp, G. 2006. Phenotypic correlates of scrounging behavior in zebra
ﬁnches: role of foraging efﬁciency and dominance. Ethology,112 ,873e878.
Bergmüller, R. & Taborsky, M. 2010. Animal personality due to social niche
specialisation. Trends in Ecology & Evolution,25, 504e511.
Brockmann, H. J. & Barnard, C. J. 1979. Kleptoparasitism in birds. Animal Behaviour,
Broom, M. & Ruxton, G. D. 2003. Evolutionarily stable kleptoparasitism: conse-
quences of different prey types. Behavioral Ecology,14,2
Broom, M., Luther, R. M., Ruxton, G. D. & Rychtár, J. 2008. A game-theoretic model
of kleptoparasitic behavior in polymorphic populations. Journal of Theoretical
Bugnyar, T. & Kotrschal, K. 2002. Scrounging tactics in free-ranging ravens, Corvus
Burnham, K. P., Anderson, D. R. & Huyvaert, K. P. 2011. AIC model selection and
multimodel inference in behavioral ecology: some background, observations,
and comparisons. Behavioral Ecology and Sociobiology,65,23e35.
Clayton, N. S. & Emery, N. J. 2007. The social life of corvids. Current Biology,17,
Coolen, I., Giraldeau, L.-A. & Lavoie, M. 2001. Head position as an indicator of
producer and scrounger tactics in a ground-feeding bird. Animal Behaviour,61,
David, M., Cézilly, F. & Giraldeau, L.-A. 2011. Personality affects zebra ﬁnch feeding
success in a producerescrounger game. Animal Behaviour,82,61e67.
Dyer, J. R. G., Croft, D. P., Morrell, L. J. & Krause, J. 2009. Shoal composition de-
termines foraging success in the guppy. Behavioral Ecology,20,165e171.
Emery, N. J., Seed, A. M., von Bayern, A. M. P. & Clayton, N. S. 2007. Cognitive
adaptations of social bonding in birds. Philosophical Transactions of the Royal
Society B,362, 489e505.
Ens, B. J. & Goss-Custard, J. D. 1984. Interference among oystercatchers, Haema-
topus ostralegus, feeding on mussels, Mytilus edulis, on the Exe Estuary. Journal
of Animal Ecology,53,217e231.
Frost, A. J., Winrow-Giffen, A., Ashley, P. J. & Sneddon, L. U. 2007. Plasticity in
animal personality traits: does prior experience alter the degree of boldness?
Proceedings of the Royal Society B,274,333e339.
Giraldeau, L.-A. 2008. Social foraging and the study of exploitative behavior. Ad-
vances in the Study of Behavior,38,59e104.
Giraldeau, L.-A. & Beauchamp, G. 1999. Food exploitation: searching for the
optimal joining policy. Trends in Ecology & Evolution,14,102e106.
Giraldeau, L.-A. & Caraco, T. 2000. Social Foraging Theory. Princeton, New Jersey:
Princeton University Press.
Giraldeau, L.-A., Soos, C. & Beauchamp, G. 1994. A test of the producer-scrounger
foraging game in captive ﬂocks of spice ﬁnches, Lonchura punctulata.Behavioral
Ecology and Sociobiology,34,251e256.
Greenberg, R. & Mettke-Hofmann, C. 2001. Ecological aspects of neophobia and
neophilia in birds. Current Ornithology,16,119e178.
Ha, R. R. & Ha, J. C. 2003. Effects of ecology and prey characteristics on the use of
alternative social foraging tactics in crows, Corvus caurinus.Animal Behaviour,
Heinrich, B. 1995. An experimental investigation of insight in common ravens
(Corvus corax). The Auk,112, 994e1003.
Heinrich, B. & Bugnyar, T. 2005. Testing problem solving in ravens: string-pulling
to reach food. Ethology,111, 962e976.
Hockey, P. A. R., Ryan, P. G. & Bosman, A. L. 1989. Age-related intraspeciﬁc klep-
toparasitism and foraging success of kelp gulls Larus dominicanus.Ardea,77,
Jolles, J. W., King, A. J., Manica, A. & Thornton, A. 2013. Heterogeneous structure
in mixed-species corvid ﬂocks in ﬂight. Animal Behaviour,85,743e750.
King, A. J., Isaac, N. J. B. & Cowlishaw, G. 2009. Ecological, social, and reproductive
factors shape producer-scrounger dynamics in baboons. Behavioral Ecology,20,
Kurvers, R. H. J. M., Eijkelenkamp, B., van Oers, K., van Lith, B., van Wieren, S. E.,
Ydenberg, R. C. & Prins, H. H. T. 2009. Personality differences explain leader-
ship in barnacle geese. Animal Behaviour,78,447e453.
Kurvers, R. H. J. M., Prins, H. H. T., van Wieren, S. E., van Oers, K., Nolet, B. A. &
Ydenberg, R. C. 2010a. The effect of personality on social foraging: shy barnacle
geese scrounge more. Proceedings of the Royal Society B,277,601e608.
Kurvers, R. H. J. M., van Oers, K., Nolet, B. A., Jonker, R. M., van Wieren, S. E.,
Prins, H. H. T. & Ydenberg, R. C. 2010b. Personality predicts the use of social
information. Ecology Letters,13, 829e837.
Lendvai, Á. Z., Barta, Z., Liker, A. & Bókony, V. 2004. The effect of energy reserves
on social foraging: hungry sparrows scrounge more. Proceedings of the Royal
Society B,271, 2467e2472.
Lendvai, Á. Z., Liker, A. & Barta, Z. 2006. The effects of energy reserves and
dominance on the use of social-foraging strategies in the house sparrow. Animal
Lessells, C. M. & Boag, P. T. 1987. Unrepeatable repeatabilities: a common mistake.
Liker, A. & Barta, Z. 2002. The effects of dominance on social foraging tactic use in
house sparrows. Behaviour,139,1061e1076.
Logan, C. J., Emery, N. J. & Clayton, N. S. 2012. Alternative behavioral measures of
postconﬂict afﬁliation. Behavioral Ecology,24,9
Marchetti, C. & Drent, P. J. 2000. Individual differences in the use of social infor-
mation in foraging by captive great tits. Animal Behaviour,60,131e140.
Martin, P. & Bateson, P. 2007. Measuring Behaviour: an Introductory Guide. Cam-
bridge: Cambridge University Press.
Morand-Ferron, J., Giraldeau, L. & Lefebvre, L. 2007a. Wild Carib grackles play a
producer scrounger game. Behavioral Ecology,18,916e921.
Morand-Ferron, J., Sol, D. & Lefebvre, L. 2007b. Food stealing in birds: brain or
brawn? Animal Behaviour,74,1725e1734.
Morand-Ferron, J., Wu, G.-M. & Giraldeau, L.-A. 2011. Persistent individual dif-
ferences in tactic use in a producerescrounger game are group dependent.
Mottley, K. & Giraldeau, L.-A. 2000. Experimental evidence that group foragers can
converge on predicted producer-scrounger equilibria. Animal Behaviour,60,
Nagy, M., Akos, Z., Biro, D. & Vicsek, T. 2010. Hierarchical group dynamics in pi-
geon ﬂocks. Nature,464, 890e893.
Norton-Grifﬁths, M. 1967. Some ecological aspects of the feeding behaviour of the
oystercatcher Haematopus ostralegus on the edible mussel Mytilus edulis.Ibis,
Robichaud, D., Lefebvre, L. & Robidoux, L. 1996. Dominance affects resource
partitioning in pigeons, but pair bonds do not. Canadian Journal of Zoology,74,
R Development Core Team. 2012. R: A Language and Environment for Statistical
Computing. Vienna: R Foundation for Statistical Computing. http://www.R-
Réale, D., Reader, S. M., Sol, D., McDougall, P. T. & Dingemanse, N. J. 2007. Inte-
grating animal temperament within ecology and evolution. Biological Reviews,
Scheid, C. & Noë, R. 2010. The performance of rooks in a cooperative task depends
on their temperament. Animal Cognition,13, 545e553.
Seed, A. M., Clayton, N. S. & Emery, N. J. 2008. Cooperative problem solving in
rooks (Corvus frugilegus). Proceedings of the Royal Society B,275, 1421e1429.
Stillman, R. A., Goss-Custard, J. D. & Caldow, R. W. G. 1997. Modelling interference
from basic foraging behaviour. Journal of Animal Ecology,66, 692e703.
Stillman, R. A., Caldow, R. W. G., Goss-Custard, J. D. & Alexander, M. J. 2000.
Individual variation in intake rate: the relative importance of foraging efﬁciency
and dominance. Journal of Animal Ecology,69, 484e493.
Theimer, T. 1987. The effect of seed dispersion on the foraging success of domi-
nant and subordinate dark-eyed juncos, Junco hyemalis.Animal Behaviour,35,
de Vries, H.1995. An improved test of linearity in dominance hierarchies containing
unknown or tied relationships. Animal Behaviour,50,1375e1389.
de Vries, H. 1998. Finding a dominance order most consistent with a linear hier-
archy: a new procedure and review. Animal Behaviour,55,827e843.
de Vries, H., Netto, W. J. & Hanegraaf, P. L. H. 1993. MatMan: a program for the
analysis of sociometric matrices and behavioural transition matrices. Behaviour,
Webster, M. M. & Ward, A. J. W. 2011. Personality and social context. Biological
Webster, M. M., Ward, A. J. W. & Hart, P. J. B. 2007. Boldness is inﬂuenced by social
context in threespine sticklebacks (Gasterosteus aculeatus). Behaviour,144,351e371.
Wilson, D. S., Coleman, K., Clark, A. B. & Biederman, L. 1993. Shy-bold continuum
in pumpkinseed sunﬁsh (Lepomis gibbosus): an ecological study of a psycho-
logical trait. Journal of Comparative Psychology,107, 250e260.
J. W. Jolles et al. / Animal Behaviour 85 (2013) 1261e1269126 8
Figure A1. Sequence of photos that shows a successful producing attempt as seen in the social-foraging experiment. After arriving above the string (a), an individual had to grab the
string with its beak (b), pull up the string (c) and stand on the string (d), and perform this sequence of behaviours up to three times (e) in order to get to the food item (f).
Ethogram of agonistic and afﬁliative interactions used to determine the dominance rank and pair bonds in the group
Displacement One individual retreats at the approach of another, which locates itself in the retreating individual’s original position
Threaten One individual makes a movement directly at, pecks at, ﬂies at or lunges at another without making contact
Pecking One individual uses its bill to peck at and make contact with another individual
Feather pulling One individual grabs the feather(s) of another individual and makes a pulling motion
Chasing A prolonged, continuous approach by one bird towards another bird that continuously moves away
Avoiding One individual moves out of the way as another individual approaches
Bill twining Two individuals interlock the mandibles of their bills, often accompanied by simultaneous displaying
Active food sharing One individual places a food item into the bill of another individual, often after the recipient makes begging calls
Sitting in contact Two individuals sit next to each other, maximum 5 cm apart
Allopreening One individual nibbles or strokes the feathers of another
Dual object manipulation Two birds manipulating the same object
Dual caching & cache recovery Two individuals caching the same object or recovering and eating a previously cached food item
Co-feeding Two individuals foraging while in proximity to each other.
Sitting in proximity Two individuals sit within one body length of each other, but more than 5 cm apart
Passive food sharing One individual places a food item in front of another individual after which they eat from it together
Developed in collaboration with Corina Logan (Logan et al. 2012).
J. W. Jolles et al. / Animal Behaviour 85 (2013) 1261e1269 1269