J. Ethol.

16:1-i3, 1998

The Function of Allogrooming in Domestic Cats

(Felis silvestris cams);
a Study in a Group of Cats Living in Confinement ~

Ruud van den B o s

Anthrozoology Institute, Dept. Biology,
University of Southampton, Southampton, UK;
Dept. of Ethics, History and Alternatives of Anirrtal Experiments,
University of Leiden, Leiden, the Netherlands

Abstract -- Grooming interactions (n=83) occurring in a group of non free-ranging adult

neutered male (n=14) and female ( n = l l ) domestic cats (Fells silvestris catus) were
analysed. Grooming was not induced by the proximity (distance < = 0 . 5 m) of another
animal. Grooming was in general directed at the head-neck area. Higher ranking animals
groomed lower ranking animals more often than the other way round. Groomers tended
to adopt 'higher' (standing, sitting upright) postures than groomees (sitting, lying). Ago-
nistic behaviour occurred in 35% of interactions. Groomers showed offensive behaviour
more often than groomees, most often after grooming a partner. Furthermore groomers
often groomed themselves after grooming a partner. The degree of relatedness of animals
did not affect the frequencies or durations of grooming. These results are consistent with
the hypothesis that allogrooming in domestic cats may be a way of redirecting (potential)
aggression in situations in which overt aggression is too costly.

Introduction to study factors and processes leading to group

living as well as the way the social organization
Depending on the distribution and abundance is formed and maintained in the family of felids
of food flee-ranging domestic cats (Felis silves- (e.g. Macdonald et al., 1987; van den Bos & de
tris catus) live either solitary or in groups (re- Vries 1996). At the same time there has been a
views: Bradshaw, 1992; Bradshaw & Brown, growing interest in cat social behaviour under
1992; Liberg & Sandeil, 1988). Females with non-free ranging conditions (i.e. confined cats;
their dependent offspring form the core of Podberscek et al., 1991; Brown, 1993; van den
groups, whereas males are only loosely attached Bos & de Cock Buning, 1994a,b; Smith et al.,
to such groups and wander between them 1994; van den Bos, 1995a,b, 1996; van den Bos
(Liberg & Sandell, 1988). Over the last decade & de Vries, 1996) partly because of animal wel-
there has been a growing interest in cat social fare related questions and partly because con-
behaviour under flee-ranging conditions (e.g. fined animals allow for studying aspects of so-
Macdonald et al., 1987; Natoli & de Vito, 1991; cial organization in more detail (e.g. van den
Brown, 1993) partly because domestic cats have Bos & de Cock Buning, 1994a,b; van den Bos
been considered to be good model animals & de Vries, 1996).

1The data were previously presented at the 29th International Congress of the International Society for Applied
Ethology (van den Bos, R. (1995) Allogrooming in domestic cats in confinement, Proceedings of the 29th Interna-
tional Congress of the International Society for Applied Ethology, S.M. Rutter et al. (Eds.), pp. 109-110)
2Author's address: Ruud van den Bos PhD; Institute of Evolutionary and Ecological Sciences, Section Theoretical
Biology, University of Leiden, P.O. Box 9516, 2300 RA Leiden, the Netherlands
 One important issue which has been addres- ships with individuals with whom it may be
sed is conflict regulation in group-living condi- advantageous to affiliate like high ranking indi-
tions (Natoli & de Vito, 1991; van den Bos & viduals m which may give support in future con-
de Cock Buning, 1994a). Cats appear to lack flictsmas has been repeatedly observed in pri-
formal dominance hierarchies: none of their mates (see e.g. Wilson, 1975; Seyfarth, 1976,
affiliative behavioural patterns shows a unique 1977, 1980; Goosen, 1980; Hemelrijk, 1990;
one-way distribution with respect to rank-orders Hemelrijk & Ek, 1991; Spruijt et al., 1992).
based on the outcome of aggressive encounters As hypothesized in a previous paper (van den
(van den Bos & de Cock Buning, 1994a,b; van Bos & de Cock Buning, 1994a) allogrooming
den Bos, 1995a,b). Still, allogrooming, i.e. may be a way of reducing tension between
grooming a conspecific, appears to be an impor- animals which are living near one another. This
tant behavioural pattern in conflict-regulation, study addresses this possibility by careful
that is it appears to play a role in forming and observation and analysis of grooming interac-
maintaining stability of groups (van den Bos & tions.
de Cock Buning, 1994a; van den Bos & de
Vries, 1996).
The following has been specifically revealed Materials and Methods
with respect to allogrooming between adults: (i)
allogrooming occurs regardless of sex Animals & Study Area
(male/female), sexual status (intact/neutered) or
living conditions (free-ranging/non flee-rang- The study colony existed since 1988. Cats
ing); (ii) the distribution of allogrooming bouts (European short hairs) were purchased from
in pairs is asymmetric; (iii) the flow of allog- different breeders and brought together while
rooming is downrank; (iv) animals receive more juvenile. Males were neutered at an age of 6-12
bouts of grooming from animals from whom months, females at an age of 12-25 months.
they also receive more bouts of offensive be- The coefficients of relatedness ranged from 0 to
haviour; (v) allogrooming is associated with be- 0.632. These coefficients were calculated from
havioural patterns expressing confidence and breeders' information on which males sired
which are more likely to be shown by high- which females (see Brown, 1993 for details on
ranking animals, such as social sniffing and snif- the genealogy of this colony). Males were
ring rear; (vi) proximity scores of pairs and fie- generally stockier and heavier than females
quencies of allogrooming in pairs correlate posi- (weight ratio male to female: 1.3; ranges: male:
tively; (vii) in groups, mean grooming rate 5-8 kg, female: 4-5.5 kg). At the start of this
(bouts/h/pair) and density (cats/sqm) correlate study eleven females and fourteen males were
positively, while mean agonistic interaction rate present. During the study, after 9 sessions, one
(interactions/h/pair) and density correlate nega- female was removed from the colony, leaving
tively, i.e. the higher the density the less ten females for the remainder of sessions.
aggression and the more allogrooming occurs This colony has been studied earlier (March-
(Brown, 1993; van den Bos & de Cock Buning, -June 1991; reported in Brown, 1993). The be-
1994a,b; van den Bos, 1995a,b; van den Bos & havioural characteristics of the animals as well
de Vries, 1996; van den Bos et al. unpublished as the rank orders for males and females are
data). therefore well documented. Cats were classified
The asymmetrical distribution of grooming in into three ranking categories (high, middle and
pairs and the effect of rank order exclude the low) based on informal observations by Brown
possibility that grooming solely serves a (1993) and formal analysis of the flow of ago-
hygienic function as has been observed in im- nistic behaviour (data in Brown, 1993; cf. van
pala's (Hart & Hart, 1992; Mooting & Hart, den Bos & de Cock Buning, 1994a).
1993). Furthermore the rank order effect ex- The group's riving quarters consisted of an in-
cludes the possibility that grooming serves as a door and outdoor enclosure. The indoor enclo-
way of getting access to and establish relation- sure consisted of two rooms (A: 23 sqm; B: 28
sqm) and a corridor (length: 10.8 m; width: 1.3 ence with theft behaviour, the cats were fami-
m; area: 14 sqm) connecting rooms A and B liarized with the presence of an observer during
and leading to the outside enclosure. Room A two 1-hour sessions before actual observations
contained shelfs with beds in which cats could started (cf. van den Bos & de Cock Buning,
sleep, lie, sit etc.. This room also contained lit- 1994a). During these sessions and throughout
ter trays (bedding: wood shavings). Room B the observations the observer did not interact
contained cages in which cats could lie, sit etc.. with the cats. The observer recorded all
The corridor contained a pole with shelfs on ongoing behaviour from one spot in the enclo-
which cats could lie, sit etc. as well as a box in sure. The camera was focussed on the area
which cats could hide. The outside enclosure most frequently visited by the cats (about one
(96 sqm) contained several elements on which third of the enclosure). Several rehabituation
cats could sit, lie (pole with shelfs) or in which sessions were included when recording sessions
they could hide (boxes). Furthermore it con- were separated too far apart in time (see be-
tained logs for cats to scratch on. The floor of low). Recording was spread out over a period
the outside enclosure was made of tiles (29 • of 6 months. Sessions were interspersed with re-
29 cm). The population density was 0.16 cording sessions in room A. The latter sessions
cat/sqm. were initiated on days different from those in
Cats had free access to the outside enclosure the outside enclosure (van den Bos, in prepara-
during working hours (08.00--17:00 h). They re- tion). The temperature in the outside enclosure
mained in the inside enclosure at night. Cats was determined after each recording session us-
were kept on a 12-12 h day-night cycle with ing an ordinary thermometer to assess the effect
lights on from 08:(D--20:00 h. They were fed dry of temperature on grooming rate (see Troisi &
(SDS; U.K.) and wet food (Whiskas (R)) once Schino, 1986).
daily in the afternoon (after observations took
place; see below). Water was available through- Interactions
out the day in bowls present at various places in
the enclosure. Temperature in the inside enclo- An interaction was considered to have started
sure was maintained at 21 degrees Celsius; when a cat clearly and unambiguously
temperature in the outside enclosure varied approached another cat--approach: moving
according to weather conditions. clearly towards another c a t - - o r invited another
cat for an interaction by e.g. rolling in front of
Observational Regimen another cat. The cat which started the interac-
tion (approaching another cat or rolling in front
For each recording session cats were moved of another cat) was labelled the initiator, its
to the outside enclosure if the weather allowed partner the recipient. If cats started to groom
so. Only during dry weather recording sessions one another while sitting or lying close together
were initiated. Furthermore the floor and the (no approach or invitation) the cat which deli-
elements on which cats could sit etc. had to be vered the first bout of grooming was called the
dry. Two litter trays were provided during re- initiator.
cording in case cats needed to urinate or defe- The distance between the cats at the start of
cate. Observations were always made between the interaction was split into two categories:
11:00 and 15:00 h. The observer was present in start from within 0.5 m (proximity: a cat's body
the enclosure and recorded the behaviour of the length; van den Bos & de Cock Buning, 1994a)
cats using a video camera recorder with zoom- and start from a distance of more than 0.5 m.
lens (Philips Explorer Autofocus CCD During the interaction four behavioural pat-
VKR6850). Previous observations showed that tems of the initiator and recipient were re-
cats behave somewhat more aggressive and less corded (Table 1A) using the programme 'The
affiliative towards one another when an obser- Observer' (v2.0; Noldus Information Technolo-
ver is present in the enclosure (Brown, 1993). gy, Wageningen, the Netherlands). One extra
Therefore, in order to avoid too much interfer- category of behaviour was added, i.e. 'other be-
 Table 1. Ethogram used in this study (after van den Bos & de Cock Buning, 1994a; UK Cat Behaviour
Working Group, 1991). In parentheses the abbreviations which are used in the tables are shown.

(A): BEHAVIOURS
Behaviour Description
allogrooming (GC) licking the fur of another cat;
autogrooming (GS) cat licking its own fur (including wiping head region with paw after licking
paw);
Offensive behaviour (OF) combination of the following elements: (slowly) approaching another cat
(lowered head); running in pursuit of another cat, erect ears swiveled to
point back on the head; hairs on back and tail erected; lashing tail; pant-
ing, salivating; cuffing, growling, yowling; narrowing of the pupils (includ-
ing mounting);
Defensive behaviour (DE) combination of the following elements: lowering of hindquarters; pilo-
erection; flattening of the ears to the head such that they tend to flush
with the top of the head; hissing, spitting; growling; rolling on the dorsum
to expose claws; cuffing; dilation of the pupils (including fleeing);

(B) POSTURES
Posture Description
standing (STA) standing on all four;
sitting upright (SITUP) sitting on hindpaws with outstretched forepaws; the weight of the body is
shifted towards hindquarters;
sitting (SIT) sitting on hindpaws and frontpaws; forepaws may be tucked underneath
body or stretched out in front of the animal; the weight of the body is
supported by front- and hindquarters;
lying (LYI) lying on the side, back etc.; the weight of the body is supported by the
body rather than the paws;

haviour', which contained all other behaviours. mined whether it was directed to the head-neck
Furthermore postures of the groomer and area, the shoulder-chest area or the a b d o m e n
groomee (Table 1B) while bouts of grooming area (see insert of Fig. 1 for the body-areas).
were exchanged were recorded.
An interaction was considered to have ended
Body Are== =nd AJlogroondng
if one of the (or both) cats moved away or fled
(distance: > 0 . 5 m). If cats remained together
(distance: < = 0.5 m; proximity) the interaction
was considered to have ended if none of the be- 100~ - i
C "
havioural patterns of Table 1A was shown by 80X;
the cats for a period of 60 s.
A bout of allogrooming was defined as a
series of licking spells not interrupted by 40g - l
another behaviour, like looking round, watch
20~ -
other cats, licking the perioral region, unclearly
directed nibblings to the fur of the other cat or o,J r
a clear short pause (cf. van den Bos & de Cock
Buning, 1994a). The same principle applies to Body =rea
the definition of bouts of autogrooming, offen- ['] Bout= I I Durltlon
sive behaviour and defensive behaviour.
Fig. 1. Distribution of the number of bouts of aUog-
rooming and of total duration of allogrooming across
Data Analysis and Statistics different body-areas; Insert: Areas into which the cat's
body was split for analysis: A: head-neck area; B:
For each allogrooming bout it was deter- shoulder-chest area; C: abdomen area.
 Postures of the groomer and groomee were Throughout the Results section M E A N + S E
analysed by preparing a matrix containing post- is reported unless otherwise stated. Reported p-
ures for the groomer (rows) and the groomee values are two-tailed unless otherwise stated, p-
(columns). The matrix was analysed using a Values equal to or smaller than 0.05 were cons-
Chi-square test (Sokal & Rohlf, 1981). idered to indicate statistical significant differ-
For each interaction the members of the in- ences.
teracting pair were denoted. Interactions were
labelled unidirectional if only one animal
groomed, otherwise bidirectional. Interactions RF~ULTS
between the same individuals, i.e. the same
pair, were added to obtain one score (frequen- Interactions: General Aspects
cy, duration) per behavioural pattern per mem-
ber of a pair. Thus for each member of a pair In total n = 2 8 sessions were recorded on
one score per behavioural pattern (frequency, video tape (23h45min) over a 6 month period.
duration) was obtained, whether the animal had Ninety-two interactions were taped (3.9+0.4
been an initiator or a recipient in an interac- interactions/h), but nine interactions (9.8%)
tion. could not be analysed, because interactions
Scores (frequency, duration) per behavioural were only partly on screen and because other
pattern per individual were totalled, i.e. for animals prevented reliable recording. No cor-
each individual one score (frequency, duration) relation existed between the temperature in the
per behavioural pattern was obtained, irrespec- outside enclosure and the number of grooming
tive with whom the individual had interacted interactions per h (rS=0.05, n--25, ns).
with as an initiator or as a recipient. Most often (n=78 interactions; 94%) interac-
The number of interacting male-male (MM), tions (duration: 103.9+15.7s, n=83; range
male-female (MF) and female-female (FF) pairs 5.7-945.4s) started with one animal approaching
was related to the maximal possible number of or inviting another animal; only 5 interactions
pairs for each category (MM,MF or FF pairs). (6%) started with animals sitting or lying
For the final number of different grooming together (distance: < = 0 . 5 m). In 89.7% of the
pairs an expected distribution was calculated approaches/invitations animals were more than
based on the maximal number of pairs per 0.5 m apart before the interaction started.
category. The observed distribution and the ex- Accordingly in total 84.3% (n=70) of interac-
pected distribution were subsequently compared tions started while animals were more than 0.5
using a Chi-square Goodness-of-Fit test (Sokal m apart.
& Rohlf, 1981). Subsequently, means + SE's for The majority of interactions (91.6%) was un-
the different behaviours of the MM, MF and idirectional. In case of bidirectional interactions
FF pairs were calculated. Differences between the distribution of grooming in terms of bouts
males and female in MF pairs were analysed us- (correlation coefficient: r=0.28, dr=5, ns) as
ing paired t-tests. well as duration ( r = - - 0 . 0 2 , df--5, ns) deviated
To assess the effects of kinship, for each pair from symmetry ( r = 1).
the overall scores of the different behavioural Allogrooming was most often directed to the
patterns (frequency, duration) were correlated (dorsal part of the) head-neck area both in
with the coefficient of relatedness using Spear- terms of number of bouts and duration (Fig. 1).
man rank correlation tests (Sokal & Rohlf In the majority of cases (n=58 interactions;
1981). Furthermore pairs were classified accord- 69.9%) interactions ended with one animal
ing to whether the animals were highly (r = > moving away (>0.5 m); in only a small number
0.25), intermediately (0.125 < = r < 0.25) or of interactions (n=15; 18.1%) animals re-
distantly (r < 0.125) related. Differences in fre- mained within 0.5 m of one another. Ten in-
quencies and durations of behavioural patterns teractions (12%) ended with the flight of an
between different categories were analysed us- animal.
ing one-way Analysis of Variance (ANOVA).
Interactions: Behavio ural patterns number of allogrooming and autogrooming
bouts as well as between their durations was
Groomers were often standing (43.6%) or sit- found for both initiators (r=0.63, dr=81,
ting upright (45.1%) whereas groomees were P<0.001 and r=0.42, df=81, P<0.001 respec-
often sitting (46.6%; Table 2). Due to the large tively) and recipients (r=0.47, df=81, P<0.001
number of cells containing values smaller than 5 and r=0.14, df=81, n.s. respectively). These
(9/16: 56.3%) a Chi-square could not be prop- same correlations held when allogrooming and
erly run. Yet, it should be noted that the cells autogrooming scores of recipients and initiators
containing the combination standing/sitting were summed to the scores for different indi-
(groomer/groomee; 23.3%) and sitting viduals ( n - 2 1 individuals: r=0.92, df=19,
uptight/sitting (18.8%) contained the highest P<0.001 and r=0.87, df=19, P<0.001 respec-
values (Table 2). tively). Bouts of allogrooming ( n = 8 3 bouts)
In 34.9% of interactions agonistic behaviour more were often followed (60.2%) than pre-
(offensive or defensive behaviour) was re- ceded (39.8%) by bouts of autogrooming when
corded. If offensive behaviour was shown by both occurred.
either an initiator or recipient of an interaction
(n--30 cases) it was more often shown by the Pairs: Males and Females
groomer (70%; n = 2 1 cases) than by the
groomee (30%; n = 9 cases). Bouts of offensive In total, 37 different pairs were observed to
behaviour shown by the groomer ( n = 2 6 bouts) groom (13.4% of all possible pairs; Table 3).
were more often preceded (80.8%; n = 2 1 The overall grooming rate was: 0.28+0.07
bouts) than followed (19.2%; n--5 bouts) by bouts/pair/h. The distribution of allogrooming
bouts of allogrooming when both occurred. across pairs was asymmetric whether in terms of
Bouts of defensive behaviour ( n = 1 8 bouts) frequencies of grooming bouts (Fig. 2A) or tot-
were more often preceded by bouts of offensive al duration of grooming (Fig. 2B).
behaviour (83.3%; n = 1 5 bouts) than by bouts The 37 pairs showed 83 interactions. Most in-
of allogrooming (16.7%; n = 3 bouts) when both teractions consisted of males interacting with
occurred. one another: MM: 54 (65.1%), MF: 26 (31.3%)
A strong positive correlation between the and FF: 3 (3.6%). In the 26 MF interactions,

Table 2. Occurrencesof combinations of postures of groomer (rows) and groomee (columns) while bouts of
grooming were delivered. Tot: totals for rows or columns. For abbreviations of behaviours see Table 1.

STA SITUP SIT LYI Tot

STA i2 i1 31 4 58
SITUP 13 16 25 6 60
SIT 0 1 2 2 5
LYI 2 0 4 4 10
Tot 27 28 62 16 133

Table 3. Total number of different pairs interacting (n=37). Since in the majority of sessions (67.9%) 10
rather than 11 females were present the values were calculated using the smaller number of females leading
to a slight overestimation of the contribution of females. %tot: number of pairs expressed as percentage of
total number of pairs (n=37); %class: observed number of pairs per class expressed as percentage of the tot-
al number of pairs per class (MM: n=91 pairs; MF: n = 140 pairs; FF: n=45 pairs); int/pair: number of in-
teractions per pair.

class observed (%tot) expected (%tot) %class int/pair

MM 18 (48.6) 12 (33.0) 19.8 3.0+0.7
MF 16 (43.2) 19 (50.7) 11.4 1.6+0.3
FF 3 (8.1) 6 (16.3) 6.7 1.0+0.0
 7
Froquenc~ of AIiogroomlng (90.4%; n--75).
In the 26 MF interactions, males delivered all
numbe~of bouW Individual B
40 allogrooming bouts in 17 cases (65.4%),
females in 6 cases (23.1%); both delivered
bouts in 3 cases (11.5%). In MF pairs (n = 16),
30 -
males appeared to deriver more allogrooming
bouts per interaction than females (3.0+0.9 vs
0.9+0.3).
2( -
In MF pairs ( n = 16), males groomed females
more often in 62.5% of cases whereas the re-
L0 - @ verse was true in 37.5% (Chi-square Goodness-
@
of-Fit, n.s.). Males tended to groom females

0
orii-i vv
'
10 @ 2JO 9 30

number of bout~ Indlvkk~ A

Duration of AIIogroomlng

l~xmd~l~lr
400
DurlUon o f / i ~ r o o m l n g

~ ~ B 300

300 +

200
250
@@
200
lO01
150

100
.IL~I-.+ L 9 J,-+.-,
vv
Z~JI'.I,_
v v i
@
- | 0,1 0,2 0,3 0,4 0,5 06
50 ~ ol v~lhK~mmm
-- @ 9
i
i@
T v -5~" vv i00 15o moo 250 300
Frequency of AIIogroomlng
llcolw4l kw:IM<~l~ A

Fig. 2. Relationships between frequencies (panel A) 50

and durations (panel B) of allogrooming of different
members (individuals A and B) of pairs (n--37 in total)
40
in which allogrooming occurred. The labels A and B
were arbitrarily assigned to members of pairs. Although
this prohibits calculating correlation coefficients properly 30
(any change in A and B leads to different results) the 00
distributions in panels A and B (for which correlation
20
coefficients are' frequency: r--0.04, df=35, ns; duration: @
r = - 0 , 0 5 , df=35, ns) make clear that there is no 1 ' 1
distribution in pairs (shown by the dotted Line under the to
angle of 45 degrees in both panels) no matter how the @ #
@@
labels A and B are assigned to individuals. Some dots re- t o ~ l ~ L e :@@, $
flect more than one data-point. These are indicated in 0o
.'.d __t_
o.i 0,2 0.3 0.4 0.5 0
the panels. ~ of t~mtK~a

Fig, 3, Relationship between coefficient of relatedness

males acted as initiator in 80.8% of cases
(r) and frequency (panel A) and duration (panel B) of
(n=21), females in 19.2%. Males behaved allogrooming in pairs (n--37). Some dots reflect more
accordingly as initiator in nearly all interactions than one data-point. These are indicated in the panels.
longer than females males (40.3+20.8s vs appeared to be low compared to other studies
3.4+1.8s; t=1.777, df=15, P<0.10); no other (van den Bos & de Cock Buning, 1994a; van
differences were observed. den Bos, 1995a: 53-95%). Although it has been
shown that animals become less affiliative and
Pairs: Rank Order more aggressive towards one another when an
observer is present (Brown, 1993) this is unlike-
In MM pairs in which differences in rank ex- ly to account for this low number. Brown
isted (n=12 of n = 1 8 pairs) higher ranking (1993) observed cats in the outside enclosure
males groomed lower ranking males more often using a closed video-circuitry in a situation in
than the other way round in 75% of pairs ( n = 9 which the cats could freely move between the
pairs). In two of three FF pairs in total in which inside and outside enclosure. Recalculating her
a rank difference existed higher ranking females data showed that also under these conditions
groomed lower ranking females more often only a small number of pairs, i.e. 15.7%, was
than the other way round. Accordingly, higher observed to groom in the outside enclosure
ranking animals groomed lower ranking animals (Brown, 1993; 157h of observation). The low
more often than the other way round in 78.6% number of pairs is also not due to observing
of pairs ( n = l l ) ; the reverse was true in 21.4% animals in the outside enclosure per se: record-
(n=3; n = 1 4 pairs in total). This is significantly ings (using a closed video-circuitry) of the anim-
different from a 50--50% distribution (Chi- als in the inside enclosure (room A) revealed a
square Goodness-of-Fit: 4.571, d r = l , P<0.05). low number of grooming pairs (5.8%; 22h55
min observation over a 4 month period, Van
Pairs: Kinship den Bos unpublished data). The low density
(0.16 cat/sqm) under which the animals were
None of the behavioural patterns showed any living may however have been a contributing
correlation with the coefficient of relatedness, factor. For, previous studies showed a positive
whether in terms of number of bouts or dura- relationship between density (0.42--0.60
tions of patterns per pair (all rS, P>0.05). Fig. cat/sqm) and the number of grooming pairs in
3 shows the data for frequencies (panel A) and group living females in confmement (53-95%;
durations (panel B) of allogrooming. Analysing van den Bos, 1995a,b), i.e. the lower the densi-
the data in terms of classes of relatedness (high, ty the lower the number of grooming pairs.
intermediate and distantly) showed the same Neutering may have been a contributing fac-
picture (all one-way Anova's: P~>0.05; not tor as well. For, it has been found that the sex-
shown). ual cycles in females affect the likelihood that
females groom one another: females rub onto
females while in oestrus and receive allogroom-
Discussion ing in return during such interactions (van den
Bos & de Cock Buning, 1994a). As to the
General males it should be noted that previous studies
showed that intact confined males also groom
This is the first detailed study on allogroom- one another (Podberscek et al., 1991; van den
ing in adult domestic cats in confinement. Until Bos & de Cock Buning, 1994b). Whether
now allogrooming has been mentioned to occur neutering facilitates the occurrence of grooming
among adults in confinement (Leyhausen, 1979; in males or not remains an open question as
Podberscek et al., 1991; Brown, 1993; van den yet. It should be noted that for females and
Bos & de Cock Buning, 1994a,b; Smith et al., males the structure of allogrooming sequences
1994; van den Bos, 1995a,b; van den Bos & de does not appear to be affected by neutering
Vries, 1996) but its function remained obscure. (see Brown, 1993; van den Bos & de Vries,
In only a few pairs allogroorning was 1996).
observed in this colony of females and males. Males appeared to be more active groomers
Especially the number of female pairs (6.7%) than females, both among themselves and in
male-female interactions. The latter observation the decrease of the stress-response of indi-
will be discussed below within the context of viduals (expressed through behavioural (Aureli
different hypotheses on grooming. The former & van Schaik, 1991) and physiological (Terry,
however seems to be surprising in view of the 1970; Boccia et al., 1989; cf. Feh & de
fact that males are normally wandering solitarily Mazieres, 1993) signs); a stress-response which
and hardly interact (Bradshaw, 1992; Bradshaw has arisen because of conflicts (cf. reconcilia-
& Brown, 1992). As indicated above neutering tion; see Kappeler & van Schaik, 1992 for re-
per se does not seem to be a decisive factor. view) or the proximity of another animal
Rather it may be suggested that confinement (hypothesis II). Different predictions arise from
per se, a situation in which animals are con- these hypotheses. Following observations in
stantly near another and in which there are am- birds (Harrison, 1964) under hypothesis I
ple opportunities for males to interact may be a grooming is expected to be directed at the
contributing factor (see below). head-neck area since it is this area at which cats
Despite the relatively low number of interact- direct an attack onto others (Leyhausen, 1979).
ing pairs and the strong involvement of males Hypothesis II does not predict a specific area in
the data allow along with those of previous stu- this respect. Under hypothesis I grooming may
dies to discriminate between different hypoth- be accompanied by aggression during interac-
eses on allogrooming in cats. tions (due to the balance between overt and re-
direction of aggression), while aggression is not
Allogrooming: Alternative Hypotheses expected to occur under hypothesis H. Fur-
thermore following observations in rats (Grant,
The present data confirm the notion that 1%3) the frequency (and duration) of auto-
allogrooming is neither a hygienic behaviour grooming (expressing the level of tension in cats
nor a way of establishing relationships with as a result of the balance between overt and re-
animals with whom it may be advantageous to direction of aggression (Maestripieri et al.,
affiliate (see Introduction for references). 1992; Spruijt et al., 1992; Willemse et al., 1994;
First, interactions were most often un- Willemse & Spruijt, 1995; van den Bos, 1996))
idirectional, and if bidirectional highly asym- and the frequency (and duration) of allogroom-
metric. This is furthermore reflected in the ing are expected to be positively correlated in
asymmetry in frequencies and durations of individuals during interactions under hypothesis
aUogrooming in pairs (cf. van den Bos & de I. In contrast, under hypothesis II the frequency
Cock Buning, 1994a). Secondly, differences ex- (and duration) of autogrooming, expressing the
isted with respect to the percentages of groom- level of the stress-response due to conflict or
ing pairs per sex class. Thirdly, higher ranking proximity (Willemse et al., 1994; Willemse &
animals directed more grooming to lower rank- Spruijt, 1995; van den Bos, 1996), are expected
ing animals than the other way round (cf. van to be decreased in the groomee, since allog-
den Bos & de Cock Buning, 1994a). Fourthly, rooming reduces the level of the stress-response
grooming was not affected by kinship. in the groomee (cf. Aureli & van Schaik, 1991;
Previously it has been hypothesized that Spruijt et al., 1992); in fact a negative correla-
allogrooming may be a way of reducing tension tion may therefore be expected in individuals
between animals which are living near one between the frequency (and duration) of auto-
another (van den Bos & de Cock Buning, grooming and the ffequencey (and duration) of
1994a). Two different hypotheses may be allogrooming received. Finally, following
formulated in this respect. First, allogrooming observations in birds (Harrison, 1964), rats
may serve as a way of redirecting (potential) (Grant, 1963) and primates (O'Brien, 1993)
aggression when overt aggression is too costly aIlogrooming is expected to flow downrank
(Grant, 1963; Harrison, 1964; Wilson, 1975; under hypothesis I, while this is expected to be
Troisi et al., 1989; see also Spruijt et al., 1992; rank-neutral or uprank under hypothesis 11.
O'Brien, 1993; hypothesis/) and thereby reduce The question arises then as to what extent
tension. Second, allogrooming may accelerate the data fit hypothesis I (redirecting (potential)
10
aggression) or hypothesis H (reducing stress-re- tion that offensive behaviour occurs often after
sponse). a groomer groomed its partner. Again this
In line with previous observations (van den argues against hypothesis H but in favour of
Bos & de Cock Buning, 1994a; van den Bos et hypothesis I.
al., unpublished data) it was found that offen- Previously, a positive correlation was
sive behaviour was more often displayed by the observed between the frequency of grooming
g r o o m e r m m o s t often after grooming bouts m and proximity scores of pairs (van den Bos &
Moreover offensive behaviour appeared in the de Cock Buning, 1994a; cf. 1994b). These data
same factor as allogrooming when the indi- suggested that grooming and proximity are
viduals' behaviour was factor analysed (van den either expressing two sides of the same coin,
Bos, unpublished data). This seems to argue i.e. both measures are expressing a mutual pre-
against hypothesis H. Furthermore, it was found ference of animals, or that one induces the
that higher ranking animals were more likely to other (cf. Troisi et al., 1989). The analysis in
show allogrooming than lower ranking animals the present study eliminated the problem of the
(cf. van den Bos & de Cock Buning, 1994a). previous study that grooming and proximity
Higher ranking cats are the more aggressive, were not assessed independently. The present
lower ranking the more fearful animals (van data show that allogrooming is neither induced
den Bos & de Cock Buning, 1994a; van den by proximity nor induces proximity. This fits in
Bos & de Vries, 1996). Together with the fact with the notion that grooming pairs do not
that males (being heavier and stockier) tended necessarily have higher proximity scores than
to show more grooming to females than the non-grooming pairs (van den Bos & de Cock
other way round in male-female pairs, it may in Buning, 1994a). Yet, it should be noted that in
general be stated that the animal with the larger the inside enclosure (room A, in which cats
Resource Holding Potential (RHP; larger sized, normally sleep) allogrooming occurred more
heavier, and potentially more threatening anim- often when animals were already near one
al) is more likely to show allogrooming than the another for some time as well as remained near
animal with the smaller RHP (van den Bos & one another more often than in the outside en-
de Vries, 1996). The fact that the groomer closure (van den Bos unpublished data). Taken
often adopted a higher posture than the together these data suggest that allogrooming
groomee while grooming (see also Fig. 1 in and proximity may rather independently express
Macdonald et al., 1987; cf. Harrison, 1964) is a mutual preference of cats for one another
consistent with this notion, i.e. the cat's higher (two sides of the same coin) than that one in-
or lower social status is reflected in its posture duces the other (see Troisi et al., 1989). The
(see van Hooff & Wensing, 1987). These data immediate causation of grooming may therefore
argue in favour of hypothesis I. The observation not be another cat's proximity with a resultant
that cats directed grooming most often towards increase in tension, which is alleviated by allog-
the head-neck area also fits this hypothesis. rooming, but rather the mere presence of a cat
The fact that frequencies (as well as dura- with which the existence of a bond or social
tions) of allogrooming and autogrooming were position (high and low rank) needs to be con-
strongly positively correlated in individuals firmed. These data argue against hypothesis H.
argues in favour of hypothesis I. ff the occurr- Overall the foregoing data strongly suggest
ence of autogrooming is taken as a sign of ten- that allogrooming between adult cats in confine-
sion in the groomer (Goosen, 1980; Troisi & ment may be a form of redirecting (potential)
Schino, 1987; cf. Willemse et al., 1994; Wil- aggression. A cost-benefit analysis for groomer
lemse & Spruijt, 1995; van den Bos, 1996; see and groomee suggests the following. The
also Maestripieri et al., 1992) the observation groomer would enjoy the benefit of not engag-
that autogrooming often followed allogrooming ing in costly overt aggression in order to main-
(cf. Grant, 1%3) suggests that tension is gra- tain its position, the groomee the benefit of not
dually building up in the groomer during the in- being attacked by its opponent (cf. O'Brien,
teraction. This fits the forementioned observa- 1993). In confinement such aggression regulat-
 11
ing mechanisms may be important. This view is Conclusion
consistent with the observed positive correlation
between the density of groups and the mean All in all the data suggest that allogrooming
rate of grooming (bouts/h/pair) in these groups, in domestic cats may be a way of redirecting
and negative correlation between density and (potential) aggression when aggression is too
the mean rate of agonistic interactions costly, i.e. they suggest that allogrooming
(interaction/h/pair): with increasing densities allows animals to remain to live in one
there is a shift towards the expression of more another's presence and establish short-term and
grooming and less overt aggression (and less long-term bonds. Future studies are directed at
rigid rank orders; van den Bos, 1995a,b). whether this is a characteristic of species with
facultative social structures like domestic cats
Allogrooming in free-ranging domestic cats (see introduction), and whether this function
holds for other group-living felids, like lions
Although studies on confined non-breeding and (male) cheetah's, as well.
animals may not be conclusive with respect to
the function of a behavioural pattern (see Kxebs Acknowledgements - The author wishes to ex-
& Davies, 1993) such studies may allow to press his gratitude to Dr. John Bradshaw (Uni-
formulate a working hypothesis for the function versity of Southampton, Southampton, U.K.)
of a behaviour, which can be subsequently for allowing him to study the colony of cats, his
tested under flee-ranging and freely breeding hospitality during his visit to Southampton, and
conditions (see also van den Bos & de Vries, his constructive criticism during the project.
1996). Furthermore the author acknowledges two
ff the hypothesis is forwarded that the func- anonymous referees for their constructive com-
tion of allogrooming is to redirect (potential) ments which helped to improve the quality of
aggression in cases where overt aggression is this paper. This study was supported by a travel
too costly and allows animals thereby to remain grant of the Netherlands Research Organization
in one another's presence (cf. Harrison, 1964), (NWO).
it may be predicted that under flee-ranging con-
ditions this behavioural pattern would occur in
groups under conditions in which it may be References
necessary to maintain bonds, e.g. in case of
cooperative defense of resources or communal Aureli, F. & C.P. van Schaik 1991 Post-conflict
nursing and raising of kittens. Under free-rang- behaviour in long-tailed macaques (Macaca
ing conditions related females with their depen- fascicularis): II. coping with the uncertainty.
dent offspring form the core of groups (Brad- Ethology 89: 101-114.
shaw, 1992; Bradshaw & Brown, 1992). In line Boccia, M.L., M. Reite & M. Laudenslager
with the foregoing prediction grooming has 1989 On the physiology of grooming in a pig-
been observed between (related) females (a tail macaque. Physiol. Behav. 45 : 667-670.
mother and her two adult daughters), which de- Bos, van den R. 1995a Enige inzichten in het
fended their resources and cooperated in raising sociale gedrag van katten. Biotechniek 34:
theft young (Macdonald et al., 1987). Furth- 27-31.
ermore grooming occurred between these (re- Bos, van den R. 1995b Recent insights in cat
lated) females and a tom cat who appeared to social behaviour. A N Z C C A R T News 8(4):5-
be the breeding tom of this group. Although 6.
further data are lacking, the observation that in Bos, van den R. 1996 Post conflict behaviour in
two colonies of free-ranging non-breeding (re- domestic cats in confinement, Proceedings of
lated and non-related) cats hardly any grooming International Society for Anthrozoology
occurred (Brown, 1993) seems consistent with Meeting in Cambridge, UK, 24-26th July
this view. 1996, p.65.
Bos, van den R. & Tj. de Cock Burring 1994a
12
Social behaviour of domestic cats (Fells lybica pp.219-252.
forma catus L.): A study of dominance in a Kappeler, P. & C.P. van Schaik 1992 Methodo-
group of female laboratory cats. Ethology 98 : logical and evolutionary aspects of reconcilia-
14-37. tion among primates. Ethology 92 9 51-69.
Bos, van den R. & Tj. de Cock Buning 1994b Krebs, J.R. & N.B. Davies 1993 An introduc-
Social and non-social behaviour of domestic tion to Behavioural Ecology, 3rd edition
cats (Fells cams L.): a review of literature Blackwell Scientific Publications, Oxford.
and experimental findings. In: (Bunyan, J Leyhausen, P. 1979 Cat Behavior; The Preda-
(ed.)) Welfare and Science Royal Society of tory and Social Behavior of Domestic and
Medicine Press London pp.53-57. Wild Cats. Garland STPM Press, New York
Bos, van den R & H. de Vries 1996 Clusters in and London.
social behaviour of female domestic cats Liberg, O. & M. Sandell, M. 1988 Spatial orga-
(Fells silvestris cams) living in confinement, J. nization and reproductive tactics in the
Ethol. 14: 123-131. domestic cats and other felids. In: The
Bradshaw, J.W.S. 1992 The Behaviour of the Domestic Cat: the Biology of its Behaviour.
Domestic Cat. C.A.B. International, Walling- (Turner, D.C. & Bateson, P., eds.) Cam-
ford, U.K. bridge University Press, Cambridge, pp.83-
Bradshaw, J.W. & S.L. Brown 1992 Social be- 98.
haviour of cats. Tijdschrift voor Diergenees- Macdonald, D.W., P.J. Apps, G.M. Carr & G.
kunde vol 117 Suppl. 1, pp.54S-56S. Kerby 1987 Social dynamics, nursing coali-
Brown, S.L. 1993 The Social Behaviour of Neu- tions and infanticide among farm cats, Felis
tered Domestic Cats (Felis catus), PhD Thesis catus. Adv. Ethology 28: 1-66.
University of Southampton. Southampton, Maestripieri, D., G. Schino, F. Aureli & A.
U.K. Troisi 1992 A modest proposal: displacement
Feh, C. & de J. Mazieres 1993 Grooming at a activities as an indicator of emotions in pri-
preferred site reduces heart rate in horses. mates. Anita. Behav. 44:967-979.
Anim. Behav. 46: 1191-1194. Mooring, M.S. & B.L. Hart 1993 Effects of re-
Goosen, C. 1980 On grooming in Old World latedness, dominance, age and association on
monkeys, PhD Thesis University of Leiden, reciprocal allogrooming by captive impala.
Leiden, the Netherlands. Ethology 94:207-220.
Grant, E.C. 1963 The analysis of the social be- Natoli, E. & E. de Vito 1991 Agonistic be-
haviour of the male laboratory rat. Behaviour haviour, dominance rank and copulatory suc-
21 : 2N)-281. cess in a large multi-male feral cat, Felis catus
Harrison, C.J.O. 1964 Allopreening as agonistic L., colony in central Rome. Anim. Behav. 42
behaviour. Behaviour 24: 161-209. :227-241.
Hart, B.L. & L.A. Hart 1992 Reciprocal allog- O'Brien, T.G. 1993 Allogrooming behaviour
rooming in impala, Aepyceros melampus. among adult female wedge-capped capuchin
Anita. Behav. 44: 1073-1083. monkeys. Anita. Behav. 46 : 499-510.
Hemelrijk, C.K. 1990 Models of, and tests for, Podberscek, A.L., J.K. Blackshaw & A.W.
reciprocity, unidirectionality and other social Beattie 1991 The behaviour of laboratory col-
interaction patterns at a group level. Anita. ony cats and their reactions to a familiar and
Behav. 39:1013-1029. unfamiliar person. Appl. Anita. Behav. Sci.
Hemelrijk, C.K. & A. Ek 1991 Reciprocity and 31:119-130.
interchange of grooming and 'support' in Seyfarth, R.M. 1976 Social relationships among
captive chimpanzees. Anita. Behav. 41:923- adult female baboons. Anim. Behav. 24:917-
935. 938.
Hooff, van J.A.R.A.M. & J.A.B. Wensing Seyfarth, R.M. 1977 A model of social groom-
1987 Dominance and its behavioral measures ing among adult female monkeys. J. Theor.
in a captive wolf pack, In: Man and Wolf Biol. 65:671-698.
(Frank, H. ed.) Junk Publishers, Dordrecht Seyfarth, R.M. 1980 The distribution of groom-
 13
ing and related behaviours among adult and social influences on autogrooming be-
female vervet monkeys. Anim. Behav. 28: haviour in a captive group of Java monkeys.
798-813. Behaviour 100: 292-302.
Smith, D.F.E., K.J. Durman, D.B. Roy & Troisi, A., G. Schino & F. Aureli 1989 Allog-
J.W.S. Bradshaw 1994 Behavioural aspects of rooming and inter-individual proximity in two
the welfare of rescued cats. J. Feline Advis- species of macaques (macaca fascicularis and
ory Bureau 31:25-28. m. nemestfina). Behaviour 111:196-207.
Sokal, R.R. & F.J. Rohlf 1981 Biometry; The UK Cat Behaviour Working Group 1991 An
Principles and Practice of Statistics in Biolo- Ethogram for Behavioural Studies of the
gical Research, 2nd edition, W.H. Freeman Domestic Cat (Felis catus L.), UFAW Anim-
& Co., New York. al Workshop Report No.l, UFAW Potters
Spruijt, B.M., J.A.R.A.M. van Hooff & W.H. Bar, U.K.
Gispen 1992 Ethology and neurobiology of Willemse, T., M. Mudde, M. Josephy & B.M.
grooming behavior. Physiol. Rev. 72:825- Spruijt 1994 The effect of haloperidol and
852. naloxone on excessive grooming behavior of
Terry, R.L. 1970 Primate grooming as a tension cats. Eur. Neuropsychopharmacol. 4:39-45.
reduction mechanism. J. Psychol. 76:129- Willemse, T. & B.M. Spruijt 1995 Preliminary
136. evidence for dopaminergic involvement in
Troisi, A. & G. Schino 1986 Diurnal and clima- stress-induced excessive grooming in cats,
tic influences on allogrooming behaviour in a Neurosci. Res. Comm. 17:203-208.
captive group of Java monkeys. Anita. Be- Wilson, E.O. 1975 Sociobiology: The New
hay. 34: 1420-1426. Synthesis. Harvard University Press, Cam-
Troisi, A. & G. Schino 1987 Environmental bridge Massachuttes

(Received 5 September 1996 ;Accepted 6 February 1998)

Published by Japan Ethological Society, Department of Zoology, Kyoto University, Sakyo, Kyoto, 606-8502, Japan