Abstract: In order to improve the housing conditions of stallions in individual boxes by offering a possibility to have more social contact, the Swiss national stud farm tested a new box system for horses, allowing increased physical contact with the neighbouring stallion. The aim of this part of the study was to investigate whether this type of housing system (named “social box”) potentially induces a change in stress reactions in stallions compared to conventional boxes. Therefore faecal cortisol metabolite (FCM) concentration was measured as a non-invasive parameter to assess endocrine responses related to this new environment.
Four groups each consisting of eight adult Freiberger breeding stallions were included in the test design. Every stallion spent three weeks in a conventional box and in a social box respectively (cross-over design). The conventional box consisted of a separation wall with a lower opaque part and an upper part with vertical barriers (5 cm between barriers), allowing visual and olfactory contact but strongly limiting tactile contact. The separating wall of the social box consisted of two lateral sections, one part being opaque to the ceiling and the second part consisting of vertical barriers (30 cm between barriers), allowing the horse to have physical contact with its neighbour or to avoid it.
In horses, FCM concentration reflects an average level of circulating cortisol over a period of approximatively 24h. Faecal samples were collected the day following integration in social / conventional boxes, reflecting the potential stress induced by increased social interactions during the integration. In order to asses potential chronical stress, faeces samples were also collected in week one, two and three after the integration into the social / conventional box (in total: 4 samples per horse and housing system). The samples were immediately stored at -20°C until they were analysed. The samples were not analysed in the laboratory until the end of the experiment, therefore the duration of conservation in the freezer varied from 40 to 429 days.
A considerable percentage of data from groups 1 and 2 was below the detection limit (<0.8 ng/g) (Tab. 1). Thus the statistical analysis was conducted with the FCM concentration from groups 3 and 4 (n horses = 16) which contained no values below the detection limit.
Tab. 1: Details about FCM values and storage time for the 4 groups of stallions
Group Storage duration [d] Proportion of data below the detection limit (<0.8 ng/g) Mean [ng/g] Median [ng/g]
Group 1 384-429 55.6 % 2.2 0
Group 2 315-360 25.5 % 5.8 6.3
Group 3 41-79 0.0 % 8.7 8.0
Group 4 40-85 0.0 % 5.8 5.4
Despite the impressive social interactions observed between the stallions directly after being introduced into the social boxes, we did not find any differences in FCM concentration between the stallions being introduced into the conventional box and the social box on the day of integration (social box: n samples = 16, mean±SD: 6.9±4.7 ng/g; conventional box: n samples = 16, mean±SD: 9.0±11.2 ng/g; Wilcoxon signed rank test V = 70, p = 0.94).
Overall the samples taken during integration and in week one, two and three did not show evidence of changes in FCM concentration in either housing system over a longer period of time (social box: n samples = 64, mean±SD: 7.9±6.2 ng/g; conventional box: n samples = 64, mean±SD: 6.6±3.4 ng/g; Linear mixed model (LMM), p = 0.56).
Our results suggest that the possibility of having physical contact with a conspecific does not induce changes in FCM concentration in breeding stallions. The considerable percentage of values below the detection limit in groups 1 and 2 seemed to correlate with the increasing duration of storage before analysis. During the IESM Network Meeting 2015, we would like to discuss possible methodological issues and the possibilities to correctly integrate these low values in the statistical analysis.

Abstract: In ethical discussions, a cost-benefit analysis requires that welfare costs associated with an activity can be reliably estimated and balanced against the potential benefits of the activity to both humans and animals. The current study applies a meta-analysis to the peer-reviewed evidence for costs and benefits of hyperflexion of the neck in horses; a practice that has attracted enormous public and scientific scrutiny over the past 15 years. A literature review identified 55 studies dealing with horses’ head and neck postures. Fourty-two of these studies examined the impact of various postures on equine welfare, for example, by assessing behavior, physiological stress parameters, health or rider-horse interaction. Thirty-five studies examined the impact of various postures on gymnastics (e.g. kinematics, shifts in weight distribution, muscle activity, airway functioning or overall workload). For the meta-analysis a dataset containing information from each of the individual studies was created. Data included information such as type, degree, duration and circumstances of hyperflexion applied in that particular study as well as information on the horses (e.g., sport discipline, level of training, breed) and on the study design (e.g., size of study and experimental or epidemiological research design). The results of the study regarding the impact of hyperflexion on a) welfare and b) gymnastics were coded as positive (1), insignificant or contradictory (0) or negative (-1). The significant majority of studies (88%) concluded that a hyperflexed head and neck posture negatively impacts welfare. Just one study suggested welfare advantages of training in a hyperflexed head and neck posture. An analysis using a generalized linear mixed model to assess the influence of the above factors collated in the dataset revealed that none of these factors significantly influenced the probability of a study to detect negative welfare implications. Thus hyperflexing the neck appears to impair horses’ welfare regardless of, for example, the duration or the way of achieving hyperflexion. A concurrent assessment of the evidence for gymnastic benefits showed that approximately one quarter of studies conclude that there may be benefits, while another quarter of the studies conclude that hyperflexion has detrimental effects on gymnastics. Thus, on the costs-side there is a clear reduction in equine welfare and some undesirable gymnastic effects, as well as likely a compromised profile of the equestrian sports in public. Benefits, on the other hand, include some desirable gymnastic effects, and potentially increased control of the horse for the rider. On balance, it appears that the costs associated with hyperflexion exceed the potential benefits of the activity to both humans and horses.

Abstract: The design of a captive environment should facilitate the expression of the natural behavioural repertoire of the species concerned. The domestic horse (Equus caballus) is a social, herd dwelling species that is often housed in isolation from con-specifics or kept in groups in paddock enclosures. Although the latter allows for social interaction and does not restrict movement per se, it may not promote natural activity patterns or group cohesion. The aim of the current study was to assess the impact of two different paddock designs on the behaviour of a stable group of horses (n=6: 2 mares, 4 geldings).
The paddock designs tested were adjacent and grassed similarly, but configured differently. A central paddock (75 x 75m) designated NT, was surrounded by a track 1.6-5.5m wide, designated T. The horses were turned out in their group into T or NT for 3 consecutive days. Their behaviour was recorded for one hour three times each day (10.00, 13.00, 16.00 hrs). Within each observation period of one hour focal sampling was used, each individual horse being observed for a 10 minute period. They were then moved to the other enclosure type for a further 3 consecutive days, followed by a repeat of each condition. When turned out the horses were fitted with a global positioning system device (Garmin Forerunner 305) to monitor distance travelled and speed for the period 10.00-17.00 hrs. The mean percentage of time spent in each behavioural state (standing alert, standing resting, walk/trot, grazing, lying, social interaction) during the periods observed was calculated. Behaviour during social interactions was classified as either affiliative (approach, follow, friendly contacts, mutual grooming) or agonistic (approach and retreat, bite, chase, head threat).
The horses travelled significantly further in T than in NT (paired samples t-test: t (5) = 11.74, p<0.001) and moved significantly faster (Wilcoxon signed rank test: z = -2.21, p=0.03). See Table 1. When the percentage of time spent in each behavioural state in T and NT was compared some significant differences were found. A significantly higher percentage of time was spent active (walking /trotting) in T than in NT (paired samples t-test: t (5) = 5.74, p=0.002). Standing alert was only recorded in T (paired samples t-test: t (5) = 3.48, p=0.02). A significantly higher percentage of time was spent grazing in NT than in T (paired samples t-test: t (5) = -3.58, p=0.016). Significantly more social interaction occurred in T than in NT (paired samples t-test: t (5) = 5.93, p=0.002). See Figure 1. In T, 91% of social interactions were affiliative and 9% agonistic, whereas in NT 29% were affiliative and 71% agonistic. No difference was found in the percentage of time spent standing resting or lying down in T and NT.
The benefits of housing horses in groups as opposed to individually have been demonstrated in previous studies. In addition to better satisfying the behavioural needs of the horse it has been found that group housed horses adapt more easily to training and display less undesirable behaviour than those housed individually (Rivera et al. 2002; Søndergaard and Ladewig 2004; Visser et al. 2008). However, individual housing is frequently selected by horse owners in preference to group housing to avoid the risk of injury during agonistic encounters. Fureix et al. (2012) suggest that management practices may well contribute to aggressiveness in horses and that the conditions under which we keep horses should be reviewed. The findings of the present study indicate that the design of the enclosure in which groups of horses are kept affects the nature of social interactions. The T paddock design resulted in reduced intra-group aggression. However, this paddock design also reduced the time spent grazing and increased vigilant behaviour. Although the results demonstrate that a paddock system including tracks may facilitate group cohesion and more natural movement patterns, the long-term impact on behaviour and welfare requires further investigation.
References:
Fureix C, Bourjade M, Henry S, Sankey C, Hausberger M. (2012). Exploring aggression regulation in managed groups of horses (Equus caballus). Applied Animal Behaviour Science 138: 216-228.
Rivera E, Benjamin S, Nielsen B, Shelle J, Zanella AJ. (2002). Behavioural and physiological response of horses to initial training: the comparison between pastured versus stalled horses. Applied Animal Behaviour Science 78: 235–252.
Søndergaard E, Ladewig J. (2004). Group housing exerts a positive effect on the behaviour of young horses during training. Applied Animal Behaviour Science 87: 105-118.

Abstract: This study replicated an unreported finding observed in a color perception experiment (Pick, Lovell, Brown, & Dail, 1994) where, after using the method of successive approximations to train a blue-gray discrimination, red-gray trials were initiated without further training. Although a gray choice had never been reinforced, the subject chose gray on the first 20 trials (p < .000001). In the study reported here, a horse was trained to approach a red feed bucket and not a green feed bucket. After the subject mastered the discrimination, a blue bucket was substituted for the previously reinforced red bucket. With double-blind controls in place, the subject chose the unreinforced green bucket on 15 out of the first 20 blue-green trials yielding a binomial p = 0.0148 that this outcome could be due to chance alone. These results are contrary to all behavioristic psychological learning theories, but consistent with prospect theory (Kahneman & Tversky, 1979). Prospect theory predicts that given a choice between two previously unreinforced stimuli, one familiar and the other novel, humans will choose the familiar. It is argued that the bias toward the familiar is the basis to a heuristic that has a genetic origin and should exist in other animals on the phylogenetic scale. The results of this study indicate that the heuristic is available at least as far down the scale as the horse. Conceptual replications using shape stimuli and sound stimuli are in progress.

Abstract: BACKGROUND: As visual attention is an intrinsic part of social relationships, and because relationships are built on a succession of interactions, their establishment involves learning and attention. The emotional, rewarding or punishing, content can modulate selective attention. In horses, the use of positive/negative reinforcement during training determines short and long-term human-horse relationships. In a recent study in horses, where either food or withers' grooming were used as a reward, it appeared that only the food-rewarded horses learned the task and show better relationship with humans. In the present study, we hypothesized that this differential effect of grooming/food rewards on learning performances could be due to attentional processes. Monitoring, gazes and behaviors directed towards the trainer revealed that the use of a food reward (FR) as positive reinforcement increased horses' selective attention towards their trainer. Conversely, horses trained with grooming reward (GR) expressed more inattentive responses and did not show a decrease of “agitated” behavior. However, individual plotting of attention vs. rate of learning performances revealed a complex pattern. Thus, while all FR horses showed a “window” of attention related to faster learning performances, GR horses' pattern followed an almost normal curve where the extreme animals (i.e., highest and lowest attention) had the slowest learning performances. On the other hand, learning was influenced by attention: at the end of training, the more attentive horses had also better learning performances. This study, based on horses, contributes to the general debate on the place of attentional processes at the interface of emotion and cognition and opens new lines of thought about individual sensitivities (only individuals can tell what an appropriate reward is), attentional processes and learning.