Abstract: Feral horses are social animals, which have to rely on survival strategies centered on the formation of cohesive social bonds within their bands. Many problems in the husbandry of social animals such as horses, are due to the fact that the limits of their adaptive abilities are exceeded. Evidence suggests that the fundamental social characteristics of domestic horses have remained relatively unchanged. The social structure, social strategies and social interactions were investigated (3 non-consecutive years, 24 hr per day for several weeks) in long term established groups of domestic horses (mares and geldings of all ages) and a few small introduced groups, kept in (semi)natural environments. The general aim was to investigate the social needs of domestic horses. The social life of domestic horses was characterised by long lasting bonds with preferred partners which were established and maintained by allogrooming, play, proximity and dominance behaviours. Bonding partners were mainly found within the same sex-age group, but adult geldings also bonded with sub-adult mares and geldings. Adult mares were clustered in a group, while the other animals formed a second group. Among the adult mares, subgroups according to reproductive state were formed. Individuals regulated their social network by interfering with interactions between other members of the herd, which in itself is complex. An intervention is a behavioural action of one animal that actively interferes with an ongoing interaction between a dyad with the apparent aim of altering that interaction. This was verified by post-hoc analyses of disturbed and undisturbed interactions. Interventions in allogrooming or play were performed significantly more often when at least one member of the initial dyad was a preferred partner of, or familiar to (within the small introduced bands) the intervener. The stronger the preferred association in allogrooming between the intervener and member(s) of the initial dyad, the higher the probability the intervener would displace one initial member and continue allogrooming with the other. Just five behaviours were extracted which reliably reflected the dominance relations among horses. Aggression with the hind quarters was used both offensively and defensively and therefore not suitable as a reliable parameter. Individual dominance relationships were related to social experience. The implications of these findings for horse husbandry were assessed. It is argued that the execution of affiliative behaviours may be rewarding in itself, and therefore always will be a highly motivated behaviour. It is shown that social positive physical interactions (allogrooming, play) with other horses is an ethological need and therefore indispensable in modern husbandry systems. Ethological needs are so important for the animal that husbandry systems that lack the possibilities to execute such behaviours will cause chronic stress. It is concluded that all horses need physical social contact, and that horses, which lack appropriate social learning experiences during ontogeny, may be hampered in their social functioning later in life. Solutions for problems, including dominance problems, in individual social housing and group housing are presented.

Abstract: I develop guidelines for assessing the precision and power of statistical techniques that are frequently used to study nonhuman social systems using observed dyadic associations. Association indexes estimate the proportion of time that two individuals are associated. Binomial approximation and nonparametric bootstrap methods produce similar estimates of the precision of association indexes. For a mid-range (0.4-0.9) association index to have a standard error of less than 0.1 requires about 15 observations of the pair associated, and for it to be less than 0.05, this rises to 50 observations. The coefficient of variation among dyads of the proportion of time that pairs of individuals are actually associated describes social differentiation (S), and this may be estimated from association data using maximum likelihood. With a poorly differentiated population (S~0.2), a data set needs about five observed associations per dyad to achieve a correlation between true and estimated association indexes of r=~0.4. It requires about 10 times as much data to achieve a representation with r=~0.8. Permutation tests usually reject the null hypothesis that individuals have no preferred associates when S2H>5, where H is the mean number of observed associations per individual. Thus most situations require substantial numbers of observations of associations to give useful portrayals of social systems, and sparse association data inform only when social differentiation is high.