Incidence and global distribution of equine housing research

Horse housing studies have steadily increased in number over the last thirty years, with 56% of studies in this review having been carried out in the last decade. As the studies represent a significant and diverse body of work, it is considered timely to review what has been reported to date, identify gaps in knowledge and suggest future research directions.

The global distribution of studies identified in this review were conducted in western nations and largely show alignment with countries that use horses for leisure and competition purposes with some application in agricultural ranch work. A global record of horse numbers by country is held by the Food and Agriculture Organisation of the United Nations (2024). Whilst data are not submitted by all countries, and a significant amount is based on estimate, the latest extensive records for 2022 show the top six most populous countries, accounting for 56% of recorded horse numbers, are the USA (10.3M; 16.6%), Mexico (6.4M; 10.3%) and Brazil (5.8M; 9.4%), Mongolia (4.8M; 7.7%), Kazakhstan (3.9M; 6.2%) and China (3.7M; 5.9%). There are also records for 36 African countries, representing 11.5% (7.2M) of the total horse numbers. Despite their large populations of horses, many of these countries do not appear to be conducting research into housing design. It may be that horses are stabled less or not at all in these areas however some of these countries do stable horses and account for a large proportion of global horse numbers. Based on this, conducting future studies in horse housing that better represent the global population of horses, with a view to having a greater impact on worldwide horse welfare, should be considered.

Housing design assessed via behavioural observation and hormonal measures

All but three of the studies in this review reported a positive impact on behaviour, physiology or both parameters, when horses are housed in groups or a more open stable design. The studies that reported a positive impact of single housing all cite improvements in behavioural parameters and could be open to alternative interpretation. It is worth noting that some studies report no change in behaviour or physiology when comparing single- and group-housed horses.

Positive findings from single-housed horses included Søndergaard and Halekoh (Reference Søndergaard and Halekoh2003) who were exploring the effects of housing young horses on horse-human interaction, with a reduced reaction to human interaction interpreted as positive. They reported decreased latency to approach a human in individually housed horses compared with horses housed in groups of three when tested at both 12 and 24 months of age. Single-housed horses approached sooner and were more easily approached by a human than group-housed horses where no effect of handling was observed. Single-housed horses expressed less restless behaviour, more explorative behaviour, and less vocalisation than group-housed horses. Although interpreted as a positive outcome of single housing for the aim of this study, it should also be considered that horses housed as individuals could be seeking interaction, regardless of whether this is conspecific or human, due to limited social contact, whereas the group-housed horses may have a reduced need to seek out this contact due to having companions housed with them. This raises the question of a possible challenge between efforts to increase the welfare standards of equine housing without impacting horse training, particularly with young horses, and warrants further investigation. It is to be hoped that any training benefit derived from individual isolation would be compensated by positive impacts of free movement and conspecific interaction on the temperament, welfare and emotional state of more socially housed horses, as indicated by Lesimple et al. (Reference Lesimple, Fureix, LeScolan, Richard-Yris and Hausberger2011).

A second study reported increased fear response to a novel object test for group-housed broodmares in comparison with stabled stallions, however the housing system did not affect the responses to the human approach test (Sanchez et al. 2020). The authors suggested that an increased fear response in group-living horses is to be expected as a species-specific behaviour of a herd animal and the ability to express their natural behaviour is a positive indicator of welfare, although the ability to make comparisons between the two different groups of horses here is limited.

The remaining study reported negative consequences of weaning foals in pairs for both behavioural and physiological indicators (Hoffman et al. Reference Hoffman, Kronfeld, Holland and Greiwe-Crandell1995) suggesting that the pairing of foals for weaning is not ideal and alternate methods and housing should be considered in this circumstance.

Positive results were frequently reported in studies that included outside access, regardless of its nature or use. This suggests the importance of natural open space, and the opportunity for horses to express locomotory behaviours. Increase in space was also reported to improve post- exercise recovery in the long term (Connysson et al. Reference Connysson, Rhodin and Jansson2019) and reduce time spent displaying stereotypic and undesirable behaviours, despite no difference in hair or plasma cortisol between study groups (Arena et al. Reference Arena, Marliani, Sabioni, Gabai, Bucci and Accorsi2021). Results have also shown higher cognitive abilities in horses when access to pasture is provided (Lesimple et al. Reference Lesimple, Fureix, LeScolan, Richard-Yris and Hausberger2011). During weaning, separation of mares into individual stables resulted in an increase in negative physical and vocalisation behaviours, when compared to mares placed in a paddock, however both groups of mares had similar elevated cortisol concentrations in response to the weaning event (Falomo et al. Reference Falomo, Gabai, Franchini, Poltronieri, Rossi and Normando2020).

Similarly, in a separate study, weanlings expressed higher vocalisation, decreased eating and aberrant behaviours when housed individually compared with those paddock housed in groups (Heleski et al. Reference Heleski, Shelle, Nielsen and Zanella2002). There was no significant difference in cortisol metabolites measured between the groups, but levels were elevated four weeks post-weaning in both groups. The four-year prospective study of Waters et al. (Reference Waters, Nicol and French2010) on 225 weanlings developed these findings further, identifying that weaning foals by confinement was associated with an increased incidence of abnormal stereotypic and redirected behaviours when compared with paddock-weaned foals.

Whilst it appears from the sample of publications in this review that group housing has clear benefits on both behavioural, physiological and environmental parameters, it has also been discussed that the novelty of an altered environment could be responsible for such changes, particularly in studies that show positive behavioural responses (Cooper et al. Reference Cooper, McDonald and Mills2000). Longer-term studies would give greater insight into habituation and adaptation when horse housing is changed and the accommodation of an adjustment period in experimental design may allow for reported behavioural and physiological changes resulting more from a change of routine rather than housing design.

Positive indicators of welfare are becoming more popular as a measure of well-being, including recumbency and rest behaviour (Giannetto et al. Reference Giannetto, Giudice, Acri and Piccione2018; Stomp et al. Reference Stomp, Leroux, Cellier, Henry, Lemasson and Hausberger2018). Horses are a prey species and are therefore vulnerable when they are recumbent. This suggests that if a domestic horse spends time lying down in the absence of illness or injury, they are comfortable in their environment. However there have been reports of isolated horses making use of time undisturbed by conspecifics to be recumbent in stables (Erber et al. Reference Erber, Wulf, Aurich, Rose-Meierhöfer, Hoffmann, von Lewinski, Möstl and Aurich2013), so care is needed when interpreting results of studies investigating rest behaviour.

Lying behaviours in relation to box size have been investigated. Raabymagle and Ladewig (Reference Raabymagle and Ladewig2006) reported that recumbency was higher in a large stable (13.1 m² × 15 m²) compared to a small stable (4.7 m² × 6 m²) with significantly more time spent in sternal recumbency in the large stable, especially if horses were housed in the small boxes first then switched to the larger housing. This is further supported by research from Burla et al. (Reference Burla, Rufener, Buchmann, Gygax, Patt and Hillmann2017) and Kjellberg et al. (Reference Kjellberg, Sassner and Yngvesson2022), which also identified that the larger the lying space given, the longer the horses spent in recumbency. Burla et al. (Reference Burla, Rufener, Buchmann, Gygax, Patt and Hillmann2017) also discovered that recumbency was significantly reduced when no bedding was available, so suitable surface provision for recumbency appears to be another factor that could affect the choice to lie down.

Measurement of locomotory behaviour has been included in several housing studies. Horses are a free-ranging species, therefore ability to roam is important and housing that offers this opportunity may benefit well-being. Walking time and distance increased in horses that spent part of their day housed inside, compared to those housed outside full time (Schmitz et al. Reference Schmitz, Tonn, A-K and Isselstein2020). It was suggested that those who were kept in stables used their opportunity to move more when outside, therefore creating rebound locomotion. Houpt et al. (Reference Houpt, Houpt, Johnson, Erb and Yeon2001) reported that pregnant mares displayed an increase in cortisol that decreased after two weeks when moved from pasture to stables, but there were no signs of adrenal exhaustion or hypertrophy. Rebound locomotory behaviours were also higher in horses who had limited free movement once turned out, as in Schmitz et al. (Reference Schmitz, Tonn, A-K and Isselstein2020).

Giannetto et al. (Reference Giannetto, Fazio, Assenza and Piccione2016) reported that Thoroughbreds showed significantly more movement in box stables compared to pasture, suggesting an increase in stress, manifesting as box walking. Nature of movement therefore needs to be considered rather than just the movement time or distance in isolation, as not all movement can be regarded as good. In this case, Thoroughbreds were compared with Standardbred and Italian Saddle horses who did not express this increase in locomotion when confined, also highlighting a likely breed and temperament effect that also warrants consideration in experimental design.

Increased cortisol and alert behaviour were reported in horses housed individually for the first time (Harewood & McGowan Reference Harewood and McGowan2005; Erber et al. Reference Erber, Wulf, Aurich, Rose-Meierhöfer, Hoffmann, von Lewinski, Möstl and Aurich2013) and those who had extended periods on full-time pasture before returning to stabling (Ruet et al. Reference Ruet, Arnould, Levray, Lemarchand, Mach, Moisan, Foury, Briant and Lansade2020). This suggests that abrupt changes in management, especially from extensive to confined housing, may be unsettling for horses. It also highlights the importance of maintaining access to turn-out or outside accommodation should housing or shelter design need to be changed, and that some time on pasture can have similar effects to full-time turn-out when it is not available (Popescu et al. Reference Popescu, Lazar, Borda, Blaga Petrean and Mitrănescu2022). Results from surveys supported this, reporting that daily access to pasture and free movement decreased the chance of stereotypic or abnormal behaviours (Bachmann et al. Reference Bachmann, Audige and Stauffacher2003; Leme et al. 2014).

To compensate for lack of turn-out, paired and group stabling is utilised. Compared to single housing, group housing has reported positive results (Søndergaard & Ladewig Reference Søndergaard and Ladewig2004; Yarnell et al. Reference Yarnell, Hall, Royle and Walker2015) however, results for paired accommodation vary. The importance of appropriate pair bonds as well as the situation should not be overlooked as evidence suggests higher aggression can be seen when weaning foals in pairs compared to single housing (Hoffman et al. Reference Hoffman, Kronfeld, Holland and Greiwe-Crandell1995). However, when stabling horses for the first time, Visser et al. (Reference Visser, Ellis and van Reenen2008) reported that stress-related behaviours like neighing, pawing, nibbling and snorting were all displayed more frequently in individually housed horses when compared with paired-housed horses. At the end of the study, 67% of the individually housed horses were seen performing one or more stereotypies. Additionally, cortisol and adrenocorticotropic hormone (ACTH) responses to a corticotrophin releasing factor challenge test were lower for horses in individually housed boxes, suggesting depression in the socially isolated animals caused by a desensitisation of the HPA axis in response to stress-induced elevations in ACTH and cortisol.

No difference in salivary cortisol levels were recorded between group-housed horses in a paddock and horses in individual stabling with access to an individual paddock or daily exercise (Aurich et al. Reference Aurich, Wulf, Ille, Erber, von Lewinski, Palme and Aurich2015). This suggests the importance of outdoor exposure as part of a stabled horse’s routine and that opportunities for movement were enough in this instance in housed horses without constant turn-out as no increase in salivary cortisol was observed when stabled.

It is worth considering that once established, STBs are only reversable within a short time-frame, after which horses may have adequate welfare but be psychologically unable to stop performing the behaviours (Vinken et al. Reference Vinken, Lidfors, Loberg, Lundberg, Lind, Jonsson, Ghirlanda and Enquist2023). Presence or absence of such behaviours may therefore not be the most appropriate method to assess equine housing unless it is a new method of housing under scrutiny and horses studied do not display signs of STB prior to being introduced to it. Yarnell et al. (Reference Yarnell, Hall, Royle and Walker2015) compared four housing designs which were individual stables with no contact between conspecifics, individual stables with visual and limited tactile contact between conspecifics, paired housing and full turn-out in a group. Results showed presence of STBs in the most restricted housing that was not observed in any of the other housing designs. An increase in faecal cortisol as the level of isolation increased was also reported, and horses were more challenging to handle as social opportunities decreased, directly impacting safety of the horse and human handler.

Lesimple et al. (Reference Lesimple, Gautier, Benhajali, Rochais, Lunel, Bensaïd, Khalloufi, Henry and Hausberger2019) found grilled stables situated inside a barn increased sleeping behaviours compared to stables without a grill, allowing horses to put their head over the door. In addition, all horses demonstrated snort behaviours (classed as a novel method of positive welfare assessment in the study) compared to those in stables with a view of the outside only, where only 42% snorted. There was no difference for time spent conducting STB, but weaving was observed and most prevalent in the more open stables. In comparison, studies have shown that stables inside a barn also reduced the risk of abnormal oral behaviours however, horses being able to see others and not touch them increased frustration behaviour two-fold (Hockenhull & Creighton Reference Hockenhull and Creighton2014). Although the results of stereotypical behaviour are contradictory between the two studies, it is worth mentioning that Lesimple et al. (Reference Lesimple, Gautier, Benhajali, Rochais, Lunel, Bensaïd, Khalloufi, Henry and Hausberger2019) had the same observer for all behavioural recordings, but Hockenhull and Creighton (Reference Hockenhull and Creighton2014) used a survey. In the second part of the Lesimple et al. (Reference Lesimple, Gautier, Benhajali, Rochais, Lunel, Bensaïd, Khalloufi, Henry and Hausberger2019) study, mares were stabled for the first time. Grilled stables resulted in an increase in foraging and resting, standing and recumbency, whereas stables facing outside increased vigilance behaviour.

Cooper et al. (Reference Cooper, McDonald and Mills2000) reported similar findings, when horses only had access to front and back door openings without a grilled window between stables. There was an increase in standing alert behaviours, but when grills were open between adjacent stables, weaving that had been observed ceased. A similar result was seen when the front door and grills were open, suggesting horses are more interested in interactions with neighbouring horses as opposed to the outside views. Allowing the front door and side grills to be open created similar architecture to that of the open treatment in the study of Lesimple et al. (Reference Lesimple, Gautier, Benhajali, Rochais, Lunel, Bensaïd, Khalloufi, Henry and Hausberger2019) yet different results were recorded. When the front door was open during the study of Cooper et al. (Reference Cooper, McDonald and Mills2000), a view of a courtyard was visible. However, in the study of Lesimple et al. (Reference Lesimple, Gautier, Benhajali, Rochais, Lunel, Bensaïd, Khalloufi, Henry and Hausberger2019), the stables opened to an arena where the horses worked. Perhaps the difference between the views lead to the difference in findings, supporting the theory that when horses are shown an open space they would like to reach but are unable to get to, they can potentially experience an increase in stress.

One stable design that increases contact opportunity between neighbouring horses is termed the ‘social-box’, which consists of two vertical bars, 2.5 m high and 0.3 m apart in a wall of a stable, allowing for interaction between adjacent horses. This design has been trialled with unfamiliar stallions driven in pairs to see if the increased opportunity of housed social interaction reduced unwanted interactions when working. Unwanted interactions during work were monitored prior to and post housing in adjacent social boxes (Gmel et al. Reference Gmel, Zollinger, Wyss, Bachmann and Freymond2022). Unwanted interactions during work reduced significantly when stallion pairs were housed in neighbouring social boxes, interpreted as a more compliant response to driver instruction during work, and continued to reduce throughout the treatment period, possibly due to habituation or adaptation.

Human interaction tests were used to assess the effect of bars being placed at the front of a stable, allowing stallions to eat haylage from the corridor and provide visual contact with conspecifics when foraging, a behaviour that would naturally occur in the wild (Søndergaard & Halekoh Reference Søndergaard and Halekoh2003). Horses in traditional housing were seen to approach the human quicker, be more approachable, and were less affected by being placed in an unfamiliar environment, however the bond between horses was stronger in the social setting. This implies that single housing may enhance the human-horse relationship and social housing has the benefit of improved social bonds between horses (Søndergaard & Halekoh Reference Søndergaard and Halekoh2003). As already mentioned, the impacts of such findings related to housing on the behaviour of horses in training or work, particularly when separated from herd mates when housed in a social setting, warrants further investigation.

Natural housing of horses

Natural housing that aims to replicate the species-specific natural habitat, encouraging movement and social interaction whilst providing control and convenience for the owner, are well established across Europe and growing in popularity in the UK. The number of studies comparing these systems with other housing methods is limited, but those that exist tend to carry out data collection over an extended period. Marlinani et al. (2020) measured the day-time activity budgets of horses in what they termed an “ethological housing system” and demonstrated that they were similar to those reported for free-roaming feral horses. Placci et al. (Reference Placci, Sabioni, Gabai, Mondo and Borghetti2019) reported positive physiological indicators of welfare in naturally managed horses, with lower cortisol to dehydroepiandrosterone (DHEA) ratio, suggesting positive long-term effects of this housing style. However, horses housed overnight with day-time turn-out showed no difference in cortisol levels when compared to natural management, it being proposed that this could be due to the sleep quality being improved when horses were inside at night thus improving well-being (Mazzola et al. Reference Mazzola, Colombani, Pizzamiglio, Cannas, Palestrini, Dalla Costa, Gazzonis, Bionda and Crepaldi2021). The most technologically advanced forms of natural housing, referred to as active systems, incorporate functional elements, such as individual, transponder-controlled automatic hay and concentrate feeders, drinking stations, access-controlled grazing and purpose-built rest areas, separated by distance or track systems to encourage movement. Hildebrandt et al. (2021) assessed the visit frequency for different functional elements, informing their placement in active systems when horse movement stimulation is a desired outcome of the design. Rose-Meierhöfer et al. (Reference Rose-Meierhöfer, Klaer, Ammon, Brunsch and Hoffmann2010) reported that the highest movement level in a paddock was lower than the minimal value of an active design system, which was set up with functional elements distributed to encourage movement. Evidence therefore suggests that natural housing systems provide significant welfare benefits for horses and that the automated and access-controlled elements that are features of active systems could serve to address some of the negative owner perceptions and management challenges sometimes associated with the group housing of horses (Hartmann et al. Reference Hartmann, Søndergaard and Keeling2012). The increased horse movement that can be stimulated as a result of active system design could also provide benefits in terms of exercise, exercise recovery, body condition management, and other health-related benefits. More studies into these systems are therefore needed.

Housing assessed via health-related parameters

Endoscope results revealed greater accumulation of tracheal mucus in horses housed in enclosed stables (Millerick-May et al. Reference Millerick-May, Karmaus, Derksen, Berthold, Holcombe and Robinson2012) and horses that were stabled experienced an increase in exhaled breath condensate pH and gaseous ammonia compared with those out in a paddock (Whittaker et al. Reference Whittaker, Love, Parkin, Duz and Hughes2010). Housing design can influence air quality; runner stables, which are designed for young horses or mares with foals, exceeded allowable bacteria levels by three times compared to individual stables (Grzyb et al. Reference Grzyb, Podstawski and Bulski2022), with results attributed to cleanliness level. Runner stables were cleaned once a month, in comparison with individual boxes that were cleaned once a week. It has been previously reported that total airborne particulate level was correlated with number of horses housed in the same barn (Rosenthal et al. Reference Rosenthal, Gruntman and Couetil2006).

Junkkari et al. (Reference Junkkari, Simojoki, Heiskanen, Pelkonen, Tulamo, Sankari and Mykkänen2017) reported no difference in the occurrence of respiratory disease in weanlings housed in stables or loose-housing systems in cold conditions. They note that the incidence of respiratory conditions was higher in the youngest weanlings regardless of housing type and access to outdoor space. They also noted that Finn horses were better suited to cold climate housing than Standardbreds in terms of maintaining body condition, highlighting the need to consider the broader housing needs of horses in relation to welfare.

A “low dust” stable free of hay and straw achieved lower endotoxin levels compared to a conventional stable with hay and straw, but not lower than a paddock (McGorum et al. Reference McGorum, Ellison and Cullen2010). These results show that air quality cannot be attributed to housing design alone but also to overall management including feeding and bedding systems. The location of stables can also influence particulate matter levels. A slightly open stable, placed next to a racetrack road and car park, and an enclosed stable next to a manure handling building, racetrack road and city road, had a higher particulate matter than an enclosed stable next to a car park and woodland (Millerick-May et al. Reference Millerick-May, Karmaus, Derksen, Berthold, Holcombe and Robinson2011). However, there are many factors that may affect these results, including foot traffic on the yard, forage type and provision method and activity of horses. These results, in part, support those of Berndt et al. (Reference Berndt, Derksen and Robinson2010) who stated certain sources may produce endotoxins, such as manure, which could be another reason for higher levels in the enclosed stable next to manure storage.

Technological developments have led to improved mechanical ventilation, bringing air in from the outside into the centre of stables and distributing it through the ceiling into boxes, resulting in a decrease in CO₂, ammonia and reduced respiratory mucus, but no difference in dust levels. Some results report seasonal variation, possibly because the ventilation is regulated by the indoor temperature of the stables, meaning it runs at a lower capacity during colder temperatures to prevent the stables falling below desired temperatures (Wålinder et al. Reference Wålinder, Riihimäki, Bohlin, Hogstedt, Nordquist, Raine, Pringle and Elfman2011). Future research could investigate combining stall architecture with mechanical ventilation to maintain higher air quality, especially for winter periods when most windows are closed.

Immunological measures have been included in some housing studies with mixed findings. The relationship between CD4:CD8 lymphocytes was significantly higher in pregnant mares living in a reserve than in individual stabling, indicating higher immune cell activity. However, variations in breed or nutrition could have confounded these findings (Krakowski et al. Reference Krakowski, Bartoszek, Krakowska, Stachurska, Piech, Brodzki and Wrona2017). In a more controlled study, it was reported that immune activation may be a result of higher endotoxins being present in the stable environment when compared with pasture (Berndt et al. Reference Berndt, Derksen and Robinson2010). However, reduced immunity could be due to an increase in stress. Relocation to single housing led to longer-lasting changes to immune cells than cortisol, suggesting stress having a stronger or more sustained effect on the immune system than the endocrine system (Schmucker et al. Reference Schmucker, Preisler, Marr, Krüger and Stefanski2022). Marr et al. (Reference Marr, Preisler, Farmer, Stefanski and Krueger2020) reported an increase in cortisol when stabling horses for 48 h. Horses also switched to a predominant left-limb preference in laterality suggesting an increase in sympathetic nervous system activation. However, IgA levels remained unchanged, potentially because the stress was not chronic enough to stimulate any immunological changes. Schmucker et al. (Reference Schmucker, Preisler, Marr, Krüger and Stefanski2022) had an extra step in their methodology which may have increased the stress, by splitting the group into two and then re-grouping before splitting into stables. The disruption of the stable grouping may have influenced stress and therefore the immune system in conjunction with the individual stabling.

Mach et al. (Reference Mach, Lansade, Bars-Cortina, Dhorne-Pollet, Foury, M-P and Ruet2021) reported changes in gut microbiota and positive behavioural responses for horses turned out to pasture for a period of one month and also identified microbiota species that appeared linked with behaviours indicating poor welfare in housed horses. Whilst the authors acknowledged that causality could not be inferred by the study, impacts of environment and/or diet change on gut microbiota composition and the effects of this on behaviour and welfare is an emerging and important area of science that warrants further investigation.

Molinari et al. (Reference Molinari, Basini, Ramoni, Bussolati, Aldigeri, Grolli, Bertini and Quintavalla2020) explored the use of a number of oxidative stress markers as measures of positive or negative welfare in different housing environments and in relation to the presence or absence of stereotypies, but there was no significant difference in the parameters measured. Mal et al. (Reference Mal, Friend, Lay, Vogelsang and Jenkins1991) attempted to identify a number of physiological markers between isolated, confined and pasture-housed mares to assist welfare evaluation. There was a variable age- and temperament-related response to a phytohaemagglutinin (PHA) skin test in the isolated mares, but no other housing or age groups. There was no difference in leukocyte response to the PHA test or adrenal response to a subsequent exogenous adrenocorticotrophin administration between any groups indicating that the housing types tested only elicited the mildest of physiological stress response, or no response at all.

Animal welfare implications

The ways in which horses are housed can have a serious impact on their welfare. A significant and growing body of work now exists that aims to better understand this. This review consolidates the findings of this work and highlights gaps and opportunities to inform the future direction of this important area of horse welfare research, with a focus on encouraging work that has a more global perspective and serves to inform practical improvements that can be made to the ways in which horses are kept.