Abstract: A short-term study was made of 2 groups of Przewalski horses, a bachelor group of 4 juvenile stallions in Ooij Polder and a harem group of 1 stallion and 4 mares. All social interactions were recorded and the nearest and farthest neighbour was noted. Correspondence analysis was used to determine what parameters determined the relationships among the horses. There was a linear hierarchy among the bachelor stallions. The dominant stallion of the group was also the oldest. The hierarchy was not linear in the harem group, and the 3-year-old stallion was subordinate to the 5-year-old mares. He was also most likely to be farthest from other horses. The mares of the same age, who had also arrived in the park at the same time, tended to be one another's nearest neighbours. The frequency of aggression is higher among Przewalski horses than among domestic horses of similar ages. Correspondence analysis revealed that head-threats and other forms of aggression accounted for more of the variance in the data than any other behaviour, but submission, play and social interactions also contributed.

Abstract: Social interactions were recorded in two bands of free-ranging Przewalski horses living on large (greater than 30 ha) pastures in the Netherlands. The average number of aggressive interactions per hour was 8.86 at Lelystad and 10.36 at Noorderheide. The most common aggressive interactions were lower intensity, lower cost displacements (17.2% of all aggressive acts at Lelystad, 13.2% at Noorderheide), threats to bite (42.3% and 40.7%, respectively) and threats to kick (15.4% and 23.9%, respectively). Analysis of aggression revealed that a clear, linear dominance hierarchy was present in each band. For each band there was a positive and highly significant correlation between the age of a horse and its rank in the hierarchy. In each band, the stallion was not the highest ranked horse. Non-agonistic behaviors exceeded the number of agonistic interactions (1253 vs. 558 for Lelystad; 1257 vs. 995 at Noorderheide). There was a negative correlation between the rank of a horse in the dominance hierarchy and the number of non-agonistic behaviors displayed. The group displaying the highest number of non-agonistic interactions were foals (48.9% of total non-agonistic behaviors at Lelystad; 51.1% at Noorderheide). The non-agonistic/agonistic ratio was greater than 1 for yearlings and the band stallion, as was also the case for foals, but was less than 1 for males.

Abstract: The Minnesota Zoological Garden established a bachelor (i.e. all-male) group of 8 Przewalski horses (Equus przewalskii) in 1984. Possible methods for mixing males emerged through preliminary introductions and review of the relevant literature. These included: (1) grouping colts (<=3 years old) after removing them from their natal bands; (2) introducing adult stallions to colts in spacious, controlled environments to reduce the risk of combat-related injury or death associated with largely unpredictable adult temperament; (3) introducing more than 1 animal at a time to established groups. Based on the frequency and orientation of agonistic interactions observed among the bachelors, a linear dominance hierarchy was discerned among the 4 adult stallions; all 4 colts were determined to be of equally low or seemingly non-ranking status. Social rank among the adults was most influenced by individual temperament, and contention for rank was limited almost exclusively to the 3 lower-ranking stallions; the alpha male participated in 73% of all agonistic interactions observed, but was identified as the recipient in less than 3% of these interactions. Intra-band associations were detected in the bachelor group. The most prevalent of these was among the 4 colts, found together in over 70% of the observations. Other affinities commonly observed were between the colts (as a sub-group) and different stallions. The dominant stallion, the colts' most frequent companion, actively discouraged the 3 lower-ranking stallions from associating with the colts. Further, the colts were subjected to significantly fewer agonistic behaviors than the non-alpha stallions (per colt=1.05+/-0.24 h-1; per non-alpha stallion=2.01+/-0.04 h-1). These findings support the hypothesis developed by other workers for ungulates that young bachelors are sheltered from aggression by association with an alpha male. With one exception, adult associations were rarely noted. Although combat-related injuries (superficial wounds and temporary lameness) were common among the bachelors, few required medical attention. No permanent disabilities or deaths have occurred since the group's inception. The existence of social organization and intra-band associations in the bachelor group support the suggestion that such groups were once a natural part of this species' social milieu. Bachelor groups are therefore recommended as an ethologically sound, space- and cost-effective means of managing “surplus” (i.e. non-harem) male Przewalski horses.

Abstract: Sixteen Quarter Horses were randomly divided into two groups after sorting by age and sex. After a 10-day preconditioning period, each animal was scored for emotionality and trainability. Each group then completed a series of learning tasks in a modified T-maze for 20 consecutive days. Group P/D was initially tested on a simple place-learning task, while Group D/P was trained in a visual discrimination task. The groups were tested alternately on the two tasks with 10-day extinction periods between each task. Upon reaching a criterion of 11 of 12 correct responses (the last 8 responses consecutive), a horse was retired for the day. If this criterion was not attained, the horse completed 20 trials. Learning occurred at a faster rate on the discrimination tasks compared to the gradual learning curves observed on place tasks. Animals learned more rapidly and reached higher levels of performance as the series of tasks progressed. Trainability and emotionality scores tended to predict the final level of learning achieved. Correlations of performance ranks within emotionality and training groups were higher between tasks of the same type than between the different tasks. Rank correlations between odd and even days on each task indicated that the within-group rankings were more consistent on the discrimination task than on the place task.