Content uploaded by Piotr Ślipiński
Author content
All content in this area was uploaded by Piotr Ślipiński on Sep 29, 2017
Content may be subject to copyright.
Content uploaded by Michał Żmihorski
Author content
All content in this area was uploaded by Michał Żmihorski on Jul 13, 2016
Content may be subject to copyright.
Insect Science (2017) 24, 842–852, DOI 10.1111/1744-7917.12354
ORIGINAL ARTICLE
Changes in the speed of ants as a result of aggressive
interactions
Piotr ´
Slipi´
nski1and Michał ˙
Zmihorski2,3
1Laboratory of Social and Myrmecophilous Insects, Museum and Institute of Zoology, Polish Academy of Sciences, Warsaw, Poland;
2Department of Ecology, Swedish University of Agricultural Sciences, Uppsala, Sweden and 3Institute of Nature Conservation, Polish
Academy of Sciences, Krak´
ow, Poland
Abstract Subordinate ant species utilize different tactics to reduce competition with the
stronger, larger and more aggressive individuals of a dominant species. In our experimental
study, we assessed the behavioral response of individual workers of 4 subordinate ant
species during their co-occurrence with workers of a single dominant species. Contrary to
most classical experiments focused on aggressive interactions, we assessed workers’ speed
as a crucial factor in the outcome of co-occurrence. Generally, there wasa large intraspecif ic
variation in the speed of the studied species—each had slow and fast individuals. Workers
of all studied species moved faster just after interaction, suggesting that contact between
2 hostile workers is a stressful stimulus, generating a behavioral reaction of increasing
speed. Also, the number of aggressive contacts experienced by a given individual positively
affected its speed. Moreover, workers which were fast when exploring territory were also
fast after interspecific interactions. The duration of agg ression was significantly reduced by
the speed and body size of a subordinate species worker—the more quickly a worker reacted
and bigger it was, the shorter was the time of cumulative aggression. To our knowledge,
this is the first study of this type to be conducted on ants and we conclude that speed is an
overlooked and important characteristic of species and also individuals, therefore it should
be considered as a driver of patterns of co-occurrence in ant assemblages.
Key words hierarchy; reaction; response; speed; survival; velocity
Introduction
To dominate is to have priority in accessing the crucial
resources necessary to live and reproduce (Clutton-Brock
& Parker, 1995). The physical supremacy of a dominant
species often leads to establishing a spatial and tempo-
ral order in terms of access to vital niche resources. Di-
verse inter- and intraspecific hierarchies are especially
common among social animals, such as wasps, ants, and
bumblebees (Wilson, 1975; Van Honk & Hogeweg, 1981;
Th´
eraulaz et al., 1992; Parr & Gibb, 2010). Among ants,
Correspondence: Piotr ´
Slipi´
nski, Laboratory of Social and
Myrmecophilous Insects, Museum and Institute of Zoology,
Polish Academy of Sciences, Wilcza 64, 00–679 Warsaw,
Poland. Tel: +48 502 309 722; fax: +48 22 629 63 02; email:
piotrs@miiz.waw.pl
such hierarchical interspecific organization is especially
well studied in Euro-Siberian and North American bo-
real forest, as well as in tropical forests (Savolainen
& Veps¨
al¨
ainen, 1988; Andersen, 1995; Davidson, 1997;
Bluethgen & Fiedler, 2004; Philpot, 2010; Parr & Gibb,
2010; Dejean et al., 2015).
In accordance with the competition hierarchy theory,
red wood ants (RWA e.g., Formi c a r u fa Linnaeus, F.
polyctena F¨
orst) are typical dominants of multispecies
ant assemblages (Savolainen & Veps¨
al¨
ainen, 1988). A
mature colony of red wood ants dominates as a soci-
ety, influencing subordinate species’ colonies inhabiting
a common niche (Punttila et al., 1996; Czechowski et al.,
2013). Single workers of red wood ants also dominate
over workers of subordinate species due to their physical
superiority (Kaspari & Weiser, 1999; Czechowski et al.,
2012). Red wood ants are active and aggressive animals
that fight all kinds of intruders encroaching their territory
C2016 Institute of Zoology, Chinese Academy of Sciences
842
Aggression in ants 843
(Reznikova & Dorosheva, 2004; Domisch et al., 2005;
V¨
a¨
an¨
anen et al., 2010). They affect the distribution and
behavior of subordinate ant species as well as of other in-
vertebrates (Reznikova & Dorosheva, 2004; Sorvari et al.,
2014; ˙
Zmihorski & ´
Slipi´
nski, 2016). Despite this, subor-
dinate species nest and forage in the territories of stronger
competitors (Czechowski & Mark´
o, 2006; Mark´
oet al.,
2013; ´
Slipi´
nski et al., 2014).
Subordinate ant species avoid direct interactions with
dominant species by utilizing different spatiotemporal
fragments of the common niche (Arnan et al., 2012; Cerd´
a
et al., 2013). However, interactions occur anyway, forcing
the smaller and weaker individuals of subordinate species
to flee or fight (Tanner & Adler, 2009). In Euro-Siberian
boreal forests, many species of Myrmicinae and Formici-
nae subfamilies co-occur in space and time with territorial
and aggressive red wood ants (V¨
a¨
an¨
anen et al., 2010;
˙
Zmihorski, 2011). The morphological and behavioral
features of species belonging to the above-mentioned
subfamilies differ significantly, which can potentially
determine their behavior during interactions with stronger
competitors (Tanner & Adler, 2009; Czechowski et al.,
2012). While workers from the Formicinae subfamily
(subgenus Serviformica) are generally fast moving, Myr-
mica workers are rather slow. The latter are protected by a
thick cuticle, which may increase their chances of survival
during antagonistic interactions. It is not clear however
how these differences in behavior (i.e., speed) and the
morphological characteristics of subordinate species af-
fect their interactions with dominants (i.e., red wood ants).
In this paper, we focused on the variation in speed
of 4 ant species’ workers of different size and ecology
during their co-occurrence with the workers of a domi-
nant species. The whole study was inspired by a common
(among many myrmecologists) conviction that workers
of F.fusca can forage in close vicinity to large groups
of red wood ants because they (F.fusca) are “dexterous
and agile” ants. Workers of both species usually inter-
act repeatedly when, for example, they locate a common
food source, yet in most situations, F. f u s c a workers are
able to “outrun” RWA workers just after an aggressive
interaction and escape. There are many similar examples
of relatively fast subordinate species in ant assemblages,
however, not many studies have tried to quantify the speed
of subordinate ant species after interactions as part of a
possible “defense reaction.” Therefore, we aimed to test 3
hypotheses based on current knowledge concerning inter-
specific interactions. First, given the obvious morpholog-
ical differences between subordinate species, we expected
to observe clear interspecific differences in the speed of
subordinate species’ workers after interactions with work-
ers of the dominant species. Second, we assume that an
increase in speed as a response to aggression—triggered
by interactions with workers of dominant species—is uni-
versal for all considered subordinate species, despite ob-
vious morphological and behavioral differences between
them. We assumed that such universal behavioral reac-
tions (i.e., speed increase) may potentially increase the
survival probability of subordinate species’ workers when
they are forced to compete with the larger and stronger
workers of dominant species. Third, we hypothesized that
the speed of subordinate species’ workers, after the in-
teraction with workers of a dominant species affects the
physical contact duration with the workers of the domi-
nant species. The reduction of contact time may be im-
portant for increasing the survival of subordinate species.
To verify the proposed hypotheses, we hosted artificial
colonies and performed a set of laboratory experiments,
video-recording the interactions among the ants.
Materials and methods
Ant species in the experiment
Four subordinate and 1 dominant ant species were used
in the experiments.
Formica polyctena was selected as a typical, territorial
species with large, strong and aggressive individuals. F.
polyctena creates large (up to several million individuals)
colonies, shaping the location, density, and fitness of
subordinate species’ colonies (Punttila et al., 1996). F.
polyctena workers permanently patrol their territory,
excluding subordinate species’ workers, as well as
influencing the size, location and fitness of subordinate
species’ colonies. Individuals of subordinate species are
excluded from food sources, injured and killed during
encounters (Savolainen & Veps¨
al¨
ainen, 1988; Pisarski
& Veps ¨
al¨
ainen, 1989). Because of the aggressive status
of this species and its physical superiority (compared to
the subordinate species in the experiment), we expected
workers from this species to be relatively fast when ex-
ploring territory and also to react quickly after aggressive
interactions for the purpose of attacking the opponent.
F.fusca Linnaeus is a subordinate species living in
F.polyctena territories. Colonies of F.fusca nesting in
F.polyctena territories are small, well concealed and in-
conspicuous (Savolainen, 1990). Shy, but fast and agile
workers forage mostly singly, do not engage in aggres-
sive interactions and avoid contact (De Biseau, 1997). We
expect F.fusca workers to be faster after the interactions
than F.polyctena workers.
A subsequent subordinate species used in the experi-
ment was F.rufibarbis Fabricius. Compared to F.fusca,
its colonies are more numerous and not so well concealed.
C2016 Institute of Zoology, Chinese Academy of Sciences, 24, 842–852
844 P. ´
Slipi´
nski & M. ˙
Zmihorski
Workers are much more predaceous and aggressive com-
pared to F. f u s c a . Because of its aggressiveness, F.ru-
fibarbis nests on the borders of F.polyctena territories
(Savolainen & Veps¨
al¨
ainen, 1988). We expect that work-
ers of F.rufibarbis may behave in a similar manner to the
workers of F.fusca.
Two species were chosen from the Myrmicinae sub-
family: Myrmica ruginodis Nylander and M.scabrinodis
Nylander. Colonies of these species are of various sizes,
up to 2 thousand individuals. Both species may show
limited aggression towards other ants during interactions
(Czechowski, 1979). Both species—as with most of the
Myrmica sp.—are stocky, move rather slowly and are
well protected by a hard cuticle (Czechowski et al., 2012,
p. 73). We hypothesized that these 2 species might
increase their speed as a response to aggression, but it
is rather unlikely that they would be able to outrun F.
polyctena, so alternative strategies could be also expected
for this species.
Colonies in the experiment
Five colonies were used (1 per each species) in the
experiment; they were taken from an area adjacent to
the protected region of the Mazovian Landscape Park
near the city of Warsaw (central Poland 52°1026N,
18°519E). Entire colonies were taken (a partial colony
in the case of F.polyctena) from the field with all
possible nest material and kept in large plastic contain-
ers (200 mm ×200 mm ×330 mm for F.polyctena,
360 mm ×230 mm ×230 mm for F.rufibarbis,
260 mm ×120 mm ×120 mm for F.fusca, and
180 mm ×180 mm ×210 mm for both species of Myr-
mica). The colonies in the field were spatially isolated, so
there was no possibility that workers from 1 colony had
had contact with worker from another colony in field con-
ditions. Such an arrangement excludes the occurrence of
the “dear-enemy phenomenon” (Pfennig & Reeve, 1989).
Each artificial nest was connected by a silicone tube to
a plastic feeding arena of 140 mm ×140 mm ×50 mm
(supplied with dead crickets, honey, and water), where
ants (foragers) were collected for the aggression test. For-
aging workers are usually more experienced comparing
to workers confined to tasks inside the colony, therefore
the aggression tests were performed only with the use of
foraging workers.
Aggression tests
Thirty aggression tests were conducted with F.fusca,
30 with F.rufibarbis, and 30 with Myrmica workers
(21 with M.scabrinodis and9withM.ruginodis); in all
cases the subordinate species interacted with different
F.polyctena individuals (total 90). After the aggression
test, ants were separated from the colony to prevent the
same individual from performing the test twice.
Each test lasted 5 min and 20 sec. Encounters were per-
formed on a Petri dish arena (9 cm in diameter). Before the
test, each Petri dish arena was placed in the container with
F.polyctena (for 5 min), allowing the workers to walk on
it and mark it with their scent. In this way, subsequent ag-
gression tests between dominant and subordinate species
were performed on the “territory” of the former. An F.
polyctena worker was always placed first into the arena—
also to ensure the effect of territorial “ownership”—the
subordinate individual was placed later (after 5–10 sec).
Workers of both species were transported with the use of
a plastic vial aspirator and placed randomly on the arena.
The temperature during the aggression test was stable (23–
24 ºC). All workers taking part in the aggression tests were
measured (from the tip of the head to the end of the ab-
domen) in Photoshop based on photographs made of live
individuals (in 1280 ×720 resolution). As body length
is associated with species identity (e.g., Fo rmic a were in
general larger than Myrmica), body length measurements
were standardized for each species independently. More
specifically, we centered the measurements by subtract-
ing the mean and scaled the measurements by dividing the
centered value by its standard deviation. As a result, the
standardized body length of individuals belonging to each
species had a mean =0 and SD =1, thus providing a mea-
sure for larger and smaller individuals of a given species
independent of between-species differences. These stan-
dardized values of body length were further used in the
statistical analyses.
Each 5-min aggression test was video recorded (Mi-
crosoft LifeCam Studio) in high resolution (1280 ×720,
15 frames per sec). Each recorded film was later broad-
casted in slow motion mode using BehaView software
(Boguszewski, 2011) in order to calculate: general speed
(dependent variable in model 1), mean postcontact speed
(model 2) and aggressive contact duration (model 3). Gen-
erally, the speed of workers was expressed as the distance
in centimeters travelled by an ant during 2 sec. The dis-
tance travelled by an ant was assessed manually by vi-
sually tracking the ant on a 1 cm ×1cmgrid(visible
through the transparent base of the Petri dish) and ex-
pressed as the total number of the grid cells that the ge-
ometrical center of the ant’s body passed during 2 sec.
We measured workers’ speed in 2 different situations: (1)
when they were exploring territory (so called exploration
speed) and (2) just after any kind of physical interaction
between workers (postcontact speed; see description of
C2016 Institute of Zoology, Chinese Academy of Sciences, 24, 842–852
Aggression in ants 845
Tab le 1 Set of explanatory variables used in the modeling.
Abbreviation Description
ANTSPEC Ant species. Categorical variable with 5 (in the case of model 1) or 4 (model 2) levels. Expectation:
different species may behave differently.
SPEEDCONTEXT Speed context. Categorical variable with 2 levels: exploration or postcontact. Used only in model 1,
where both types of speed were considered. Expectation: movement speed may differ between
exploration and after contact with F.polyctena.
Exploration speed: speed of a worker measured every 20 sec (each time for 2 sec). Represents the speed
of a subordinate species worker while “exploring” F.polyctena territory and is, at present, devoid of
any interspecific interactions. F.polyctena exploration speed is the speed of the dominant species
“exploring” its own territory. If there was any interaction between workers, the measurement of
exploration speed was omitted and measured again 20 sec later.
Postcontact speed: Speed of a worker (subordinate and dominant species) measured just after any type
of physical contact (interaction) between the 2 individuals. Speed was measured each time for 2 sec,
separately for the subordinate species’ workers and for F.polyctena.
TIMEINTERV Time interval. Continuous variable. Refers to the time intervals (20 sec) in which a given measurement
was taken, from 1 (beginning of aggression test) to 16 (end). Expectation: ants may change their
movement speed over time.
AGGRCONT Aggressive contacts. Continuous variable. Cumulative number of aggressive contacts between the 2
ants participating in a given aggression test. Each aggressive contact started with visible aggression
between the ants and ended when the ants became spatially separated, the number of such aggression
contacts was summarized for each individual ant. Expectation: number of former contacts, that is,
experience of a given individual, may change its behavior.
EXPLSPEEDRWA Exploration speed of Red Wood Ant. Continuous variable. Mean exploration speed of an F.polyctena
individual participating in a given aggression test. Expectation: exploration speed of RWA may force
workers of a subordinate species to behave in a different way (e.g., move slower/faster).
POSTCONTSPEEDRWA Postcontact speed of Red Wood Ant. Continuous variable. Mean postcontact speed of a F.polyctena
individual participating in a given aggression test. Expectation: postcontact speed of F. polyctena
may force workers of a subordinate species to behave in a different way (e.g., move slower/faster).
EXPLSPEED Exploration speed. Continuous variable. Mean exploration speed of a subordinate species’ individual
participating in a given aggression test. Expectation: specific exploration speed for a given individual
of a subordinate species may somehow affect its postcontact speed, contact duration, etc.
POSTCONTSPEED Postcontact speed. Continuous variable. Mean postcontact speed of a subordinate species’ individual
participating in a given aggression test. Expectation: specific postcontact speed for a given individual
of a subordinate species may somehow affect its exploration speed, contact duration, etc.
BODYLENGTHRWA Body length of Red Wood Ant worker. Continuous variable. F. polyctena body length.
BODYLENGTHSUB Body length of subordinate worker. Continuous variable. F.fusca,F.rufibarbis,M.ruginodis,andM.
scabrinodis body length.
variables used in the modeling in Table 1). The 2-sec
period used during preliminary observations showed that
the behavior (i.e., speed) of subordinate species’ individ-
uals after contact with F.polyctena distinctly changed
during the following 2 sec.
Statistical analysis
Speed differences between individuals We investi-
gated the differences in speed between individuals within
a given species. For this purpose, we used the Kruskal–
Wallis test implemented independently for each species
and speed context (exploration and postcontact).
Drivers of general speed (model 1) The f irst analy-
sis was performed to verify the effect of the 4 selected
factors on the average speed of a worker (both explo-
ration and postcontact speed) of all studied species. We
considered the 4 explanatory variables: species of ant
(ANTSPEC), time elapsed since the beginning of the exper-
iment (TIMEINTERV), number of aggressive contacts that
C2016 Institute of Zoology, Chinese Academy of Sciences, 24, 842–852
846 P. ´
Slipi´
nski & M. ˙
Zmihorski
took place between the 2 individuals (AGGRCONT), and
type of a given speed measure (SPEEDCONTEXT: postcon-
tact vs. exploration). Moreover, we considered the inter-
actions between ANTSPEC and the 3 explanatory variables
(TIMEINTERV,AGGRCONT,SPEEDCONTEXT), assuming that
the effect of these 3 variables can be different for a par-
ticular species of ant. The explanatory variables showed
no strong collinearity (r<0.42 in all cases). We used a
general linear mixed model (GLMM) with Gaussian error
distribution and identity link on all available measures of
exploration and postcontact speed for 172 individuals of
all 5 species (n=2241, i.e., single speed measurement
was a unit in this model). As we usually had more than 1
measurement for a given individual, we used the ID of the
ant individual within a given species as a random categor-
ical factor. We used the lme4 package (Bates et al., 2014)
implemented in R environment (R Core Team, 2015).
Drivers of the mean postcontact speed (model 2)
This analysis was performed to examine which factors
influence the speed of subordinate species’ workers after
contact with the dominant F.polyctena workers. We in-
vestigated the variability of the individual-specific mean
postcontact speed (i.e., average values from all measure-
ments taken among 73 subordinate and 73 dominant in-
dividuals) for all individuals of 4 subordinate species.
We considered the 6 explanatory variables: ant species
(ANTSPEC), mean exploration and mean postcontact speed
of F. polyctena in a given aggression test (EXPLSPEEDRWA
and POSTCONTSPEEDRWA, respectively), mean explo-
ration speed of a given ant (EXPLSPEED) and standard-
ized body length of F.polyctena and subordinate species’
workers (n=146) (BODYLENGTHRWA and BODYLENGTH-
SUB, respectively). The explanatory variables showed no
strong collinearity (r<0.44 in all cases). We used a gen-
eral linear model (GLM) implemented in R with Gaus-
sian error distribution and identity link for the investi-
gation of the effects of ANTSPEC as well as the effects
of the 3 explanatory variables (EXPLSPEEDRWA , P OST-
CONTSPEEDRWA , EXPLSPEED), and all the interactions be-
tween these variables and ANTSPEC.
Drivers of the aggressive contact duration (model 3)
The last analysis was performed for the purpose of ex-
amining how the speed of subordinate species workers
and their body length influence the duration of aggres-
sive contacts with workers of the dominant species. We
investigated the pooled duration time in seconds of all
aggressive contacts between 2 ants (n=73 individuals of
subordinate species and 73 of F.polyctena) occurring in a
given aggression test in relation to 6 potential predictors:
mean exploration and postcontact speed of F. polyctena in
a given test (EXPLSPEEDRWA and POSTCONTSPEEDRWA ,
respectively), mean exploration and postcontact speed of a
given subordinate ant (EXPLSPEED and POSTCONTSPEED,
respectively), standardized body length of F.polyctena
and subordinate species workers (BODYLENGTHRWA and
BODYLENGTHSUB, respectively) as well as interactions
between ANTSPEC and the 4 explanatory variables (EX-
PLSPEEDRWA , P OSTCONTSPEEDRWA , E XPLSPEED,POST-
CONTSPEED). The explanatory variables showed no strong
collinearity (r<0.55 in all cases). We used a general
linear model (GLM) with Gaussian error distribution and
identity link implemented in R (R Core Team, 2015).
Model selection We used an information-theoretic ap-
proach to infer the importance of particular predictors
for the variability of the observed speed and aggressive
contact duration. For this purpose, we used Akaike’s infor-
mation criterion (AIC). For each model (1–3), all possible
models were compared based on AIC scores. Model av-
eraging was conducted within ࢞AIC ࣘ4 (Burnham &
Anderson, 2002) with the help of the MuMIn package
(Barto´
n, 2013) in R (R Core Team, 2015) and presented
in Table S1. For each model, we used averaged parameter
estimates to infer the biological importance of the vari-
ables for the observed ant behavior. Averaged parameter
estimates with 95% confidence intervals not overlapping
zero were considered “significant” and important for ex-
plaining the dependent variables.
Results
Speed differences between individuals We recorded
significant differences in the mean exploration speed be-
tween individuals belonging to the same species. The dif-
ferences were recorded for M.ruginodis,M.scabrinodis
and F.polyctena, showing that speed variance between
individuals was greater than the variance within individu-
als. There were no such differences in the case of For m i ca
fusca, and F. rufibarbis (the latter on the verge of signif-
icance, see Fig. S1). A similar pattern was observed for
postcontact speed, but in this case, the differences between
individuals were significant for all species (Fig. S1).
General speed (model 1) In general, ants were mov-
ing faster after the contact than during exploration (ef-
fect of SPEEDCONTEXT, Table 2, Fig. 1). However, the
change of speed after the contact differed between the
studied species (interaction between ANTSPEC and SPEED-
CONTEXT). Moreover, the number of aggressive contacts
between 2 individuals positively affected the speed of both
ants (effect of AGGRCONT, Fig 2A). But the speed of both
ants showed a decreasing trend over time (i.e., during a
C2016 Institute of Zoology, Chinese Academy of Sciences, 24, 842–852
Aggression in ants 847
Tab le 2 Averaged parameter estimates explaining the speed of ants (models 1–2) and aggressive contact duration between ants (model
3) in response to experimental conditions.
Response variable Predictor Estimate 95% CI
Model 1 TIMEINTERV –0.040 –0.056; –0.025
General speed AGGRCONT 0.074 0.040; 0.1087
SPEEDCONTEXT =EXPLORATION SPEED –0.919 –1.072; –0.766
ANTSPEC =F.rufibarbis 0.676 0.375; 0.977
=F.fusca 0.241 –0.030; 0.513
=M.ruginodis –0.098 –0.527; 0.329
=M.scabrinodis –1.549 –1.868; –1.230
ANTSPEC =F.rufibarbis:SPEEDCONTEXT –0.570 –0.913; –0.228
=F.fusca:S
PEEDCONTEXT –0.366 –0.686; –0.046
=M.ruginodis:S
PEEDCONTEXT 0.035 –0.401; 0.471
=M.scabrinodis:S
PEEDCONTEXT 0.910 0.600; 1.221
Model 2 ExplSpeed 0.209 0.045; 0.372
Mean postcontact PostContSpeedRWA –0.183 –0.313; –0.053
speed ExplSpeedRWA 0.131 –0.063; 0.327
BodyLengthRWA 0.448 –2.041; 2.938
BodyLengthSub –0.154 –0.349; 0.040
ANTSPEC =M.ruginodis –0.422 –1.076; 0.232
=M.scabrinodis –1.768 –2.287; –1.248
=F.rufibarbis 0.409 –0.108; 0.928
Model 3 POSTCONTSPEED –0.135 –0.229; –0.042
Aggressive contact EXPLSPEED 0.040 –0.229; –0.042
duration POSTCONTSPEEDRWA 0.031 –0.599; 5.227
EXPLSPEEDRWA –0.101 –0.319; 0.115
BodyLengthRWA 2.314 –0.599; 5.227
BodyLengthSub –0.323 –0.561; –0.085
Note: Averaging was conducted on the basis of ωAIC. Estimates that do not overlap zero are typed in bold. In the explanatory variable
ANTSPEC,theFormica polyctena species was used as the reference level in model 1, For m ica f u s ca in model 2, whereas in SPEEDCONTEXT
level, POSTCONTACT was used as the reference level and parameter estimates for these levels are not provided, as it was denoted as 0.
5-min period, the effect of TIMEINTERV). The speed of ants
differed significantly among species. M. scabrinodis was
the slowest species, whereas F.rufibarbis was the fastest,
F.fusca was also very fast.
Mean postcontact speed (model 2) The mean post-
contact speed of an ant was positively correlated with
the exploration speed of this individual, which means
that workers exploring with high speed are also fast af-
ter the interaction with a worker of the dominant species.
The mean postcontact speed of a subordinate ant was
negatively affected by the postcontact speed of the F.
polyctena individual in a given aggression test (POSTCON-
TSPEEDRWA, Fig. 2B). Moreover, we recorded differences
among the subordinate species—M. scabrinodis was the
slowest after the contact whereas M. rufibarbis was the
fastest (Table 2).
Aggressive contact duration (model 3) The duration
of all aggressive contacts between 2 ants in a given ag-
gression test was negatively influenced by the postcontact
speed (POSTCONTSPEED) of the subordinate individual and
its size (BODYLENGTHSUB)—the slower and smaller the
subordinate worker, the longer the total time of aggres-
sion experienced by that worker (Figs. 2C and D).
Discussion
The experiments revealed that the speed of subordinate
ants differs distinctly, not only between species but also
between individuals of a given species. Each species has
slow and fast individuals, but workers of the majority
of species move faster after a contact than during explo-
ration. We recorded that the postcontact speed of subordi-
nates was affected negatively by the speed of F.polyctena,
C2016 Institute of Zoology, Chinese Academy of Sciences, 24, 842–852
848 P. ´
Slipi´
nski & M. ˙
Zmihorski
Fig. 1 Exploration and postcontact speed of the 5 ant species
expressed with kernel density estimation (irregular polygons)
and boxplots (white dot, median; black box, interquartile range).
but positively by aggression between individuals. More-
over, the slower the ant, the longer was the aggression
time. Subordinate workers of F.rufibarbis and F.fusca
are in general faster compared to the dominant workers
of F.polyctena, while Myrmica workers are slower. The
whole study was based on the use of only 1 nest for each
study species, so more replicates are needed to confirm
the patterns; however, the differences between species
were large and most likely robust in spite of the lack
of replication. Moreover, using different time window for
the postaggression speed measurements (in our study: 2
sec) may affect the obtained results as ants change behav-
ior continuously over time that elapsed since aggression.
This should be kept in mind when comparing our results
to other investigations in the future. Clear methodological
recommendations in this issue are lacking which could be
the reason for why so few studies examine speed of ants.
However, basing on the preliminary observations we as-
sessed that 2-sec period covers the visible ant behavioral
response to aggression and thus we recommend using sim-
ilar time window for the future studies. Below we discuss
the role of workers’ speed as a driver of patterns of ant
co-occurrence.
Speed as an individual response
Each species has individuals exploring at limited
speeds, which we may generally term “slow” (see Fig.
S1). On the other hand, each species has individuals ex-
ploring at a high speed, which can be called “fast” (up
to ca. 9 cm/sec in the case of Formicinae workers). A
similar pattern is observed for postcontact speed variabil-
ity within each studied species (Fig. S1). The observation
that each species has slow and fast individuals is also con-
firmed by model 2: we recorded that subordinate workers
which are fast during exploration are also fast after di-
rect contact with dominant workers (Table 2, model 2).
This suggests that the speed of a given individual is—
at least to some extent—repeatable in different situations
(exploration and postcontact). The existence of, for ex-
ample, bold and shy explorers has been well recognized,
mainly in birds (Drent et al., 2002). Our studies, however,
were not devoted to addressing the issue of differences
between individuals in intrinsic motivation, willingness
to explore and physiological state (Trullas & Skolnick,
1993; Schowalter, 2011; Hui & Pinter-Wollman, 2014;
Kraij-Fiˇ
ser & Schuett, 2014), thus leaving questions con-
cerning personality open.
Speed as result of co-occurrence
The analysis revealed significant differences in speed
between species, with workers of both Myrmica species
being the slowest,subordinate Formicinae (F.fusca and F.
rufibarbis) the fastest, whereas the workers of territorial F.
polyctena were moving at moderate speed (Fig. 1). Inter-
estingly, speed after contact was negatively related to the
mean postcontact speed of F.polyctena—the faster the F.
polyctena workers, the slower the workers of subordinate
species (Fig. 2B). In the laboratory conditions of a closed
Petri dish, workers usually detect the presence of other
workers at a distance of about 1–2 cm. This observation
is consistent with the literature, for example, Camponotus
sp. recognize a nest-mate from a distance of 1 cm (Brand-
staetter et al., 2008). At this distance, workers often stop
or slow down, probably to compare the hydrocarbon pro-
file with their own profile (Howard & Blomquist, 2005).
After such a comparison, F.polyctena workers usually
decide to attack. This strategy is utilized by most of the
behaviourally dominant species, most probably because
initiating a fight can increase the likelihood of winning a
contest (Mcauley et al., 1998; Tsutsui et al., 2003; Tan-
ner & Adler, 2009). The seconds following a contact are
crucial—if the subordinate worker decides to flee—then
reaction time and speed within the first few seconds of the
interaction are vital in terms of survival (De la Mora et al.,
2008). An attacking individual needs time during physical
contact to find a suitable place to bite the opponent. Work-
ers of a similar size usually cannot directly grab the other
ant on the head, thorax or abdomen and they usually try
to grasp the opponent on the leg or antennae (Dietrich &
Wehner, 2003). Also, our results (model 3) specified that
C2016 Institute of Zoology, Chinese Academy of Sciences, 24, 842–852
Aggression in ants 849
Fig. 2 Visualization of the relationship between speed and aggressive contact duration of ants and the most important explanatory
variables used in models 1–3 (see Table 2). Dots are partial residuals obtained by dropping the given explanatory variable from the
model while including all other variables. Lines are given to help with the visual interpretation of the relationships while significance
of the relationships and parameter estimates are given in Table 2.
the duration of all aggressive contacts occurring between
2 ants was negatively influenced by the individual size
of the subordinate—the smaller the subordinate worker,
the longer the total time of aggression. Contact between 2
hostile workers is a stressful stimulus, generating a behav-
ioral reaction of increasing speed by both dominant and
subordinate workers. This is unexpected, as most studies
of behavioral stress measured the stress effect in subor-
dinate species or individuals, only a few suggested that
the dominant species is also significantly stressed (Creel
et al., 1996; Kotrschal et al., 1998; Summers, 2002).
Speed as a response to aggression
The nature of the interactions among competing ant
workers from different species may create a series of
stressful situations, including aggression (Cerd´
a, 2001;
Summers, 2002; Czechowski & Mark´
o, 2006). There-
fore, during the analysis, we also considered the effect
of the number of aggressive contacts and time elapsed
from the beginning of the experiment (i.e., TIMEINTERV in
Table 1). Both factors significantly influenced ant speed
(model 1). An increased number of aggressive contacts
resulted in increased speed, which may indicate that a
worker already engaged in an aggressive interaction is
more stressed and disturbed after each subsequent con-
tact (Fig. 2A). On the other hand, the speed of observed
individuals decreased over time, which may be related to
increasing tiredness of co-occurring ants or other factors.
Strong competitive interactions between ecologically
similar species lead to asymmetric niche compression
with the most suitable localities occupied by the domi-
nant species. This rule can be observed, for example, in
guilds of avian predators, where the subordinate species
shifts its habitat use to avoid contact with the domi-
nant species (Vrezec & Tome, 2004). Similarly, strong
intraguild competition was noted for many epigeic inver-
tebrates, for example, survival of 1 species of spiders was
affected by the density of another species (Denno et al.,
2004). One may assume that in the case of ants, subor-
dinate species try to shorten the duration of aggressive
contacts with F. polyctena as much as possible, as fight-
ing with a stronger competitor decreases their chances
for survival. Model 3 showed clearly that the duration of
aggression was reduced by the speed of a subordinate in-
dividual after contact (Fig. 2C). In other words, the more
C2016 Institute of Zoology, Chinese Academy of Sciences, 24, 842–852
850 P. ´
Slipi´
nski & M. ˙
Zmihorski
quickly a worker reacts and flees after contact, the shorter
the cumulative time of aggression. A similar behavior was
observed in Carabidae beetles exploring the territory of F.
polyctena (Reznikova & Dorosheva, 2004). This experi-
ment showed that different Carabids use sets of behavioral
tactics to avoid aggression with ants in ant-dominated ter-
ritory. The largest individuals, such as Carabus regalis,
increased their speed of movement when sighting ants.
Another species, Pterostichus oblongopunctatus, actively
maneuvered among ants, while individuals of P.magus
preferred to stop and wait until the ants passed by, which
may suggest that this species uses a different strategy
in contacts with F. Polyctena (Reznikova & Dorosheva,
2004). A similar strategy may be used by Myrmica—
which was the slowest species in our experiment. Sub-
ordinate workers of Myrmica rugulosa usually “freeze”
during an interaction or in some cases, bend their legs
inward under the body (the so called subordinate pose or
transport pose) and are passively transported, which may
potentially reduce the aggression (Czechowski, 1979).
Conclusions
Considering the 3 hypotheses declared in the introduction,
we confirmed clear between-species differences in terms
of speed, reflecting the differences in the various species’
morphology and behavior. However, at the same time, we
observed differences in the speed of subordinate species’
workers after contact with the workers of the dominant
species. Contrary to our predictions, M. scabrinodis does
not increase its speed after the contact, showing that
particular subordinate species may use different tactics to
avoid interaction with a worker of the dominant species.
We also recorded that fast individuals have shorter
cumulative aggression time with dominant workers of red
wood ants. We find this result to be especially important,
as it directly shows that a worker’s speed ability can
benefit it directly, and thus should be considered in future
studies as an important adaptation to co-occurrence with
dominant species.
Acknowledgments
We are grateful for the helpful comments of 3 anonymous
referees, which greatly improved the quality of this article.
We are also indebted to Barbara Przybylska for linguis-
tic revision. This research was financed by a grant from
Iuventus Plus (IP2011 064771) funded by the Ministry
of Science and Higher Education and also by an Internal
Grant for Young Scientists from the Museum and Institute
of Zoology, Polish Academy of Sciences.
Disclosure
All authors declare no conflict of interest.
References
Andersen, A.N. (1995) A classification of Australian ant com-
munities, based on functional groups which parallel plant
life-forms in relation to stress and disturbance. Journal of
Biogeography, 22, 15–29.
Arnan, X., Cerd´
a, X. and Retana, J. (2012) Distinctive life traits
and distribution along environmental gradients of dominant
and subordinate Mediterranean ant species. Oecologia, 170,
489–500.
Barto´
n, K. (2013) MuMIn: multi-model inference. RPackage
Version 1.9.13. http://CRAN.R-project.org/package=MuMIn
Bates, D., Maechler, M., Bolker, B. and Walker,S. (2014) _lme4:
Linear mixed-effects models using Eigen and S4_. RPackage
Version 1.1. http://CRAN.R-project.org/package=lme4>7
Bluethgen, N. and Fiedler, K. (2004) Competition for composi-
tion: Lessons from nectar-feeding ant communities. Ecology,
85, 1479–1485.
Boguszewski, P. (2011) BehaView 0.0.11c. http://www.
pmbogusz.net/software/behaview/
Brandstaetter, A.S., Endler, A. and Kleineidam, C.J. (2008)
Nestmate recognition in ants is possible without tactile in-
teraction. Naturwissenschaften, 95, 601–608.
Burnham, K.P. and Anderson, D.R. (2002) Model Selection and
Multimodel Inference: A Practical Information–Theoretic Ap-
proach. 2nd edn. Springer, New York.
Cerd´
a, X. (2001) Behavioural and physiological trails to thermal
stress tolerance in two Spanish desert ants. Etologia, 9, 15–27.
Cerd´
a, X., Arnan, X. and Retana, J. (2013) Is competition a sig-
nificant hallmark of ant (Hymenoptera: Formicidae) ecology?
Myrmecological News, 18, 131–147.
Clutton-Brock, T.H. and Parker, G.A. (1995) Punishment in
animal societies. Nature, 373, 209–216.
Creel, S., Creel, N.M. and Monfort, S.L. (1996) Social stress
and dominance. Nature, 379, 212.
Czechowski, W. (1979) Competition between Lasius niger (L.)
and Myrmica rugulosa Nyl. (Hymenoptera, Formicidae). An-
nales Zoologici, 34, 437–451.
Czechowski, W. and Mark´
o, B. (2006) Uncomfortable protec-
tion: Formica polyctena F¨
orst. shelters Form i c a fus c a L. from
Formica sanguinea Latr. (Hymenoptera: Formicidae). An-
nales Zoologici, 56, 539–548.
Czechowski, W., Radchenko, A., Czechowska, W. and
Ve ps ¨
al¨
ainen K. (2012) The Ants of Poland with the
Reference to the Myrmecofauna of Europe, 496 pp. Fauna
Poloniae, Warsaw.
Czechowski, W., Mark´
o, B., Radchenko, A. and ´
Slipi´
nski,
P. (2013) Long-term partitioning of space between two
C2016 Institute of Zoology, Chinese Academy of Sciences, 24, 842–852
Aggression in ants 851
territorial species of ants (Hymenoptera: Formicidae) and
their effect on subordinate species. European Journal of En-
tomology, 110, 327–337.
Davidson, D.W. (1997) The role of resource imbalances in the
evolutionary ecology of tropical arboreal ants. Biological
Journal of the Linnean Society, 61, 153–181.
De Biseau, J.C., Quinet, Y., Deffernez, L. and Pasteels, J.M.
(1997) Explosive food recruitment as a competitive strategy
in the ant Myrmica sabuleti (Hymenoptera: Formicidae). In-
sectes Sociaux, 44, 59–73.
Dejean, A., Az´
emar, F., C ´
er´
eghino, R., Leponce, M., Corbara,
B., Orivel, J. and Compin, A. (2015) The dynamics of ant
mosaics in tropical rainforests characterized using the self-
organizing map algorithm. Insect Science. doi: 10.1111/1744-
7917.12208.
De la Mora, A., P´
erez-Lachaud, G. and Lachaud, J.P.
(2008) Mandible strike: the lethal weapon of Odontomachus
opaciventris against small prey. Behavioural Processes, 78,
64–75.
Denno, R.F., Mitter, M.S., Langellotto, G.A., Gratton, C. and
Finke, D.L. (2004) Interactions between a hunting spider and
a web-builder: consequences of intraguild predation and can-
nibalism for prey suppression. Ecological Entomology, 29,
566–577.
Dietrich, B. and Wehner, R. (2003) Sympatry and allopatry in
two desert ant sister species: how do Cataglyphis bicolor and
C.savignyi coexist? Oecologia, 136, 63–72.
Domisch, T., Fin´
er, L. and Jurgensen, M.F. (2005) Red wood ant
mound densities in managed boreal forests. Annales Zoologici
Fennici, 42, 277–282.
Drent, P.J., van Oers, K. and van Noordwijk, A.J. (2002) Re-
alized heritability of personalities in the great tit (Paru s ma -
jor). Proceedings of the Royal Society of London B, 270,
45–51.
Howard, R.W. and Blomquist, G.J. (2005) Ecological, behav-
ioral, and biochemical aspects of insect hydrocarbons. Annual
Review of Entomology, 50, 371–393.
Hui, A. and Pinter-Wollman, N. (2014) Individual variation in
exploratory behaviour improves speed and accuracy of col-
lective nest selection by Argentine ants. Animal Behaviour,
93, 261–266.
Kaspari, M. and Weiser, M.D. (1999) The size-grain hypothesis
and interspecific scaling in ants. Functional Ecology, 13, 530–
538.
Kotrschal, K., Hirschenhauser, K. and M¨
ostl, E. (1998) The
relationship between social stress and dominance is seasonal
in greylag geese. Animal Behaviour, 55, 171–176.
Kraij-Fiˇ
ser, S. and Schuett, W. (2014) Studying personality vari-
ation in invertebrates: why bother? Animal Behaviour, 91,
41–52.
Mcauley, D.G., Clugston, D.A. and Longcore, J.R. (1998) Out-
come of aggressive interactions between American black
ducks and mallards during the breeding season. Journal of
Wildlife Management, 62, 134–141.
Mark´
o, B., Czechowski, W. and Radchenko, A. (2013) Com-
bining competition with predation: drastic effect of Lasius
fuliginosus (Latr.) on subordinate ant species at the north-
ern limit of its distribution. Annales Zoologici, 63, 107–
111.
Parr, C.L. and Gibb, H. (2010) Competition and the role of dom-
inant ants. Ant Ecology (eds. L. Lach, C. Parr & K. Abbot), pp.
77–96. Oxford University Press, Oxford, United Kingdom.
Pfennig, D.W. and Reeve, H.K. (1989) Neighbour recognition
and context-dependent aggression in a solitary wasp, Sphecius
speciosus (Hymenoptera: Sphecidae). Ethology, 80, 1–18.
Philpott, S.M. (2010) A canopy dominant ant affects twig-
nesting ant assembly in coffee agroecosystems. Oikos, 119,
1954–1960.
Pisarski, B. and Veps¨
al¨
ainen, K. (1989) Competitive hierarchy
in ant communities (Hymenoptera, Formicidae). Annales Zo-
ologici, 42, 321–329.
Punttila, P., Haila, Y. and Tukia, H. (1996) Ant communities
in taiga clearcuts: habitat effects and species interactions.
Ecography, 19, 16–28.
R Core Team (2015) R: A Language and Environment for Sta-
tistical Computing. R Foundation for Statistical Computing,
Vienna, Austria. http://www.R-project.org/.
Reznikova, Z. and Dorosheva, H. (2004) Impacts of red wood
ants Formica polyctena on the spatial distribution and be-
havioural patterns of ground beetles (Carabidae). Pedobiolo-
gia, 48, 15–21.
Savolainen, R. and Veps¨
al¨
ainen, K. (1988) A competition hier-
archy among boreal ants: impact on resource partitioning and
community structure. Oikos, 51, 135–155.
Savolainen, R. (1990) Colony success of the submissive ant
Form i c a fus c a within territories of the dominant Formica
polyctena.Ecological Entomology, 15, 79–85.
Schowalter, T.D. (2011) Insect Ecology: An Ecosystem Ap-
proach, 633 pp. Academic Press, San Diego.
Sorvari, J., Huhta, E. and Hakkarainen, H. (2014) Survival
of transplanted nests of the red wood ant Formica aquilo-
nia (Hymenoptera: Formicidae): The effects of intraspecific
competition and forest clear-cutting. Insect Science, 21, 486–
492.
Summers, C.H. (2002) Social interaction over time, implica-
tions for stress responsiveness. Integrative and Comparative
Biology, 42, 591–599.
´
Slipi´
nski, P., Czechowski, W., Mark ´
o, B., Rzeszowski, K.
and Babik H. (2014) Lasius fuliginosus (Hymenoptera:
Formicidae) shapes local ant assemblages drastically. North-
Western Journal of Zoology, 10, 404–412.
Tanner, C.J. and Adler, F.R. (2009) To fight or not to fight:
context-dependent interspecific aggression in competing ants.
Animal Behaviour, 77, 297–305.
C2016 Institute of Zoology, Chinese Academy of Sciences, 24, 842–852
852 P. ´
Slipi´
nski & M. ˙
Zmihorski
Th´
eraulaz, G., Gervet, J., Thon, B., Pratte, M. and Semenoff-
Tian-Chansky, S. (1992) The dynamics of colonyorganization
in the primitively eusocial wasp Polistes dominulus (Christ).
Ethology, 91, 177–202.
Trullas, R. and Skolnick, P. (1993) Differences in fear motivated
behaviors among inbred mouse strains. Psychopharmacology,
111, 323–331.
Tsutsui, N.D., Suarez, A.V. and Grosberg, R.K. (2003) Ge-
netic diversity, asymmetrical aggression, and recognition in
a widespread invasive species. Proceedings of the National
Academy of Sciences of the United States of America, 100,
1078–1083.
Van Honk, C. and Hogeweg, P. (1981) The ontogeny of the social
structure in a captive Bombus terrestris colony. Behavioral
Ecology and Sociobiology, 9, 111–119.
V¨
a¨
an¨
anen, S., Veps¨
al¨
ainen, K. and Savolainen, R. (2010) In-
direct effect in boreal ant assemblages: territorial wood ants
protect potential slaves against enslaving ants. Annales Zoo-
logici, 60, 57–67.
Vrezec, A. and Tome, D. (2004) Habitat selection and patterns
of distribution in a hierarchic forest owl guild. Ornis Fennica,
81, 109–118.
Wilson, E.O. (1975) Sociobiology: The New Synthesis, 697 pp.
Harvard University Press, Cambridge.
˙
Zmihorski, M. (2011) Does the decline of red wood ants after
clear-cutting favour epigeic arthropods? European Journal of
Entomology, 108, 425–430.
˙
Zmihorski, M. and ´
Slipi´
nski, P. (2016) The importance of diurnal
and nocturnal activity and interspecific interactions for space
use by ants in clear-cuts. Ecological Entomology, 41, 276–
283.
Revised February 21, 2016 Accepted March 3, 2016
Supporting Information
Additional Supporting Information may be found in the
online version of this article at the publisher’s web-site:
Tabl e S1 Three sets of competing models included in
model averaging based on delta AICc.
Fig. S1 Mean exploration and postcontact speed of
ant individuals in increasing order (with 95% confidence
intervals) and results of Kruskal–Wallis test of the dif-
ferences in mean speed between individuals of a given
species.
C2016 Institute of Zoology, Chinese Academy of Sciences, 24, 842–852