Physiology & Behavior 101 (2010) 628–634

Contents lists available at ScienceDirect

Physiology & Behavior

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / p h b

Relationship between rank and plasma testosterone and cortisol in red deer
males (Cervus elaphus)
Luděk Bartoš a,⁎, Dieter Schams b, George A. Bubenik c, Radim Kotrba a, Milan Tománek d
a
Department of Ethology, Institute of Animal Science, Praha 10-Uhříněves, Czech Republic
b
Institute of Physiology, University of Munich-Weihenstephan, Freising-Weihenstephan, Germany
c
Department of Integrative Biology, University of Guelph, Guelph, Ontario, Canada
d
Biology of Reproduction, Institute of Animal Science, Praha 10-Uhříněves, Czech Republic

a r t i c l e i n f o a b s t r a c t

Article history: The objective of this study was to assess the effect of a change in the social composition in a group of red deer
Received 12 April 2010 males on the relationship between their rank and testosterone. A group of twelve adult red deer males (Cervus
Received in revised form 17 September 2010 elaphus) was tested in two social settings. From April 15 to June 9 (Period 1) this group was kept separately in
Accepted 20 September 2010
an enclosure. On June 10, nine 3-year-old males were added to that group of adult males. They were kept
together until August 31. We performed 10 observations of the group when the agonistic interactions of the
Keywords:
Red deer
males were recorded and we took 9 blood samples per male in Period 1; 11 observations were made and 10
Cervus elaphus samples were taken in Period 2. Concentrations of testosterone and cortisol were later determined in plasma.
Testosterone Adding much younger and smaller sparring partners into the experimental group of adult males in Period 2
Cortisol altered the agonistic behaviour of the adults even though this did not trigger any change in rank position of
Rank the experimental males except one. Adult males targeted preferentially their attacks on individuals much
Dominance lower in the hierarchy. Experimental male deer with higher social rank had lower levels of testosterone in
Social structure Period 1; in Period 2 it was just the opposite. In Period 1 the animals had higher cortisol levels than in Period 2.
As controls we used four adult (5 years old) males sharing the enclosure with four 3-year-old males. No
changes in hormone concentrations were observed in the control group. Thus, changing the social
environment of adult red deer males resulted in change of the relationship between rank and testosterone
and cortisol concentrations.
© 2010 Elsevier Inc. All rights reserved.

1. Introduction As subordinates are usually subjected to higher rates of stress they

Social interactions can profoundly affect the hypothalamo-pituitary-
adrenal axis [1]. As shown in the review by Sapolsky [2], no consensus
exists whether dominant or subordinate animals are more physiolog-
ically “stressed” [3,4]. Studies on olive baboons suggested that although
physiological reactions arise from rank, they are sensitive not only to the
rank but also to the social setting in which the rank occurs [5,6]. So, in
general, it can be stated that during a major hierarchical reorganization
within the given group, dominant individuals at the centre of social
tensions typically experience the greatest amounts of physical and
psychological stress. As a result, during such reorganization or soon after
a group formation among species, dominant individuals exhibit the
greatest physiological signs of stress. Once hierarchies stabilize,
subordination becomes associated with the greatest physiological
indices of stress [7].
⁎ Corresponding author. Tel.: +420 731650801.
E-mail address: bartos@vuzv.cz (L. Bartoš).
show higher levels of cortisol [3]. In some cases, however, there are
also dominant individuals who show this type of reaction. This
includes species where dominant individuals repeatedly have to
physically reassert their rank (e.g., dwarf mongooses, African wild
dogs, female ring-tailed lemurs, and male chimpanzees) [8,9] those
who are cooperative breeders (wolves and captive marmosets and
tamarins) [3,10] and those with transient periods of major rank
instability (free ranging baboons and captive populations of talapoin,
squirrel monkeys, and rhesus monkeys) [7]. In contrast, a higher level
of stress is seen among subordinate individuals in species where a
high rank is maintained through nonphysical intimidation and the
hierarchy is stable (e.g. feral male baboons and captive populations of
squirrel and rhesus monkeys, tree shrews, rats, and mice) [7,11,12]. It
is also observed in cases where subordinates are exposed to frequent
social stressors amid low availability of social support and minimal
presence of kin (e.g. feral ring-tailed lemurs and captive populations
of male rhesus or female talapoin monkeys) [3,13]. This situation also
exists where animals are located in an enclosure too small to allow
subordinate individuals to evade dominant ones [14]. In various
species of gregarious deer, the social relationships between group

0031-9384/$ – see front matter © 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.physbeh.2010.09.011
 L. Bartoš et al. / Physiology & Behavior 101 (2010) 628–634
members during the period of antler growth may be difficult to assess.
It depends on how often individual members meet and how far away
they are from each other during that time [15].
In our previous studies with red and fallow deer all agonistic
interactions of top-ranking males were typically of dominant nature,
whereas bottom-ranking males were always submissive [16,17].
Nevertheless, in a more recent study performed on a group of
white-tailed deer, bucks developed an unstable hierarchy with
relatively frequent changes in dominance ranking, particularly in
the upper half of the hierarchy. Bucks of all ranks were attacked by
others including some subordinates and in comparison with previous
studies on red and fallow deer, the whitetails displayed fewer
agonistic interactions and more triangular relationships in hierarchy.
Under these social conditions, our results describing the relationship
between rank position and antler cycle timing of white-tailed deer
[18] were in apparent contrast with previous studies on red deer
[19,20]. Nevertheless, even in red deer, we have proposed the
importance of the social structure of the group. In our previous
study we have observed that the higher the level of aggression among
the males, the closer the relationship between rank and timing of the
antler cycle [21]. In situations with a linear hierarchy, rank position
and antler casting are highly correlated. However, in complicated
hierarchies with multiple triangular relationships, the correlation
dropped below the level of significance [20]. Interpretation of our
previous results was based on presumption that testosterone plays
the main role in antlerogenesis [22] and that the dominant position,
usually based on aggressive behaviour, often tends to be related to
elevated androgen level, while subordinate status seems to be
associated with lower androgen secretion and increasing levels of
glucocorticoids [20]. By investigating the social relationships and
endocrine feedback, our ultimate goal was the elucidation of
regulation of antler development [22,23], elements not investigated
in this study. Understanding the relationship between rank and
testosterone levels is crucial for the interpretation of our previous
results that showed a link between rank and antler growth in deer
[17]. Recent meta-analysis failed to confirm a consistent pattern of
agonistic behaviour affecting testosterone [4]. A similarly confused
pattern in the relationship between aggression, rank and testosterone
has been documented among primates [3,24]. The inconsistency of
the results of the deer studies may be caused by the difference how
the dominant individuals themselves perceive the dominant position,
how aggressive they are and whom they attack.
We manipulated experimentally the social composition of a group
of red deer male. The objective of our study was to modify agonistic
behaviour of the group members and consequently the relationship
between rank and testosterone and cortisol. By adding young males to
a group of adult red deer males we expected to achieve a change in
agonistic behaviour of the adults. This could happen because adult
males will get a chance to attack much smaller youngsters rather than
fighting with equally sized rivals. If this resulted in calming down the
social frictions among the adult males then we also could expect
decrease in their mean cortisol concentrations. Initial high cortisol
concentrations in higher-ranking adult males would suppress their
testosterone levels and should result in no or a negative relationship
between rank and testosterone levels in adult males. After adding the
youngsters to the group of adults and calming down their interindi-
vidual interactions, mean cortisol concentrations should decrease. A
positive relationship between rank and testosterone should then be
reached.
2. Materials and methods
2.1. Animals, observation and procedure
Experimental subjects, twelve adult (5 to 9 years old) red deer
males, were tested in two social situations. From April 15 to June 9
629
(Period 1) the group was kept alone in a 0.75 ha enclosure at the
animal facility of the Institute of Animal Science. Control subjects
were composed of four adult (5 years old) males which were also kept
in a 0.75 ha enclosure at the same facility. They were sharing the
enclosure with six 3-year-old males. The animals had been set up in
these exact groups about three months prior to the start of the
experiment to establish their dominance relationships. Introducing
subordinate animals should not represent social instability for the
experimental individuals. The deer were tame and were kept
outdoors. Handling the deer in a crush was used for routine
manipulations. Therefore, all deer involved in the experiment were
used to this procedure and had been handled in a crush repeatedly
before the start of the experiment. No chemical restrain was used in
addition to the physical restraint. The deer were fed on grass
supplemented with barley (0.75 kg/animal/day) and hay. Starting
on June 10 until August 31 (Period 2) nine 3-year-old males were
added to the experimental group as “sparring partners” giving the
experimental subjects an opportunity to redirect their aggression to
smaller animals, which may be associated with stress-reduction in the
aggressors. In order to assess dominance relationship among males,
we made ten observations in the experimental group and 14 in the
control group in Period 1. Furthermore, we made 11 observations in
the experimental group as well as 19 observations in the control
group in Period 2. Observation periods started when food was
provided and ended when all animals left the feeding site. We
recorded any occurrence of an approach of one male to another, any
attack, threat gesture, or sparring and fighting, which caused an
apparent displacement of the approached individual. Then we
calculated a dominance index based on win–loss scores using ranking
according to David [25]. During each observation session we counted
the number of attacks of each subject and noted whom the subject
attacked. In the experimental group all attacks were standardized for
1 h of observation. It has been recorded previously that typical
dominant red deer male usually wins all agonistic interactions [16].
Therefore, we analyzed dyadic relationships between all males in the
group. The proportion of dyads over each period between two males
when all agonistic interactions were victorious exclusively in favour of
one of them was taken as a measure of social stability. The higher the
proportion of victorious dyads, the more socially stable is the group.
We also calculated the difference in rank between attacking and
attacked individuals in the experimental group. Based on the
dominance index, the individuals were ranked in such a way that
the one having the highest dominance index value was labelled 1, the
second highest dominance index value was labelled 2, etc. The
difference in rank was calculated by subtracting the value of the
attacked from the value of the attacking animal.
The experimental males were handled weekly and the control
males once a month in a crush for blood sampling and weighing. Our
main goal for establishing a control group was to check whether or not
the seasonal effect would be detected. Therefore, we made monthly
sampling of the control deer only. Nine blood samplings for the
experimental group and three blood samplings for controls were thus
available for Period 1 and 10 samplings for experimental and three
blood samplings for control groups in Period 2. During Period 2, the 3-
year-old males were also sampled for blood in the experimental group,
while these young males were sampled in both Periods in the control
group. Samples were taken at the same time of the day (09:00).
2.2. Hormone analyses
Concentrations of testosterone and cortisol were determined in
plasma by enzyme immunoassay (EIA). For each hormone, all samples
were assayed at the same time.
Testosterone was measured by EIA [26] with a double antibody
technique after extraction of 1 ml plasma with 3 ml tertiary butyl-
methylether/petrolether 30/70 (v/v). After freezing, the solvent fraction
630 L. Bartoš et al. / Physiology & Behavior 101 (2010) 628–634

was decanted and evaporated in a water bath at 60 °C. The extract was To estimate the rate of stress among adult males in the two Periods
dissolved in a 100 μl assay buffer, and 20 μl/tube was analyzed. The assay in comparison with that in 3-year-old males (the third step), we
used a polyclonal antibody raised in a rabbit against testosterone-11- constructed another GLMM model with cortisol as dependent variable
hemisuccinate-BSA, and the label was testosterone-3-carboxy methyl- and applied it on a data set involving adult males and 3-year-old
oxime-horseradish peroxidase. Cross reactivities were: testosterone males. Cortisol levels LSMEANs were computed for each Period and
100%, 5a-dihydrotestosterone 10%, androstendione 2%, oestradiol b0.1% male age category.
and progesterone b0.1%. The standard curve ranged from 0.2 to 25 pg/ In the last, fourth step, we tested whether cortisol concentrations
tube, corresponding to a sensitivity of 10 pg/ml plasma or 1 pg/ml influenced testosterone levels and vice versa. For the former, we
plasma after concentration (dissolving in 100 μl buffer only). The ED50 constructed the GLMM model for with the Box–Cox-transformed
of the standard curve was 2 pg/tube. The intra- and inter-assay cortisol as the dependent variable and Box–Cox-transformed testos-
coefficients of variation were an average 6.7% and 10.5% respectively. terone as another fixed effect completing the set of above listed fixed
Cortisol (4-pregnene-11ß-17α-21-triol-3, 20-dione, Steraloids, effects. For the latter we constructed the GLMM model with the Box–
Wilton, N. H., USA) concentrations were measured with a competitive Cox-transformed testosterone as the dependent variable and Box–
EIA [27]. The antibody was raised in a rabbit against cortisol-21- Cox-transformed cortisol as another fixed effect.
hemisuccinate-BSA. Cross reactivities were: cortisol 100%, cortisone The experimental protocol was approved by the Institute of
8%, corticosterone 9.5%, and prednisolone 18%. Blood plasma was Animal Science and Central Commission for Animal Welfare (Ministry
extracted with a 10 fold volume of tert-butyl methyl ether. After of Agriculture of the Czech Republic) Review Committee for the use of
freezing at −60 °C, extracts were decanted, evaporated and recon- Animal Subjects (protocol code G 403).
stituted. Interassay CV was between 7 and 14%.
3. Results
2.3. Statistics
3.1. Social stability
All data were analyzed with the aid of the SAS System (SAS,
version 9.1) in four steps. Experimental and control groups were Social hierarchy structure in experimental and control groups in
analyzed by separate models. Periods 1 and 2 is shown in Fig. 1. As expected, adult males in the
In the first step we analyzed changes in agonistic behaviour. For Experimental group maintained a stable hierarchy over the whole
each experimental subject and each Period we calculated the mean period. The only change in the rank position in Period 2 was detected
number of attacks per hour each male made and the mean number of in comparison to Period 1. Three-year-old males introduced in Period
attacks per hour he received. We also calculated the mean differences 2 never dominated their adult conspecifics. For unknown reasons,
in rank between attacking and attacked individuals for adult males in counter to expectation, the control group appeared socially unstable.
each Period when the subject was attacking and when he was Stabilization of the hierarchy was not completed even after 5 months.
attacked. Comparisons between the two Periods were performed Triangle relationships were common. The 3-year-old males expected
using the Wilcoxon match pair test. to be subordinated to the adult males, challenged and often
Further three steps were performed using the multivariate dominated some of the adults. Social stability (i.e., proportion of
Generalized Linear Mixed Model (GLMM). For testosterone and dyads with all agonistic interactions victorious exclusively in favour of
cortisol concentrations we found an appropriate power transforma- the dominant male) in Periods 1 and 2, was in the experimental group
tion of the non-normally distributed observations/data by computing 81.82% and 98.48% (number of dyads = 66, difference between the
Box–Cox family of power transformations [28] using PROC TRANSREG. Periods, Fisher's exact probability test P = 0.002) while in the controls
To account for the repeated measures on the same individuals, all 21.43% and 53.57% respectively (number of dyads = 56, difference
analyses were performed using a mixed model analysis with between the Periods, Fisher's exact probability test P = 0.013).
individual males in an interaction of the month as a random effect,
using PROC MIXED. Fixed factors were classes (‘Period’—Period 1 or 2;
and ‘Month’—April, May, June, July, and August) and continuous
variables rank (dominance index), age and body weight. The
significance of each fixed effect in the GLMM model was assessed by
the F-test, on sequential dropping of the non-significant effect,
starting with a full model. All interaction terms were tested, but
were not reported unless statistically significant. Associations be-
tween hormone concentrations (testosterone or cortisol) and rank,
body weight or concentrations of the other hormone were estimated
by fitting a random coefficient model using PROC MIXED as described
by Tao et al. [29]. With this random coefficient model we calculated
predicted Box–Cox-transformed hormone concentrations and plotted
them against the dominance index, body weight or concentrations of
the other hormone with predicted regression lines. Where appropri-
ate, we calculated least-squares means (LSMEANs) by computing the
mean of each treatment and averaging the treatment means. These
means of means are then used to compare the factors. In this way, the
means are adjusted for the number of observations in each treatment.
This estimate is unbiased because the unequal number of observa-
tions is taken into account [30]. We used a CONTRAST statement in
PROC MIXED to make comparisons between LSMEANs.
To assess the relationship between testosterone and rank (the
second step), we analysed associations between testosterone con-
centrations of adult males and their rank with Box–Cox-transformed Fig. 1. Social hierarchy structure in experimental and control groups in Periods 1 and 2
testosterone as the dependent variable. (grey dashed lines indicate a change in rank position in Period 2 compared to Period 1).
 L. Bartoš et al. / Physiology & Behavior 101 (2010) 628–634 631

3.2. Hierarchy of the adult experimental males and their

agonistic behaviour

There was no difference between the periods in the number of

attacks per hour either for attacking (the first step, Wilcoxon pair test,
T = 13, P = 0.339, Fig. 2 top left) or attacked experimental adult males
(Wilcoxon pair test, T = 18, P = 0.176, Fig. 2 top right). The difference
in rank between attacking and attacked individuals in the group of
experimental adult males was higher for attacking (Wilcoxon pair
test, T = 31, P = 0.01, Fig. 2 bottom left) but equal for attacked adult
males (Wilcoxon pair test, T = 16, P = 0.169, Fig. 2 bottom right).

3.3. Association between testosterone and rank of experimental and

control adult males

The GLMM model (the second step) showed that in experimental

adult males testosterone concentrations were dependent on the
dominance index nested within the period (GLMM, F 2,80.9 = 4.77,
P b 0.01, Fig. 3 top) and on body weight (GLMM, F 1,77.3 = 23.32,
P b 0.0001, Fig. 4 top). No interaction was found between the
dominance index and body weight. In contrast, testosterone con-
centrations in control adult males did not show any significant
dependency on the dominance index nested within the period
(GLMM, F 2,29 = 0.41, P = 0.669, Fig. 3 bottom) nor on body weight
(GLMM, F 1,29 = 0.87, P = 0.360, Fig. 4 bottom).

3.4. Cortisol of the experimental and control males

Cortisol concentrations of experimental adult males (the third

step) were different between the periods but did not differ from those
of the 3-year-old males in Period 2 in experimental group [GLMM, Fig. 3. Predicted values of Box–Cox transformed testosterone (ng/ml) plotted against
category of males (adult males, 3-year-old males) nested within dominance index of adult red deer males in Periods 1 and 2 in experimental (top) and
control subjects (bottom).
period, F 1,136 = 4.73, P = 0.031, Fig. 5 left]. In the control group, cortisol
concentrations nested within the period did not show any significant

Fig. 2. Comparison of experimental adult red deer males according to number of attacks
per hour (mean ± SE, top) and mean (± SE) difference in rank between attacking and Fig. 4. Predicted values of Box–Cox transformed testosterone (ng/ml) plotted against
attacked individuals (bottom). All comparisons were made either when the male was body weight of red deer males in Periods 1 and 2 in experimental (top) and control
attacking (left column) or was attacked (right column). adult males (bottom).
632 L. Bartoš et al. / Physiology & Behavior 101 (2010) 628–634

Fig. 5. Box–Cox-transformed cortisol least square means (± SE) of adult and 3-year-old
red deer males in Periods 1 and 2 in experimental (left) and control groups (right).

variation (GLMM, F 3,31 = 0.94, P = 0.433, Fig. 5 right). However, they

were dependent on body weight (GLMM, F 1,31 = 19.92, P b 0.001,
similar trend as in Fig. 4, but not shown).

3.5. Cortisol concentrations influencing testosterone levels

In the experimental group, the GLMM model (the fourth step)

showed that in adult males testosterone concentrations were depen-
dent on the cortisol concentrations (GLMM, F1,102 = 6.58, P b 0.01, Fig. 6),
Period (GLMM, F1,98.8 = 87.01, P b 0.0001, not shown, no interaction
between the cortisol concentrations and Period was detected), and
interaction between cortisol concentrations and body weight (GLMM,
F1,101 = 7.07, P b 0.01, Fig. 7 top). The fixed effect of body weight alone
did not appear significant (GLMM, F1,88.6 = 1.72, NS). In the control
group, testosterone concentrations of adult males were not dependent
on cortisol concentrations (GLMM, F1,15 = 2.61, NS), but were depen-
dent on body weight (GLMM, F1,15 = 4.47, P = 0.052, not shown), Period
(GLMM, F1,15 = 14.74, P b 0.01, not shown), and the interaction between
cortisol concentrations and body weight nearly reached a level of
significance (GLMM, F1,15 = 3.46, P = 0.08, Fig. 7 bottom). In neither the
experimental nor the control group did we find dependency of cortisol
on any single fixed effect including testosterone (GLMM, experimental
group F1,137 = 0.13, NS; control group F1,15 = 2.62, NS).
4. Discussion
Adding much younger and smaller red deer males into the
experimental group of adult males in Period 2 altered the agonistic
behaviour of the adults. Although we did not observe any changes in the
Fig. 6. Predicted values of Box–Cox transformed testosterone (ng/ml) plotted against
Box–Cox transformed cortisol (ng/ml) of experimental adult red deer males in Periods
1 and 2.
Fig. 7. Predicted values of Box–Cox transformed testosterone (ng/ml) plotted against
Box–Cox transformed cortisol (ng/ml) and body weight of adult red deer males in
experimental (top) and control subjects (bottom).
rank position except one, social stability considerably improved. This
was indicated by a significant increase of dyads with all agonistic
interactions victorious exclusively in favour of the dominant male.
General involvement of the adult males in agonistic encounters did not
change because the number of attacks remained the same. Also in other
species the rate of aggression was not higher during unstable periods
than during stable ones [31], thus suggesting that frequencies of
aggression per se may have little effect on the rank. On contrary, in
Period 2 our experimental adult males targeted their attacks on
individuals much lower in the hierarchy; i.e. in most cases on
youngsters. In Period 1 the experimental adult males had higher cortisol
levels than in Period 2. This suggests that among adult deer maintaining
rank position was an uncertain venture causing a stress reaction
comparable to that seen in subordinate young males, who were
otherwise exposed to permanent attacks from the dominant adults in
Period 2. Social stress in adult males was the cost of dominance, not a
consequence of subordination [10]. Our high-ranked individuals had the
greatest physiological signs of stress when living in a competitive social
situation with equally sized adult conspecifics. The presence of smaller
and obviously submissive individuals created a situation in which
dominant red deer males attacked others with low risk of failure [16]
and “a typical” physiological response occurring when subordinates
exhibited relatively higher cortisol levels than the dominant ones [3,7].
This helped high-ranking males to maintain a high social position
through nonphysical intimidation and the hierarchy thus became stable.
 L. Bartoš et al. / Physiology & Behavior 101 (2010) 628–634
In experiments like this it is difficult to set up an appropriate control.
Contrary to our expectation, the controls exhibited higher cortisol
concentrations than the experimental males. This discrepancy was
obviously caused by unusual social instability of the control group.
Despite the fact that no additional individuals were introduced to the
control group, more than five months was not sufficient time to
establish stable social hierarchy within this control group. Moreover, a
significant decrease of dyads with all agonistic interactions victorious
exclusively in favour of the dominant male was recorded. This could
reflect increased aggressiveness associated with the seasonal increase in
testosterone concentrations [23]. Not only was the social structure of the
control adult males unstable, the presumed subordinate young
conspecifics frequently challenged the adult males, often even being
even dominant over them. This long-lasting social instability and
uncommon rank relationships were unexpected; differing considerably
from our previous experience with this species. However, we believe
this group will still serve as a conclusive control as it was not showing
any seasonal changes in the relationship between rank and testosterone
concentrations when social composition had not been changed
throughout the Periods 1 and 2. In addition, this group clearly showed
that permanent social frictions among the group members caused
physiological stress reflected by considerably increased cortisol con-
centrations in both dominant and subordinate as well as adults and 3-
year-olds.
Unlike various other studies in ungulates [32], we found no age
effect on the hormone concentrations of the adult males. That was
caused by the fact that most of the adult males, if not all, had already
reached the end of their physical development.
In conclusion, as hypothesized, changing the social environment of
adult red deer males resulted in a change of the relationship between
rank and testosterone concentrations. While in Period 1 there was a
negative correlation between dominance and testosterone, in Period 2 it
was just the opposite. No such change was detected in the control group.
Despite the fact that we found a significant effect of testosterone
on cortisol concentrations in our experimental males, no clear trend
was apparent (Fig. 6). Nevertheless, both groups showed a similar
tendency in testosterone dependency on the interaction between
cortisol concentrations and body weight. Decreased concentrations of
testosterone were associated with high cortisol levels and low body
weight. On the other hand, we did not find any dependency of cortisol
on testosterone. This suggests an effect of cortisol on testosterone and
not vice versa. As in our case, it has been well established that chronic
stress can cause a drop in testosterone levels, which is at least partially
driven by an increase of cortisol levels following the impact of the
stressor [33–36]. However, we have to admit, that the in various
species testosterone metabolites were significantly correlated with
cortisol [37–40]. Our 3-year-old subordinate males did not show any
difference in cortisol in comparison to adult dominant males. These
data further support the contention that social subordinance and
stress are not inevitably linked [41–43].
Acknowledgments
The authors gratefully acknowledge the assistance of Jiří Šiler,
Soběslav Losos, Petr Janovský, Vratislav Kšáda, and Ludmila Švecová.
We are grateful to two anonymous reviewers for their constructive
comments and discussion. This study was supported by the Czech
Science Foundation (grant number 523/08/0808); and the Ministry of
Agriculture of the Czech Republic (grant number MZe 0002701404).
References
[1] DeVries AC, Glasper ER, Detillion CE. Social modulation of stress responses. Physiol
Behav 2003;79:399–407.
[2] Sapolsky RM. The influence of social hierarchy on primate health. Science
2005;308:648–52.
633
[3] Abbott DH, Keverne EB, Bercovitch FB, Shively CA, Medoza SP, Saltzman W, et al.
Are subordinates always stressed? A comparative analysis of rank differences in
cortisol levels among primates. Horm Behav 2003;43:67–82.
[4] Hirschenhauser K, Oliveira RF. Social modulation of androgens in male
vertebrates: meta-analyses of the challenge hypothesis. Anim Behav 2006;71:
265–77.
[5] Sapolsky RM. Endocrine aspects of social instability in the olive baboon (Papio
anubis). Am J Primatol 1983;5:365–79.
[6] Sapolsky RM. Testicular function, social rank and personality among wild baboons.
Psychoneuroendocrinology 1991;16:281–93.
[7] Sapolsky RM. The physiology of dominance in stable versus unstable social
hierarchies. In: Mason WA, Mendoza SP, editors. Primate social conflict. New York:
State University of New York Press; 1993. p. 171–204.
[8] Creel S, Creel NM, Monfort SL. Social stress and dominance. Nature 1996;379:212.
[9] Masataka N, Ishida T, Suzuki J, Matsumura S, Udono S, Sasaoka S. Dominance and
immunity in chimpanzees (Pan troglodytes). Ethology 1990;85:147–55.
[10] Sands J, Creel S. Social dominance, aggression and faecal glucocorticoid levels in a
wild population of wolves, Canis lupus. Anim Behav 2004;67:387–96.
[11] Barnett SA. Competition among wild rats. Nature 1955;175:126–7.
[12] Eberhart JA, Keverne EB, Meller RE. Social influences on circulating levels of
cortisol and prolactin in male talapoin monkeys. Physiol Behav 1983;30:361–9.
[13] Cavigelli SA. Behavioural patterns associated with faecal cortisol levels in free-
ranging female ring-tailed lemurs, Lemur catta. Anim Behav 1999;57:935–44.
[14] McLeod PJ, Moger WH, Ryon J, Gadbois S, Fentress JC. The relation between urinary
cortisol levels and social behaviour in captive timber wolves. Can J Zool 1996;74:
209–16.
[15] Bartoš L, Perner V. Integrity of a red deer stag social group during velvet period,
association of individuals, and timing of antler cleaning. Behaviour 1985;95:314–23.
[16] Bartoš L. Dominance and aggression in various sized groups of red deer stags.
Aggress Behav 1986;12:175–82.
[17] Bartoš L, Losos S. Response of antler growth to changing rank of fallow deer buck
during the velvet period. Can J Zool 1997;75:1934–9.
[18] Bartoš L, Miller KV, Osborn D. Relationship between dominance and antler cycle in
white-tailed deer (Odocoileus virginianus). Acta Theriol 2004;49:517–25.
[19] Bartoš L. The date of antler casting, age and social hierarchy relationships in the
red deer stag. Behav Process 1980;5:293–301.
[20] Bartoš L. Social status and antler development in red deer. In: Bubenik GA,
Bubenik AB, editors. Horns, pronghorns and antlers: evolution, morphology and
social significance. New York: Springer-Verlag; 1990. p. p442–59.
[21] Bartoš L. Relationships between behaviour and antler cycle in red deer. Ethology
1986;71:305–14.
[22] Bartoš L, Schams D, Kierdorf U, Fischer K, Bubenik GA, Šiler J, et al. Cyproterone
acetate reduced antler growth in surgically castrated fallow deer. J Endocrinol
2000;164:87–95.
[23] Bartoš L, Schams D, Bubenik GA. Testosterone, but not IGF-1, LH, prolactin or
cortisol, may serve as antler-stimulating hormone in red deer stags (Cervus
elaphus). Bone 2009;44:691–8.
[24] Honess PE, Marin CM. Behavioural and physiological aspects of stress and
aggression in nonhuman primates. Neurosci Biobehav Rev 2006;30:390–412.
[25] David HA. Ranking from unbalanced paired-comparison data. Biometrika
1987;74:432–6.
[26] Blottner S, Hingst O, Meyer HHD. Seasonal spermatogenesis and testosterone
production in roe deer (Capreolus capreolus). J Reprod Fertil 1996;108:299–305.
[27] Sauerwein H, Dürsch I, Meyer HHD. Quantitation of glucocorticoid receptors in
bovine skeletal muscle: topographical distribution, sex effect and breed
comparisons. J Steroid Biochem Mol Biol 1991;39:941–5.
[28] Timm NH, Mieczkowski TA. Univariate and multivariate general linear models:
theory and application using SAS software. Cary, NC, USA: SAS Institute Inc.; 1997.
[29] Tao J, Littel R, Patetta M, Truxillo C, Wolfinger R. Mixed model analyses using the
SAS system course notes. Carry, NC, USA: SAS Institute Inc.; 2002.
[30] Welsh S, Bohannon T, Cohen R. Statistics II: ANOVA and regression course notes.
Cary, NC, USA: SAS Institute Inc.; 2000.
[31] Sannen A, Van Elsacker L, Heistermann M, Eens M. Urinary testosterone-
metabolite levels and dominance rank in male and female bonobos (Pan paniscus).
Primates 2004;45:89–96.
[32] Pelletier F, Bauman J, Festa-Bianchet M. Fecal testosterone in bighorn sheep (Ovis
canadensis): behavioural and endocrine correlates. Can J Zool 2003;81:1678–84.
[33] Cummings DC, Quigley ME, Yen SSC. Acute suppression of circulating testosterone
levels by cortisol in men. J Clin Endocrinol Metab 1983;57:671–3.
[34] Orr TE, Mann DR. Role of glucocorticoids in the stress-induced suppression of
testicular steroidogenesis in adult male rats. Horm Behav 1992;26:350–63.
[35] Moe RO, Bakken M. Effect of repeated blood sampling on plasma concentrations of
cortisol and testosterone and on leucocyte number in silver fox vixens (Vulpes
vulpes). Acta Agric Scand 1996;46:111–6.
[36] Mehta PH, Jones AC, Josephs RA. The social endocrinology of dominance: basal
testosterone predicts cortisol changes and behavior following victory and defeat. J
Pers Soc Psychol 2008;94:1078–93.
[37] Mooring MS, Patton ML, Lance VA, Hall BM, Schaad EW, Fetter GA, et al.
Glucocorticoids of bison bulls in relation to social status. Horm Behav 2006;49:
369–75.
[38] Dittami J, Katina S, Moestl E, Eriksson J, Machatschke IH, Hohmann G. Urinary
androgens and cortisol metabolites in field-sampled bonobos (Pan paniscus). Gen
Comp Endocrinol 2008;155:552–7.
[39] Lidgard DC, Boness DJ, Bowen WD, Mcmillan JI. The implications of stress on male
mating behavior and success in a sexually dimorphic polygynous mammal, the
grey seal. Horm Behav 2008;53:241–8.
634 L. Bartoš et al. / Physiology & Behavior 101 (2010) 628–634

[40] Oliveira T, Gouveia MJ, Oliveira RF. Testosterone responsiveness to winning and
losing experiences in female soccer players. Psychoneuroendocrinology 2009;34:
1056–64.
[41] Bubenik GA, Reyes Toledo E. Plasma levels of cortisol, testosterone and growth
hormone in pudu (Pudu puda Molina) after ACTH administration. Comp Biochem
Physiol 1994;107:523–7.
[42] Bercovitch FB, Clarke AS. Dominance rank, cortisol concentrations, and reproductive
maturation in male rhesus macaques. Physiol Behav 1995;58:215–21.
[43] Barrett GM, Shimizu K, Bardi M, Asaba S, Mori A. Endocrine correlates of rank,
reproduction, and female-directed aggression in male Japanese Macaques
(Macaca fuscata). Horm Behav 2002;42:85–96.