Status discrimination through fertility signalling allows ants to regulate reproductive conflicts

Abstract

Dominance hierarchies allow group-living animals to regulate the partitioning of reproduction, but the
recognition systems underlying dominance interactions remain equivocal. Individual recognition, a
cognitively complex recognition system, is often posited as an important mechanism for the regulation
of linear dominance hierarchies because of its high level of precision. However, providing it actually
allows a fine-scale discrimination of the individuals’ statuses, status discrimination may offer an alternative,
simpler, recognition system allowing the same level of precision while saving the memory-related
costs associated with individual recognition. With the aim of disentangling the cognitive
mechanisms underlying the formation and maintenance of hierarchies, we here studied the within-group
recognition systems in the ant Neoponera apicalis, where orphaned workers compete over male
parentage in a linear hierarchical structure. Overall, we found that status discrimination abilities were in
fact sufficient for the establishment and stabilization of linear hierarchies. The observed level of accuracy
allowed fine-scale discrimination of all top rankers’ hierarchical status, and thus translated into a
functional individual discrimination of all competing workers at the top of the hierarchy. Low-ranking
workers did not exhibit such fine-scale status discrimination. We moreover showed that a putative
signal of fertility, 13-methylpentacosane, precisely labelled the workers’ position in the hierarchy,
thereby providing the recognition cue likely to explain the individuals’ discrimination abilities. This
signal could therefore play a key role in the regulation of the reproductive conflict in this species. In
contrast with the traditional view, our study shows the implication of a cognitively simple but equivalently
efficient recognition system during the emergence and stabilization of a linear dominance
hierarchy.

Full text

Status discrimination through fertility signalling allows ants to
regulate reproductive conflicts
Boris Yagound
a
,
*
, Pierre Blacher
a
, Dominique Fresneau
a
, Chantal Poteaux
a
,
Nicolas Châline
a
,
b
a
Laboratoire d’Ethologie Expérimentale et Comparée, E.A. 4443, Université Paris 13, Sorbonne Paris Cité, Villetaneuse, France
b
Department of Experimental Psychology, Institute of Psychology, University of São Paulo, São Paulo, Brazil
article info
Article history:
Received 6 January 2014
Initial acceptance 10 February 2014
Final acceptance 25 March 2014
Published online
MS. number: 14-00012R
Keywords:
cuticular hydrocarbons
dominance hierarchy
individual recognition
Neoponera (formerly Pachycondyla)apicalis
recognition system
status discrimination
Dominance hierarchies allow group-living animals to regulate the partitioning of reproduction, but the
recognition systems underlying dominance interactions remain equivocal. Individual recognition, a
cognitively complex recognition system, is often posited as an important mechanism for the regulation
of linear dominance hierarchies because of its high level of precision. However, providing it actually
allows a fine-scale discrimination of the individuals’statuses, status discrimination may offer an alter-
native, simpler, recognition system allowing the same level of precision while saving the memory-
related costs associated with individual recognition. With the aim of disentangling the cognitive
mechanisms underlying the formation and maintenance of hierarchies, we here studied the within-
group recognition systems in the ant Neoponera apicalis, where orphaned workers compete over male
parentage in a linear hierarchical structure. Overall, we found that status discrimination abilities were in
fact sufficient for the establishment and stabilization of linear hierarchies. The observed level of accuracy
allowed fine-scale discrimination of all top rankers’hierarchical status, and thus translated into a
functional individual discrimination of all competing workers at the top of the hierarchy. Low-ranking
workers did not exhibit such fine-scale status discrimination. We moreover showed that a putative
signal of fertility, 13-methylpentacosane, precisely labelled the workers’position in the hierarchy,
thereby providing the recognition cue likely to explain the individuals’discrimination abilities. This
signal could therefore play a key role in the regulation of the reproductive conflict in this species. In
contrast with the traditional view, our study shows the implication of a cognitively simple but equiv-
alently efficient recognition system during the emergence and stabilization of a linear dominance
hierarchy.
Ó2014 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
The existence of recognition systems is a central feature of group
living. Recognition is used in a wide range of social interactions,
thereby allowing group members to adapt their behaviour ac-
cording to the age, sex, kinship, group membership, hierarchical
status, reproductive status, species and neighbourhood of the in-
dividuals with which they interact (Sherman, Reeve, & Pfennig,
1997; Thom & Hurst, 2004; Tibbetts & Dale, 2007). Understanding
the exact nature of the recognition mechanisms across taxa, their
contexts and associated costs and benefits is therefore a major
challenge in the biological sciences (Wiley, 2013).
Dominance hierarchies are widespread throughout the animal
kingdom. These hierarchies are characterized by asymmetries
among group members in the partitioning of resources (Zanette &
Field, 2009), and can induce important fitness consequences by
mediating access to reproduction, food resources or susceptibility
to diseases (Ellis, 1995). Nevertheless, the overt aggression often
associated with these hierarchies can also bear important costs in
terms of time, energy, physicalinjuries or vulnerability to predators
(Hsu, Earley, & Wolf, 2006; Rutte, Taborsky, & Brinkhof, 2006).
Reducing these costs may imply the use of ritualization mecha-
nisms, as is frequently observed in hierarchical contests (Hemelrijk,
2000; Hsu et al., 2006; Tibbetts & Dale, 2007). These mechanisms
allow the individuals to adapt their behaviour towards encountered
nestmates without the need for overt aggressive interactions, and
therefore play a key role in the stabilization of dominance
hierarchies.
*Correspondence: B. Yagound, Laboratoire d’Ethologie Expérimentale et Com-
parée, E.A. 4443, Université Paris 13, Sorbonne Paris Cité, 99 avenue J.-B. Clément,
93430 Villetaneuse, France.
E-mail address: boris.yagound@leec.univ-paris13.fr (B. Yagound).
Contents lists available at ScienceDirect
Animal Behaviour
journal homepage: www.elsevier.com/locate/anbehav
http://dx.doi.org/10.1016/j.anbehav.2014.04.014
0003-3472/Ó2014 The Association for the Study of Animal Behaviour. Published by Elsevier Ltd. All rights reserved.
Animal Behaviour 93 (2014) 25e35

Numerous empirical and theoretical studies have proposed a
variety of intrinsic and extrinsic factors that may be responsible
for the formation and maintaining of dominance hierarchies
(Dugatkin & Earley, 2004; Hsu et al., 2006). These factors include
pre-existing differences between competing individuals (Parker,
1974), the value of the contested resource (Maynard Smith &
Parker, 1976) and the influence of previous experiences on the
outcome of future encounters (Dugatkin & Earley, 2004; Hsu et al.,
2006; Rutte et al., 2006). Game-theoretical studies have shown
that hierarchy formation could rely on self-organizing processes,
such as winner and loser effects (Dugatkin & Earley, 2004; Hsu
et al., 2006; Rutte et al., 2006), without the need for any partic-
ular recognition mechanism. In this case the outcome of past
encounters influences the chance of winning or losing in future
interactions in a self-reinforcing manner, i.e. regardless of the
identity or rank of the opponent. However, dominance in-
teractions are often highly directed (Chase & Seitz, 2011; Hsu et al.,
2006; Tibbetts & Dale, 2007), indicating that individuals actually
recognize the status of their opponents, through either direct or
indirect (i.e. memory-based) rank perception (Hemelrijk, 2000;
Tibbetts & Dale, 2007). Recognition systems are therefore an
important feature of dominance interactions, although not
mutually exclusive with self-organizing processes. However, the
recognition systems underlying dominance interactions remain
equivocal (Hsu et al., 2006), particularly since they very often
translate into a linear hierarchical structure.
Individual recognition has often been posited as an important
mechanism for the regulation and stabilization of linear dominance
hierarchies (d’Ettorre & Heinze, 2005; Dale, Lank, & Reeve, 2001;
Thom & Hurst, 2004; Tibbetts, 2002; Tibbetts & Dale, 2007). In
this indirect rank perception system (Hemelrijk, 2000), individuals
remember earlier interactions with specific group members and
adjust their dominance behaviour in subsequent encounters with
these same individuals (Dale et al., 2001; Tibbetts, 2002). Despite
the complexity of this cognitive mechanism, recognizing individual
identity is therefore supposed to provide high benefits by matching
the level of precision required for the maintenance of linear hier-
archies (Thom & Hurst, 2004).
However, linear hierarchies can also theoretically emerge and
be maintained through direct rank perception (i.e. status recog-
nition; Hemelrijk, 2000). Individuals in this case base their de-
cisions on the characteristics signalling an opponent’s absolute
fighting abilities (resource-holding potential; Parker, 1974), such as
age, size, weight or dominance badge (Chase & Seitz, 2011). In
contrast to individual recognition, there is thus no need for the
opponents to be familiar (Tibbetts & Dale, 2007). Status recogni-
tion could therefore save the costs of memory characterizing in-
dividual recognition (Thom & Hurst, 2004). However, a critical
assumption for the involvement of such a recognition system in
the formation and stabilization of linear hierarchies is that it al-
lows a fine-scale discrimination of ranks, but this has never been
demonstrated.
Dominance hierarchies are commonly found in social insects
(e.g. ants: Cuvillier-Hot, Lenoir, Crewe, Malosse, & Peeters, 2004;
Heinze, Hölldobler, & Peeters, 1994; Heinze, Stengl, & Sledge,
2002; Liebig, Peeters, Oldham, Markstädter, & Hölldobler, 2000;
Monnin & Peeters, 1999;bees:Ayasse, Marlovits, Tengö,
Taghizadeh, & Francke, 1995; Bull, Mibus, Norimatsu, Jarmyn, &
Schwarz, 1998; wasps: Sledge, Boscaro, & Turillazzi, 2001;
Tibbetts, 20 02), and this is particularly true when the colonies
comprise several individuals with equivalent reproductive po-
tentials competing to gain access to reproduction. Workers in
hopelessly queenless colonies thus typically compete with one
another over male parentage (Bourke, 1988; Ratnieks, Foster, &
Wenseleers, 2006), with a resulting linear or near-linear hierar-
chical structure of dominance relationships regulating the
partitioning of reproduction (Heinze et al., 1994; Heinze et al.,
2002; Peeters & Liebig, 2009), as in the Neotropical ant Neo-
ponera (formerly Pachycondyla;Schmidt & Shattuck, in press)
apicalis (Oliveira & Hölldobler, 1990). This species shares all
the traits typically characterizing Ponerinae ants, i.e. small
societies, a limited queeneworker dimorphism and a high po-
tential for worker reproduction (Fresneau, 1994), and is therefore
a good model system for studying the recognition mechanisms
involved in the formation and maintenance of dominance
hierarchies.
A previous study has shown that low-ranking individuals are
able to discriminate top-ranking from low-ranking workers,
suggesting a capacity to recognize the social status of their
nestmates (Blacher, Lecoutey, Fresneau, & Nowbahari, 2010).
However, these recognition abilities have never been investigated
in top-ranking workers. Since they are the individuals actually
involved in the reproductive competition, the costs of mistaking
ranks for those of adjacent-ranking nestmates are, in contrast to
low rankers, potentially high. We could hypothesize that a more
precise recognition system (e.g. individual recognition) is neces-
sary for an efficient discrimination among top-ranking individuals
(Tibbetts & Dale, 2007), but this could also be achieved without a
necessarily greater level of cognitive complexity in the eventu-
ality of fine-scale status discrimination. Assessing top rankers’
cognitive abilities therefore remains a crucial step in under-
standing the recognition mechanisms underlying the formation
and stabilization of the hierarchical structure in these social
groups (Elwood & Arnott, 2012; Wiley, 2013). Here we tested the
possibility of fine-scale status discrimination without the need for
individual recognition by studying the cognitive abilities of
N. apicalis top- and low-ranking workers. Furthermore, the nature
of the recognition cues involved in these dominance interactions
remains unknown, but they probably involve chemical commu-
nication. Chemical signals, mainly cuticular hydrocarbons, are
widely acknowledged to be of primary importance in the
communication of dominance and especially reproductive status
in social insects (Liebig, 2010; Monnin, 2006). We therefore also
analysed the individuals’chemical profile to investigate the na-
ture of the putative recognition cues at the basis of these domi-
nance interactions.
METHODS
Ants
Colonies of N. apicalis were collected in the Kérenroch forest,
Petit Saut (5

04
0
15.8
00
N, 53

02
0
36.3
00
W), French Guiana in March
2007 and have been kept in the laboratory in France ever since.
Ants were housed in plaster nests (18 14 cm) connected to a
foraging area of the same dimensions, where food (crickets and
honey/apple mixture) was provided twice a week and water ad
libitum. Each colony had a queen, more than 70 workers and brood
at every developmental stage. Nests were maintained at a tem-
perature of 27 2

C, a relative humidity of 60 5% and a 12:12 h
light:dark cycle. Ant collection, husbandry and experimental pro-
cedures used in this study fulfilled all the legal requirements con-
cerning insect experimentation of France.
Dominance Hierarchy
From our stock colonies, we created six experimental colonies
by isolating 40 randomly chosen workers and placing them in a
B. Yagound et al. / Animal Behaviour 93 (2014) 25e3526

new nest. Taking the workers away from the influence of the queen
induces the formation of a dominance hierarchy by means of
ritualized agonistic behaviours (Blacher et al., 2010; Oliveira &
Hölldobler, 1990). All ants were individually labelled with
numbered tags and dots of paint to allow the individual monitoring
of their behaviour. Housing and feeding conditions were the same
as above.
Each experimental colony was then observed 1 h a day during a
14-day period, during which we recorded all behavioural acts
linked to the establishment of the dominance hierarchy, i.e. ritu-
alized biting and antennal boxing (antennal strokes on another
ant’s body) (Cuvillier-Hot et al., 2004; Heinze et al., 2002; Oliveira
& Hölldobler, 1990). All observations started the same day the
workers were isolated from the queen, and were performed
through a red plastic film to avoid disturbances that may affect the
ants’behaviour. All agonistic interactions (performed and
received) were then compiled in a matrix and arranged in an order
minimizing the number of inconsistencies (i.e. when an individual
is given a lower rank than an individual it dominates). This
allowed us to reconstruct the dominance hierarchy and to assign a
hierarchical rank to each individual (see Blacher et al. (2010) and
references therein for a detailed description of the method used).
Ants at the top of the hierarchy that collectively performed more
than 75% of the agonistic acts (mean SE: 11.5 0.2 individuals,
N¼6 colonies; Table 1) were considered high-ranking individuals.
Two other classes of individuals were additionally determined,
namely middle-ranking individuals (remaining ants performing
up to 95% of the agonistic acts with the exclusion of high-ranking
workers, 11.8 1.6 individuals) and low-ranking individuals
(remaining ants at the bottom of the hierarchy, 13.8 1. 6
individuals).
HabituationeDiscrimination Procedure
To test the cognitive abilities of high- and low-ranking
workers, we used a habituationediscrimination paradigm, a
classical procedure in cognitive studies (Ferguson, Young, & Insel,
2002) consisting of two consecutive phases. The habituation
phase consists of four consecutive exposures (4 min each with a
5 min interval) of a stimulus (a nestmate ant) to a tested indi-
vidual. The tested ant thus becomes familiar with the proposed
stimulus. The stimulus ant is CO
2
-anaesthetized to avoid any in-
fluence of its behaviour on the tested ant’s response. Following a
5 min interval, the discrimination phase consists of a single test
(4 min) in which the tested ant is confronted with two stimuli: a
familiar stimulus (the anaesthetized ant previously used during
the habituation phase) and an unfamiliar stimulus (another
anaesthetized nestmate). Discrimination is typically manifested
by a longer duration of the behavioural response towards the
unfamiliar stimulus compared with the familiar stimulus (Wiley,
2013). Note that both stimuli are nestmates of the tested ant,
and therefore both have potentially already interacted with it. The
use of ‘familiar’and ‘unfamiliar’thus refers to the habituatione
discrimination procedure only, since these terms are classically
used in these experiments.
Each test was performed in a neutral arena (diame-
ter ¼5.3 cm) with externally black-covered walls and Fluon-
coated sides to prevent the ant from escaping. To enable famil-
iarization with the device, the tested ant was gently placed in it
30 s before the test began. The stimulus ant was then introduced
into the centre of the arena, in which a filter paper had been
placed as a substrate (the paper was replaced between each test
to avoid any odour residues). All tests were videorecorded for
subsequent analyses of the tested ants’behaviour. During these
analyses, we measured the duration of antennal contacts with the
stimulus (commonly taken as a measure of an ant’s interest to-
wards a social stimulus; Boulay, Hefetz, Soroker, & Lenoir, 2000)
with EthoLog 2.2 software (Ottoni, 2000). In addition, we inves-
tigated the possibility that the behavioural response of high- and
low-ranking workers could be influenced by differences in their
overall reaction or motivation in the experimental device by
quantifying their mobility pattern in 58 randomly selected
habituation tests derived from all four habituation tests and all
colonies using EthoVision 3.1.16 (Noldus Information Technology,
Wageningen, Netherlands). Three behavioural variables were
quantified: total distance moved, mean angular velocity and
duration of mobility. We also quantified the duration close to the
stimulus ant, which reflects the general attraction/avoidance of
the tested ant towards the proposed stimulus. Observations were
performed twice and blind.
On day 15, four different experiments were carried out using
this procedure in which tested and stimuli ants were either low- or
high-ranking individuals as previously defined. All tested and
stimuli ants were only used in a single habituationediscrimination
test.
Experiment 1: status discrimination by high rankers
In the first experiment, the tested ant (N¼21, two for colony B,
three for colony A, four for colonies C, D, E and F) was a high
ranker and was confronted during the discrimination phase with
stimuli belonging to different rank classes, i.e. a low- and a high-
ranking individual (mean gap in their respective rank of
22.5 0.6). This experiment was aimed at verifying whether high-
ranking workers are capable of status discrimination, an ability
that has already been shown in low-ranking workers (Blacher
et al., 2010). To avoid the stimulus rank class affecting the tested
ant’s behaviour, the familiar stimulus was a low-ranking individ-
ual for half of the tested ants (N¼11) and a high-ranking indi-
vidual for the other half (N¼10).
Experiment 2: fine-scale discrimination of high-ranking individuals
by high rankers
The second experiment was aimed at testing the possibility of
fine-scale discrimination among high-ranking workers. High
rankers (tested ants, N¼19, two for colony E, three for colonies B, C
and D, four for colonies A and F) were confronted during the
discrimination phase with two high-ranking workers (stimuli ants)
Table 1
Dominance hierarchy characteristics
Colony Number of
top rankers
Linearity
*
Correlation with
dominance index
Correlation with
ovarian index
A11 K¼0.85
P<0.0001
r
s
¼0.68
P<0.0001
r
s
¼0.75
P<0.0001
B12 K¼0.91
P<0.0001
r
s
¼0.80
P<0.0001
r
s
¼0.85
P<0.0001
C11 K¼0.45
P¼0.031
r
s
¼0.40
P¼0.014
r
s
¼0.69
P<0.0001
D12 K¼0.69
P<0.0001
r
s
¼0.69
P<0.0001
r
s
¼0.75
P<0.0001
E12 K¼0.94
P<0.0001
r
s
¼0.89
P<0.0001
r
s
¼0.61
P<0.0001
F11 K¼0.92
P<0.0001
r
s
¼0.86
P<0.0001
r
s
¼0.81
P<0.0001
*
See Appleby (1983). Note that the smaller values of Kfor colony C and to a lesser
extent colony D are mainly due to more missing values for some dyads (11 and 3 for
colonies C and D respectively) compared with the other colonies (0 or 1) (de Vries,
1995).
B. Yagound et al. / Animal Behaviour 93 (2014) 25e35 27

located higher in the hierarchy and separated by a single rank
(mean gap in their respective rank of 1.0 0.0).
Experiment 3: fine-scale discrimination of low-ranking individuals
by high rankers
In the third experiment, high rankers (tested ants, N¼21, three
for colonies B, E and F, four for colonies A, C and D) were confronted
during the discrimination phase with two low-ranking workers
(stimuli ants) with again very similar ranks (mean gap in their
respective rank of 1.7 0.3), but this time located at the bottom of
the hierarchy.
Experiment 4: fine-scale discrimination of high-ranking individuals
by low rankers
Finally, the fourth experiment was done in order to compare
the behavioural discrimination of high- and low-ranking
workers. Low rankers have already been shown to discriminate
high- and low-ranking individuals, but not low-ranking in-
dividuals (Blacher et al., 2010). However, their ability to
discriminate high-ranking workers has never been investigated.
Low rankers (tested ants, N¼24, four in each colony) were
therefore confronted with two high-ranking workers (stimuli
ants, mean gap in their respective rank of 1.1 0.1) in the
discrimination phase.
Fertility Measurement
To link the ants’hierarchical rank with their reproductive
dominance, all workers were frozen for dissection at the end of
the habituationediscrimination procedure, and their fertility was
determined. As an ovarian index, we measured the total size of
the six basal oocytes. We then determined three classes of in-
dividuals depending on their number of developed oocytes (i.e.
size >0.5 mm; Fresneau, 1994): highly fertile individuals (five to
six developed oocytes, 11.8 2.4 individuals, N¼6 colonies),
moderately fertile individuals (one to four developed oocytes,
13.3 2.7 individuals) and infertile individuals (no developed
oocytes, 12.0 2.3 individuals). Some individuals (two in col-
onies A, D and F, three in colonies C and E, five in colony B) died
before the experiments and were therefore not included in the
analyses.
Chemical Profiles
We finally investigated the nature of the putative recognition
cues involved in the dominance interactions of the reproductive
hierarchy by analysing the cuticular hydrocarbon profile of all
individuals. This procedure allowed us to study how ants
diverged in their chemical signature according to their social
rank and fertility state. We sampled the cuticular hydrocarbons
of a total of 223 ants. Extraction was performed by placing an ant
in 400
m
l of pentane containing 8 ng/
m
l of an internal standard
(n-C
17
) for 20 min. We then transferred 100
m
lintoa200
m
lglass
insert. Following evaporation, 20
m
l of pentane were added to the
200
m
l glass insert. We then manually injected 2
m
l of the extract
into an Agilent 7890A gas chromatograph, equipped with an HP-
5MS capillary column (30 m 25
m
m0.25
m
m) and a splite
splitless injector, coupled to an Agilent 5975c mass spectrometer
with 70 eV electron impact ionization. The carrier gas was heli-
um at 1 ml/min. The temperature program was as follows: an
initial hold at 70

C for 1 min, then 70e180

Cat30

C/min, then
180 e320

Cat5

C/min, then hold at 320

Cfor5min.Theareas
of 34 peaks present in all ant cuticular extracts (Appendix
Fig. A1) were integrated with the Agilent ChemStation
software. Hydrocarbons were identified on the basis of their
mass spectra and retention times, and compared with known
standards.
Statistical Analyses
Following hierarchy reconstruction, we calculated the Kindex
of linearity varying from 0 (no linearity) to 1 (linear hierarchy)
and tested the statistical significance of linearity according to
Appleby (1983). The test of linearity was performed on high-
ranking workers only, because middle- and particularly low-
ranking workers performed far fewer agonistic interactions,
thus creating incomplete information, which is known to under-
estimate the values of linearity (de Vries, 1995). Using the
Spearman rank correlation test, we calculated the correlation
between the hierarchical rank and (1) the dominance index
(proportion of agonistic acts performed) and (2) the ovarian in-
dex. We compared the proportion of agonistic acts that high
rankers performed towards other high-ranking workers of
consecutive ranks with the Friedman test followed by post hoc
exact permutation tests with the BonferronieHolm method
(Holm, 1979).
For each experiment, we compared the duration of antennal
contacts with the stimulus between the first and the fourth
habituation test, and the antennation duration towards the familiar
and the unfamiliar stimulus in the discrimination test using exact
permutation tests for paired samples. Each mobility variable was
compared between high and low rankers confronted with high-
ranking nestmates, and between high rankers confronted with
high- and low-ranking nestmates using exact permutation tests for
independent samples.
For chemical analyses, we arcsine-transformed (Sokal & Rohlf,
2012) the relative quantities of 34 compounds common to all in-
dividuals. We then performed a discriminant function analysis to
investigate how individuals diverge in their chemical profile ac-
cording to their hierarchical rank (i.e. high-, middle- and low-
ranking individuals), and their fertility level (i.e. highly fertile,
moderately fertile and infertile individuals). We finally investigated
the existence of a putative fertility signal (Monnin, 2006)by
comparing the absolute quantity of all compounds between the
three fertility groups using one-way ANOVAs followed by post hoc
exact permutation tests with the BonferronieHolm method. Ab-
solute quantities of the compounds with the highest contribution
to the discrimination were also correlated with the individuals’
fertility and social rank for all individuals using the Spearman rank
correlation test. The Monte Carlo procedure was used when
appropriate to deal with large sample sizes (Metropolis & Ulam,
194 9).
All statistical analyses were performed with StatXact 8.0 (Cytel
Software Corporation, Cambridge, MA, U.S.A.) and Statistica 8.0
(StatSoft, Tulsa, OK, U.S.A.). Statistical significance was set at
P<0.05.
RESULTS
Dominance Hierarchy
During the 14-day observation period of dominance/subordi-
nate relationships, we recorded a total of 11808 agonistic acts
(1968.0 248.2 per colony), which allowed us to determine the
hierarchical rank of the ants successfully. The linearity or near-
linearity of the hierarchy was significant in all colonies (Table 1).
Furthermore, the individuals’hierarchical rank was highly corre-
lated with their dominance index (Table 1). Among high rankers,
B. Yagound et al. / Animal Behaviour 93 (2014) 25e3528

agonistic interactions were not randomly directed towards other
high-ranking workers, but instead were highly biased towards the
individuals with the closest lower ranks in the hierarchy (Friedman
test on the proportion of agonistic acts that high rankers performed
towards individuals with immediate consecutive ranks:
T
F, 4
¼53.73, N¼69, P<0.0001; Fig. 1).
Agonistic interactions directed towards dominant individuals
(inconsistencies) were very rare among high rankers (2.56 0.84%
per colony). In addition, the formation of a linear hierarchy was
very quick, since the proportion of inconsistencies was already
small during the first 24 h of isolation (12.76 8.09% per colony).
HabituationeDiscrimination Procedure
Habituation, manifested by a decrease in the duration of
antennal contacts with the stimulus ant between the first and the
last habituation test (Wiley, 2013), occurred in the four experi-
ments (permutation tests: all P<0.021; Appendix Fig. A2). In the
first experiment, habituation occurred whether the tested ant
was familiarized with a high- or a low-ranking nestmate (N¼10,
P¼0.041 and N¼11, P¼0.003, respectively; Appendix Fig. A2a).
High and low rankers showed similar mobility patterns in the
experimental device (N¼39, all P>0.18; Appendix Table A1), as
did high rankers towards high- and low-ranking individuals
(N¼39, all P>0.18; Appendix Table A1). Furthermore, duration
of contacts with the stimulus ant was similar for each of the four
habituation tests and all tests combined whether the stimulus
was a high- or a low-ranking individual (N¼21, all P>0.22).
Tested ants were thus able to familiarize themselves with a
nestmate, the social status of this nestmate having no influence
on this process.
In the discrimination test of the first experiment, tested ants
spent more time antennating unfamiliar than familiar ants (N¼21,
P¼0.0004; Fig. 2), irrespective of the status of the habituation
stimulus (high-ranking nestmates: N¼10, P¼0.020; low-ranking
nestmates: N¼11, P¼0.018). There was no significant difference
in the duration of antennal contacts between high- and low-
ranking familiar stimuli (N¼21, P¼0.96) and between high- and
low-ranking unfamiliar stimuli (N¼21, P¼0.97). High rankers
were thus well able to discriminate the social status of their nest-
mates, as has already been shown in low rankers (Blacher et al.,
2010).
In the second experiment, duration of antennal contacts to-
wards unfamiliar ants was higher than towards familiar ants
(N¼19, P¼0.037; Fig. 2). Since stimuli ants were separated by a
single rank, this clearly demonstrates that high rankers were
capable of fine-scale status discrimination of other high-ranking
workers.
In the third experiment, duration of antennation towards un-
familiar and familiar nestmates was not significantly different
(N¼21, P¼0.42; Fig. 2). High rankers therefore did not show a
behavioural discrimination of low-ranking nestmates with very
similar statuses. This result further confirms that the differential
response of the tested ants towards unfamiliar stimuli compared
with familiar stimuli in the first and second experiments was based
on status discrimination.
In the fourth experiment, low-ranking tested ants did not spend
significantly more time antennating unfamiliar or familiar nest-
mates (N¼24, P¼0.15; Fig. 2). Furthermore, there was no signif-
icant difference between low- (experiment 4) and high-ranking
tested ants (experiment 2) in the duration of antennal contacts
towards high-ranking familiar stimuli (N¼43, P¼0. 79) and in the
duration of antennal contacts towards high-ranking unfamiliar
stimuli (N¼43, P¼0.36). In contrast to high rankers, low rankers
thus did not exhibit fine-scale status discrimination of high-ranking
nestmates.
Fertility Measurement
Hierarchical rank was strongly correlated with the ovarian index
in all colonies (Table 1). This corroborates the well-known rela-
tionship between fertility and social status in insect societies
(Blacher et al., 2010; Cuvillier-Hot et al., 2004; Heinze et al., 2002;
Monnin & Peeters, 1999).
60
Familiar stimulus
1
N = 21
2
N = 19
3
N = 21
Ex
p
eriment
4
N = 24
***
*NS
NS
Unfamiliar stimulus
50
40
30
Antennal contacts (s)
20
10
0
Figure 2. Duration of antennal contacts (s) towards the familiar and unfamiliar stimuli
during discrimination tests in all experiments. In experiments 1e3, tested ants were
high rankers and were confronted with a low- and a high-ranking nestmate (experi-
ment 1), two high-ranking nestmates (experiment 2) or two low-ranking nestmates
(experiment 3). In experiment 4, tested ants were low rankers and both stimuli were
high-ranking nestmates. Box plots represent 10th, 25th, 50th (median), 75th and 90th
percentiles. Sample sizes of individuals are indicated below each box plot.
***P<0.001; *P<0.05.
35
30
25
20
15
10
a
bbc
cd d
5
0i−1
N = 69
i−2
N = 69
i−3
N = 69
Hierarchical rank
Agonistic acts performed (%)
i−4
N = 69
i−5
N = 69
Figure 1. Percentage of agonistic acts performed by high rankers (rank i) towards the
individuals with the closest lower ranks in the hierarchy (rank i-1 to rank i-5). Box
plots represent 10th, 25th, 50th (median), 75th and 90th percentiles. Sample sizes of
individuals are indicated below each box plot. Different letters denote statistical
differences.
B. Yagound et al. / Animal Behaviour 93 (2014) 25e35 29

Chemical Profiles
Workers could clearly be separated on the basis of their cutic-
ular hydrocarbon profiles according to their hierarchical rank
(Wilks’s
l
¼0.304, F
40,402
¼8.19, P<0.0001; Appendix Fig. A3a)
and even more strongly to their ovarian development (Wilks’s
l
¼0.130, F
52,390
¼13.33, P<0.0001; Appendix Fig. A3b). This in-
dicates that the chemical signature provides a reliable cue for an
ant to discriminate the fertility and social rank of its nestmates.
From the 34 compounds constituting the shared chemical pro-
file of N.apicalis workers (Appendix Fig. A1), 14 displayed signifi-
cant differences in their amounts between individuals of varying
fertility (Appendix Table A2). Among these compounds, 13-
methylpentacosane (13-MeC
25
) had the highest contribution to
the discrimination of all individuals according to their ovarian
development (partial Wilks’s
l
¼0.799, F
2,195
¼24.39, P<0.0001)
and hierarchical rank (partial Wilks’s
l
¼0.867, F
2,201
¼15.47,
P<0.0001) in the discriminant function analyses. Quantity of 13-
MeC
25
was further strongly correlated with both fertility
(Spearman rank correlation: r
s
¼0.77, N¼223, P<0.0001; Fig. 3a;
Appendix Table A2), and hierarchical rank (r
s
¼-0.73, N¼223,
P<0.0001; Fig. 3b).
DISCUSSION
Our results clearly show that N. apicalis workers in hopelessly
queenless colonies establish a linear dominance hierarchy in
which agonistic behaviour, fertility, cuticular hydrocarbon profile
and social rank are all closely correlated. Dominance interactions
were highly directed, with top-ranking individuals performing
most agonistic acts towards other high rankers with immediate
consecutive ranks. The vast majority of hierarchical relationships
were clearly and quickly established. Furthermore, it has been
shown in N. apicalis and several other species that once the hier-
archy has emerged, dominance relationships can be maintained
for extended periods of time, but with a dramatic decrease in
agonistic behaviour (Blacher et al., 2010; Cuvillier-Hot et al., 2004;
Monnin & Peeters, 1999). Although this does not exclude the in-
fluence of self-organizing processes, at least in the beginning
(Dugatkin & Earley, 2004; Hsu et al., 2006; Rutte et al., 2006), it
unambiguously indicates the implication of recognition
mechanisms in the formation and maintenance of the hierarchical
structure.
All ants were able to discriminate top-ranking from low-
ranking nestmates, thus confirming and expanding the exis-
tence of status discrimination abilities in this species (Blacher
et al., 2010). Furthermore, top rankers were able to discriminate
two high-ranking nestmates separated by a single rank in the
hierarchy. This ability of fine-scale status discrimination among
top rankers is beneficial for the regulation of reproductive
dominance. It allows each competing ant to adapt its behaviour
according to the social status of all encountered nestmates, i.e.
dominantly towards a lower ranker and submissively towards a
higher ranker. Avoiding recognition errors decreases the costs
associated with dominance interactions in terms of colony pro-
ductivity, and therefore enhances the individuals’inclusive fitness
(Gobin, Heinze, Strätz, & Roces, 2003). Note that both stimuli in
the discrimination phase were dominant for the tested ant,
meaning that individuals did not merely discriminate an ant
higher than itself from a lower ant in the hierarchy. This em-
phasizes that recognizing all the competing individuals’statuses
can be adaptive in the eventuality of a hierarchy disruption (Hart
& Monnin, 2006), as has been shown to occur (Oliveira &
Hölldobler, 1990).
It has been shown in two related species, N. villosa and
N. inversa, that unrelated co-founding queens establishing domi-
nance hierarchies seem capable of individual recognition (d’Ettorre
& Heinze, 2005; Dreier, van Zweden, & d’Ettorre, 2007). In our
study, top rankers did not behaviourally discriminate low rankers
of virtually identical status. This could partly stem from a low
motivation for accomplishing this task, as low-ranking nestmates
are not involved in the reproductive competition. The habituatione
discrimination procedure was, however, used to reduce the influ-
ence of motivation on the tested ants’response. Although the
context of hierarchy formation is different in N. apicalis, and
unambiguously demonstrating individual recognition abilities is
particularly challenging (Wiley, 2013), our results nevertheless
suggest an absence of identity-based discrimination in workers.
Within the scope of status discrimination abilities, and as we
discuss in more depth below, this absence of behavioural discrim-
ination between two low rankers could most likely be explained by
an absence of recognition cues allowing an unequivocal
discrimination.
70
(a) (b)
60
50
40
30
20
10
5
1
70
60
50
40
30
20
10
5
1
0246
Ovarian develo
p
ment (mm) Hierarchical rank
Quantity of 13-MeC25 (µg)
8 10 0 5 101520253035
Figure 3. Relationship between quantity of 13-MeC
25
(
m
g) and (a) ovarian development (mm) and (b) hierarchical rank. Note log scale on ordinates.
B. Yagound et al. / Animal Behaviour 93 (2014) 25e3530

If our results fail to provide any strong evidence in favour of
individual recognition, they, however, clearly indicate that linear
hierarchies can arise without such a recognition system. Status
discrimination, a cognitively simpler mechanism (Wiley, 2013),
has in this species at least the level of accuracy enabling a fine-
scale discrimination of the individuals’statuses without the ne-
cessity of recognizing their identity. This recognition system thus
appears suitable for precise regulation of dominance interactions
without the need for aggressive behaviours, and more impor-
tantly avoids the cognitive costs linked to individual recognition.
Indeed, individual recognition relies on the memory of each op-
ponent’s distinctive features, and of their history of encounters in
the context of dominance interactions. It thus requires active
learning and an accurate memory, both of which are costly pro-
cesses in terms of time and energy expenditure (Burns, Foucaud,
& Mery, 2011; Dukas, 2008). By contrast, status discrimination is a
cognitively less demanding task, as it relies on the direct
perception of each encountered individual’s rank. Carefully
investigating alternative hypotheses thus remains a crucial step
when studying animal cognition (Elwood & Arnott, 2012). Overall,
these results show that direct rank perception is probably a
critical factor in the establishment and stabilization of the hier-
archical structure.
Whereas high and low rankers showed very similar behav-
ioural reactions in the experimental device in terms of their
proximity and duration of antennal contacts towards stimuli in
both the habituation and the discrimination phase, low rankers
failed to discriminate at a fine scale ants belonging to the same
rank class, be they high- or low-ranking nestmates (this study;
Blacher et al., 2010). This difference with top rankers could be due
to an absence of motivation since all high-ranking nestmates are
by definition much higher in the hierarchy and could induce the
same submissive behaviours from low rankers. In contrast to
other species (Hart & Monnin, 2006), low rankers indeed play no
role in top ranker replacements (Oliveira & Hölldobler, 1990), and
thus in this context they have no benefits from discriminating
two high rankers. Alternatively, the difference between high- and
low-ranking workers could also be due to differences in their
cognitive abilities. Indeed, top and low rankers can differ in a
number of physiological characteristics. For example, neuroen-
docrine activities (levels of brain biogenic amines) can vary
depending on the dominance and reproductive status. Top
rankers have higher levels of octopamine than low rankers in the
bumble bee Bombus terrestris (Bloch, Simon, Robinson, & Hefetz,
2000), and octopamine levels are correlated with reproductive
activity in the queenless ant Streblognathus peetersi (Cuvillier-Hot
&Lenoir,2006). Octopamine further acts as a neuromodulator
which is known to affect cognitive processes such as learning and
memory (Farooqui, 2007; Verlinden et al., 2010). Different inter-
nal states according to the individuals’hierarchical status could
thus theoretically mediate various levels of recognition abilities.
Intraspecific variation in recognition abilities has recently begun
to be explored (e.g. Injaian & Tibbetts, 2014), and future in-
vestigations are therefore required to examine the existence of
actual differences in cognitive abilities depending on the in-
dividual’s social status.
Cuticular hydrocarbon profiles diverged between reproductive
and nonreproductive individuals, and probably constitute the
recognition cues used in dominance interactions. The chemical
nature of cuticular hydrocarbons signalling fertility can differ
markedly according to species, but the occurrence of fertility
signals seems to be a general phenomenon in social insects
(Cuvillier-Hot et al., 2004; Heinze et al., 2002; Liebig, 2010; Liebig
et al., 2000; Monnin, 2006; Sledge et al., 2001). This can be
explained by the fact that being permanently informed about the
fertility state of the egg-layer(s) provides inclusive fitness ben-
efits to all colony members (Keller & Nonacs, 1993). Here,
amounts of 13-methylpentacosane (13-MeC
25
) were highly
correlated with the individuals’ovarian activity. This compound
may therefore have the role of a putative fertility signal (Liebig,
2010; Monnin, 2006). Hydrocarbons are synthesized in the
oenocytes, cells associated with the epidermis and the fat bodies
(Martins & Ramalho-Ortigão, 2012). It is generally assumed that
common endocrinological mechanisms (e.g. gonadotropic hor-
mones) underlie the biosynthesis of cuticular hydrocarbons and
the activity of the ovaries (Cuvillier-Hot et al., 2004; Liebig, 2010;
Peeters & Liebig, 2009), thus explaining the close link between
reproductive activity and chemical signals. Such an intrinsic
causal link would mean the recognition system could not be
faked, and fertility signals would be evolutionarily stable (i.e.
honest; Laidre & Johnstone, 2013; Maynard Smith & Harper,
1995).
The quantity of the putative fertility signal was also highly
correlated with the individuals’social rank, because of the close
link between fertility and hierarchical status. The exponential
form of the relationship between amounts of 13-MeC
25
and hi-
erarchical rank means that two individuals with very different
social statuses (i.e. a high- and a low-ranking worker) would have
a large difference in their amounts of 13-MeC
25
, therefore
allowing status discrimination. Two nestmates having close but
none the less different hierarchical statuses (i.e. two top rankers
in this case) would have a lower but yet significant difference in
their amounts of 13-MeC
25
. In this case, status discrimination
becomes precise enough to allow a fine-scale discrimination of
their ranks, corresponding to a functional individual discrimina-
tion. In contrast, two individuals having a very similar social
status (i.e. two low rankers) would have a very small difference in
their amounts of 13-MeC
25
, thus making any discrimination
probably difficult, possibly beyond the workers’sensory and
information-processing capabilities. According to this hypothesis,
13-MeC
25
is likely to form the proximate signal at the basis of the
individuals’status discrimination, and could in this respect
constitute a chemical badge of status (Guilford & Dawkins, 1995).
Such signals, traditionally linked with the individuals’resource-
holding potential (Johnstone & Norris, 1993), could allow in-
dividuals in this context to select the best egg-layer in the colony.
Similar mechanisms have already been suggested (Cuvillier-Hot
et al., 2004), and could form a general rule in the regulation of
dominance hierarchies in insect societies. The dynamics of
fertility signalling during the establishment of the hierarchical
structure remains unknown, but it is conceivable that prefertility
differences at the onset of orphaning could strongly influence
dominance interactions and therefore the determination of hier-
archical ranks. Such investigations should thus be conducted in
future studies.
In conclusion, we have shown here that status discrimination
based on a putative fertility signal is able to generate a linear
dominance hierarchy in N. apicalis ants. This single cuticular hy-
drocarbon appears to act as a badge of status by precisely labelling
the individual’s position in the hierarchy, and therefore regulates
the conflict over male parentage in this species. Whereas the for-
mation of linear hierarchies is often assumed to rely on complex
cognitive processes such as individual recognition, our results
suggest on the contrary that simpler recognition mechanisms can
be sufficient to regulate dominance interactions efficiently. By
mutually benefiting all members of the nest, this recognition sys-
tem is thus very likely to have been selected for by both individual-
and colony-level selection pressures.
B. Yagound et al. / Animal Behaviour 93 (2014) 25e35 31

Acknowledgments
We thank R. S. Ferreira and P. Devienne for collecting the ants, C.
Leroy for technical assistance with the chemical analyses, and S.
Chameron and two anonymous referees for helpful suggestions on
the manuscript. This study was supported by the FAPESB/CNPq
(PNX0011/2009 PRONEX).
References
Appleby, M. C. (1983). The probability of linearity in hierarchies. Animal Behaviour,
31,600e608.
Ayasse, M., Marlovits, T., Tengö, J., Taghizadeh, T., & Francke, W. (1995). Are there
pheromonal dominance signals in the bumblebee Bombus hypnorum L (Hy-
menoptera, Apidae)? Apidologie, 26,163e180 .
Blacher, P., Lecoutey, E., Fresneau, D., & Nowbahari, E. (2010). Reproductive hierar-
chies and status discrimination in orphaned colonies of Pachycondyla apicalis
ants. Animal Behaviour, 79,99e105 .
Bloch, G., Simon, T., Robinson, G. E., & Hefetz, A. (2000). Brain biogenic amines and
reproductive dominance in bumble bees (Bombus terrestris). Journal of
Comparative Physiology A, 186,261e268.
Boulay, R., Hefetz, A., Soroker, V., & Lenoir, A. (2000). Camponotus fellah colony
integration: worker individuality necessitates frequent hydrocarbon exchanges.
Animal Behaviour, 59,1127e113 3 .
Bourke, A. F. G. (1988). Worker reproduction in the higher eusocial Hymenoptera.
Quarterly Review of Biology, 63,291e311.
Bull, N. J., Mibus, A. C., Norimatsu, Y., Jarmyn, B. L., & Schwarz, M. P. (1998).
Giving your daughters the edge: bequeathing reproductive dominance in a
primitively social bee. Proceedings of the Royal Society B: Biological Sciences,
265,1411e1415.
Burns, J. G., Foucaud, J., & Mery, F. (2011). Costs of memory: lessons from
‘mini’brains. Proceedings of the Royal Society B: Biological Sciences, 278,
923e929.
Chase, I. D., & Seitz, K. (2011). Self-structuring properties of dominance hierarchies:
a new perspective. Advances in Genetics, 75,51e81.
Cuvillier-Hot, V., & Lenoir, A. (2006). Biogenic amine levels, reproduction and social
dominance in the queenless ant Streblognathus peetersi.Naturwissenschaften,
93,149e153.
Cuvillier-Hot, V., Lenoir, A., Crewe, R., Malosse, C., & Peeters, C. (2004). Fertility
signalling and reproductive skew in queenless ants. Animal Behaviour, 68,
1209e1219.
Dale, J., Lank, D. B., & Reeve, H. K. (2001). Signaling individual identity versus
quality: a model and case studies with ruffs, queleas, and house finches.
American Naturalist, 158,75e86.
Dreier, S., van Zweden, J. S., & d’Ettorre, P. (20 07). Long-term memory of individual
identity in ant queens. Biology Letters, 3, 459e462.
Dugatkin, L. A., & Earley, R. L. (2004). Individual recognition, dominance hierarchies
and winner and loser effects. Proceedings of the Royal Society B: Biological Sci-
ences, 271, 1537e1540.
Dukas, R. (2008). Evolutionary biology of insect learning. Annual Review of Ento-
mology, 53,145e160 .
Ellis, L. (1995). Dominance and reproductive success among nonhuman animals: a
cross-species comparison. Ethology and Sociobiology, 16,257e333.
Elwood, R. W., & Arnott, G. (2012). Understanding how animals fight with Lloyd
Morgan’s canon. Animal Behaviour, 84, 1095e11 02 .
d’Ettorre, P., & Heinze, J. (2005). Individual recognition in ant queens. Current
Biology, 15,2170e2174.
Farooqui, T. (2007). Octopamine-mediated neuromodulation of insect senses.
Neurochemical Research, 32, 1511e1529.
Ferguson, J. N., Young, L. J., & Insel, T. R. (2002). The neuroendocrine basis of social
recognition. Frontiers in Neuroendocrinology, 23,200e224.
Fresneau, D. (1994). Biology and social behaviour of a Neotropical ponerine ant
(Pachycondyla apicalis)(Unpublished doctoral dissertation). Villetaneuse,
France: Université Paris 13.
Gobin, B., Heinze, J., Strätz, M., & Roces, F. (2003). The energetic cost of reproductive
conflicts in the ant Pachycondyla obscuricornis.Journal of Insect Physiology, 49,
747e752.
Guilford, T., & Dawkins, M. S. (1995). What are conventional signals? Animal
Behaviour, 49, 1689e1695.
Hart, A. G., & Monnin, T. (2006). Conflict over the timing of breeder replacement in
vertebrate and invertebrate societies. Insectes Sociaux, 53,375
e389.
Heinze, J., Hölldobler, B., & Peeters, C. (1994). Conflict and cooperation in ant so-
cieties. Naturwissenschaften, 81, 489e497.
Heinze, J., Stengl, B., & Sledge, M. F. (2002). Worker rank, reproductive status and
cuticular hydrocarbon signature in the ant, Pachycondyla cf. inversa.Behavioral
Ecology and Sociobiology, 52,59e65.
Hemelrijk, C. K. (2000). Towards the integration of social dominance and spatial
structure. Animal Behaviour, 59, 1035e1048.
Holm, S. (1979). A simple sequentially rejective multiple test procedure. Scandi-
navian Journal of Statistics, Theory and Applications, 6,65e70.
Hsu, Y., Earley, R. L., & Wolf, L. L. (2006). Modulation of aggressive behaviour by
fighting experience: mechanisms and contest outcomes. Biological Reviews, 81,
33e74.
Injaian, A., & Tibbetts, E. A. (2014). Cognition across castes: individual recognition in
worker Polistes fuscatus wasps. Animal Behaviour, 87,91e96.
Johnstone, R. A., & Norris, K. (1993). Badges of status and the cost of aggression.
Behavioral Ecology and Sociobiology, 32,127e134 .
Keller, L., & Nonacs, P. (1993). The role of queen pheromones in social insects: queen
control or queen signal? Animal Behaviour, 45, 787e794.
Laidre, M. E., & Johnstone, R. A. (2013). Animal signals. Current Biology, 23,
R829eR833.
Liebig, J. (2010). Hydrocarbon profiles indicate fertility and dominance status in ant,
bee, and wasp colonies. In G. J. Blomquist, & A.-G. Bagnières (Eds.), Insect hy-
drocarbons: Biology, biochemistry, and chemical ecology (pp. 254e281). Cam-
bridge: Cambridge University Press.
Liebig, J., Peeters, C., Oldham, N. J., Markstädter, C., & Hölldobler, B. (2000). Are
variations in cuticular hydrocarbons of queens and workers a reliable signal of
fertility in the ant Harpegnathos saltator?Proceedings of the National Academy of
Sciences of the United States of America, 97,4124e4131.
Martins, G. F., & Ramalho-Ortigão, J. M. (2012). Oenocytes in insects. Invertebrate
Survival Journal, 9,139e
152 .
Maynard Smith, J., & Harper, D. G. C. (1995). Animal signals: models and termi-
nology. Journal of Theoretical Biology, 177, 305e311.
Maynard Smith, J., & Parker, G. A. (1976). The logic of asymmetric contests. Animal
Behaviour, 24,159e175 .
Metropolis, N., & Ulam, S. (1949). The Monte Carlo method. Journal of the American
Statistical Association, 44,335e341.
Monnin, T. (2006). Chemical recognition of reproductive status in social insects.
Annales Zoologici Fennici, 43,515e530.
Monnin, T., & Peeters, C. (1999). Dominance hierarchy and reproductive conflicts
among subordinates in a monogynous queenless ant. Behavioral Ecology, 10,
323e332.
Oliveira, P. S., & Hölldobler, B. (1990). Dominance orders in the ponerine ant
Pachycondyla apicalis (Hymenoptera, Formicidae). Behavioral Ecology and So-
ciobiology, 27, 385e393.
Ottoni, E. B. (2000). EthoLog 2.2: a tool for the transcription and timing of behavior
observation sessions. Behavior Research Methods, Instruments, & Computers, 32,
446e449.
Parker, G. A. (1974). Assessment strategy and the evolution of fighting behaviour.
Journal of Theoretical Biology, 47, 223e243.
Peeters, C., & Liebig, J. (2009). Fertility signaling as a general mechanism of regu-
lating reproductive division of labor in ants. In J. Gadau, & J. Fewell (Eds.), Or-
ganization of insect societies: From genome to socio-complexity (pp. 220e242).
Cambridge, MA: Harvard University Press.
Ratnieks, F. L. W., Foster, K. R., & Wenseleers, T. (2006). Conflict resolution in insect
societies. Annual Review of Entomology, 51, 581e608.
Rutte, C., Taborsky, M., & Brinkhof, M. W. G. (2006). What sets the odds of winning
and losing? Trends in Ecology & Evolution, 21,16e21.
Schmidt, C. A., & Shattuck, S. O. (in press). The higher classification of the ant
subfamily Ponerinae [Hymenoptera: Formicidae], with a review of ponerine
ecology and behavior. Zootaxa.
Sherman, P. W., Reeve, H. K., & Pfennig, D. W. (1997). Recognition systems. In
J. R. Krebs, & N. B. Davies (Eds.), Behavioural ecology: An evolutionary approach
(4th ed.) (pp. 69e96). Oxford, U.K.: Blackwell Science.
Sledge, M. F., Boscaro, F., & Turillazzi, S. (2001). Cuticular hydrocarbons and
reproductive status in the social wasp Polistes dominulus.Behavioral Ecology and
Sociobiology, 49,401e409.
Sokal, R. R., & Rohlf, F. J. (2012). Biometry: The principles and practice of statistics in
biological research (4th ed.). New York: Freeman.
Thom, M. D., & Hurst, J. L. (20 04). Individual recognition by scent. Annales Zoologici
Fennici, 41,765e787.
Tibbetts, E. A. (2002). Visual signals of individual identity in the paper wasp
Polistes fuscatus.Proceedings of the Royal Society B: Biological Sciences, 269,
1423e1428.
Tibbetts, E. A., & Dale, J. (2007). Individual recognition: it is good to be different.
Trends in Ecology & Evolution, 22, 529e537.
Verlinden, H., Vleugels, R., Marchal, E., Badisco, L., Pflüger, H.-J., Blenau, W., et al.
(2010). The role of octopamine in locusts and other arthropods. Journal of Insect
Physiology, 56, 854e867.
de Vries, H. (1995). A n improved test of linearity in dominance hierarchies
containin g unknown or tied relationshi ps. Animal Behaviour, 50,
1375 e1389 .
Wiley, R. H. (2013). Specificity and multiplicity in the recognition of individuals:
implications for the evolution of social behaviour. Biological Reviews, 88,179e
195.
Zanette, L., & Field, J. (2009). Cues, concessions, and inheritance: dominance
hierarchies in the paper wasp Polistes dominulus.Behavioral Ecology, 20,
773e780.
B. Yagound et al. / Animal Behaviour 93 (2014) 25e3532

Appendix
Table A1
Mobility pattern of tested ants in the experimental device
Total distance
moved (cm)
Mean angular
velocity (rad/s)
Duration of mobility (s) Duration close to
the stimulus ant (s)
High-ranking tested ants confronted with
high-ranking stimuli (1, N¼20)
28.725.48 18.143.62 28.145.28 8.121.73
Low-ranking tested ants confronted with
high-ranking stimuli (2, N¼19)
39.675.78 14.622.22 35.864.97 11.873.01
High-ranking tested ants confronted with
low-ranking stimuli (3, N¼19)
22.214.76 12.192.22 22.094.29 8.801.89
P(1) vs (2) 0.18 0.42 0.29 0.31
P(1) vs (3) 0.38 0.18 0.38 0.79
Data are presented as mean SE.
Table A2
Chemical differences between workers of varying fertility
Compound Highly fertile
individuals
(N¼72)
Moderately fertile
individuals
(N¼80)
Infertile
individuals
(N¼71)
Unidentified 23.576.38 24.708.11 28.033.82
10-MeC
19
2.680.23 a 2.700.20 a 3.650.29 b
n-C
20
24.171.50 20.101.21 23.271.57
Unidentified 3.101.04 2.971.26 1.870.23
C
21:1
132.1434.10 45.9011.84 108.6247.33
C
21:1
42.1233.76 a 183.5979.26 a 802.62178.15 b
C
21:1
12.500.63 10.750.57 69.3059.80
C
21:1
79.319.56 67.668.42 67.569.76
n-C
21
2400.67106.70 a 2215.7679.44 ab 2023.28113.08 b
11-MeC
21
3.210.50 a 4.642.15 a 0.650.12 b
9-MeC
21
3.920.52 a 2.770.33 a 1.560.28 b
C
22:2
189.6214.08 184.6010.75 202.4814.92
n-C
22
214.879.78 203.736.85 199.568.37
11-MeC
22
10.761.09 9.180.74 8.530.97
9-MeC
22
1.800.19 2.020.23 2.420.46
C
23:2
7333.15388.67 ab 7548.17286.85 a 6265.83444.56 b
C
23:1
261.4833.84 220.0227.47 277.6565.43
n-C
23
2300.9272.74 2272.4070.29 2409.5282.63
11-MeC
23
75.456.59 a 61.354.79 a 28.633.98 b
C
24:2
38.403.12 37.921.81 36.292.38
n-C
24
19.020.62 17.730.56 19.240.62
C
25:2
485.8741.50 a 395.5627.64 a 219.1527.79 b
n-C
25
288.1412.40 258.259.96 263.7113.81
13-MeC
25
19.501.61 a 9.961.06 b 1.250.14 c
11-MeC
25
4.420.55 a 4.120.42 a 2.670.32 b
n-C
26
22.491.13 20.130.93 19.131.23
C
27:2
31.194.13 a 20.192.05 b 5.900.60 c
n-C
27
444.3025.12 a 378.1118.45 ab 354.5821.98 b
n-C
28
27.251.51 26.611.10 29.151.87
2-MeC
28
88.354.64 87.933.86 98.246.52
C
29:2
18.551.47 a 14.270.85 b 11.130.72 c
C
29:1
12.910.71 12.170.62 11.470.73
n-C
29
338.7923.86 340.4017.10 380.6622.18
n-C
31
43.923.82 a 59.544.64 b 86.395.70 c
Data correspond to absolute quantities (
m
g) of 34 cuticular hydrocarbons common to all individuals, and are presented as mean SE. Different letters (a, b, c) denote statistical
differences.
B. Yagound et al. / Animal Behaviour 93 (2014) 25e35 33

Abundance
Retention time (min)
8 1012141618202224262830
30
33
34
32
29
28
27
26
25
23
22
21
20
19
18
17
16
15
9
8
4
3
2
111
1.107
2.107
3.107
4.107
0
10 14 24
31
7
6
51213
Figure A1. Chromatogram of the cuticular hydrocarbon profile of a moderately fertile N. apicalis worker. Peaks used in the statistical analysis are indicated by numbers:
1¼unidentified; 2 ¼10-MeC
19
;3¼n-C
20
;4¼unidentified; 5 ¼C
21:1
;6¼C
21:1
;7¼C
21:1
;8¼C
21:1
;9¼n-C
21
;10¼11-MeC
21
;11¼9-MeC
21
;12¼C
22:2
;13¼n-C
22
;14¼11 -
MeC
22
;15¼9-MeC
22
;16¼C
23:2
;17¼C
23:1
;18¼n-C
23
;19¼11-MeC
23
;20¼C
24:2
;21¼n-C
24
;22¼C
25:2
;23¼n-C
25
;24¼13-MeC
25
;25¼11-MeC
25
;26¼n-C
26
;27¼C
27:2
;
28 ¼n-C
27
;29¼n-C
28
;30¼2-MeC
28
;31¼C
29:2
;32¼C
29:1
;33¼n-C
29
;34¼n-C
31
.
High-ranking stimulus
N = 24
T1 T2 T3 T4
0
20
40
60
80
High-ranking stimulus
N = 19
T1 T2 T3 T4
0
20
40
60
80
**
High-ranking stimulus
N = 10
T1 T2 T3 T4 T1 T2 T3 T4
Antennal contacts (s)
0
20
40
60
80
100 *
**
Low-ranking stimulus
N = 21
T1 T2 T3 T4
0
20
40
60
80
100 *** *
Low-ranking stimulus
N = 11
(a) (b)
(c) (d)
Figure A2. Duration of antennal contacts (s) towards the stimulus nestmate during habituation tests 1e4 (4 min each with a 5 min interval) in all experiments (see text). (a)
Experiment 1: a high ranker was confronted with either a high- or a low-ranking nestmate. (b) Experiment 2: a high ranker was confronted with a high-ranking nestmate. (c)
Experiment 3: a high ranker was confronted with a low-ranking nestmate. (d) Experiment 4: a low ranker was confronted with a high-ranking nestmate. Box plots represent 10th,
25th, 50th (median), 75th and 90th percentiles. Sample sizes of individuals are indicated for each experiment. ***P<0.001; **P<0.01; *P<0.05.
B. Yagound et al. / Animal Behaviour 93 (2014) 25e3534

−4−3−2−10123456
Discriminant function 1: 86.46%
−4
−3
−2
−1
0
1
2
3
4
5
6
Discriminant function 2: 13.54%
High-ranking individuals
Middle-ranking individuals
Low-ranking individuals
(a)
−4−3−2−10123456
Discriminant function 1: 92.96%
−4
−3
−2
−1
0
1
2
3
4
5
6
Discriminant function 2: 7.04%
Highly fertile individuals
Moderately fertile individuals
Infertile individuals
(b)
Figure A3. Discriminant function analyses showing the differences in the chemical profiles of 223 workers according to their (a) hierarchical rank and (b) ovarian development.
Ellipses represent 90% confidence intervals around centroids. The percentage of variance explained is depicted on each axe.
B. Yagound et al. / Animal Behaviour 93 (2014) 25e35 35