EE

Ca

Rb

c

C

a

AA

KHSACMS

0d

Applied Animal Behaviour Science 138 (2012) 216– 228

Contents lists available at SciVerse ScienceDirect

Applied Animal Behaviour Science

journa l h o me pag e: www.elsev ier .com/ locate /applan im

xploring aggression regulation in managed groups of horsesquus caballus

arole Fureixa,∗, Marie Bourjadec, Séverine Henryb, Carol Sankeyb, Martine Hausbergera,b

Université Rennes 1, UMR CNRS 6552, laboratoire Ethologie Animale et Humaine, Campus de Beaulieu, bâtiment 25, 263 avenue du Général Leclerc, 35042ennes Cedex, FranceUniversité Rennes 1, UMR CNRS 6552, laboratoire Ethologie Animale et Humaine, Station Biologique, 35380 Paimpont, FranceUniversité de Provence – Aix-en-Provence, Département de Psychologie développementale et différentielle, Centre de Recherche en Psychologie de laonnaissance, du Langage et de l’Émotion PSYCLE EA 3273, 29 Av. Schuman, 13621, Aix-en-Provence Cedex 1, France

r t i c l e i n f o

rticle history:vailable online 10 April 2012

eywords:orsesocial behaviourggression regulationoping in groupanagement

ocial development

a b s t r a c t

Horses are highly social animals that have evolved to live in social groups. However, in mod-ern husbandry systems, single housing prevails where horses experience social isolation,a challenge-to-welfare factor. One major reason for this single housing is the owners’ con-cerns that horses may injure each other during aggressive encounters. However, in naturalconditions, serious injuries due to aggressive encounters are rare. What could thereforeexplain the claimed risks of group living for domestic horses? Basing our questioning onthe current knowledge of the social life of horses in natural conditions, we review differ-ent practices that may lead to higher levels of aggression in horses and propose practicalsolutions.

Observations of natural and feral horses mostly indicate a predominance of low frequen-cies and mild forms of aggression, based on subtle communication signals and ritualizeddisplays and made possible by group stability (i.e. stable composition), dominance hierar-chy and learning of appropriate social skills by young horses. Obviously, adults play a majorrole here in canalizing undesirable behaviours, and social experience during development,associated with a diversity of social partners, seems to be a prerequisite for the young horseto become socially skilled.

Given the natural propensity of horses to have a regulation of aggression in groups, thetendency to display more aggression in groups of domestic horses under some manage-ment practices seems clearly related to the conditions offered. We therefore review themanaging practices that could trigger aggressiveness in horses. Non social practices (space,resource availability) and social practices (group size, stability of membership, compositionand opportunities for social experiences during development) in groups of domestic horsesare discussed here.

Finally, we propose simple practical solutions leading to more peaceful interactions ingroups of domestic horses, based on the knowledge of horses’ natural social life which

therefore should be enhanced (e.g. ensuring roughage availability, favouring group stability,introducing socially experienced adults in groups of young horses, etc.). The state of theart indicates that many questions still need to be answered. Given the importance of the

issues

associated welfare required, which could ben

∗ Corresponding author. Tel.: +33 2 23 23 61 55; fax: +33 2 23 23 69 27.E-mail address: carole.fureix@gmail.com (C. Fureix).

168-1591/$ – see front matter © 2012 Elsevier B.V. All rights reserved.oi:10.1016/j.applanim.2012.02.009

and the consequences on the use of horses, further research is

efit horses. . . and humans.

© 2012 Elsevier B.V. All rights reserved.

Behavio

C. Fureix et al. / Applied Animal

1. Introduction

Horses (Equus caballus) are highly social animals thathave evolved to live in groups (e.g. Linklater, 2000; Waring,2003). However, group housing remains limited in thedomestic situation, especially in the case of valuable sporthorses.

One reason is the owners’ concerns that horsesmay injure each other during aggressive encounters(e.g. Grogan and McDonnell, 2005; Hartmann et al.,2009, 2011; McDonnell and Haviland, 1995). Thus, inmodern husbandry systems, single box housing pre-vails where horses experience both social isolationand confinement. The prevalence of weaving especiallyhas been shown to relate to lack of social contact(Benhajali et al., 2010; Cooper et al., 2000), and a mir-ror or a poster of a horse has proved to diminishthe frequency of this undesirable behaviour (McAfeeet al., 2002). Social deprivation also leads to aggressivebehaviours when horses are confronted again to con-specifics, and to undesirable behaviours at work (e.g.biting, kicking humans during training, Rivera et al.,2002; Sondergaard and Halekoh, 2003; Sondergaard andLadewig, 2004).

Given the importance of the practical and welfareissues associated with social isolation, an evaluation of thecosts and benefits of group housing in domestic horsesis needed. Inspections of a semi-feral herd of ponies on4 occasions over a period of 28 months revealed thatall injuries and blemishes were minor, most likely fromglancing contact of hooves or teeth rather than frankbite or kick wounds (Grogan and McDonnell, 2005). InFeist’s (1971) 6-months study of Pryor Mountain horses,35 out of 270 animals died from infected injuries, onlythree of which were potentially due to complications ofinjuries from aggressive interactions. In 509 h of obser-vation on 78 feral horses, Berger (1977) revealed thatnearly half of the 20 inter-band agonistic interactionsdid not even lead to a fight. In natural conditions, seri-ous injuries due to aggressive encounters are rare evenin bachelor male groups or during encounters betweenfamily stallions and intruders (Feh, 2005; Tilson et al.,1988). Low levels of injury are also reported in domes-tic horses in semi-natural conditions (e.g. kept in stablesocial groups in appropriately spacious fields with forag-ing opportunities and watering sites), even if there areless quantitative data in these populations. For instance,McDonnell and Haviland (1995) kept together in pasturegroups of 3–20 mature pony stallions and reported remark-ably few and generally minor injuries despite sometime“spectacular” inter-male interactions (see also Christensenet al., 2002a).

In light of the relatively low levels of aggression innaturally living horses, how can the perceived risks ofgroup living for domestic horses, especially stallions, beexplained? Basing our questioning on the current knowl-edge of the social life of horses (wild, feral and domestic)

in natural conditions, we thereafter review the differ-ent factors that may lead to higher levels of aggressionin some domestic situations and propose practical solu-tions.

ur Science 138 (2012) 216– 228 217

2. The social organization of horses in naturalconditions

Horses are group-forming equids, with a long last-ing bond observed between adults (e.g. Waring, 2003)and a constant type of social organization across popu-lations despite different environmental and demographiccharacteristics (e.g. Linklater, 2000). The socioecology ofhorses has been extensively described and we are justgiving a brief account here. Feral and Przewalski horsesare organized in long-lasting non-territorial reproductiveassociations, the so-called family band. Family band sizehas been reported to vary from 2 to 30 individuals, themost common size being 4–6 individuals (Waring, 2003).Under natural conditions horses (including non breed-ing stallions) tend to associate in groups characterized bytheir high stability. Overall, aggression rates are rather lowin natural conditions (Table 1) and as mentioned above,wounds are rare. Four factors can be identified that maycontribute to explain these findings.

2.1. Stable group composition

Family bands are made up of one to several unre-lated adult mares, one to a few stallions (long term stablenucleus), and their immature offspring (foals and one- tothree-year old young horses) (e.g. Berger, 1986; Boyd andKeiper, 2005; King, 2002; vanDierendonck et al., 1996;review in Boyd and Keiper, 2005 and Feh, 2005). In familygroups where more than one stallion is present, one is dom-inant with a privileged access to breeding and the others aresubordinate (e.g. Linklater and Cameron, 2000; Salter andHudson, 1982). However, tolerance between them is highand some authors have suggested that they form alliancesto protect the group (Berger, 1986; Feh, 1999; Keiper, 1976)although this remains controversial (Linklater et al., 1999).Adult horses tend to form dyadic social bonds (mare–mare,stallion–mare and stallion–stallion, Feh, 1999) mostlycharacterized by spatial proximity (preferential closestneighbours) but also reciprocal positive interactions (e.g.allogrooming). Young horses of both sexes leave their natalband at sexual maturity, usually around 2–3 years old,thus remaining with their mares beyond physical depen-dence. Non-breeding males can be solitary but are moreoften observed in bachelor groups of 2 to more than 15stallions (e.g. Berger, 1986; Feh, 1999). Bachelor groupsare less stable than family bands. To our knowledge, theage at the time of first acquisition of a harem is around 5years (reviewed in Boyd and Keiper, 2005); suggesting thatnon-breeding males are prone to stay 2–3 years in a bache-lor band. However, long-term dyadic associations betweenstallions have been reported (e.g. Berger, 1977; Bourjadeet al., 2009a; Feh, 1999; Salter and Hudson, 1982). Finally,family bands may group temporarily with other bands,especially in winter when risks of predation may increase.The stability of group membership (at least between adults)allows each horse to know the social status of others and

to behave accordingly, sometimes without the need fora dominant to give any aggressive signal (Berger, 1977;Heitor et al., 2006; Wells and Goldschmidt-Rothschild,1979).

218C.

Fureix et

al. /

Applied

Anim

al Behaviour

Science 138 (2012) 216– 228

Table 1Mean frequencies of agonistic behaviours (per horse per hour) and potential factors of variation of aggression rates in various horse populations gathered from multiple published studies. Studied horses arenatural population, domestic groups of horses in natural conditions and domestic groups of horses in paddock. Standard errors are presented when available. Rates of aggression are low both in stable naturalgroups and domestic groups kept under semi-natural conditions. However, rates of aggression can be higher in large groups, under high density, experiencing modifications of the group composition and/or whenaccess to resources is limited. Note that these conditions are not limited to domestic conditions but can be more common in domestic groups of horses, especially in paddock, compared with natural populations.

Study Life condition Rates of aggressivebehaviour (per horseh−1)

Group size (number ofhorses)

Density (number ofhorses ha−1)

Group composition Group compositionmodifications(human managing)

Restrictedresources

Weeks et al. (2000) Domestic – paddock 6.3 ± 4.1 h−1 14 1.4–127.7 Breeding mares with foals Part of theobservations madeduring feeding ofsupplemental grain

Montgomery (1957) Domestic – paddock 3.2 h−1 11 Na Mares and geldings 1 injured mareremoved duringthe course of thestudy

Benhajali et al. (2008) Domestic – paddock 2.5 ± 1.4 h−1 44 200.0 Arab breeding mares(without foals)

No foragingopportunities (nograss, no hay)

Keiper and Sambraus (1986) Natural population 2.4 h−1 Ponies from AssateagueIsland

Confined by artifi-cial/topographicalbarriers →restrictedresources?

Clutton-Brock et al. (1976) Natural population 1.9 h−1 17 Na Highland ponies (marewith and without foals,juvenile mares andgelding)

1 mare removedduring the courseof the study

Christensen et al. (2002b) Natural population 1.5 h −1 13 0.2 Przewalski bachelor groupsHeitor et al. (2006) Domestic – semi

natural1.2 ± 0.7 h−1 11 0.6–2 Family band of Sorraia

horsesStallion placedwith mares forbreeding at thebeginning of thestudy

Wells andGoldschmidt-Rothschild(1979)

Natural population 0.9 ± 0.3 h−1 ≈20 ≈0.1 Family band of Camarguehorses

Herd stallion addedto the group at thebeginning of thestudy

Jorgensen et al. (2011) Domestic – paddock 0.9 ± 0.8 h−1 3–6 1.7–33.3 Mares and geldingsFeh (1988) Semi natural

population0.8 ± 0.2 h−1 4–5 0.25–1.25 Przewalski bachelor group

and family bandvanDierendonck et al. (1995) Domestic –

semi-natural0.7 h−1 31 22.1 Icelandic horses (geldings,

mares, juvenile stallionsand mares) + a few ponies

Some adults absentfor some weeks(breeding)

Houpt and Keiper (1982) Natural population 0.2 ± 0.0 h−1 7 Na Family band of ponies fromAssateague Island

Bourjade et al. (2009b) Natural population 0.2 ± 0.1 h−1 9 0.02 Przewalski bachelor groupsHausberger et al. (unpublished) Domestic – semi

natural0.2–1.1 h−1 2–5 0.7–2.5 Stallions and geldings

Sigurjonsdottir et al. (2003) Domestic – seminatural

0.1 h−1 34 4.25 Adult mares, immaturehorses, adult geldings ingroups

Behavio

C. Fureix et al. / Applied Animal

2.2. Stable dominance hierarchy

Another major factor involved in the regulation ofaggression is the dominance hierarchy, where the knowl-edge of each other’s dominance status makes novelchallenges useless (Berger, 1977; Heitor et al., 2006; Wellsand Goldschmidt-Rothschild, 1979). As much broadlydefined, the function of dominance would be to regu-late conflicts over coveted resources of any kind, loweringaggressiveness and social tensions as the result of the pre-dictability of contest outcomes (Rutberg and Greenberg,1990), thus reducing the risk of injury while interacting(Bourjade et al., 2009a; Keiper and Sambraus, 1986).

In natural horse societies, stable linear dominance hier-archies are formed with occasional reversals and triangles(Feh, 2005; feral horse: Houpt and Keiper, 1982; Keiperand Sambraus, 1986; Tyler, 1972; Wells and Goldschmidt-Rothschild, 1979; Przewalski horse: Boyd and Houpt, 1994;Feh, 1988). Dominance gives a priority of access to limitedresources (water, food, etc.) and therefore may be diffi-cult to observe in environments where food (e.g. grass) isuniformly distributed or water easily accessible. Moreover,once established, the dominance relationships are main-tained by both spontaneous avoidances by subordinatesand low intensity threats by dominants (Berger, 1977;Wells and Goldschmidt-Rothschild, 1979). Behaviours suchas “bite-threats”, “head-threats” and “avoid” appear as bet-ter indicators of dominance than actual “attack”, “kick” and“kick threats” which show more rarity or directional incon-sistency (Heitor et al., 2006; vanDierendonck et al., 1995).

Dominance rank in horses does not depend upon sizeor weight (Feh, 1990). Age and anteriority in the group (i.e.higher length of residency in the group) seem to be majordeterminants (Monard et al., 1996). Stallions may or maynot be dominant over mares (e.g. Berger, 1977; Keiper andSambraus, 1986; Stevens, 1988; Wells and Goldschmidt-Rothschild, 1979). Stallions’ herding behaviour, where theychase back their females to the group, has been inter-preted as an aggressive act and sign of dominance by someauthors (McDonnell and Haviland, 1995; Miller, 1981)but rather as a sexual behaviour by others (Keiper andSambraus, 1986). As in other species (e.g. pigs: Bolhuis et al.,2005, carnivorous: Sands and Creel, 2004, birds: Poisbleauet al., 2006; Verbeek et al., 1999), dominance rank is notor very weakly correlated with the relative aggressive-ness of horses (e.g. Berger, 1986; vanDierendonck et al.,1995; Wells and Goldschmidt-Rothschild, 1979; Przewal-ski horse: Bourjade et al., 2009b; Feh, 1988), meaningthat dominant horses are not necessarily the most aggres-sive individuals. Dominance rank differs from “leadership”where some individuals are more followed than otherstowards resources (Bourjade et al., 2009b) and are notrelated.

2.3. Ritualized communication

Given this long term associations between group

members, horses have developed a complex socialand communication system based on close associationsbetween few partners (Feh, 2005). Field observationssuggest individual recognition (e.g. Feh, 1999; Feist and

ur Science 138 (2012) 216– 228 219

McCullough, 1976), an ability now supported by experi-mental data on domestic horses (Lemasson et al., 2009;Proops et al., 2009). Horses have a repertoire of gradedvisual signals that enable other animals to assess the inten-tions of the emitter (Feh, 2005; Kiley-Worthington, 1976;Waring, 2003). Examples of displays and theirs grada-tions are given in Waring (2003), while an ethogram ofaggressive behaviours has been proposed by McDonnelland Haviland (1995). Ritualized displays may, as in otherspecies (e.g. Sebeok, 1979), be useful substitutes to realaggressions. This is especially the case for stallion–stallionencounters that involve ritualized interactive sequences(McDonnell and Haviland, 1995; Waring, 2003).

Aggression in horses varies from low to high intensityand takes the forms of non-contact aggressions repre-sented by threats to bite or threats to kick, and physicalaggressions such as bites, kicks, chases or attacks. Patternsof aggressive interactions are characterized by a fixed earsbackward position, sometimes reaching the flattening ofears against the top of the neck (Feist and McCullough,1976). Agonistic interactions act at increasing distancebetween two opponents either through a spontaneous dis-placement of one of them or through the aggression ofone opponent upon the other (Berger, 1986; Feist andMcCullough, 1976). Displacements, which can be eitherspontaneous or a response to aggressions, take the forms ofavoiding by moving a part of the body away, moving awayby walking or fleeing by trotting or cantering.

On the basis of the available literature, Waring (2003)concludes that horses usually display the minimal amountof aggression the situation requires. Threats are hence themost common channel used to direct aggression towardsconspecifics in long-term stable groups (Table 2).

2.4. Development of social skills

As other social species, horses need social experiencein order to develop appropriate social skills (e.g. Bourjadeet al., 2009a). Of course, the first social bond a foal formsis with its mother: a short time before giving birth, themare stays away from the group for one to a few days ina place where the mare will give birth (Estep et al., 1993;vanDierendonck et al., 2004) and establishes an exclusivebond with her foal. The mother remains a foal’s preferredpartner for quite a long time (e.g. at 6 months of age, foalsare still spending 40% of the time with their dam as closestneighbour, Crowell-Davis and Weeks, 2005; Tyler, 1969).However, the foal starts interacting with other foals and itsfather after 2–3 weeks (Feh, 2005; Tyler, 1969). There is aswitch of interest over time with more and more time spentwith other foals or other family members (Boyd, 1988;Crowell-Davis, 1986; McDonnell and Poulin, 2002; Tyler,1972). Weaning occurs on average 15 weeks (but up some-times to 24 h) before the birth of the next foal, when themother generally prevents the foal from suckling (Welsh,1975). The young remains nevertheless close to the dam,paying special attention to the newborn.

After weaning, interacting with peers becomes evenmore frequent, and Tilson et al. (1988) reported that thedominant stallion tended to stay close to the 1–2 year oldother males and protect them from other group members.

220 C. Fureix et al. / Applied Animal Behaviour Science 138 (2012) 216– 228

Table 2Proportions of aggressive behaviours in agonistic interactions in function of their intensity in (a) feral populations and groups of domestic horses in semi-natural conditions and (b) domestic horses in paddock. Low intensity: displacement, non-contact aggressions (simple threats, i.e. ears backward only;threats to bite; to kick). High intensity: physical aggressions (bites; kicks). In long-term stable groups, low intensity aggressions appear to be the mostcommon channel used to direct aggression towards conspecifics. Note that, even if numerous other studies mentioned such a result, quantitative data arenot always available, especially in domestic populations.

Low intensity High intensity

Displacement Simple threat, bite threats Kick threat Bite Kick

(a) Feral and semi-natural populationsa

Berger (1977) 71% (mares)48% (stallions)

Berger (1986) 92% 8% <0.01%Keiper and Sambraus (1986) 48% 27% 8%vanDierendonck et al. (1995) 49% 27% 6% 12% 6%Christensen et al. (2002b) ≈27% ≈27% ≈13% ≈27% ≈7%Heitor et al. (2006) 33% 60% 3% 3% 0.6%

(b) Domestic populations in paddockb

Araba and Crowell-Davis (1994) 53% 18% 21% 2%Montgomery (1957) 10% ≈50% 6%c ≈25% 6%c

a Berger (1977, 1986): feral horses, Grand Canyon, band size range: 3–6 individuals, including 1 stallion; Christensen et al. (2002b): Przewalski bachelorgroup (n = 13 stallions, Ukraine); Heitor et al. (2006): Sorraia horses, 10 adult mares and 1 adult stallion; Keiper and Sambraus (1986): feral ponies,A donck ea mara).

s, Georg

r(2prtba(2oopstla

rahiKp1(saapsatdoff

ssateague Island, 3–28 individuals, including 1–2 stallion(s); van Dierennd 2 juvenile mares) and 5 ponies (breeds: Shetland, New Forest, Conneb Araba and Crowell-Davis (1994): 15 Belgian brood-mares and 10 foalc Kick threats and kick counted together in this study.

Young horses receive overall little aggression. Foalsarely get injured when they approach other adult maresCrowell-Davis and Weeks, 2005; Grogan and McDonnell,005). One mechanism may be the use of snapping. In thisarticular display, the foal pulls the lips back and claspshythmically the teeth, often with an arched back and aucked in tail (Feh, 2005; Waring, 2003). Snapping haseen considered as a “submission” display (e.g. McDonnellnd Haviland, 1995) or a sign of conflict of motivationapproach/withdraw from this impressive adult; Waring,003), the result being apparently that little aggression isbserved towards young individuals (as in juvenile signalsf dogs puppies, Bekoff, 1977). Data are lacking on thisotential appeasing effect of snapping but more than 60% ofnapping directed to mares occurred after she threatenedhe foal and most snapping displayed by foals towards stal-ions occurred without any threat from the adult in Wellsnd von Goldsmith-Rothschild’s study (1979).

However, behaving appropriately in group (withegards to aggressiveness here) clearly also relates oncquisition of social skills. For instance, while adult feralorses displayed only 0.3% of their threats towards a dom-

nant conspecific (see also Clutton-Brock et al., 1976: 4.9%;eiper and Sambraus, 1986: 7.3%), the proportion of “inap-ropriate” threats directed up the hierarchy reached up0–20% of yearlings’ threats, and 50% of foals’s threatsWells and Goldschmidt-Rothschild, 1979). These resultsuggest a major role of experience on social competencescquisition. At all developmental stages, the young horsesre confronted to a variety of social partners: siblings,eers, related and unrelated adults, including mares andtallions. Social experience and especially adult modelling,s in other social species (e.g. West et al., 2003), appearherefore to play a crucial role in order for the young to

evelop the typical low level/low rate of aggression typicalf natural social groups. In a study performed on naturalamily bands of Przewalski horses, Bourjade et al. (2009a)ound that the overall rate of aggression towards peers in 1

t al. (1995): 26 Icelandic horses (6 geldings, 16 mares, 2 juvenile stallions

ia; Montgomery (1957): 11 pleasure riding horses (7 geldings, 4 mares).

and 2 year old horses was negatively correlated with theadult-young ratio. When this ratio was low (less adultspresent), the young horses showed more aggressiveness,more segregation from adults and more bonding with sameage partners. Moreover, the mother’s rank tends to havean influence on its offspring’s rank both prior and afterweaning (Tyler, 1972; Wells and Goldschmidt-Rothschild,1979).

2.5. Conclusion

Mild forms and low frequencies of aggression havevaluable advantages for horses, since physical aggressioninvolves energetic costs, increases the risk of injury andis known to decrease reproductive success, by reducingrates of conception and increasing rates of foetal and foalmortality (e.g. Berger, 1986; Linklater et al., 1999). Thepredominance of low frequencies and mild forms of aggres-sion, based on subtle communication signals and ritualizeddisplays is made possible in natural horses populationsthanks to group stability, dominance hierarchy and learn-ing of appropriate social skills by young horses. Obviously,adults play a major role in canalizing socially undesirablebehaviours of the immature horses.

3. Impact of management on aggression and itsregulation in the domestic/captive situation

3.1. Horses’ aggressiveness in the domestic situation:myth or reality?

Studies of feral horses living under natural condi-tions indicate that the species-specific behaviour of equidshas remained relatively unaffected by the domestication

process (e.g. Klingel, 1982; Linklater, 2000; Tyler, 1972).Nevertheless, rates of aggression may be affected by thedomestic life (Table 1). A consequence of increased aggres-siveness may be an increase of injuries (Houpt et al.,

Behavio

C. Fureix et al. / Applied Animal

1978). Thus, in a survey involving 58 Thoroughbred andQuarter Horse farms that managed 1987 weanlings andyearlings, Gibbs and Cohen (2001) observed that injurieswere reported in 64% of the farms and mostly attributed byowners to horses “playing too hard”, which probably corre-sponds to the interpretation given by owners of aggression,confused with play fighting. Although a few studies aresufficiently precise to enable a correct evaluation of aggres-sion rates, environmental conditions offered to domestichorses, rather than being a domestic animal per se, may beat stake when high level of aggression are reported. Indeed,many domestic horses are kept in restricted conditions ofspace, food and social opportunities and these factors cantrigger aggression. Note that feral populations experienc-ing for instance modifications of their group composition(due to human management) may show temporary higherlevels of aggression, while stable domestic horses kept insemi-natural conditions show low frequencies of aggres-sion (e.g. islandic horses studied by Sigurjonsdottir et al.,2003 and vanDierendonck et al., 1995) (Table 1). Thus, nonsocial and social factors may explain some high levels ofaggressions sometimes encountered.

3.2. Non social factors

3.2.1. SpaceConfinement, including forced proximity and and/or

high density may increase aggressions in domestic animals(Archer, 1970). Stebbins (1974) observed that aggressionswere more frequent and intense in Appaloosa horses whenkept in paddocks rather than in pastures. According to Skiff(1982, cited in Keiper, 1986), the number of aggressionsshown by Przewalski horses increased as the size of theenclosure decreased. Limited space (i.e. limited resources?)on Assateage Island may explain the higher rates of aggres-sion observed by Keiper and Sambraus (1986) in feralponies (Table 1). An increase of aggressions and derivedinjuries may be observed in Przewalski horses kept incaptivity in zoos (Boyd, 1988), sometimes leading to abnor-mally high aggressiveness (Keiper, 1986).

Group density could also affect social behaviour (seeBenhajali et al., 2008). Observing groups of horses in pad-docks of about 100 m2 per horse up to 75,000 m2 per horse,Jorgensen et al. (2009) reported that horses with the small-est space allowance showed the highest mean number ofaggressive interactions (4.6 h−1) as compared to all otherbatches (1.3 h−1). Invasion of another horse personal spaceis a common source of aggression (Heitor et al., 2006;Keiper and Sambraus, 1986; Tyler, 1972). Davidson (1999,cited by Christensen et al., 2002b) noted that when living ina herd, conflicts are typically avoided through spatial distri-bution of the horses since individuals spread out and allowonly a few others to be within their personal space. Tightspaces, that do not allow submissive animals to escape, aremore common in domestic group situations than in semi-feral groups (Grogan and McDonnell, 2005).

3.2.2. Resources availabilityIn general, limited access to resources induces social

competition and increased levels of aggression (e.g. Berger,1977; Clutton-Brock et al., 1976; Grogan and McDonnell,

ur Science 138 (2012) 216– 228 221

2005; Montgomery, 1957; vanDierendonck et al., 1995).Thus, when additional food was brought to free ranginghighland ponies, agonistic behaviours increased almosttwofold, i.e. from 1.9 to 3.3 aggressive behaviours perhour per horse (Clutton-Brock et al., 1976). Note that, innatural conditions where foraging is possible on widelydistributed food resources, competition can neverthelessoccur around water, shelters, branches, etc. Deprivation ofsuch resources may certainly lead to increased agonisticinteractions when again made available.

Care must be given to the ways of providing food, min-erals and water. With regard to the contexts in whichaggression occurred, vanDierendonck et al. (1995) reportedin a herd of Icelandic horses in captivity that aggressionoccurred on the pasture but was more common around themineral supply, the drinking bowls, and in the transitioncorridor (see also Montgomery, 1957). In pastured domes-tic horses, it is more common to provide water from asingle concentrated source than from a stream or pond andto provide supplemental feeds that are highly palatable.The resulting food- and water-related aggression seemsmuch more frequent and intense than water- or forage-related aggression in naturally foraging herds (Groganand McDonnell, 2005). The automatic feeding systems, byallowing only one horse at a time to enter the feeding box,seem also to trigger more aggressions, especially high levelaggressions that may lead to injuries (Zeitler-Feicht et al.,2010).

Foraging is a major activity in a horse’s life. In all pop-ulations studied, free ranging horses spent 60–80% of theirtime grazing (e.g. Boyd and Keiper, 2005; Waring, 2003).Conversely, domestic/captive horses daily receive a lim-ited number of high energy meals and a limited amountof roughage (e.g. Harris, 2005; Nicol et al., 2002). Whendomestic horses are placed as a group in a bare pad-dock and given a variety of enrichments including straw,branches, object, they showed a time limited interest inobjects and the only efficient enrichment appeared to bethe straw (providing thus foraging opportunity) (Jorgensenet al., 2011). In this study, presence of straw was asso-ciated, amongst other aspects, with less agonistic andmore friendly social interactions, while in all other cases(controls, poles, play objects, etc.), horses exhibited moreagonistic than friendly interactions.

The importance of foraging opportunities in regulat-ing social interactions had been experimentally tested byBenhajali et al. (2008, 2009). In a first study, the authorsdrew up the behavioural repertoire and time-budget ofa dense (200 mares/ha) group of 44 Arab breeding mareshoused in individual boxes at night but kept 6 h a day in abare paddock. They found that the behavioural repertoireand the time-budget were affected: neither lying down,nor rolling were observed while locomotion was abnor-mally high. They also observed a very limited social life: nopreferred partner, no allogrooming and only a few socialinteractions exclusively agonistic interactions (Benhajaliet al., 2008). At that stage, both the high density and the

lack of foraging opportunity could be responsible for theseresults.

The authors then tested the impact of foraging oppor-tunity per se by dividing one hundred breeding mares from

2 Behavio

ts1taswpmethmatiw

inbiiwaut

3

3

gpfI(ratbhawt

3

ladGilUgsomg

22 C. Fureix et al. / Applied Animal

he same facility into two groups of fifty mares, kept in theame conditions as mentioned above (but with a density of15 mares/ha). However one group was given the oppor-unity to forage in the paddock, as 50 hay nets were hunglong the fences (Benhajali et al., 2009). All mares had theame total amount of hay, as the control group received thehole amount in the box at night. Therefore, only the tem-oral distribution varied. As expected, the experimentalares spent most of their time foraging at hay, but they also

xhibited a larger overall behavioural repertoire. However,he most interesting was probably that they also showedigher rates of positive interactions (e.g. allogrooming),ore social cohesion (preferred partners) as well as less

ggression (0.9 ± 0.7 aggressions/mare/h) than the con-rol mares (0.2 ± 0.2 aggressions/mare/h). Thus aggressiventeractions dropped by half when foraging opportunities

ere provided in the paddock (Benhajali et al., 2009).Two studies therefore converge towards a same major

mpact of roughage availability on the prevalence of ago-istic behaviours. This is especially interesting as differentreeds (Warmbloods/Purebred Arab horses) and types (rid-

ng – sport horses/breeding mares) of horses are involved,n very different climates (Scandinavia/Tunisia). This may

ell reflect a species-specific trend due to the discomfort ofn empty stomach for hours (Harris, 2005), making horsesnfriendly or leading to an unfulfilled time-budget leavingime for undesirable behaviours.

.3. Social factors

.3.1. Group sizeDomestic groups are often larger than natural social

roups (vanDierendonck et al., 2004), which can be aotential source of social tensions and aggression. Indeed,requencies of aggression per feral pony mare at Assateaguesland were higher in larger bands than in smaller bandsRutberg and Greenberg, 1990). Christensen et al. (2002b)eported an aggression rate of 1.46 per horse per hour in

group of 13 Przewalski stallions, which is twice as muchhan the rate observed in a small group of 4 Przewalskiachelor stallions in Feh’s (1988) study (0.76 per horse perour). Group size is highly variable in the domestic situ-tion and its influence on the prevalence of aggressionsould need worth further investigation, paying attention

o the distinction between group size and density effects.

.3.2. Group compositionIn free ranging populations, mares tend to perform

ess agonistic behaviours than stallions, but both in feralnd domestic situations, stallions are not necessarilyominant over females (Houpt et al., 1978; Wells andoldschmidt-Rothschild, 1979). Groups of males do occur

n natural conditions and grouping of stallions does notead to increased injuries (e.g. Christensen et al., 2002a,b).nfortunately in the domestic situation, stallions areenerally kept in single boxes, which may impair their

ocial skills (see further). More information towards horsewners is needed here. Allotting 66 adult horses into oneare group, one gelding group and one mixed gender

roup, Jorgensen et al. (2009) reported no significant effect

ur Science 138 (2012) 216– 228

of gender composition on social interactions or on the veryfew observed injuries (all superficial).

Reproductive status of mares, and in particular foal-ing, can impact on aggressiveness towards conspecifics.Thus, mares tend to become aggressive after parturition,preventing the approach of herd members towards thefoal (vanDierendonck et al., 1995; Wells and Goldschmidt-Rothschild, 1979). vanDierendonck et al. (2004) investi-gated the social behaviour of mares before and after foalingin a herd of Icelandic horses consisting of adult mares, adultgeldings and juvenile fillies, geldings and colts. Mares withfoals separated off into a distinct subgroup and becamemore aggressive, but the presence of adult geldings inthe herd during the foaling season did not prevent theexpression of characteristic species-specific behaviour andallowed the development of long-term stable social rela-tionships (vanDierendonck et al., 2004). Family groups ofdomestic horses, when kept in semi-natural conditions,seem to function as do feral families (see further). Breed-ing groups where stallions stay with the mares and foalsare obviously not a source of risk (Grogan and McDonnell,2005). Although infanticide has been mentioned in nat-ural conditions, it remains a rare phenomenon, possiblyrestricted to major social changes (Feh, 1990).

As mentioned before, age is a crucial element in thedominance hierarchy and older individuals tend to havehigher ranks than immature horses both in groups ofdomestic and feral/wild horses (Houpt et al., 1978; Houptand Wolski, 1980, see part2). In general, aggressions arerare amongst young animals and in a food competition test,all yearlings shared food (Houpt et al., 1978).

There is no evidence that breed differences do occurin aggressiveness. Observational data in stable groups givesimilar low prevalence for different breeds (Table 1). How-ever, according to Grogan and McDonnell (2005), Shetlandponies may be less injured during aggressive interactions,partly because of their appearance: being small, stocky andfat, means that they may “cushion” the kicks and maybegive less powerful kicks. Although physical characteristicsof horses (height, weight) cannot be used to predict domi-nance or aggressiveness (Houpt and Wolski, 1980), furtherstudies are certainly needed to investigate the potentialimpact of some breed characteristics in group composition.

To our knowledge there is no experimental data test-ing aggressiveness as a temperamental trait in horses(Hartmann et al., 2011). Some horses are reported to haveabnormally high levels of aggressiveness, but then thismay be pathological (Keiper, 1986). In paired feeding tests,dominance (access to buckets) may be correlated with thefrequency of aggressive behaviours (Houpt and Wolski,1980), but observational studies mostly do not show sucha correlation (see Section 2.3). Hormonal status (e.g. ges-tation in mares) or age have been shown to influenceboth aggressiveness and dominance rank, suggesting lowindividual consistency over time and across situations, apre-requisite for a temperamental trait (Hausberger andRichard-Yris, 2005).

Changes in herd composition can disrupt the socialorganization of a herd (Keiper and Sambraus, 1986),meaning that changes in individual status may occur. Illus-trations can be extracted from a case study (Hausberger

C. Fureix et al. / Applied Animal Behaviour Science 138 (2012) 216– 228 223

Table 3Characteristics of the populations studied by our research group. Period 1: 9 ponies observed for 30 h in August 1993 on a 2-ha pasture. The group had beenstable for 2 years (except EZ, which arrived 6 months earlier); period 2: 10 ponies observed for 70 h in October–December 1994 on a 2 ha paddock, with around-baller of hay. The group had been stable for 6 months; period 3: 5 ponies observed for 110 h between March and June 2001 on natural pastures (6 ato 2 ha). The group had been stable for 4 years. In all cases, water was provided ad libitum, natural shelters were available. Apart from the filly, the poniesworked either as riding school ponies (periods 1 and 2) or leisure horses (period 3). They were generally ridden together, mostly for 2 days in the week(4–6 h in periods 1 and 2, 1–3 h at the most for period 3). Experiments involving these animals complied with current French laws (Centre National de laRecherche Scientifique) related to animal experimentation and were in accordance with the European directive 86/609/CEE. No licence/permit/institutionalethical approval was needed. Animal husbandry and care were under the management of their private owner, as this experiment involved only horses inthe “field” (no laboratory animals).

Horse Breed Sexa Period 1 (1993) Period 2 (1994) Period 3 (2001)N = 9 N = 10 N = 5Age Age Age

A Unregistered G 5 6 12P Fjord G 6 7 13U Welsh G 7 8PB Unregistered G 14 15Q Merens F 11 –S Connemara F 10 –F Unregistered G 5 6E Pottok G 17 –EZ Welsh × unregistered F 1 2B Dartmoor G 5D Haflinger × unregistered G 10G Haflinger G 11V Fjord G 7T Haflinger M 15H Appaloosa × pony M 6

Ha Haflinger × Connemara G

a G: gelding, F: female, and M: Male.

et al., unpublished data), in which we followed given indi-viduals, and two especially in different group compositions(Table 3). In period 1 (Fig. 1), P and A appeared at the topof the hierarchy for aggressions given, avoidances receivedand success at the paired feeding test, while U was at thebottom of the hierarchy. One year later (period 2, Fig. 1),after the group composition had changed, P was still themost aggressive but he was less avoided than others andranked only second in the feeding test. U went up andbecame the second most aggressive pony, the first to beavoided and the first to access the bucket, while A was stillavoided but became less aggressive than before. Only P andA could be followed over a longer time period. In period3 (Fig. 1), they had been with three other individuals forfour years, living all year round, night and day, togetheron a pasture. In this situation, A has clearly become a sub-ordinate, whatever the parameter chosen, while P has anintermediate status. The relative status of P and A has alsochanged over time (e.g. feeding test in period 1/period 3),showing that despite a long term relationship, status wasnot definitively acquired. The results show that the stabilityof an individual’s status may depend upon group stabilityrather than on intrinsic individual characteristics.

3.3.3. Group stabilityIn the domestic situation, group composition often

changes according to the use of horses (Grogan andMcDonnell, 2005). In stable groups of domestic horses,the dominance relationships are mainly maintained by

spontaneous avoidances by subordinates like in naturalpopulations (e.g. Arnold and Grassia, 1982; Jorgensen et al.,2009). Subordinates rarely “contest”: Heitor et al. (2006)report for instance that most offensive interactions (84.8%)

6

elicit an avoidance/withdrawal reaction and adult horsesdisplay only 1.9% of aggression directed up the hierarchy.Therefore, a stable group hierarchy has been reported to bean important factor in prevention of kick and bite injuries(Furst et al., 2006; Knubben et al., 2008). Actual scientificdata are scarce and further work is needed.

In free ranging groups, aggressive displays are mostlikely directed towards the newcomer in a group(Rutberg, 1990; Rutberg and Greenberg, 1990; Wellsand Goldschmidt-Rothschild, 1979). Young transitionalfemales have been reported to suffer from higherinjury/blemish grades than long-time resident haremmares (Grogan and McDonnell, 2005). Arnold and Grassia(1982) observed in a group of domestic horses that thehorses that received the most of aggression were all horsesintroduced in the paddock later than the rest of the group.A horse’s dominance rank tends to be inversely correlatedto its length of residency in the herd (e.g. Heitor et al., 2006;McDonnell and Haviland, 1995). When adult stallions wereintroduced with young colts, Tilson et al. (1988) observedan increase of offensive behaviours (threats, bites, kicks)from the stallions, which generally decreased or even van-ished after 1 h.

Artificially inducing familiarity by exposing residenthorses to newcomers in adjacent boxes may reduce aggres-sive interactions (Hartmann et al., 2009), although morestudies are needed here as the very limited time ofpre-exposure (5 min in this study) may have led to under-estimate the potential relevance of such a familiarization.

Introducing the unfamiliar animal with two rather than oneresident does not seem to change the aggressive behaviourand this introduction does not seem to modify the fre-quency of social interactions in the group (Hartmann et al.,

224 C. Fureix et al. / Applied Animal Behaviour Science 138 (2012) 216– 228

Fig. 1. Variations in individual’s status function of group composition. Three studies were performed (in 1993: period 1, in 1994: period 2 and in 2001:period 3) where we followed several horses (in bold) and two especially (in bold and grey) in different group compositions. For each study, here are thes of avoidf roup obt here hos ther tha

2bh

3

ota

ociograms based on (a) the number of aggression given, (b) the numbereeding tests. Agonistic interactions were noted ad libitum during focal or gests were performed in an outdoor limited enclosure (10 m diameter) wtability of an individual’s status may depend upon group composition ra

011). Further work is clearly needed on all these aspectsut prior familiarization seems to be a promising line toelp reducing aggression towards newcomers.

.3.4. Developmental issues

As mentioned in part 2, the developmental trajectory

f young horses in natural conditions involves the acquisi-ion of social skills throughout a development that implies

diversified social environment (mother but also other

ance received and (c) the time spent eating in the bucket during pairedservation sessions performed in the horses’ usual pastures. Paired feedingrses were confronted to a single bucket with barley. Results show that

n on intrinsic individual characteristics.

adults). The domestic situation is highly different: foalsdevelop in a restricted social environment (no stallions,no older peers), which becomes even more impoverishedas they grow older (same sex and same age groups). Theyundergo human management that involves early handling

and early weaning (both alimentary and social). All theseprocedures may have an impact on the development ofthe young’s social skills and therefore its later abilities toregulate aggressions.

Behavio

C. Fureix et al. / Applied Animal

3.3.4.1. The mother–young bond. As in free managedhorses (Wells and Goldschmidt-Rothschild, 1979), domes-tic mares have a major influence on their offspring,influencing their foals’ dominance rank (Houpt and Wolski,1980; vanDierendonck et al., 1995) and their offspring’saggressiveness (Araba and Crowell-Davis, 1994; Weekset al., 2000). Although genetic influence is possible, it islikely that the mother has a modelling influence on thefoal’s behaviour through her behaviour. Thus, the dam alsoinfluences the foal’s choice of preferred associate: foalstend to associate with the offspring of their mother’s pre-ferred partner (Weeks et al., 2000). Human interferenceswith the mare-foal bond are numerous and of course a char-acteristic of the domestic situation. Human interventionsare especially crucial at two time periods (events): birthand weaning.

It has become popular to handle foals at birth, with thebelief that it would create a long-term memory of humans’“dominance” or partnership or of objects being harmless,until scientific studies converged to show little or no sucheffects (e.g. Simpson, 2002; Williams et al., 2002). A morerecent study has revealed that this so-called “imprintingprocedure” (handling the foal intensively on the groundfor the first hour after birth) had a series of negativeconsequences, one of them being an unsecure attachmentto their dam (Bowlby, 1969). These foals tended to stayclose to the dam, had less social interactions with theirpeers and reacted very strongly to weaning. When laterobserved at 2 years, these young horses exhibited socialwithdrawal, but also a higher frequency of aggressivebehaviour towards peers as compared to control horses(Henry et al., 2009). Early experience, even limited (seealso Hausberger et al., 2007), may therefore have longterm consequences on a horse’s social profile.

Human intervention is also most influential at wean-ing. Under domestic conditions, weaning can differ fromthe natural process in several important ways: it is oftenabrupt and tends to take place earlier, typically between4 and 6 months of age when the foal is still in close rela-tion with its dam (for a review: Crowell-Davis and Weeks,2005). Weaning of foals results not only in the breakingof the mare-foal bond and deprivation of maternal care,but feeding practices, housing and social environment maybe deeply altered. Under such conditions, weaning resultsin high levels of stress (Waran et al., 2008) and increasedaggressiveness may be visible for several days after wean-ing.

Environmental conditions, either physical or social, dohave a strong influence on stress reactions of foals. Thehighest reactions are observed for foals weaned singlyin a box. When paired with a peer, foals exhibit lessstress reactions but aggressions do arise and can lead toinjuries (Hoffmann et al., 1995). Other practices, such asthe progressive retrieval of mares from the group (Hollandet al., 1996), keeping foals at pasture in groups (Heleskiet al., 2002) or still more so introducing adults withthe weanlings (Henry et al., in press) all lead to low-

ered expressions of stress, including lowered levels ofaggression. Space, dispersed resources (grass) and socialdiversity may help foals go through this forced and arti-ficial separation from the dam without developing higher

ur Science 138 (2012) 216– 228 225

levels of aggressiveness. Since foals, before weaning, asso-ciate with their dam’s preferred partner’s offspring, it mayalso be interesting to observe groups and preserve bondswhen preparing weaning groups (Waran et al., 2008).

3.3.4.2. The importance of social diversity. As other socialspecies, horses need social experience in order to developappropriate social skills (Waran et al., 2008). Anecdotalreports mention that motherless foals or foals raised onlywith their dam may lack the snapping behaviour (whichcould be a mechanism involved in regulation of aggressiontowards young horses, see part 2). They also often mentionthat these same foals may lack appropriate social skills andbe either excessively withdrawn or aggressive when placedin groups. As mares are tolerant towards their young, youngmay then not learn from her how to respect social rules andadult’s dominance, a general feature in natural conditions(e.g. Bourjade et al., 2009a).

Christensen et al. (2002a) have showed that young stal-lions deprived of social contact during their developmentexhibited more agonistic behaviours towards other youngstallions when regrouped at the age of two years than peersraised in a group all along. They may need to learn not onlyto produce appropriate social signals but also to producethem in an appropriate context. It is likely that keepingstallions with the family, a very rare practice nowadays,would further help the foal’s social development, as familystallions tend to interact a lot with their offspring in fieldsituations (see part 2).

In most farms, young horses are kept in same-age andsame-sex groups. Observations of such groups of 1 and 2year-old horses revealed that these animals had restrictedbehavioural repertoires, little social cohesion but high fre-quencies of agonistic interactions (Bourjade et al., 2008).The introduction of two unknown/unrelated adults in suchgroups resulted in an enlarged behavioural repertoire, inparticular the emergence of adult-like behaviours, a highersocial cohesion between the young animals and an increaseof affiliative behaviours, associated with a decrease ofaggressiveness (Bourjade et al., 2008). Interestingly, snap-ping was only observed after the arrival of adults.

Many questions still need to be answered and furtherresearch is required. Given the importance of the adult-young ratio on free ranging young animals, further researchshould investigate what would be the ideal proportion ofadults in social groups, whether the adults’ sex or age areimportant and whether they should or should not be famil-iar to the foals. In any case, the importance of adults inpromoting the use of appropriate skills, regulating aggres-sion and promoting social cohesion has been demonstratedenough, both in field and domestic conditions, to deserveimmediate consideration in breeding facilities. Experimen-tal research is also needed on the potential appeasingfunction of snapping.

4. Conclusion: towards better practices

Given the natural propensity of horses, as a socialspecies, to have a regulation of aggression in groups,the higher level of aggressiveness that can sometimes beobserved under given management conditions (see part

2 Behavio

3e“m

sasmlamSmantaobfealsweAmaref

dowrtw

ttalrpaar

alaaiaihgs

26 C. Fureix et al. / Applied Animal

) seems clearly related to the conditions offered. This isspecially the case if interactions become injurious, whichmay indicate deficits in stable design, space allowance andanagement” (Furst et al., 2006).Practices based on a better knowledge of horses’ natural

ocial life are not necessarily time consuming or difficult topply. Some inappropriate situations may rather be a con-equence of insufficient knowledge. One example is stallionanagement, where the predominant belief is that stal-

ions are naturally aggressive towards other horses. As theyre generally kept singly and only occasional attempts areade to put them in a group, the belief becomes reality. . .

tallions raised in familial or bachelor groups develop nor-al dominance hierarchies, overall low aggression levels

nd may be very protective of their foals. Of course, inon breeding contexts (i.e. no mating necessity), separa-ion from mares may be a necessity as they would then, as

limited resource, be a source of aggressiveness. Breedersr riders may benefit from keeping stallions in families orachelor groups, which in turn would increase their wel-are and make them more manageable (see also Riverat al., 2002; Sondergaard and Halekoh, 2003; Sondergaardnd Ladewig, 2004). Group stability should be enhanced,eading to more peaceful interactions. Although this mayeem a constraint for riding centers, keeping horses whichork regularly as a group outside work may also favour

asier management of horses and riders during lessons.s hierarchy and familiarity are established outside work,uch less attention has to be given to the regulation of

ggression at work, which may lower the attention to theiders’ action and therefore the horses’ welfare (Lesimplet al., 2010). This in turn may also be a source of aggressionrom horses . . . to humans (Fureix et al., 2010).

Keeping horses in paddocks or pastures may make aifference. Grazing is a crucial element that may lowerverall social interactions, including aggressions. An easyay of promoting such an effect in paddock is to provide

oughage while the group stays in the paddock, ensuring athe same time appropriate time budgets and physiologicalell-being.

Food is clearly a central element of a horse’s life andherefore a source of competition. Attention has to be giveno apparatus that limit food distribution to one animalt a time, while simple devices, such as a partition or aine above buckets may enable subordinates to eat withouteceiving aggressions (Houpt and Wolski, 1980). Favouringastures over paddocks when possible, ensuring roughagevailability, multiplying the sources of food and ensuring

reasonable density are interesting ways of lowering theisks of aggressions.

Data are still missing on the ideal group compositionnd little is known about the impact of group size but it isikely that a limited size (4–6 in feral families) promotes

more harmonious social life. Preferred partnershipsre a major feature of horses’ sociability, and observingndividual characteristics may also help ensuring anppropriate group composition, which may differ accord-

ng to the horses’ use. Individual observations of adultorses’ affinities and behaviour may help determiningroup composition: for a same horse, the level of aggres-iveness differs according to the individual encountered,

ur Science 138 (2012) 216– 228

which suggests an adjustment to each other’s behaviouralresponses (Hartmann et al., 2009). Observation of affinities,spatial proximities and agonistic interactions (throughrepeated scan sampling of inter-individual distances, focalsampling of positive/aggressive social interactions) maythus reveal social partners that are better to keep togetherwithin a group, indicate withdrawn animals that may bekept in smaller groups or animals that do not get alongover a longer time period. Group composition appearsmore a question of individual social characteristics than ofgender or age.

Individual behaviour may vary over time and across sit-uation as it is strongly influenced by life conditions. Analtered welfare may be associated with a higher aggressive-ness, as mentioned for instance in Benhajali et al. (2008)’sstudy. Horses may well develop tendencies for general-ized aggressiveness when in poor welfare conditions, asriding school horses were found to generalize negativebehaviours from familiar to unfamiliar persons (Fureixet al., 2009) but also from other horses to humans, bothbeing correlated (Fureix et al., unpublished data). Moreconvincing still is the finding that aggressiveness towardshumans was higher in horses that suffered from vertebraldisorders, probably leading to chronic pain (Fureix et al.,2010). Attention to social and non social management pro-moting welfare is crucial and group management wouldcertainly be easier (e.g. lower risk of injury) if horses allbenefitted from good welfare conditions. The opportunityfor social contact is crucial, keeping in mind that socialbehaviour is both a source and consequence of welfareconditions.

In this context, developmental issues are essential anddecreasing human interferences at some crucial pointswould be an easy way of diminishing some aggressionproblems. Choosing appropriate parents and especially themother, as an important social model, may be interestingwhile ensuring better conditions of weaning seems cru-cial. The important and durable influence of mares on theiroffspring has been clearly demonstrated in experimentalstudies where they were used as a mediator for favouringhuman–foal relationships (Henry et al., 2005). Protectivemothers induced distrust in foals: tendency to behaveaggressively may therefore be transmitted the same way.However, attention and interest should also be given tothe presence of unrelated adults in groups of young at allages. They are essential for the development of appropri-ate social skills. Introducing (socially experienced) adultsin groups of weanlings or yearlings appears as an easy wayto ensure more aggression regulation in young animals.Diversifying the social environment of young horses maybe an easy future development in breeding facilities.

There are broad lines for which studies converge andthat seem to be reachable goals on a daily basis for a farmor a riding school. Many questions still remain to be inves-tigated in more details and further studies should involvework on the limit in the grouping intervals that make agroup stable or unstable, the appropriated group size and

density, breed differences, the appropriated adult – youngratio amongst others. Future studies would undoubtedlyreveal the whole set of factors that are involved in theregulation of aggression in domestic horses.

Behavio

C. Fureix et al. / Applied Animal

Conflict of interest

None declared.

Acknowledgements

The authors are grateful to V. Joinel, I. Tremblay, G.Trou and A. Ouvrard for their help on data collection,and to Dr Ann Cloarec for correcting the English of thismanuscript. Our research group is indebted to the COST“Haras Nationaux” and the Caisse Centrale de la MutualitéSociale Agricole for their financial supports.

References

Araba, B.D., Crowell-Davis, S.L., 1994. Dominance relationships andaggression of foals (Equus caballus). Appl. Anim. Behav. Sci. 41,1–25.

Archer, J., 1970. Effects of population density on behaviour in rodents. In:Crook, J.H. (Ed.), Social Behaviour in Birds and Mammals. AcademicPress, New York.

Arnold, G.W., Grassia, A., 1982. Ethogram of agonistic behaviour for Thor-oughbred horses. Appl. Anim. Ethol. 8, 5–25.

Bekoff, M., 1977. Social communication in canids, evidence for the evolu-tion of a stereotyped mammalian display. Science 197, 1097–1099.

Benhajali, H., Richard-Yris, M.A., Ezzaouia, M., Charfi, F., Hausberger, M.,2009. Foraging opportunity: a crucial criterion for horse welfare? Ani-mal 3, 1308–1312.

Benhajali, H., Richard-Yris, M.A., Ezzaouia, M., Charfi, F., Hausberger, M.,2010. Reproductive status and stereotypies in breeding mares: a briefreport. Appl. Anim. Behav. Sci. 128, 64–68.

Benhajali, H., Richard-Yris, M.A., Leroux, M., Ezzaouia, M., Charfi, F., Haus-berger, M., 2008. A note on the time budget and social behaviour ofdensely housed horses—a case study in Arab breeding mares. Appl.Anim. Behav. Sci. 112, 196–200.

Berger, J., 1977. Organization systems and dominance in feral horses inthe grand Canyon. Behav. Ecol. Sociobiol. 2, 131–146.

Berger, J., 1986. Wild Horses of the Great Basin. University of Chicago Press,Chicago.

Bolhuis, J.E., Schouten, W.G.P., Schrama, J.W., Wiegant, V.M., 2005. Indi-vidual coping characteristics, aggressiveness and fighting strategiesin pigs. Anim. Behav. 69, 1085–1091.

Bourjade, M., Moulinot, M., Richard-Yris, M.A., Hausberger, M., 2008. Couldadults be used to improve social skills of young horses, Equus caballus?Dev. Psychobiol. 50, 408–417.

Bourjade, M., des Roches, A.D., Hausberger, M., 2009a. Adult–young ratio,a major factor regulating social behaviour of young: a horse study.Plos One 4, e4888.

Bourjade, M., Tatin, L., King, S.R.B., Feh, C., 2009b. Early reproductive suc-cess, preceding bachelor ranks and their behavioural correlates inyoung Przewalski’s stallions. Ethol. Ecol. Evol. 21, 1–14.

Bowlby, J., 1969. Attachement et perte. Collection Le fil rouge Presses,Universitaires de France.

Boyd, L., 1988. Time budgets of adult Przewalski horses—effects ofsex, reproductive status and enclosure. Appl. Anim. Behav. Sci. 21,19–39.

Boyd, L., Houpt, K.A., 1994. Przewalski’s Horse, the History and Biology ofan Endangered Species. State University of New York Press, New York.

Boyd, L., Keiper, R., 2005. Behavioural ecology of feral horses. In: Mills,D., McDonnell, S. (Eds.), The Domestic Horse. The Evolution, Develop-ment and Management of its Behaviour. Cambridge University Press,Cambridge, pp. 55–82.

Christensen, J.W., Ladewig, J., Sondergaard, E., Malmkvist, J., 2002a. Effectsof individual versus group stabling on social behaviour in domesticstallions. Appl. Anim. Behav. Sci. 75, 233–248.

Christensen, J.W., Zharkikh, T., Ladewig, J., Yasinetskaya, N., 2002b. Socialbehaviour in stallion groups (Equus przewalskii and Equus caballus)kept under natural and domestic conditions. Appl. Anim. Behav. Sci.76, 11–20.

Clutton-Brock, T.H., Greenwood, P.J., Powell, R.P., 1976. Ranks and rela-

tionships in Highland ponies and Highland cows. Zeitschrift FurTierpsychologie—J. Comp. Ethol. 41, 202–216.

Cooper, J.J., McDonald, L., Mills, D.S., 2000. The effect of increasing visualhorizons on stereotypic weaving: implications for the social housingof stabled horses. Appl. Anim. Behav. Sci. 69, 67–83.

ur Science 138 (2012) 216– 228 227

Crowell-Davis, S., Weeks, J., 2005. Maternal behaviour and mare-foalinteraction. In: Mills, D.S., Mc Donnell, S.M. (Eds.), The Domestic Horse,the Evolution, Development and Management of its Behaviour. Cam-bridge University Press, Cambridge, pp. 126–138.

Crowell-Davis, S.L., 1986. Spatial relations between mares and foals of theWelsh pony (Equus caballus). Anim. Behav. 34, 1007–1015.

Estep, E.Q., Crowell-Davis, S.L., Earl-Costello, S.A., Beatey, S.A., 1993.Changes in the social behavior of drafthorse (Equus caballus)mares coincident with foaling. Appl. Anim. Behav. Sci. 35, 199–231.

Feh, C., 1988. Social-behavior and relationships of Przewalski horses inDutch semi-reserves. Appl. Anim. Behav. Sci. 21, 71–87.

Feh, C., 1990. Long term paternity data in relation to different aspects ofrank for Camargue stallions Equus caballus. Anim. Behav. 40, 995–996.

Feh, C., 1999. Alliances and reproductive success in Camargue stallions.Anim. Behav. 57, 705–713.

Feh, C., 2005. Relationships and communication in socially natural horseherd. In: DS, M., SM, M. (Eds.), The Domestic Horse, The Origins,Development and Management of its Behaviour. Cambridge Univer-sity Press, Cambridge, pp. 83–93.

Feist, J.D., 1971. Behavior of Feral Horses in the Pryor Mountain Wild HorseRange. University of Michigan.

Feist, J.D., McCullough, D.R., 1976. Behaviour patterns and communicationin feral horses. Zeitschrift für Tierzuchtung und Zuchtungsbiologie 4,337–373.

Fureix, C., Jego, P., Sankey, C., Hausberger, M., 2009. How horses (Equuscaballus) see the world: humans as significant objects. Anim. Cogn.12, 643–654.

Fureix, C., Menguy, H., Hausberger, M., 2010. Partners with bad temper:reject ou cure? A study of chronic pain and aggression in horses. PloSOne 5, e12434.

Furst, A., Knubben, J., Kurtz, A., Auer, J., Stauffacher, M., 2006. Group hous-ing of horses: veterinary considerations with a focus on the preventionof bite and kick injuries. Pferdeheilkunde 22, 254–258.

Gibbs, P.G., Cohen, N.D., 2001. Early management of race-bred weanlingsand yearlings on farms. J. Equine Vet. Sci. 21, 279–283.

Grogan, E.H., McDonnell, S.M., 2005. Injuries and blemishes in a semi-feralherd of ponies. J. Equine Vet. Sci. 25, 26–30.

Harris, P., 2005. Nutrition, behaviour and the role of supplements forcalming horses: the veterinarian’s dilemma. Vet. J. 170, 10–11.

Hartmann, E., Christensen, J.W., Keeling, L.J., 2009. Social interactions ofunfamiliar horses during paired encounters: effect of pre-exposureon aggression level and so risk of injury. Appl. Anim. Behav. Sci. 121,214–221.

Hartmann, E., Keeling, L.J., Rundgren, M., 2011. Comparison of 3 methodsfor mixing unfamiliar horses (Equus caballus). J. Vet. Behav.—Clin. Appl.Res. 6, 39–49.

Hausberger, M., Henry, S., Larose, C., Richard-Yris, M.A., 2007. First suck-ling: a crucial event for mother–young attachment? an experimentalstudy in horses (Equus caballus). J. Comp. Psychol. 121, 109–112.

Hausberger, M., Richard-Yris, M.A., 2005. Individual differences in thedomestic horse, origins, development and stability. In: Mills, D.S.,McDonnell, S.M. (Eds.), The Domestic Horse: The Origins, Develop-ment and Management of its Behaviour. Cambridge University Press,Cambridge, pp. 33–52.

Heitor, F., Oom, M.D., Vicente, L., 2006. Social relationships in a herd ofSorraia horses Part 1. Correlates of social dominance and contexts ofaggression. Behav. Process. 73, 170–177.

Heleski, C.R., Shelle, A.C., Nielsen, B.D., Zanella, A.J., 2002. Influence ofhousing on weanling horse behavior and subsequent welfare. Appl.Anim. Behav. Sci. 78, 291–302.

Henry, S., Hemery, D., Richard, M.A., Hausberger, M., 2005. Human–marerelationships and behaviour of foals toward humans. Appl. Anim.Behav. Sci. 93, 341–362.

Henry, S., Richard- Yris, M.A., Tordjman, S., Hausberger, M., 2009. Neonatalhandling affects durably bonding and social development. PLoS One4, e5216.

Henry, S., Zanella, A.J., Sankey, C., Richard-Yris, M.A., Marko, A., Bled, F.,Hausberger, M. Unrelated adults may be used to alleviate weaningstress in domestic foals (Equus caballus). Physiol. Behav., in press.

Hoffmann, R.M., Kronfeld, D.S., Holland, J.L., Greiwecrandell, K.M., 1995.Preweaning diet and stall weaning method influences on stressresponse in foals. J. Anim. Sci. 73, 2922–2930.

Holland, J.L., Kronfeld, D.S., Hoffman, R.M., Griewe-Crandell, K.M., Boyd,

T.L., Cooper, W.L., Harris, P.A., 1996. Weaning stress is affected bynutrition and weaning methods. Pferdeheilkunde 12, 257–260.

Houpt, K.A., Keiper, R., 1982. The position of the stallion in the equinedominance hierarchy of feral and domestic ponies. J. Anim. Sci. 54,945–950.

2 Behavio

H

H

J

J

K

K

K

K

K

K

K

L

L

L

L

L

M

M

M

M

M

M

N

P

P

R

28 C. Fureix et al. / Applied Animal

oupt, K.A., Wolski, T.R., 1980. Stability of equine hierarchies and theprevention of dominance related aggression. Equine Vet. J. 12, 15–18.

oupt, K.A., Wolski, T.R., Welkovitz, N., 1978. The effect of manage-ment practices on horse behaviour. In: Proceedings of the 1st WorldCongress on Ethology as Applied to Zootechnics, Madrid, pp. 27–35.

orgensen, G.H.M., Borsheim, L., Mejdell, C.M., Sondergaard, E., Boe, K.E.,2009. Grouping horses according to gender—effects on aggression,spacing and injuries. Appl. Anim. Behav. Sci. 120, 94–99.

orgensen, G.H.M., Liestol, S.H.O., Boe, K.E., 2011. Effects of enrichmentitems on activity and social interactions in domestic horses (Equuscaballus). Appl. Anim. Behav. Sci. 129, 100–110.

eiper, R., 1986. Social structure. In: Crowell-Davis, S.L., Houpt, K. (Eds.),Behavior, Veterinary Clinics of North America: Equine Practices.

eiper, R., 1976. Social organization of feral ponies. Proc. PennsylvaniaAcad. Sci. 50, 69–70.

eiper, R., Sambraus, H.H., 1986. The stability of equine dominance hier-archies and the effects of kinship, proximity and foaling status on thehierarchy rank. Appl. Anim. Behav. 16, 121–130.

iley-Worthington, M., 1976. The tail of ungulates, canids, and felids withparticular reference to their causation and function as displays. Behav.LVI, 69–115.

ing, S.R.B., 2002. Home range and habitat use of free ranging Przewalskihorses at Hustai National Park, Mongolia. Appl. Anim. Behav. Sci. 78,130-113.

lingel, H., 1982. Social organization of feral horses. J. Reprod. Fert. Suppl.32, 89–95.

nubben, J.M., Furst, A., Gygax, L., Stauffacher, M., 2008. Bite and kickinjuries in horses: prevalence, risk factors and prevention. Equine Vet.J. 40, 219–223.

emasson, A., Boutin, A., Boivin, S., Blois-Heulin, C., Hausberger, M., 2009.Horse (Equus caballus) whinnies: a source of social information. Anim.Cogn. 12, 693–704.

esimple, C., Fureix, C., Menguy, H., Hausberger, M., 2010. Human DirectActions May Alter Animal Welfare, a Study on Horses (Equus caballus).Plos One 5 (4), e10257.

inklater, W.L., 2000. Adaptive explanation in socio-ecology: lessons fromthe Equidae. Biol. Rev. 75, 1–20.

inklater, W.L., Cameron, E.Z., 2000. Tests for cooperative behaviourbetween stallions. Anim. Behav. 60, 731–743.

inklater, W.L., Cameron, E.Z., Minot, E.O., Stafford, K.J., 1999. Stallionharassment and the mating system of horses. Anim. Behav. 58,295–306.

cAfee, L.M., Mills, D.S., Cooper, J.J., 2002. The use of mirrors for the controlof stereotypic weaving behaviour in the stabled horse. Appl. Anim.Behav. Sci. 78, 159–173.

cDonnell, S.M., Haviland, J.C.S., 1995. Agonistic ethogram of the equidbachelor band. Appl. Anim. Behav. Sci. 43, 147–188.

cDonnell, S.M., Poulin, A., 2002. Equid play ethogram. Appl. Anim. Behav.Sci. 78, 263–290.

iller, R., 1981. Male aggression, dominance and breeding behaviour inred desert feral horses. Z. Tierpsychol. 57, 340–351.

onard, A.M., Duncan, P., Boy, V., 1996. The proximate mechanisms ofnatal dispersal in female horses. Behaviour 133, 1095–1124.

ontgomery, G.G., 1957. Some aspects of the sociality of the domestichorse. Trans. Kansas Acad. Sci. 60, 419–424.

icol, C.J., Davidson, H.P.D., Harris, P.A., Waters, A.J., Wilson, A.D., 2002.Study of crib-biting and gastric inflammation and ulceration in younghorses. Vet. Rec. 151, 658–662.

oisbleau, M., Fritz, H., Valeix, M., Perroi, P., Dalloyau, S., Lambrechts,M.M., 2006. Social dominance correlates and family status in win-tering dark-bellied brent geese, Branta bernicla bernicla. Anim. Behav.71, 1351–1358.

roops, L., McComb, K., Reby, D., 2009. Cross modal individual recognitionin domestic horses (Equus caballus). Proc. Natl. Acad. Sci. U.S.A. 106,

947–951.

ivera, E., Benjamin, S., Nielsen, B., Shelle, J., Zanella, A.J., 2002. Behavioraland physiological responses of horses to initial training: the compar-ison between pastured versus stalled horses. Appl. Anim. Behav. Sci.78, 235–252.

ur Science 138 (2012) 216– 228

Rutberg, A.T., 1990. Inter-group transfer in Assateague pony mares. Anim.Behav. 40, 945–952.

Rutberg, A.T., Greenberg, S.A., 1990. Dominance, aggression frequenciesand modes of aggressive competition in feral pony mares. Anim.Behav. 40, 322–331.

Salter, R.E., Hudson, R.J., 1982. Social organization of feral horses in west-ern canada. Appl. Anim. Ethol. 8, 207–223.

Sands, J., Creel, S., 2004. Social dominance, aggression and faecal glucocor-ticoid levels in a wild population of wolves, Canis lupus. Anim. Behav.67, 387–396.

Sebeok, T.A., 1979. How Animals Communicate. Indiana University Press,Bloomington.

Sigurjonsdottir, H., Van Dierendonck, M.C., Snorrason, S., Thorhallsdottir,A.G., 2003. Social relationships in a group of horses without a maturestallion. Behaviour 140, 783–804.

Simpson, B., 2002. Neonatal foal handling. Appl. Anim. Behav. Sci. 78,303–317.

Sondergaard, E., Halekoh, U., 2003. Young horses’ reactions to humans inrelation to handling and social environment. Appl. Anim. Behav. Sci.84, 265–280.

Sondergaard, E., Ladewig, J., 2004. Group housing exerts a positive effecton the behaviour of young horses during training. Appl. Anim. Behav.Sci. 87, 105–118.

Stebbins, M.C., 1974. Social Organization in Free Ranging AppaloosaHorses. Pocatello, p. 306.

Stevens, E.F., 1988. Contests between bands of feral horses for access tofresh water—the residents wins. Anim. Behav. 36, 1851–1853.

Tilson, R.L., Sweeny, K.A., Allen Binczik, G., Reindl, N.J., 1988. Buddies andbullies: social structure of a bachelor group of Przewalski horses. Appl.Anim. Behav. Sci. 21, 169–185.

Tyler, S.J., 1969. The Behaviour of a Population of New Forest Ponies. Cam-bridge University Press, Cambridge.

Tyler, S.J., 1972. The behaviour and social organization of the New Forestponies. Anim. Behav. Monogr. 5, 85–196.

vanDierendonck, M.C., Bandi, N., Batdorj, D., Dugerlham, S., Munkhtsog, B.,1996. Behavioural observations of reintroduced Takhi or Przewalskihorses (Equus ferus przewalskii) in Mongolia. Appl. Anim. Behav. Sci.50, 95–114.

vanDierendonck, M.C., De Vries, H., Schilder, M.B.H., 1995. An analysis ofdominance, its behavioral parameters and possible determinants in aherd of icelandic horses in captivity. Netherlands J. Zool. 45, 362–385.

vanDierendonck, M.C., Sigurjonsdottir, H., Colenbrander, B., Thorhallsdot-tir, A.G., 2004. Differences in social behaviour between late pregnant,post-partum and barren mares in a herd of Icelandic horses. Appl.Anim. Behav. Sci. 89, 283–297.

Verbeek, M.E.M., De Goede, P., Drent, P.J., Wiepkema, P.R., 1999. Individualbehavioural characteristics and dominance in aviary groups of greattits. Behaviour 136, 23–48.

Waran, N.K., Clarke, N., Famworth, M., 2008. The effects of weaning on thedomestic horse (Equus caballus). Appl. Anim. Behav. Sci. 110, 42–57.

Waring, G., 2003. Horse Behavior, second edition. Noyes Publica-tions/William Andrew Publishing, Norwich, New York.

Weeks, J.W., Crowell-Davis, S.L., Caudle, A.B., Heusner, G.L., 2000. Aggres-sion and social spacing in light horse (Equus caballus) mares and foals.Appl. Anim. Behav. Sci. 68, 319–337.

Wells, S.M., Goldschmidt-Rothschild, B., 1979. Social behavior and rela-tionships in a herd of Camargue horses. Z. Tierpsychol. 49, 363–380.

Welsh, D.A., 1975. Population, Behavioural and Grazing Ecology of theHorses of Sable Island, Nova Scotia. Dalhousie University, Halifax,Nova Scotia.

West, M.J., King, A.P., White, D.J., 2003. The case for developmental ecol-ogy. Anim. Behav. 66, 617–622.

Williams, J.L., Friend, T.H., Collins, M.N., Toscano, M.J., Sisto-Burt, A., Nevill,C.H., 2002. The effects of early training sessions on the reactions of

foals at 1, 2 and 3 month of age. Appl. Anim. Behav. Sci. 77, 105–114.

Zeitler-Feicht, M.H., Streit, S., Dempfle, L., 2010. Automatic feeding sys-tems for horses in group housing systems with regard to animalwelfare. Part 1: feeding stalls versus automatic feeding systems. Tier-aerztliche Praxis Ausgabe Grosstiere Nutztiere 38, 363–370.