Abstract: The aim of this study was to investigate the social behaviour and spacing between horses when grouped according to gender in homogeneous- or heterogeneous groups, and to see if the amount and severity of aggression differed in relation to gender composition.
A total of 66 horses divided in six batches were used. In each batch, horses were allotted into one mare group, one gelding group and one mixed gender group, with most groups consisting of three or four animals. Three batches were performed on Icelandic horses and three batches on riding horses and -ponies. After 4-6 weeks of acclimatisation, a trained observer recorded all social interactions using direct, continuous observation one hour in the morning and one hour in the afternoon for three consecutive days. In addition, the nearest neighbour of each horse was recorded using instantaneous sampling every 10 minutes. The horses were inspected for injuries before grouping, day one after grouping and after 4-6 weeks. Furthermore, a novel object test and a handling test (taking horse out of group) were conducted after the behavioural observations were completed.
No significant effect of gender composition was found on social interactions, spacing or injuries. Most of the aggressive interactions recorded were threats, and not aggressive interactions involving physical contact. Aggression seemed to increase with decreased space per animal and more horses per feeding place. Very few injuries were found and most were superficial. Horses showed a low degree of fearfulness in the novel object test and most horses were easy to handle in the “horse out of group test”.
In conclusion, gender composition does not seem to have any effect on aggression level, spacing or injuries in social groups of horses.

Abstract: Objective: In the context of the ongoing discussion about keeping horses in individual stable boxes vs. in herds the lack of relaxation of the horse as a flight animal is an argument often put forward against individual housing. The long-term objective of our investigations is to determine the sleep phases in various housing systems in order to find a substantiated answer to that issue. For that purpose, the sleep waves measured by EEG have to be defined beforehand and allocated to the individual stages of sleep. The experiments described here are intended to support this effort. The objective is to establish a method which by means of a portable polysomnograph allows to capture the sleeping behaviour of horses for the purpose of defining the individual stages of sleep. It was investigated which stages of sleep horses undergo, and to what extent they may be comparable to those of humans.
Animals/materials/methods: Given the high technical effort involved, somnographic examination of large animals has so far been difficult and mostly required the animals to be sedated. Meanwhile, however, instruments such as the Somnoscreen by Somnomedics have become available. This is a completely portable polysomnograph with up to 28 channels and wireless online signal transmission and synchronous video transmission to a PC.
Using this instrument, the sleep profile of 5 Icelandic ponies and 10 horses (different race) aged 5-10 years was recorded and evaluated for 4 or 5 nights per horse. The following parameters were assessed: EEG (electroencephalogram), EOG (electrooculogram), EMG (electromyogram), ECG (electrocardiogram), thoracic and abdominal breathing motions as well as identification of the body posture. Synchronous video recordings were made. EEG recordings were obtained through gold-coated disk electrodes with long flexible cables, applied and secured to the scalp.
Results: As with humans, various stages of sleep can also be defined for horses using the above methods of recording. The waking condition is characterised by alpha waves, which just like in humans are within a range of 8-12 Hz. Typical REM phases as in humans were also detected, although not only when stretched completely on their side, as has hitherto always been described, but also when lying on their chest. Phases of deep sleep (stage 4) can also be measured, with the animals mostly in a standing position. The multi-stage human sleeping pattern, which is made up of 4-6 repeat phases of sleep (waking stage eyes open – waking stage eyes closed – REM phase – stage 1 – stage 2 – stage 3 – stage 4 – return to REM phase etc.) was found to be similar in horses in individual sequences. However, the sleep phases are shorter and more frequently interrupted by waking phases.
Conclusions: Horses are flight animals, which is why they have to be “on eye” in every situation so as to be able to flee in the face of danger. In a natural herd lying positions are only assumed if one or more members watch over the herd. In some publications the REM phase is treated as equivalent to the deep sleep phase. Although the REM phase is a phase of total muscle relaxation it is at the same time the dream phase and due to the high frequencies and the low amplitudes in the EOG resembles Stage I. This means that the sleeping horse can be awakened very quickly from this REM phase so as to be able to react to any dangerous situation. It therefore makes sense for the horses to assume a lying position during REM phases as the muscles are relaxed, yet a waking condition can be reached very quickly. A standing position seems to be preferred during deep sleep phases, where waking takes rather long, so that at least the position will not have to be changed. Whether the sleeping behaviour changes depending on age and race has yet to be investigated.

Abstract: The management of the domestic horse often requires them to be kept in isolation from con-specifics. Installing a picture of a horse (generally head and neck view) with a view to providing surrogate companionship has been shown to reduce the negative impact of this isolation. This study aimed firstly to compare the spontaneous response of horses (N=10) to a 2-D image of a horse’s face (FP) with their response to a comparable abstract 2-D image (AP). Secondly, the spontaneous response of horses (N=20) to a 2-D image of a horse’s face with the ears forward (PFP positive) was compared with the response to a 2-D image of a horse’s face with the ears back (NFP negative). The posters were A1 sized and displayed in the horse’s own stable. In study 1, one poster was displayed for 5 minutes and the horse’s behaviour video-recorded. This was removed and the second poster was displayed for 5 minutes and the behaviour video-recorded. FP was displayed first for 5 of the horses and AP displayed first for the other 5. The video footage was observed and the behaviour of the horses and number of times they touched the poster recorded. For the purpose of identifying the area of the poster that was touched by the horse it was divided into 4 equal quarters (TL, TR, BL, BR). In FP the nose of the horse in the 2-D image was located in BL, eyes and ears in TL, chest and lower neck in BR and upper neck in TR. In AP each area contained similar but unique abstract patterns of comparable colour to FP. Differences in behaviour were found according to which poster was displayed. FP was touched significantly more than AP (p=0.001) and was looked at more often (p=0.008). With FP the horses spent significantly longer with their ears forward (p=0.008) and licking and chewing (p=0.016). When the number of touches per poster area was compared (FP and AP) a significant difference was found in the number of times that BL (nose) and BR (chest/lower neck) were touched (p=0.011). Both areas were touched more frequently on FP, with BL being touched the most. In study 2 the same experimental protocol was used to compare responses to positive (PFP) or negative (NFP) 2-D images of a horse’s face (same horse in both PFP and NFP). Again, differences in behaviour were found in response to the two posters. PFP was touched significantly more than NFP (p=0.002) and on both posters the area BL (nose) was touched more frequently than the other areas (PFP: p=0.02, NFP: p=0.01). More ears back behaviour (p<0.001) and more ear locked on behaviour (p=0.008) was shown with NFP. The results of these studies indicate that horses can recognize 2-D images as con-specifics as well as responding to differences in facial expression. There is now the potential for further investigation into the importance of other visual cues in recognition and social interaction as well as the application of findings to enhance equine welfare.

Abstract: Some behaviours remain a mystery and their determinants are still uncertain. Here we propose a critical interpretation of two puzzling behaviours, i.e. yawning and adult play, recorded in large samples of domestic horses kept in sub-optimal conditions (e.g. time-restricted feeding practices, social isolation). Yawning is involved in behavioural state changes, especially in quiet contexts of motor relaxation, but seems also to be triggered by stress and emotional contexts. Thus, we investigated specific correlates of yawning and stereotypic behaviours in two large samples of working horses (n1 = 87, n2 = 59). A clear co-occurrences of yawning and stereotypic behaviours appeared: stereotypic horses yawned more than the non-stereotypic horses (Fisher test, p < 0.02), yawning increased at the same time periods as stereotypic behaviours did (Wilcoxon tests, p < 0.05 to 0.001) and yawning frequency was even sometimes positively correlated with stereotypic behaviour frequencies (Spearman correlations, p < 0.005). Play, more typical of young stages in healthy individuals, occurs rarely at adult stages but then more often in captive/domestic animals (which can be challenging to welfare) than in animals living in natural conditions. Thus, we tested the hypothesis that adult play may reflect altered welfare states in horses, where, as in several species, play rarely occurs at adult stages in natural conditions. We observed the behaviour (in particular social play) of 29 adult riding school horses during occasional outings in a paddock and measured several stress indicators (health-related, behavioural and physiological) when these horses were in their home individual boxes. Results revealed that the number of horses and rates of adult play appeared very high compared to field report data and that most stress indicators measured differed between “players” and “non-players”, revealing that most “playful” animals were more prone to suffer from chronic stress contrarily to “non-playful” horses (Fisher, Chi-square, Mann-Whitney and Spearman correlation tests, p < 0.05 to 0.001). Altogether, these results show that yawning and adult play can both reflect altered welfare states in horses, highlighting that these puzzling behaviours should not be considered, as currently thought, as reliable indicators of good welfare states and / or expression of positive emotions.

Abstract: In polygynous species with high reproductive skew in males, mothers often show differential investment between sons and daughters; however consistent sex differential investment has not been found by previous studies in horses. We investigated sex differences in mother-offspring relationships in nutritionally independent sub-adult semi-feral Carneddau Welsh mountain ponies Equus ferus caballus. Mothers and their sub-adult sons had consistently closer relationships than mothers and daughters. Stronger affiliative bonds between mothers and sons were quantified by their maintenance of closer proximity, higher rates of affiliative interactions and more frequent suckling bouts. These measures of affiliation were temporally associated with higher aggression levels directed towards sub-adults by other band members, particularly stallions. We suggest that aggression may serve as the proximate mechanism driving male dispersal in feral horses and that the stronger mother-son bond signals an attempt to delay their dispersal, highlighting conflict within the band concerning dispersal timing. Since males become increasingly central within the band over time, with mature stallions requiring excellent social skills in order to both acquire and keep a band of mares, we propose that delaying colts' dispersal allows for further development of these skills in a relatively safe environment. This additional investment is expected to maximise their reproductive success. This study illustrates how social network theory can be used to quantify individuals' social experiences, contributing to a greater understanding of the evolution of group living. It also gives us further insight into the mechanisms underlying dispersal in wild and semi-wild horse populations and how conflict often arises when individual needs differ.