Non-human animals often produce different types of vocalisations in negative and positive contexts (i.e. different valence), similar to humans, in which crying is associated with negative emotions and laughter is associated with positive ones. However, some types of vocalisations (e.g. contact calls, human speech) can be produced in both negative and positive contexts, and changes in valence are only accompanied by slight structural differences. Although such acoustically graded signals associated with opposite valence have been highlighted in some species, it is not known if conspecifics discriminate them, and if contagion of emotional valence occurs as a result. We tested whether domestic horses perceive, and are affected by, the emotional valence of whinnies produced by both familiar and unfamiliar conspecifics. We measured physiological and behavioural reactions to whinnies recorded during emotionally negative (social separation) and positive (social reunion) situations. We show that horses perceive acoustic cues to both valence and familiarity present in whinnies. They reacted differently (respiration rate, head movements, height of the head and latency to respond) to separation and reunion whinnies when produced by familiar, but not unfamiliar individuals. They were also more emotionally aroused (shorter inter-pulse intervals and higher locomotion) when hearing unfamiliar compared to familiar whinnies. In addition, the acoustic parameters of separation and reunion whinnies affected the physiology and behaviour of conspecifics in a continuous way. However, we did not find clear evidence for contagion of emotional valence. Horses are thus able to perceive changes linked to emotional valence within a given vocalisation type, similar to perception of affective prosody in humans. Whinnies produced in either separation or reunion situations seem to constitute acoustically graded variants with distinct functions, enabling horses to increase their apparent vocal repertoire size.

Mushroom is a unique coat color phenotype in Shetland Ponies characterized by the dilution of the chestnut coat color to a sepia tone and is hypothesized to be a recessive trait. A genome wide association study (GWAS), utilizing the Affymetrix 670K array (MNEc670k) and a single locus mixed linear model analysis (EMMAX), identified a locus on ECA7 for further investigation (P_corrected = 2.08 × 10⁻¹⁰). This locus contained a 3 Mb run of homozygosity in the 12 mushroom ponies tested. Analysis of high throughput Illumina sequencing data from one mushroom Shetland pony compared to 87 genomes from horses of various breeds, uncovered a frameshift variant, p.Asp201fs, in the MFSD12 gene encoding the major facilitator superfamily domain containing 12 protein. This variant was perfectly concordant with phenotype in 96 Shetland Ponies (P = 1.15 × 10⁻²²), was identified in the closely related Miniature Horse for which the mushroom phenotype is suspected to occur (f_mu = 0.02), and was absent in 252 individuals from seven additional breeds not reported to have the mushroom phenotype. MFSD12 is highly expressed in melanocytes and variants in this gene in humans, mice, and dogs impact pigmentation. Given the role of MFSD12 in melanogenesis, we propose that p.Asp201fs is causal for the dilution observed in mushroom ponies.

In domestic conditions, adult stallions are mostly housed individually in internal stables to reduce the risk of injuries during social interactions. Social deprivation in horses results in physiological stress and behavioural problems. The aim of this study was to test the “social box” (SB), which allows closer physical contact between neighbouring horses. Eight pairs of stallions (n = 16) were filmed over a 24 h period in the SB and in their usual box stables, “conventional boxes” (CB), which strongly restrict tactile contact. The effect of housing in the SB on behaviour and the occurrence and characteristics of injuries was investigated. The total duration of active social interactions was significantly higher in the SB than in the CB (51.1 vs. 4.9 min, p < 0.0001). Positive interactions accounted for about 71% of the total duration of interactions in SB and CB stabling. The stallions interacted significantly more often in the SB than in the CB (113.5 vs. 23.8 social interaction sequences over 24 h, p < 0.0001). No grievous injuries were recorded. The social box appears to be a suitable solution to give adult stallions the possibility of having physical interactions. Therefore, it can be considered a substantial environmental enrichment for singly housed horses.

In order to improve the housing conditions of stallions in individual boxes, we tested a so-called “social box” allowing increased physical contact between neighbouring horses. This study investigated whether housing stallions in social boxes changes the number of social interactions during carriage driving. We hypothesised that the stay in social boxes would decrease the number of unwanted social interactions between stallions when driven in pairs. Eight Franches-Montagnes breeding stallions were observed when driven in pairs with a “neutral” stallion housed in a so-called “conventional box”, strongly limiting physical contact. They were driven on a standardised route over the course of four days before, during, and after being housed in social boxes. The type and frequency of behaviours of the pairs and the interventions of the groom and the driver during the test drives were assessed live and using video recordings. Results from linear mixed-effect models show that unwanted social interactions decreased during and after the stallions were housed in the social box (p < 0.001). Stallions’ interactions also decreased over the four days (p < 0.01), suggesting a habituation to the test conditions by learning not to interact, or by subtly settling dominance. The social box tended to decrease unwanted social behaviours of stallions driven in pairs and could therefore be used as an environmental enrichment for horses.

Finding feeding strategies that meet horses’ needs without compromising health is essential for optimising welfare, particularly in group housing, where limited hay availability increases aggression and injury risks. Recently, two strategies have emerged: portioning daily intake into smaller, frequent meals using time-controlled hay racks, or slowing intake with “slow-feeders.” However, the effects of such management practices on horse behaviour remain underexplored. We conducted a cross-over study with 18 mares divided into four groups to compare three feeding strategies: “traditional” (3 of 2-hours meals during daylight, TD), “portioned” (6 of 1-hour meals spread over 24h, PO) and “slow-feeding” (ad libitum hay covered by a net, SF). Each treatment included 3 weeks of habituation and 2 weeks of data collection. We continuously recorded social interactions for 15 hours and noted the position and activity of all horses every 15 minutes. We also recorded injuries periodically and measured the lying behaviour using accelerometers. We analysed the effects of treatment on agonistic and affiliative behaviour within groups using generalised mixed model and selected the best model using AIC. We used the same procedure for the injuries and lying behaviour at the individual level. Horses in SF exhibited activity time budgets resembling natural conditions, while TD and PO resulted in time budgets similar to box-stall systems, despite the loose-housing system. Surprisingly, our results suggest that PO may be more frustrating for the horses than TD. Indeed, there was no significant reduction of agonistic behaviours during feeding times in PO compared to TD and lying behaviour tended to be impaired in PO (−11.3 min/day, 95% CI [−25.8; 3.1]) compared to SF and TD (37.5 min/day on average). In our study, portioning into smaller, more frequent meals did not reduce the stress in horses. This highlights the need for further research on portioning strategies to find optimal feeding management. In addition, slow-feeding was a more suitable feeding strategy for horses than portioning. However, more research is required to substantiate the initial findings on the efficacy of ad libitum slow-feeding on the horse’s health and behaviour.

Horse sports face public scrutiny, particularly regarding concerns about horse welfare. Understanding how the physical positioning of the horse's head and neck while being ridden affects behavior is crucial for improving welfare standards. This study aimed to investigate the influence of head–neck position (HNP) in ridden horses–specifically, the vertical and poll angles–on conflict behavior (CB). Furthermore, it aimed to identify a critical threshold at which HNP significantly affects behavioral outcomes. Elite dressage horses were observed during warm-ups and tests at competitions, and horses presented in educational videos were analyzed to examine the relationship between HNP and CB. A total of 191 rides were analyzed, HNP parameters (angle to the vertical of nasal plane, poll angle, and shoulder angle) and CB indicators (e.g., unusual oral behaviors and tail swishing) were recorded over 3-minute intervals using Observer XT (Noldus). Data were analyzed using general linear mixed-effects models, with CB (sum of all behaviors) as the response variable and horse-rider-ID as a random effect. Both poll and vertical angles had significant negative effects on CB, indicating that horses exhibited more CB when these angles decreased (p < 0.001, df = 139.64/137.28, respectively). Shoulder angle also had a significant impact on CB in one model (p = 0.014, df = 136.61). A vertical angle of -7.5° behind the vertical was identified as the cut-off value associated with an increase in CB. Stallions exhibited more CB than mares and geldings (p < 0.001, df = 73.16). The models demonstrated good fit, with consistently non-significant effects of age, breed, and bit type. Optimizing HNP during riding is crucial for enhancing behavioral well-being and the overall welfare of dressage horses in competitive settings.