Effects of feeding management and group composition on agonistic behaviour of group-housed horses

• Individually separated feeding places decreased agonistic behaviour during feeding.
• Horses fed with ‘hay nets’ showed the lowest probabilities for aggressive behaviour.
• Aggressive behaviour decreased with increasing duration of hay availability.

Under natural conditions, horses spend 12–18 h of a day with synchronous grazing, but for stabled horses, the amount of forage and thus duration of feed intake usually are restricted. In group-housed horses, therefore, feeding times are often situations with increased levels of agonistic behaviour. Our aim was to evaluate how forage provision, feeding duration and group composition influence agonistic behaviour. The study was conducted on 50 groups of 4–21 adult horses (n = 390) housed in group housing systems. Each group was observed for 30 min before and for the first 30 min after a hay feeding. Feeding systems were ‘floor’, ‘fodder rack’, ‘feed fence’, ‘net’, ‘feed stall with contact’, ‘feed stall no contact’ or a ‘combination’ of these. Hay was provided mostly 2–3 times per day but the duration of hay availability varied from 1.5–24 h per day, whereas straw was mostly available ad libitum  . Results showed that the proportion of horses showing agonistic behaviour was generally higher before than during feeding. Aggressive behaviour – bearing the risk of injuries by physical contact – was highest in ‘floor’ and lowest in ‘net’ (χ62 = 13.0, p = 0.043). Aggressive behaviour decreased substantially during feeding with an increasing duration of hay availability (χ12 = 7.1, p = 0.008). Threatening behaviour was highest in ‘floor’, ‘fodder rack’ and ‘feed fence’ and lowest in ‘feed stalls’ (χ62 = 25.4, p < 0.001). Threatening behaviour increased with an increasing proportion of mares in the group (χ12 = 5.0, p = 0.025). Displacements occurred most in ‘floor’ and least in ‘feed stalls’ (χ62 = 37.37, p < 0.001). Additionally, displacements decreased substantially during feeding with an increasing duration of straw availability (χ12 = 4.4, p = 0.035). Furthermore, horses fed at time-bound feeding times tended to show more locomotor activity before feeding than horses with no time-bound feeding times, whereas no difference was found during feeding (F1,321 = 3.62, p = 0.058). In conclusion, feeding places that are either individually separated by partitions or distant from each other lead to a reduced occurrence of agonistic behaviour, most likely because horses in such systems are able to maintain their perceived individual distances. Furthermore, it is highly recommended – from an ethological point of view – to provide not only straw but also hay over an unlimited period, regardless of the feeding system.