ArticlePDF Available

Abstract and Figures

A dominance hierarchy is an important feature of the social organisation of group living animals. Although formal and/or agonistic dominance has been found in captive wolves and free-ranging dogs, applicability of the dominance concept in domestic dogs is highly debated, and quantitative data are scarce. Therefore, we investigated 7 body postures and 24 behaviours in a group of domestic dogs for their suitability as formal status indicators. The results showed that high posture, displayed in most dyadic relationships, and muzzle bite, displayed exclusively by the highest ranking dogs, qualified best as formal dominance indicators. The best formal submission indicator was body tail wag, covering most relationships, and two low postures, covering two-thirds of the relationships. In addition, both mouth lick, as included in Schenkel's active submission, and pass under head qualified as formal submission indicators but were shown almost exclusively towards the highest ranking dogs. Furthermore, a status assessment based on changes in posture displays, i.e., lowering of posture (LoP) into half-low, low, low-on-back or on-back, was the best status indicator for most relationships as it showed good coverage (91% of the dyads), a nearly linear hierarchy (h' = 0.94, p
Content may be subject to copyright.
Dominance in Domestic Dogs: A Quantitative
Analysis of Its Behavioural Measures
Joanne A. M. van der Borg
*, Matthijs B. H. Schilder
, Claudia M. Vinke
, Han de Vries
1Wageningen University Behavioural Ecology Group, Department of Animal Sciences, P.O. Box 338, 6700
AH, Wageningen, The Netherlands, 2Utrecht University Faculty of Veterinary Medicine, Department of
Animals in Science & Society, P.O. Box 80166, 3508 TD, Utrecht, The Netherlands, 3Utrecht University
Animal Ecology Group, Department of Biology, Padualaan 8, f3584 CH, Utrecht, The Netherlands
A dominance hierarchy is an important feature of the social organisation of group living ani-
mals. Although formal and/or agonistic dominance has been found in captive wolves and
free-ranging dogs, applicability of the dominance concept in domestic dogs is highly
debated, and quantitative data are scarce. Therefore, we investigated 7 body postures and
24 behaviours in a group of domestic dogs for their suitability as formal status indicators.
The results showed that high posture, displayed in most dyadic relationships, and muzzle
bite, displayed exclusively by the highest ranking dogs, qualified best as formal dominance
indicators. The best formal submission indicator was body tail wag, covering most relation-
ships, and two low postures, covering two-thirds of the relationships. In addition, both
mouth lick, as included in Schenkels active submission, and pass under head qualified as
formal submission indicators but were shown almost exclusively towards the highest rank-
ing dogs. Furthermore, a status assessment based on changes in posture displays, i.e.,
lowering of posture (LoP) into half-low,low,low-on-back or on-back, was the best status
indicator for most relationships as it showed good coverage (91% of the dyads), a nearly
linear hierarchy (h= 0.94, p<0.003) and strong unidirectionality (DCI = 0.97). The associ-
ated steepness of 0.79 (p<0.0001) indicated a tolerant dominance style for this dog group.
No significant correlations of rank with age or weight were found. Strong co-variation
between LoP,high posture, and body tail wag justified the use of dominance as an inter-
vening variable. Our results are in line with previous findings for captive wolves and free-
ranging dogs, for formal dominance with strong linearity based on submission but not
aggression. They indicate that the ethogram for dogs is best redefined by distinguishing
body postures from behavioural activities. A good insight into dominance hierarchies and
its indicators will be helpful in properly interpreting dog-dog relationships and diagnosing
problem behaviour in dogs.
PLOS ONE | DOI:10.1371/journal.pone.0133978 August 26, 2015 1/18
Citation: van der Borg JAM, Schilder MBH, Vinke
CM, de Vries H (2015) Dominance in Domestic Dogs:
A Quantitative Analysis of Its Behavioural Measures.
PLoS ONE 10(8): e0133978. doi:10.1371/journal.
Editor: Odile Petit, CNRS (National Center for
Scientific Research), FRANCE
Received: September 19, 2014
Accepted: July 5, 2015
Published: August 26, 2015
Copyright: © 2015 van der Borg et al. This is an
open access article distributed under the terms of the
Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any
medium, provided the original author and source are
Data Availability Statement: Data are available in
the paper itself and the Appendix (its Supporting
Information files).
Funding: No specific funding was received for this
Competing Interests: The authors have declared
that no competing interests exist.
Generally stated, living in social groups can be beneficial for individual and species level survival
for several reasons and in several circumstances, both in the short and long term (see [1]).
Wolves and domestic dogs are Canidea species known for their high degree of sociality [2,3],
but there is little quantitative data concerning the dynamics of their social organisation, domi-
nance hierarchy and dominance style, affiliative relationships, coalition/alliance formationf and
reconciliation behaviour. Such behaviour, if the social organisation in wolves can be taken as an
example, appears to assist in coping with continuous change in several areas. For example, they
contribute to hunting efficiency: prey size, availability of prey, prey detection, and to the care
offered in a pack: learning opportunities, alloparental care; and territory defence [2,4]. At the
same time, living in social groups may enhance competition for highly valued resources such as
food and mates [5]. This competition can lead to conflicts that may compromise group stability
and may even result in chronic stress and physical harm [6]. It can be alleviated by dominance
hierarchies built from stable dyadic dominance relationships, allowing regulation of priority of
access to highly valuable resources, and preventing fierce or recurring conflicts. The peacefulness
of interactions within wild wolf packs impressed Mech, an experienced wolf observer [7]. Never-
theless, Mechs remarks to this effect has led to a series of articles in the ongoing debate on
whether or not dominance plays a role in the society of the wild wolf and consequently, whether
dominance could constitute an important element in structuring social relationships between
dogs and also between dogs and their owners (for review [8]). Unfortunately, in this debate the
term dominance is widely used, often without reference to an underlying model or definition.
In our research, we rely on the model devised by van Hooff and Wensing [9], which has
been applied to primates by de Waal [10] and used in many other studies of social living ani-
mals (free-ranging dogs: [11], bonobos: [12], macaque species: [13], wolves: [14], plains zebras:
[15], Icelandic horses: [16], domestic pigs: [17]). The model runs as follows:
Members of a social group may differ in many aspects, including asymmetries regarding
physical power, stamina, personality, weight, weaponry, age, and so on (see for detailed descrip-
tions on criteria:[18]). These differences in personal properties are likely to influence the rela-
tionships between individuals (see also [19]) and may be stable for some time. Stable
relationships between individuals may be correlated with more or less predictable differences in
behaviours and predictable outcomes of conflicts. However, motivation may interfere and this
may lead to some variation in the outcomes of conflicts over resources. It is thus only useful to
speak of a dominance relationship between two individuals when a number of (behavioural)
asymmetries correspond. Thus a number of different behaviours exchanged within each pair of
animals should show corresponding main directions: e.g. individual A shows some relevant
dominance related behaviours more frequently towards individual B than vice versa, and conse-
quently some submission related behaviours consistent with these main directions could be
shown more frequently by B towards A. If this is the case, then dominance may be regarded as a
so called intervening variable that summarizes a set of behavioural differences between individ-
uals [20,21]. If it then turns out that the individual group members can be ranked according to
this intervening variable, the concept of dominance as defined above is applicable.
It should be made clear that animals do not need to have a notion of the concept of domi-
nance in order to establish a dominance relationship and with it, a hierarchy. As computer simu-
lations have shown, rank orders may arise automatically [22], when a few simple rules of giving
or taking precedence are followed. Self-organisation is an underestimated aspect of social organi-
sation in animal species. It arises through repeated encounters among group members, which are
opportunities to gain information on the actual strength of opponents and help to avoid losing
fights in the future [23]; thus learning plays a role in the formation of dominance relationships.
Dominance in Domestic Dogs
PLOS ONE | DOI:10.1371/journal.pone.0133978 August 26, 2015 2/18
Its important here to understand what dominance actually is. The definition of dominance
by Drews [24] is analogous to the original definition by Schelderup-Ebbe [25]: the outcomes of
agonistic dyadic interactions result in consistent winners being dominant and losers being sub-
ordinate. But dominance based on winning conflicts in agonistic contexts is not the only way
to view it. Two more types of dominance, distinguished by primatologist de Waal [10], are
based on either formal dominance or competitive ability. Formal dominance develops via the
exchange of status information through ritualized and/or greeting signals that are independent
of context. Competitive ability considers the motivation of animals to obtain or to possess
resources. In canids, this has been measured using pairwise competition tests over bones or
toys [26]. Competitive orders based on priority of access to food or water, however, are not
necessarily in agreement with agonistic dominance [27] or formal dominance [28,11], although
these are usually correlated.
The ritualized communication patterns that may serve a relative riskless establishment of
dominance relationships (primates: [10], wolves: [29]) are of specific interest in the study on
dominance and hierarchy formation. Such communication signals seem to deescalate conflicts
and reduce the risk of physical injuries or worse [10]. A variety of ritualized signals, mostly
body postures and facial expressions, were found to be shown in only one direction in dyadic
encounters. These are described in primate species to serve as formal signals of dominance or
submission (e.g. bare-teeth display in rhesus monkeys: [28]), and the same has been seen in
wild wolves, captive wolves and free-ranging dogs [3,3036]. This one-directional pattern is
also seen in Van Hooff & Wensing [9], who were the first to study body postures as dominance
indicators in a captive pack of wolves. Their findings showed that in a captive wolf pack, two
postural displays (namely high posture and low posture), accompanied by seven agonistic and
affiliate behaviours, were displayed in the dyads in mainly one direction and were therefore
better indicators of formal dominance than the seven behaviours per se. On the basis of these
two postures, it was possible to construct a rank order that was not only transitive but also lin-
ear (Landaus linearity index h>0.9). In this way, formal dominance was proven and the signif-
icant correspondence between the rank orders constructed on the basis of several formal status
indicators justified the application of the dominance concept in a captive wolf pack.
A different set of indicators was used by Bradshaw and co-workers [37] in their unpublished
qualitative study in which they examined a semi-permanent all-male group of 19 neutered
domestic dogs, searching for a dominance hierarchy based on confident(e.g. growl,inhibited
bite,stand over,stare at,chase,bark at,mount) and submissive(e.g. crouch,avoid,displace-
ment lick/yawn,run away) behaviours. These behaviours were presumed to be useful as rank
indicators but not investigated for their usefulness, as Van Hooff & Wensing had done in their
captive wolf study [9]. Since they found no linear hierarchy, Bradshaw et al. [37] concluded
that dominance does not play a role in the domestic dog and that dogs do not strive for a domi-
nant position. Their study excluded posture as a behavioural variable previously shown to be
an important variable for dominance relationship assessment in wolves [9].
Dominance hierarchies based on aggression and submission were found in two packs of
Indian free-ranging dogs [38], but the properties linearity, directional consistency and coverage
were not assessed. A later recalculation of these properties for these behaviours ([39], p.78),
showed high levels of linearity in the two packs. Linearity is also shown in a study on Italian free-
ranging dogs [11] in which three types of dominance were investigated and compared: the formal
dominance, agonistic dominance and competitive ability [10]. In that study, presumed behav-
iours belonged to one of the three clusters (aggressive, dominance and submissive behaviour) and
these clusters were investigated to determine whether they could be used to fit dogs into a linear
dominance hierarchy. The findings showed that (1) the agonistic-dominance hierarchy was sub-
stantially linear, (2) the submissive-affiliative patterns fulfilled the criteria of formal submission
Dominance in Domestic Dogs
PLOS ONE | DOI:10.1371/journal.pone.0133978 August 26, 2015 3/18
signals (but were not observed among all dog pairs) and (3) the competitive rank order (based on
gaining access to food) was predicted reasonably well by the agonistic rank order.
So far, nearly linear hierarchies constructed from systematic, quantitative data on submissive
(but not agonistic) behavioural measures, appear in one study on captive wolves [9] and in two
studies on free-ranging dogs [11,39] but not in group housed domestic dogs [37]. These contra-
dictory results concerning the role of dominance in canids have led us to investigate this aspect
of social organisation in a group of domestic dogs in more depth. We hypothesize that the dom-
inance model as sketched above is applicable in our group of domestic dogs, just as in captive
wolves [9] and in free-ranging dogs [11,39]. More specifically, we expect that postural commu-
nication and submissive behaviours play a major role in status communication, since these have
been shown to be the best indicators in wolves and free-ranging dogs. Since wolves and dogs are
genetically very close [40,41], despite domestication we expect to find the same social organisa-
tion, meaning a (nearly) linear hierarchy based on formal dominance with stable relationships.
In the present research, we firstly address the question of the usefulness of postural and
behavioural variables as status indicator in domestic dogs. To this end, we strictly distinguished
postures (including ear, tail, and body positions) from all the other behavioural variables (e.g.
bark,growl or pilo-erection). Subsequently, we compared these behaviours and postures with
respect to their qualities as indicators of dominance or submission in dyadic relationships and
characterised the constructed rank orders in terms of linearity. As an additional element, we
investigated the steepness of the one rank order that stood out in terms of linearity, transitivity
and coverage [42,43]. Steepness provides a measure of the strength of the asymmetry between
neighbouring ranked individuals concerning a relevant behavioural measure for instance, the
summed dyadic proportions with which a group member receives submissive acts or wins
dyadic encounters indicates overall individual dominance success. Finally, cluster analysis on a
subset of postures and behaviours was used to reveal clusters possibly indicative of different
aspects of dominance and aggression that would justify the use of the concept of dominance as
an intervening (= summarizing) variable [9,20].
Materials and Methods
Ethics Statement
This study complies with Dutch regulations regarding the ethical treatment of laboratory domestic
dogs. Research permission to conduct the study was granted by the Faculty of Veterinary Medi-
cine and the Faculty of Biology of the Utrecht University. Research protocol aimed at self-regula-
tion of conflicts in the dog group. Serious fights that could possibly inflict wounds were prevented
by human interventions, in some cases even leading to removal of the aggressor from the group.
Study group and housing
In this study, we reanalysed data originally gathered from a newly formed group of domestic
dogs at the dog kennel of the Faculty of Veterinary Medicine of the Utrecht University, the
Netherlands. The group consisted in total of sixteen dogs of different breeds and age: three
adults, two sub adults, seven juveniles (of which four were litter mates) and three pups. Their
individual features such as breed, gender, age and weight are listed in Table 1. All dogs were
sexually intact. Of these 16 dogs, ten (referred to as the core group) were present during the
total observation period of 12 weeks. In week 5, three of the 13 original dogs were removed
from the group: A (subadult female) and F (adult female) were removed due to a severe fight,
and J (adult female) was removed due to advanced pregnancy. In week 10 of the observation
period, three pups (G, R, Y) were introduced into the group. With exception of these pups, the
dogs were housed individually in kennels outside observation hours.
Dominance in Domestic Dogs
PLOS ONE | DOI:10.1371/journal.pone.0133978 August 26, 2015 4/18
The indoor dog kennels were connected by a corridor to an outdoor enclosure of 273 m
(10.5 m. x 26 m.) covered with low undergrowth vegetation, grass and sand. On observation
days (Monday through Friday), the dogs were brought into the outdoor enclosure at dawn (5
6 am). At the end of the observations the dogs were brought back to their own kennels.
The dogs were fed regular dog food pellets twice a day: in the morning by the observers at
three fixed feeding places in the outdoor enclosure and in the afternoon by the dog caretaker in
their own kennels. Some old bones and a white plastic platform (height 35 cm with 1m by
0.5m) were available to play with in the outdoor enclosure. Drinking water was available ad
libitum from a drinking trough.
During the weekends, dogs were regularly taken home by students for socialisation.
Data collection
Preliminary observations during a two months observation period were carried out in order 1)
to describe the ethogram and learn the codes for the behavioural variables, and 2) to practise
the observation and recording methods until more than 90% agreement among the two observ-
ers was reached during live registrations. Data collected during this period were excluded from
analyses. Systematic observations by two observers started two months after the formation of
the group and were conducted from 11 June until 31 August 1984, Mondays through Fridays,
totalling 60 observation days spread over 12 weeks.
Quantitative recording was conducted during a total of 323 hours. Of these, 107 hours of
the last four weeks were used for analyses after relationships were considered to be stabilized:
58 hours focal animal sampling and 49 hours ad libitum sampling [44]. Observers used both
Table 1. The composition of the dog group by name, code, breed, sex, age and weight.
Name Code Breed Sex Age
(months) Weight
Adults Flets F Beagle F 78 10.2
(>18 months) Juultje J Beagle F 44 11.5
Issie I
Cairn terrier F 24 6.3
Subadults Pasha P
Malinois F 12.5 25.3
(9 months to 18 months) Astarte A Great Dane F 9.3 44.9
Juveniles Vlek V
Beagle M 6.7 11.0
(3 months to 9 months) Zwart Z
Beagle F 6.7 11.9
Streep S
Beagle F 6.7 10.2
Kraag K
Beagle F 6.7 9.9
Witband W
Beagle M 5.7 10.6
Tanja T
Doberman Pinscher F 5 16.4
Umpie U
Dalmatian F 4.8 16.1
Bodo B
Labrador Retriever F 3.4 14.5
Pups Geel G German Shepherd F 2.5 10.3
(<3 months) Reu R Dutch Shepherd M 2.2 8.7
Yazzoo Y German Pointer F 2 6.2
= at the start of the introduction in the group
= core group of 10 dogs during the total observation period of 12 weeks
= littermates; offspring of adult female Flets
= son of adult female Juultje
F = female
M = male
Dominance in Domestic Dogs
PLOS ONE | DOI:10.1371/journal.pone.0133978 August 26, 2015 5/18
methods equally often, changing methods between them on a daily basis. The unit of focal ani-
mal time per dog was 10 minutes. Every dog in the pack was sampled twice daily following a
fixed observation order, starting each day with the dog next in the order. The time between two
consecutive observations on the same animal depended on the number of animals in the
group. Ad libitum sampling was focussed on play and agonistic interactions, specifically
between dogs that seldom interacted.
Behaviour variables
As in wolves, different body postures in dogs can be assessed on the basis of specific combina-
tions of postural components: position of the tail and head, the ears and the bending of the
hind legs and the straightness of the back [30,34]. In the studies on wolves [9,14] three distinct
postures were used: high posture (head up and tail upright, ears pricked, straight back and
straight legs), low posture (head low, tail down or tugged between the legs and somewhat bent,
and ears folded backwards) and neutral posture (as an intermediate posture). By adding four
additional postures, we employed a more detailed postural ethogram distinguishing in total 7
postures (Table 2). The postures half-high and half-low were added to capture intermediate
positions. The postures neutral and low were scored, if performed in a dorsal to lateral lying
position, as back and low-on-back. Posturing was scored relative to breed, because breed affects
posture, e.g. a neutral posture for a Beagle looks different from that of a Labrador retriever.
Postures were consistently recorded first and then the accompanying behaviour elements
were recorded. Thus descriptions of postures excluded behaviours, with the exceptions of body
tail wag and pass under head; tail position was also included in the definition to characterize
behaviours accurately. From the original 75 elements in the ethogram, 24 elements were
researched for their suitability as status indicator. Their inclusion was based on their corre-
spondence with behaviours described in the ethograms for captive wolves and free-ranging
dogs, and on the traditional presuppositions in the domestic dog literature that they are related
to dominance or subordination [37,45].
Table 2. Ethogram for tail and ear positions for 7 postures.
Tail Ears
High maximum highest carriage maximally erected (standing) or held
forward (hanging)
Half-high partially highest carriage and held above the
horizontal line of the back
partly erected or hanging forward,
higher than Neutral
Neutral follows line of hind quarter and held around the
horizontal line of the back
held relaxed, partly sideward
Back as in Neutral but in a dorsal or lateral lying
as in Neutral
Half-low lower than Neutral but not held against or
between the hind-legs
partly retracted into the neck, lower than
Low the upper side of tail against hind quarter and s-
shaped, or lower tugged between the hind-legs
maximally retracted into the neck
(standing) or held backwards (hanging)
as in Low but in a dorsal or lateral lying position as in Low
The range of tail carriage (from high to low) differs strongly between breeds and have been taken into
account in assessing the posture.
Dominance in Domestic Dogs
PLOS ONE | DOI:10.1371/journal.pone.0133978 August 26, 2015 6/18
Recording of dyadic interactions
While three types of interactions were recorded (dyadic, triadic and polyadic), to assess the use-
fulness of dominance or subordinate status indicators we analysed only dyadic interactions,
which followed the basic format: ActorPosture and Behaviour(s)RecipientPosture and
For an assessment of status in dyadic relationships, the observers also compared the begin-
ning and ending of interactions, for changes in posture displays. This was done directly after
the dogs ended their interaction, in the following three contexts:
1. Spontaneous: a dog spontaneously lowered its posture into a lower posture as shown previ-
ously, namely one of the low postures (= half-low,low or low-on-back)oron-back, towards
a recipient that showed no signs of aggression.
2. Agonistic context: in reaction to opponents aggressive behaviour (stare,pilo-erection,growl,
show teeth,snap,lunge,bite,muzzle bite,fight,stand over,chase,bark) a dog lowered its pos-
ture into one of the low postures or on-back, or, if the dog was already showing one of the
low postures or on-back, the dog started to signal behaviours indicative of fear, submission
or avoidance (tongue flick,freeze,look away,body tail wag,lick mouth,pass under head,high
pitch vocalisations,flee,retreat,shrink back).
3. Competitive context: a dog lowered its posture into one of the low postures or on-back
towards an opponent that tried to take or took an object or bone, or pushed the dog away
from a feeding place or burying site, or the dog controlling the resource showed a neutral or
higher posture.
A change in posture display as described above by one of the interactors was labelled Lower-
ing of Posture (= LoP).
Statistical analysis
Observational data gathered by the focal animal sampling method were directly typed into a
handheld computer. Ad libitum sampled data were noted on special check sheets and typed
into a computer that same day.
Based on preliminary analysis of dominance and the development of the rank order, we
decided to split the observation period in three periods of 4 weeks. All behavioural variables
were regarded as point events and frequencies were calculated per period of 4 weeks. For analy-
sis of dominance only dyadic interactions of the last 4 weeks were used, because relationships
appeared to be stable during this period.
The extent to which the postures and behaviours are suitable indicators of dominance was
assessed by calculating three properties for each of the seven postures and 24 behaviours: lin-
earity, directional consistency and coverage, using MatMan 1.1 (Noldus technology, Wagenin-
gen, the Netherlands). For comparison with earlier studies on canids, we used the improved
linearity index (h)[46], and determined the rank order most consistent with a linear hierarchy
via the I&SI method [47]. Linearity is considered to be strong with a linearity index of 0.9 or
higher [48]. The directional consistency index is S(H
), where summation is
across all dyads i,j(i>j), and H
is the highest frequency count and L
the lowest frequency
count within dyad i,j[9,12]. It is a measure of the overall within-dyad unidirectionality (0 indi-
cates complete symmetry, 1 indicates complete unidirectionality). Coverage is the proportion
of non-zero dyads in the matrix [9]. All matrices with submission related variables were trans-
posed, i.e. mirrored around the diagonal.
Dominance in Domestic Dogs
PLOS ONE | DOI:10.1371/journal.pone.0133978 August 26, 2015 7/18
Calculation of the steepness of a rank order was based on the Davids score (DS) [49], con-
sidered to be a suitable measure of overall individual dominance success which takes the rela-
tive strength of other individuals into account [43,50]. Subsequently, we converted the DS into
normalized DS (NormDS), based on P
, i.e. the dyadic proportions of received changes in pos-
ture display (based on the exchanged LoP frequencies in dyads). The NormDS values were
plotted against the ordinal ranks of these NormDS values (with equal NormDS values assigned
different ranks) [42]. With the statistical R package steepness[51] ordinary least-squares linear
regression was used to find the best-fitting straight line and the slope of this line was used as a
measure of the steepness of the dominance hierarchy [42]. With a randomization test proce-
dure (20,000 randomizations) the significance of the right-tailed P value was obtained by calcu-
lating the proportion of times that a randomly generated steepness under the null hypothesis is
greater than or equal to the actually observed steepness.
Correlation between the LoP rank order based on NormDS (using P
; for comparison with
cluster analysis using NormDS based on D
see S1 File) and the individual factors age and
weight was tested with Spearman rank correlation.
To identify behavioural variables that co-vary, a cluster analysis with average linkage clus-
tering method and the Pearson product moment correlation as similarity measure was per-
formed on the NormDS based on P
of 2 postures and 11 behaviour variables, that were
selected for having significant hand less than 25% blank relationships [52].
Body postures as suitable status indicator
The 7 postures (N= 8,873 observations) recorded for the 10 core group dogs were examined
for linearity, directional consistency and coverage (Table 3). High posture had strong linearity
(h= 0.93), was highly unidirectional (DCI = 0.90), covered most relationships (84.4%) and
therefore qualified best as status indicator for dominance. Low and low-on-back did not show
significant linearity due to low coverage (53.3% and 37.8% respectively), but were highly unidi-
rectional (DCI>0.95). Combined, these two low postures (N= 186) showed significant linear-
ity (h= 0.71) and very high unidirectionality (DCI = 0.99); consequently the coverage
increased to 66.6%. A low posture, whether shown in a standing (low) or a lying position (low-
on-back), qualified as best formal status indicator for dyadic relationships. Although the pos-
tures half-high,neutral and half-low all showed significant linearity and nearly 100% coverage,
they were not good indicators because they were shown in 75% or more relationships in both
directions (DCI = 0.40, 0.38 and 0.76, respectively).
The assessment of status within dyadic relationships based on LoP display (Table 4)
revealed strong linearity (h= 0.94, p<0.0001), high unidirectionality (DCI = 0.97) and high
coverage (91%). Lowering of posture during an interaction was thus the status indicator best
suited to assessing dominance in this group of domestic dogs.
Behaviours as suitable status indicator
A selection of 24 behaviours (for ethogram see Table 5) were included in our analysis, assum-
ing to reflect sufficiently the packs behavioural repertoire related more or less to dominance
and subordinate behaviour. Table 6 shows the linearities, directional consistencies and cover-
ages of these 24 behaviours.
Body tail wag was the most reliable status indicator of subordination for most dyadic rela-
tionships, with strong linearity (h= 0.82), high unidirectionality (DCI = 0.96) and good cover-
age (75.6%). Neither mouth lick nor pass under head showed significant linearity, due to low
coverage (44.4% and 35.6% respectively), but both were highly unidirectional (DCI 0.97)
Dominance in Domestic Dogs
PLOS ONE | DOI:10.1371/journal.pone.0133978 August 26, 2015 8/18
and therefore both qualified as indicators of subordination for some dyadic relationships.
Combining these two behaviours (comparable to active submission as described by [9,11,31])
did little to increase the values of the properties. Further inspection of matrices revealed that
these two behaviours were almost exclusively shown towards the male W and female P (138
out of 142 and 181 out of 193, respectively).
Five aggressive behaviours (stare,pilo-erection,growl,show teeth,snap) were highly bidirec-
tional (0.50 DCI 0.74), but did show significant linearity (p 0.04). Interestingly, with
increasing intensity of aggression, unidirectionality increased, with the more intense aggressive
behaviour lunge showing the highest unidirectionality (DCI = 0.94). Inspection of this matrix
revealed that lunge was almost exclusively shown towards the two females T and U (106 of
124) by lower ranking dogs showing lower postures than their receivers, indicating this type of
aggression is likely to be protest or fear motivated.
The most intense aggressive behaviours bite and fight had low frequencies (28 and 18,
respectively) indicating low levels of extreme aggression (wounds inflicted) in this group after
exclusion of the one pair that fought. Because of low unidirectionalities (DCI = 0.43 and 0.11,
respectively), these two behaviours are not suitable as status indicators. The most intense fear
behaviour flee was highly unidirectional (DCI = 0.98), but had very low frequency (N= 19).
Stand over and muzzle bite, both assumed to be related to dominance, revealed no significant
linearity (h= 0.32 and h= 0.33, respectively) due to low coverage (20% and 22.2%, respectively).
Besides Stand over, with the highest unidirectionality (DCI = 1), had very low frequency (N=18),
Table 3. Properties of 7 postures: frequency (N), improved linearity index (h), directional consistency index (DCI), coverage (Unknown), unidirec-
tionality (1-Way), bidirectionality (2-Way) and number of ties (Tied), over the last 4 weeks of observations.
High 249 0.93*(p = 0.0001) 0.90 7 (15.6%) 28 (62.2%) 10 (22.2%) 2 (4.4%)
Half-high 3 904 0.85(p = 0.0002) 0.40 0 (0%) 0 (0%) 45 (100%) 0 (0%)
Neutral 2 814 0.63(p = 0.006) 0.38 2 (4.4%) 9 (20.0%) 34 (75.6%) 4 (8.9%)
Half-low 1 616 0.95(p = 0.0001) 0.76 0 (0%) 9 (20.0%) 36 (80.0%) 0 (0%)
Low 132 0.61(p = 0.03) 121 (46.7%) 24 (53.3%) 0 (0%) 0 (0%)
Low-on-back 54 0.38(p = 0.3) 0.96 28 (62.2%) 16 (35.6%) 1 (2.2%) 0 (0%)
On-back 104 0.38(p = 0.3) 0.79 20 (44.4%) 17 (37.8%) 8 (17.8%) 1 (2.2%)
Improved Linearity index [46]
Directional consistency index [9]
Number and percentage of unknown relationships
Number and percentage of one-way relationships
Number and percentage of two-way relationships
Number and percentage of tied relationships
*Indexes 0.9 are in bold.
Table 4. Properties of Lowering of Posture (LoP): frequency (N), improved Landaus linearity (hindex), direction consistency (DCI), coverage
(Unknown), unidirectionality (1-Way), bidirectionality (2-Way) and number of ties (Tied), over the last 4 weeks of observations.
LoP 552 0.94 *(p = 0.0001) 0.97 4 (8.9%) 34 (75.6%) 7 (15.6%) 3 (6.7%)
: see legend below Table 3)
*Indexes of 0.9 are in bold.
Dominance in Domestic Dogs
PLOS ONE | DOI:10.1371/journal.pone.0133978 August 26, 2015 9/18
and therefore was disqualified as a status indicator. By showing very high unidirectionality
(DCI = 0.98) and sufficient frequency (N=101)muzzle bite qualified as a useful status indicator
of dominance for some relationships. Inspection of this matrix revealed that muzzle bite was
exclusively shown by the three highest ranking dogs, but mostly by female P (87 of 101).
The behaviours tongue flick,look-away and freeze showed significant and high linearity
(h>0.80), but due to insufficiently strong unidirectionality (DCI <0.80) were not useful as
status indicators. Miscellaneous behaviours such as take away object,bark,paw on,approach
and tail wag were not useful as status indicators due to insufficient coverage and/or low
Table 5. Ethogram for 24 behaviours in dogs (adapted from Zimen [34] and van Hooff and Wensing
Mouth lick Licking repeatedly with fast movements directed to the recipients mouth corners
Body tail wag Accelerated, irregular movement of the tail, often also the hindquarter is moving, in a
neutral or lower posture (posture is included to distinguish from normal tail wag, see
Pass under head Passing from the lateral side closely underneath the head of the recipient, often short
nose-chin contact with the recipient, in a neutral or lower posture
Stare Intense xating look towards recipient with tensed body, for a minimal duration of 2
Pilo-erection Raising the hair on one or more upper parts of the body (neck, shoulder, hindquarter)
and/or tail base
Growl Low-pitched rumbling, fairly monosyllabic vocalization from the dogs throat
Show teeth Baring of the teeth, which become partly or totally visible.
Snap Attempt to bite while moving not more than 1 or 2 steps (about ½meter) in the
direction of the recipient, without physical contact
Lunge Attempt to bite while moving over a distance between ½to 3 meters in the direction of
the recipient, without physical contact
Bite Taking any part of the recipients body between the jaws with sufcient pressure that
could cause harm to the recipient
Fight Severe, offensive aggressive interaction between two dogs, including aggressive
elements like lunge and bite
Shrink back Accelerated movement directed away from the recipient over a distance up to 1 meter
Retreat Accelerated movement directed away from the recipient over a distance from 1 to 3
Flee Running away from the recipient over a distance of 3 meters or more, with head in
opposite direction of the recipient
Stand over Standing over the recipients body, with four paws on the ground, in a neutral or higher
Muzzle bite Inhibited biting over the recipients snout from above or from the side
Tongue ick Showing one or more brief licking movements with tongue directed towards nose and
head oriented towards recipient, without physical contact
Look away Turning only the head away from the recipient, while staying on the same spot
Freeze General rigidity of the body, with exception of the tail, and no staring towards the
Approach In normal pace walking (not accelerated) towards the recipient up to a distance of 1
meter or less
Take away object Taking away object or bone that is in possession of the recipient
Bark Loud and repetitive barking (characteristic for dogs) directed towards the recipient
Tail wag Non accelerated, regular sideward movements of the tail, about in one plane
Paw on Placing one or both front paws on the recipients head or back
Dominance in Domestic Dogs
PLOS ONE | DOI:10.1371/journal.pone.0133978 August 26, 2015 10 / 18
Steepness of LoP rank order
As LoP qualified best for ordering the dogs into a formal linear hierarchy, we assessed the
steepness of this rank order (Fig 1) based on the normalized DS values of the LoP matrix (S1
Appendix). The steepness of the observed matrix is 0.79, which differs significantly (right-
tailed P= 0.0001) from the steepness value 0.32 expected under the null hypothesis.
The hierarchy in this group of dogs turned out to consist of three layers: first the highest
ranking male W and two females P, T with NormDS values between 8 and 7, then a group of
four middle ranking dogs V, I, U, B with NormDS values between 5 and 4, and finally a group
of three lowest ranking dogs K, Z, S with NormDS values between 2 and 1. Most dyadic rela-
tionships in the lowest ranking subgroup (females K, Z, S; all from the same litter) remained
unresolved since we hardly observed LoP,high posture or body tail wag, although they
approached each other often (except dogs Z and S) (see matrices in the S1 Appendix).
Correlation between rank and individual factors
No significant correlation was found between rank order (based on the NormDS of LoP) and
weight (Spearman rank correlation: r
= 0.46, N= 10, P= 0.09) or age (Spearman rank
Table 6. Properties of 24 behavioural elements: frequencies (N), improved Landaus linearity (hindex), direction consistency (DCI), coverage
(Unknown), unidirectionality (1-Way), bidirectionality (2-Way) and number of ties (Tied), over the last 4 weeks of observations.
Nh' index
Mouth lick 193 0.4(p = 0.18) 0.97*25 (55.6%) 19 (42.2%) 1 (2.2%) 0 (0%)
Body tail wag 316 0.82(p = 0.0003) 0.96 11 (24.4%) 30 (64.4%) 5 (11.1%) 1 (2.2%)
Pass under head 142 0.45(p = 0.16) 129 (64.4%) 16 (35.6%) 0 (0%) 0 (0%)
Stare 996 0.85(p = 0.0002) 0.50 1 (2.2%) 7 (15.6%) 37 (82.2%) 1 (2.2%)
Pilo-erection 295 0.66(p = 0.009) 0.68 10 (22.2%) 15 (33.3%) 20 (44.4%) 1 (2.2%)
Growl 608 0.52(p = 0.04) 0.70 1 (2.2%) 17 (37.8%) 27 (60%) 3 (6.7%)
Show teeth 337 0.55(p = 0.03) 0.74 9 (20.0%) 22 (48.9%) 14 (31.1%) 2 (4.4%)
Snap 231 0.57(p = 0.03) 0.74 10 (22.2%) 22 (48.9%) 13 (28.9%) 1 (2.2%)
Lunge 124 0.59(p = 0.026) 0.94 18 (40.0%) 24 (53.3%) 3 (6.7%) 1 (2.2%)
Bite 28 0.29(p = 0.49) 0,43 33 (73.3%) 7 (15.6%) 5 (11,1%) 3 (6.7%)
Fight 18 0.24(p = 0.60) 0.11 38 (84.4%) 2 (4.4%) 5 (11.1%) 5 (11.1%)
Shrink back 155 0.63(p = 0.01) 0.66 10 (22.2%) 18 (40.0%) 17 (37.8%) 4 (8.9%)
Retreat 116 0.50(p = 0.07) 0.76 12 (26.7%) 22 (48.9%) 11 (24.4%) 6 (13.3%)
Flee 19 0.38(p = 0.27) 0.98 30 (66.7%) 14 (31.1%) 1 (2.2%) 1 (2.2%)
Stand over 18 0.32(p = 0.41) 136 (80.0%) 9 (20.0%) 0 (0%) 0 (0%)
Muzzle bite 101 0.33(p = 0.40) 0.98 35 (77.8%) 9 (20.0%) 1 (2.2%) 0 (0%)
Tongue ick 557 0.83(p = 0.0001) 0.72 1 (2.2%) 17 (37.8%) 27 (60.0%) 4 (8.9%)
Look away 989 0.92(p = 0,0001) 0.78 0 (0%) 12 (26.7%) 33 (73.3%) 1 (2.2%)
Freeze 1 198 0.94(p = 0.0001) 0.78 0 (0%) 14 (31.1%) 31 (68.9%) 2 (4.4%)
Approach 1 744 0.47(p = 0.05) 0.35 0 (0%) 0 (0%) 45 (100%) 1 (2.2%)
Take away object 123 0.53(p = 0.05) 0.61 18 (40.0%) 18 (40.0%) 9 (20.0%) 1 (2.2%)
Bark 361 0.43(p = 0.14) 0.49 0 (24.4%) 18 (40.0%) 16 (35.6%) 4 (8.9%)
Tail wag 507 0.86(p = 0.0001) 0.55 0 (0%) 11 (24.4%) 34 (75.6%) 4 (8.9%)
Paw on 292 0.34(p = 0.29) 0.53 13 (28.9%) 12 (26.7%) 20 (44.4%) 6 (13.3%)
: see legend below Table 3)
*Indexes 0.9 are in bold
Dominance in Domestic Dogs
PLOS ONE | DOI:10.1371/journal.pone.0133978 August 26, 2015 11 / 18
correlation: r
= 0.14, N= 10, P= 0.35). With regard to sex and rank, no statistical conclusion
could be drawn, because only two males were in the group.
Co-variation between behaviour variables
To assess the usefulness of dominance as an intervening variable, we analyzed the co-variation
between the variables that met the criteria of 25% coverage and significant linearity: postural
variables LoP,high posture, and 11 behaviours. The cluster analysis of the Pearson correlation
matrix of these variables revealed three clearly distinct clusters that strongly co-varied and
some loosely linked behaviours (S1 Fig).
In the first cluster, the behaviours freeze,tongue flick,look away and show teeth strongly cor-
responded. These behaviours indicated either conflict (tongue flick and look away) or agonism
(freeze and showing teeth). We labelled this group Agonistic-conflict behaviours. The second
cluster included the behaviours body tail wag and high posture, as well as lowering of posture
performed in the three different contexts (matrices shown in S1 Appendix) that qualified as
best status indicators. The NormDS values of these postures and behaviour correlated strongly:
LoP with high posture (r= 0.84, N= 10, P<0.001) and with body tail wag (r= 0.92, N= 10,
P<0.001), and body tail wag with high posture (r= 0.84, N= 10, P<0.001). Following van
Hooff and Wensing [9] we labelled this group dominance-subordination status signals. The
three aggression related behaviours, growl,snap and pilo-erection, clearly clustered apart from
the other two clusters. Finally, tail wag loosely linked to this Aggression cluster, while stare and
shrink back were loosely linked to the two aforementioned clusters.
Fig 1. Steepness of rank order. The normalized Davids scores (NormDS based on P
) for the LoP matrix
plotted against the rank of 10 dogs, ranked from dog W (highest NormDS = rank 1) to dog S (lowest
NormDS = rank 10). The steepness of this rank order (i.e. the absolute value of the slope of the fitted line) is
Dominance in Domestic Dogs
PLOS ONE | DOI:10.1371/journal.pone.0133978 August 26, 2015 12 / 18
This study in a group of domestic dogs is the first that quantitatively researched postural and
behavioural measures in dyads for their suitability as formal status indicator. We found one
posture (high posture) and one behaviour (muzzle bite) suitable as formal signals for expressing
dominance. Furthermore, two low postures (low and low-on-back) and three behaviours (body
tail wag,mouth lick and pass under head) were found to be formal status signals of submission.
However, only high posture and body tail wag covered most relationships. The assessment
based on the change in postural display (lowering of posture, LoP) was the best formal indicator
for expressing submission. The high correlation between the rank orders for LoP,body tail wag
and high posture justifies the use of the term dominance as the intervening variable in domestic
dogs. Moreover, for the first time, steepness [42] has proven its usefulness when describing the
social organisation of a group of domestic dogs in terms of dominance style [10].
Status indicators
Postural display is an important feature in intraspecific relationships in wolves. As predicted,
we found the postural display high posture to be suitable in our dog group as a formal status
signal for dominance and the two low postures low and low-on-back to signal submission. This
is in line with the quantitative findings for captive wolves [9], in whom both dominant and
subordinate roles are signaled by postural displays that might be considered a form of meta-
communication [9]. Moreover, we found our determination of dominance relationships was
reliable for most group members when we assessed the change of postural display during
dyadic interactions, i.e. lowering of posture (LoP): changing to a lower posture (e.g. half-low,
low,low-on-back,on-back) during an interaction. LoP was the best variable for determining lin-
ear rank order (linearity h= 0.94): lowering of posture can be seen as context-independent, rit-
ualized postural signaling expressing a subordinate status.
In addition to postural displays, we also found certain behaviours to be suitable as status
indicators. The strongest of these appears to be body tail wag, representing a ritualized signal-
ing of acceptance of subordinate status and at the same time communicating friendly inten-
tions. This body tail wag, along with mouth lick and pass under head, included in Schenkels
description of active submission [31], stand out in our study group as formal status indicators
of subordination. Mouth lick and pass under head did not occur in all relationships in our
study, but were almost exclusively received by the highest ranking male and female. This find-
ing strongly corresponds to what has been observed in free-ranging dogs, where mouth lick
and tail wagging with low tail(comparable to body tail wag) were found to be formal submis-
sion signals not covering all relationships [11]; the same was also seen in the study on 24
domestic dogs in a daycare center, where muzzle licking behaviour was found to meet most
dominance criteria but not to cover all relationships [53]. We stress that analyzing separately
the constituent elements of active submission (mouth lick,body tail wag and pass under head)
enabled us to reveal differentiation regarding the display of these status indicators in the social
organisation of domestic dogs.
In descriptive wolf studies, the behaviour muzzle bite is claimed to be useful as an indicator
of dominance shown by the breeding pair or parents forcing offspring to the ground [31,54],
although a quantitative analysis of this behaviour in a captive wolf group did not reveal this
behaviour to be a reliable status indicator [9]. Commonly it is assumed that dogs lost this dom-
inance assertion behaviour during domestication [34,55]. However, in our mix breed dog
group, muzzle bite behaviour was shown most frequently by the highest ranking female P
towards the middle ranking female dogs T, U and B. Also the highest ranking male W dis-
played this specific behaviour towards the middle ranking females (U and B). Our analysis
Dominance in Domestic Dogs
PLOS ONE | DOI:10.1371/journal.pone.0133978 August 26, 2015 13 / 18
showed muzzle bite to be a formal signal of dominance (cf [56]), and moreover, we may even
categorize this behaviour as an exclusive status signaller, as it was only employed by the highest
ranking male and female towards lower ranking dogs and has also been described in family
groups of wild wolves [54].
Most wolf researchers indicated the tail posture to be most indicative for assessing domi-
nance (e.g. [30,31,35,57]), with the first quantitative evaluation of the tail posture in wolves
done by Van Hooff and Wensing [9]. This metacommunication by means of postural displays
occurs in all contexts. As every behaviour is accompanied by a posture, postures add an extra
signal to the behaviour corresponding the relative status of the individuals involved. Therefore,
as an example, to identify dominance related forms of aggression from other forms of aggres-
sion, the accompanying posture (tail position) should definitely be taken into account for reli-
able further interpretation of the performed behaviour.
None of the other behaviours investigated in the present study were found to be useful as
status indicators, since they were highly bidirectional or did not have sufficient coverage or did
not show significant linearity. Although aggressive behaviours like stare,growl,show teeth and
bite are labelled as dominant behaviour by some authors [37,38], in the current study none of
these behaviours met the criteria for suitable rank indicator, especially because of insufficient
unidirectionality. This compares to the outcomes of the more detailed analyses of aggressive
wolfwolf interactions, where no link between bared teeth and rank position could be revealed
unless the retraction of the lips was taken into account [58]. Moreover, in our study aggressive
behaviours have been observed to occur simultaneously with lowered postures, which contra-
dicts their interpretation as dominant behaviours.
Steepness and dominance style
Alongside linearity, as a quantitative measure to characterize a dominance hierarchy, we used
steepness [42]. A steep rank order indicates large asymmetries between individuals. This in
turn is a characteristic of a despotic, less tolerable society. In a more tolerable society, smaller
asymmetries will be found between individuals, and consequently a less steep rank order
should be found. In macaque species, steepness has been used, along with asymmetry, intensity
of aggression and the use of status signals, to distinguish four types of dominance styles [13]:
(1) despotic, (2) tolerant, (3) relaxed, and (4) egalitarian. Using steepness as an indicator and
following these labels, the dominance style in our group can be best indicated as tolerantfor
the following reasons: (a) the degree of steepness itself (0.79 when based on Pij; 0.67 when
based on Dij) (0.79 when based on Pij; 0.67 when based on Pij)) fits within the variation of
steepness found for styles 2 and 3 in macaque species [59]; (b) large dyadic asymmetries in pos-
tural displays (specifically LoP and high posture) were reinforced with mild to moderate aggres-
sion (e.g. stare,pilo-erection,growl,show teeth,snap); (c) both dominant and subordinate dogs
signaled their roles using formal status signals; (d) many relationships were formalized, while
some relationships (especially those in the middle ranking group and in the lowest ranking
group) were unresolved, but they were not avoiding each other (see approach matrix in the S1
Appendix) and (e) aggressive behaviours of low intensity had high bidirectional exchange. The
steepness of the hierarchy, research on macaque species suggests, covaries with dominance or
social style [60]. High steepness indicates strong asymmetries between the individuals and
large absolute rank distances between high and low ranking individuals. Investigation of the
relations between steepness and dyadic aggression characteristics such as bidirectional aggres-
sion, counter-aggression, aggression intensity and conflict resolution efficacy or reconciliation,
therefore, should be the focus of future research in dog groups.
Dominance in Domestic Dogs
PLOS ONE | DOI:10.1371/journal.pone.0133978 August 26, 2015 14 / 18
Concept of dominance as intervening variable
To assess the usefulness of the concept of dominance as intervening variable, we determined
the agreement between distributions of different postural and behavioural variables. The vari-
ables lowering of posture,high posture and body tail wag, all status indicators at group level,
were found to show a high covariation. This finding underlines existing asymmetries between
individuals and shows dominance to be a useful intervening variable in explaining a social
organisation. Also, this finding corresponds closely with wolf behaviours (low,active submis-
sion and high) in the dominance-subordination domain as identified in the Burgers zoo wolves
([9], p 236). Aggressive behaviours conventionally associated with the concept of dominance,
like growl,show teeth and snap, clearly stand apart from the dominance and subordination sta-
tus signals in our dog group as discussed above. This confirms the findings in the Burgers zoo
wolf study, that threat/assault clearly stands apart from the dominance-subordination domain.
In conclusion, we showed that the concept of dominance, as shown in captive wolves and free-
ranging dogs, is applicable in our group of dogs and is reflected foremost by postural displays,
and not by aggressive displays.
This systematic study on the applicability of the concept of dominance revealed formal domi-
nance to be applicable to domestic dogs. As a direct application of our findings, we could char-
acterize the dominance style in this study group of dogs as tolerant. These findings are useful
in correctly interpreting the relationship between dogs at the dyadic level regarding status as
well as for the position in the group (rank order). The method described herein could be
applied to determine breed-related differences, and moreover these findings could be helpful in
correctly diagnosing the status in dog-dog or dog-human relationships in case of behavioural
problems, based on formal signals and not aggression.
The question can be raised whether the tolerant dominance style found in our group can be
generalized to other dog groups. In macaque species, for example, not only are there variations
in dominance style within (female vs males) and between groups of the same species, but also
between species [60]. As dog breeds vary highly in morphology, aggressive tendencies and tem-
perament or personality [55,61,62,63] we expect to see variation in dominance style that will be
highly dependent on the breed composition of the group. Bold breeds (e.g. Rottweilers, Mali-
nois) might show a more despotic dominance style, whereas, shy breeds (e.g. Cavalier King
Charles Spaniels, Labrador retrievers) might exhibit a more relaxed or even an egalitarian dom-
inance style. In mixed breed groups the dominance style of the highest ranking male / female
would reasonably be expected to heavily influence the steepness of the hierarchy. In view of
this potentially great variation regarding behavioural tendencies both between breeds and
within breeds [64], it is to be expected that all types of dominance styles will be found in dogs.
Thus while the characterization of our study group cannot be generalized to other dog groups,
the method used can be, and future research should reveal the extent to which different domi-
nance styles are present in group housed domestic dogs of different breeds and breed types.
Supporting Information
S1 Appendix. Three dyadic dominance related matrices ordered by the normalized Davids
score; (a-c): LoP,high posture and body tail wag. NDS = Normalized Davids score based on P
The individuals in the fourth matrix (d), approach, are ordered in LoP rank order, for compari-
son purposes.
Dominance in Domestic Dogs
PLOS ONE | DOI:10.1371/journal.pone.0133978 August 26, 2015 15 / 18
S1 Fig. Cluster analysis (average linkage method) of Lowering of posture and High posture
and 11 behaviours. The Pearson correlation between NormDS values are used as similarity
measure. The correlations are rescaled to a distance measure which varies between 0 and 25,
such that the ratios of these distances are identical to the ratios of the original correlations (or,
to the ratios of the average correlations between pairs of behaviours in different clusters).
S1 File. Comparison cluster analysis using NormDS based on Pij with cluster analysis
based on Dij.
Special thanks are due to Jurriaan F Slegers for his help in gathering and analysing the data.
We also thank emeritus professor JARAM van Hooff for his support to publish this study and
we are grateful to Linda McPhee for language editing the manuscript.
Author Contributions
Conceived and designed the experiments: JvdB. Performed the experiments: JvdB. Analyzed
the data: JvdB HdV MS CV. Wrote the paper: JvdB MS CV HdV.
1. MacFarland D. Animal Behaviour: psychobiology, ethology and evolution. Pitman, University of Cali-
fornia; 1985.
2. Creel S, MacDonald D. Sociality, group size and reproductive suppression among carnivores. Adv
Study Behav 1995; 24:203257.
3. Feddersen-Petersen DU. Hundepsychologie. Sozialverhalten und Wesen, Emotionen und Individuali-
tät. Kosmos, Stuttgart, Germany; 2004.
4. Mech LD, Boitani L. Wolf social ecology. In: Mech LD, Boitani L, editors. Wolves. Behavior, Ecology
and Conservation. University Chicago Press, USA; 2003. p. 134.
5. Packard JM. Wolf behavior: Reproductive, Social, and Intelligent. In: Mech LD, Boitani L, editors.
Wolves. Behavior, Ecology and Conservation. University Chicago Press, USA; 2003. p. 3565.
6. Sapolsky R. Social Status and Health in Humans and Other Animals. Annu Rev Anthropol 2004;
7. Mech LD. Alpha status, dominance, and division of labor in wolf packs. Can J Zool 1999; 77:1196
8. Schilder MBH, Vinke CM, van der Borg JAM. Dominance in domestic dogs revisited: Useful habit and
useful construct? J Vet Behav 2014; 9:184191.
9. van Hooff JARAM, Wensing JAB (1987) Dominance and its behavioral measures in a captive wolf
pack. In: Frank H, editor. Man and Wolf. Dr. W. Junk Publishers, Dordrecht, the Netherlands; 1987. p.
10. de Waal FBM. Dominance style and primate social organisation. In: Standen V, Foley RA, editors.
Comparative socioecology: the behavioural ecology of humans and other animals. Oxford: Blackwell
Scientific Publications; 1989. p. 243263.
11. Cafazzo S, Valsecchi P, Bonanni R, Natoli E. Dominance in relation to age, sex, and competitive con-
texts in a group of free-ranging domestic dogs. Behav Ecol 2010; 21(3):443455.
12. Vervaecke H, de Vries H, van Elsacker L. Dominance and its behavioral measures in a captive group of
bonobos (Pan paniscus). Int J Primatol 2000; 21:4768.
13. Flack JC, de Waal FBM. Dominance style, social power, and conflict management in macaque socie-
ties: A conceptual framework. In: Thierry B, Singh M, Kaumanns W, editors. Macaque Societies: a
model for the study of social organization. Cambridge University Press; 2004. p. 157185.
14. Derix R. The Social Organisation of Wolves and African wild dogs: an empirical and model-theoretical
approach. Dissertation. University of Utrecht; 1994.
Dominance in Domestic Dogs
PLOS ONE | DOI:10.1371/journal.pone.0133978 August 26, 2015 16 / 18
15. Schilder MBH. Dominance relationships between adult plains zebra stallions in semi-captivity. Behav-
iour 1988; 104:300319.
16. van Dierendonck MC, de Vries H, Schilder MBH. An analysis of dominance, its behavioural parame-
ters and possible determinants in a herd of Icelandic horses in captivity. Neth J Zool 1995; 45(3
17. Langbein J, Puppe B. Analysing dominance relationships by sociometric methodsa plea for a more
standardised and precise approach in farm animals. Appl Anim Behav Sci 2004; 87:293315.
18. Maynard Smith J, Parker GA. The logic of asymmetrical contests. Anim Behav 1976; 24:159175.
19. Bernstein IS. Dominance; the baby and the bathwater. Behav Brain Sci 1981; 4:419457.
20. Hinde RA. Dominance and roletwo concepts with dual meanings. J Social Biol Struct 1978; 1:2738.
21. Rowell TE. The Concept of Social Dominance. Behav Biol 1974; 11:131154. PMID: 4367951
22. Hemelrijk CK. Self organization and natural selection in the evolution of complex despotic societies.
Biol Bull 2002; 202:283288. PMID: 12087001
23. Chase ID, Seitz K. Self-structuring properties of dominance hierarchies: a new perspective. Adv Genet
2011; 75:5281.
24. Drews C. The concept and definition of dominance in animal behaviour. Behaviour 1993; 125:283311.
25. Schjelderup-Ebbe T. Beiträge zur Sozialpsychologie des Haushuhns. Z Psychol 1922; 88:225252.
26. Scott JP, Fuller JL. Genetics and the Social Behavior of the Dog. Chicago: The University of Chicago
Press; 1965.
27. Syme GJ. Competitive orders as measures of social dominance. Anim Behav 1974; 22:931940.
28. de Waal FBM, Luttrell LM. The formal hierarchy of rhesus macaques: An investigation of the bared-
teeth display. Am J Primatol 1985; 9:7385.
29. Harrington FH, Asa CS (2003) Wolf communication. In: Mech LD, Boitani L, editors. Wolves. Behavior,
Ecology and Conservation. University Chicago Press, USA; 2003. p. 66103.
30. Schenkel R. Ausdruckstudien an Wölfen. Behaviour 1947; 1:81129.
31. Schenkel R. Submission, its features and function in the wolf and the dog. Am Zool 1967; 7:319329.
32. Fox MW. The anatomy of aggression and its ritualization in canidae: a developmental and comparative
study. Behaviour 1969; 35(34):242258.
33. Fox MW. A comparative study of the development of facial expressions in canids; wolf, coyote and
foxes. Behaviour 1970; 36(12):473.
34. Zimen E. Wölfe und Königspudel, Vergeleichende Verhaltensbeobachtungen. Dissertation. Institut für
Haustierkunde, Christian-Albrechts-Universität, Kiel Munich, Germany; 1971.
35. Zimen E. Social dynamics of the wolf pack. In: Fox MW, editor. The wild canids: their systematics,
behavioral ecology and evolution. Van Nostrand Reinhold Co, New York, US. 1975. p. 336368.
36. Spotte S. Societies of wolves and free-ranging dogs. Cambridge: Cambridge University Press; 2012.
37. Bradshaw JWS, Blackwell EJ, Casey RA. Dominance in domestic dogsuseful construct or bad
habit? J Vet Behav 2009; 4(3):135144.
38. Pal SK, Ghosh B, Roy S. Agonistic behaviour of free-ranging dogs (Canis familiaris) in relation to sea-
son, sex and age. Appl Anim Behav Sci 1998; 59:331348.
39. Bonanni R, Cafazzo S. The Social Organisation of a Population of Free-Ranging Dogs in s Suburban
Area of Rome: A Reassessment of the Effects of Domestication on DogsBehaviour. In: Kaminski J,
Sarah Marshall-Pescini (editors). The Social Dog. Behaviour and Cognition. Academic Press, Elsevier,
Amsterdam; 2014. p. 65104.
40. vonHoldt BM, Pollinger JP, Lohmueller KE, Han E, Parker HG, Quignon P, et al. Genome-wide SNP
and haplotype analyses reveal a rich history underlying dog domestication. Nature 2010; 464
(7290):898902. doi: 10.1038/nature08837 PMID: 20237475
41. Lindblad-Toh K, Wade CM, Mikkelsen TS, Karlsson EK, Jaffe DB, Kamal M, et al. Genome sequence,
comparative analysis and haplotype structure of the domestic dog. Nature 2005; 438(7069):803819.
PMID: 16341006
42. de Vries H, Stevens JMG, Vervaecke H. Measuring and testing the steepness of dominance hierar-
chies. Anim Behav 2006; 71:585592.
43. Gammell MP, de Vries H, Jennings DJ, Carlin CM, Hayden TJ. Davids score: a more appropriate domi-
nance ranking method than Clutton-Brock et als index. Anim Behav 2003; 66:601605.
44. Altmann J. Observational study of behaviour. Sampling methods. Behaviour 1974; 49:227265. PMID:
Dominance in Domestic Dogs
PLOS ONE | DOI:10.1371/journal.pone.0133978 August 26, 2015 17 / 18
45. De Palma C, Viggiano E, Barillari E, Palme R, Dufour AB, Fantini C et al. Evaluating the temperament
in shelter dogs. Behav 2005; 142:13131334.
46. de Vries H. An improved test of linearity in dominance hierarchies containing unknown or tied relation-
ships. Anim Behav 1995; 50:13751389.
47. de Vries H. Finding a dominance order most consistent with a linear hierarchy: a new procedure and
review. Anim Behav 1998; 55:827843. PMID: 9632471
48. Martin P, Bateson PPG. Measuring behaviour: an introductory guide. Cambridge: Cambridge Univer-
sity Press; 1993.
49. David HA. Ranking from unbalanced paired-comparison data. Biometrika 1987; 74(2):432436.
50. Balasubramaniam KN, Berman CM, De Marco A, Dittmar K, Majolo B, Ogawa H et al. Consistency of
dominance rank order: A comparison of Davids scores with I&SI and Bayesian methods in macaques.
Am J of Primatol 2013; 75(9): 959971.
51. Leiva D, de Vries H. steepness: An R package for measuring and testing steepness in dominance hier-
archies. The R project forstatistical computing, version 2.13.2; 2012. Available:
52. Klass K, Cords M. Effect of unknown relationships on linearity, steepness and rank ordering of domi-
nance hierarchies: Simulation studies based on data from wild monkeys. Behav Processes 2011;
88:168176. doi: 10.1016/j.beproc.2011.09.003 PMID: 21945822
53. Trisko RK. Dominance, Egalitarianism and Friendship at a Dog Daycare Facility. Dissertation. Univer-
sity of Michigan; 2011. Available:
54. Mech LD, Cluff HD. Prolonged intensive dominance behavior between grey wolves, Canis lupus. Can
Field Nat 2010; 124: 215218.
55. Goodwin D, Bradshaw JWS, Wickens SM. Paedomorphosis affects agonistic visual signals of domestic
dogs. Anim Behav 1997; 53:297304.
56. Bauer EB, Smuts BB. Cooperation and competition during dyadic play in domestic dogs, Canis famil-
iaris. Anim Behav 2007; 73:489499.
57. Zimen E. On the regulation of Pack Size in Wolves. Z. Tierpsychol 1976; 40:300341.
58. Fatjó J, Feddersen-Petersen D, Ruiz de la Torre JL, Amat M, Mets M, Braus B et al. Ambivalent signals
during agonistic interactions in a captive wolf pack. Appl Anim Behav Sci 2007; 105:274283.
59. Balasubramaniam KN, Dittmar K, Berman CM, Butovskaya M, Cooper MA, Majolo B et al. Hierarchical
steepness, Counter-aggression, and Macaque Social Style Scale. Am J of Primatol 2012; 74(10):915
60. Thierry B. Unity in diversity: lessons from Macaque societies. Evol Anthropol 2007; 16:224238.
61. Svartberg K, Tapper I, Temrin H, Radestater T, Thorman S. Consistency of personality traits in dogs.
Anim Behav 2005; 69:283291.
62. Duffy DL, Hsu Y, Serpell JA. Breed differences in canine aggression. Appl Anim Behav Sci 2008;
63. Feddersen-Petersen DU. Social behaviour of dogs and related canids. In: Jensen P, editor, The beha-
vioural biology of dogs. CAB International, Wallingford, UK; 2007. p. 105119.
64. Mehrkam LR,Wynne CDL. Behavioral differences among breeds of domestic dogs (Canis lupus famil-
iaris): Current status of the science. Appl Anim Behav Sci 2014. Available: doi: 10.1016/j.applanim.
2014.03.005 PMID: 24904190
Dominance in Domestic Dogs
PLOS ONE | DOI:10.1371/journal.pone.0133978 August 26, 2015 18 / 18
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
This paper updates and extends Dewsbury's (1982) review of the literature on dominance and reproductive success (RS). The findings from approximately 700 studies are included, over two thirds of which were unavailable to Dewsbury. In order to give a highly condensed and yet meaningful overview, the main findings are represented in four tables, one for male nonprimates, one for female nonprimates, one for male primates, and one for female primates. In the tables for males, findings are analyzed in terms of six different indicators of RS, and in the tables for females, in terms of eight RS indicators.Outside the primate order, evidence largely supported the hypothesis that high-ranking males enjoy greater RS than do subordinate males. For females, studies are more evenly divided between those supporting the hypothesis that high rank and RS are positively correlated and those indicating no significant rank-RS relationship. This may reflect both the lower saliency of hierarchical relationships among females, as well as the lower variability in RS among females, relative to males.Among primates, a complex picture has emerged, especially in the case of males. Much of the complexity appears due to the importance of age and seniority in affecting dominance rank. Also, in some primate species, female preferences for sex partners seem to have little to do with the male's dominance rank, at least at the time mating takes place. Nevertheless, the majority of studies suggest that high- to middle-ranking males have at least a slight lifetime reproductive advantage over the lowest ranking males.
THE FIRST REAL BEGINNING to our understanding of wolf social ecology came from wolf 2204 on 23 May 1972. State depredation control trapper Lawrence Waino, of Duluth, Minnesota, had caught this female wolf 112 km ( 67 mi) south of where L. D. Mech had radio-collared her in the Superior National Forest 2 years earlier. A young lone wolf, nomadic over 100 km2 (40 mi2) during the 9 months Mech had been able to keep track of her, she had then disappeared until Waino caught her. From her nipples it was apparent that she had just been nursing pups. "This was the puzzle piece I needed," stated Mech. "I had already radio-tracked lone wolves long distances, and I had observed pack members splitting off and dispersing. My hunch was that the next step was for loners to find a new area and a mate, settle down, produce pups, and start their own pack. Wolf 2204 had done just that."
This dissertation examines dominance, play, affiliation and social preferences in a group of 24 domestic dogs that socialize regularly at a dog daycare facility. In Chapter 1, we analyzed the frequencies and directions of aggression, submission, and dominance displays. The distribution of submission and aggression resulted in a significantly linear hierarchy in the group, and submission was the best indicator of dominance relationships. Age was significantly correlated with dominance rank with older dogs out-ranking younger dogs. Muzzle-licking met most of the criteria for a formal display of submission in dogs, but was displayed in only 18% of relationships. Only 29% of all possible pairs had discernable dominance relationships. In Chapter 2, we examined the relationship between agonism, play and affiliation. Twenty-two percent of all possible pairs had known dominance relationships and also exchanged friendly behaviors (formal relationships), 21% of all pairs exchanged friendly behavior but had no discernable dominance relationship (egalitarian relationships), 50% of all pairs were never observed exchanging friendly or agonistic behaviors (non-interactive relationships), and 7% of pairs had known dominance relationships but did not exchange friendly behavior (agonistic relationships). Friendly behavior was much more frequent than agonistic behavior, and we found a complex association between the two. Egalitarian relationships as well as play and affiliation were more common between males and females than between same-sex pairs. Relationship affinity (an index combining play and affiliation) did not significantly differ in formal versus egalitarian relationships, but the latter were significantly more equitable and playful. In Chapter 3, we investigated patterns of interventions during dyadic play. Interveners directed either (1) a playful behavior at one of the dogs (the target), (2) an affiliative behavior at the target, or rarely, (3) an aggressive behavior at the target. Individual rank did not influence individual interventions rates. Rank relationships between interveners, targets and non-targets did not influence playful interventions. Dogs tended to target higher-ranking dogs during affiliative interventions and target lower-ranking dogs during aggressive interventions, but the latter were too infrequent to apply statistical analyses. Dogs tended to target their preferred partners (“friends”) during play more than support them.
Wolves are charismatic emblems of wilderness. Dogs, which descended from wolves, are models of urbanity. Do free-ranging dogs revert to pack living or are their societies only reminiscent of a wolfish heritage? Focusing on behavioral ecology, this is the first book to assess societies of both gray wolves and domestic dogs living as urban strays and in the feral state. It provides a comprehensive review of wolf genetics, particularly of New World wolves and their mixture of wolf, coyote and dog genomes. Spotte draws on the latest scientific findings across the specialized fields of genetics, sensory biology, reproductive physiology, space use, foraging ecology and socialization. This interdisciplinary approach provides a solid foundation for a startling and original comparison of the social lives of wolves and free-ranging dogs. Supplementary material, including a full glossary of terms, is available online at
This book provides an up-to-date description of the behavioural biology of dogs. It is written for students of animal behaviour or veterinary medicine at advanced levels and dog owners. This book is divided into 4 parts and 14 chapters. The first part (chapters 1-3) focuses on the evolution and development of the dog. The second part (chapters 4-8) deals with the basic aspects of animal behaviour with particular emphasis on dogs. The third part (chapters 9-12) places the modern dog in its present ecological framework in the niche of human coexistence. A broad overview of the behavioural aspects of living close to humans is given. The fourth part (chapters 13 and 14) focuses on behavioural problems, their prevention and cure.
The evolutionary continuity between humans and other animals suggests that some dimensions of personality may be common across a wide range of species. Unfortunately, there is no unified body of research on animal personality; studies are dispersed across multiple disciplines and diverse journals. To review 19 studies of personality factors in 12 nonhuman species, we used the human Five-Factor Model plus Dominance and Activity as a preliminary framework. Extraversion, Neuroticism, and Agreeableness showed the strongest cross-species generality, followed by Openness; a separate Conscientiousness dimension appeared only in chimpanzees, humans' closest relatives. Cross-species evidence was modest for a separate Dominance dimension but scant for Activity. The comparative approach taken here offers a fresh perspective on human personality and should facilitate hypothesis-driven research on the social and biological bases of personality.