Basic Needs in Horses?—A Literature Review

Simple Summary All animals have requirements that are essential for their welfare, and when these basic needs are not met, the animal suffers. In horses, it is claimed that these needs include social contact, social companionship, free movement and access to roughage in the form of grass, hay and/or straw. To validate this claim, this review examines 38 studies that reported on horses’ responses when one or more of these factors are restricted. We categorised the type of responses investigated: (a) Stress (e.g., increased stress hormones), (b) Active (e.g., increased aggression), (c) Passive (e.g., depressive-like behaviour) and (d) Abnormal Behaviour (e.g., stereotypies), and analysed the frequencies with which the investigated responses were shown. Overall, the studies reported that horses did react to restrictions in the described basic needs, especially to combinations of restricted social contact, free movement and access to roughage. The observation of passive responses and the development of abnormal behaviour provided compelling evidence that horses were suffering under these restrictions, and existing abnormal behaviours indicated that they had suffered at some time in the past. We conclude that the literature supports the claim that social contact, free movement and access to roughage are basic needs in horses and need to be taken into consideration to ensure their mental and physical welfare in management and training. Abstract Every animal species has particular environmental requirements that are essential for its welfare, and when these so-called “basic needs” are not fulfilled, the animals suffer. The basic needs of horses have been claimed to be social contact, social companionship, free movement and access to roughage. To assess whether horses suffer when one or more of the four proposed basic needs are restricted, we examined several studies (n = 38) that reported behavioural and physiological reactions to these restrictions. We assigned the studies according to the four types of responses investigated: (a) Stress, (b) Active, (c) Passive, and (d) Abnormal Behaviour. Furthermore, the number of studies indicating that horses reacted to the restrictions were compared with the number of studies reporting no reaction. The limited number of studies available on single management restrictions did not allow conclusions to be drawn on the effect of each restriction separately, especially in the case of social companionship. However, when combinations of social contact, free movement and access to roughage were restricted, many of the horses had developed responses consistent with suffering. Passive Responses, indicating acute suffering, and Abnormal Behaviour, indicating suffering currently or at some time in the past, were especially clearly demonstrated. This provides further evidence of the usefulness of assessing behavioural parameters in combination with physiological measurements when evaluating horse welfare. This meta-analysis of the literature confirms that it is justified to claim that social contact, free movement and access to roughage are basic needs in horses.


Introduction
Every animal species has particular environmental requirements that are essential for its welfare [1][2][3] and these are described as basic needs. As a general assumption, it has been claimed that social contact, social companionship, free movement and access to roughage are horses' basic needs [4][5][6][7][8][9][10]. Horses are said to need social contact because in a natural setting they live in large groups, with about 200-400 horses comprising a herd. These herds are divided into subgroups of harems (usually composed of one to five males, several females and their offspring) and bachelor bands (composed of males of different ages) [4,5,7,9,10]. Furthermore, horses are assumed to need social companionship because about one third of all horses form stable social bonds with members of their subgroup. Bonded animals mutually protect each other and their offspring, as well as protecting resources such as food, water and resting places [4,5,9]. In addition, it has been claimed that horses need free movement because under natural conditions, they cover between 3 to 30 km daily [4,[6][7][8]. Finally, horses have been said to need access to roughage as, in nature, they feed on grass for 12 to 16 h per day [4,5,7,8].
However, it must be evaluated whether keeping horses under human management conditions that restrict their basic needs compromises welfare. Therefore, several studies have set out to assess whether horses suffer and, if so, which responses demonstrate suffering, when one or more of these needs are not fulfilled.
Behavioural disorders have also been assessed in horses [17,44], including stereotypies [8,26,[45][46][47][48] and self-harming (i.e., redirected behaviour [8]). It has been debated whether abnormal behaviour is actually harmful or is rather a behavioural adaptation to a poor environment [47]. In this respect, most authors agree that stereotypies can be considered maladaptive behaviour indicating that horses are suffering under their housing or training conditions, and they give various reasons for this conclusion. Firstly, horses develop stereotypies in housing in which movement is reduced and there is little social contact. Secondly, free ranging, feral horses have never, to date, been observed showing stereotypic behaviour [7]. Thirdly, animals that displayed redirected behaviour and stereotypies also showed clinical signs of reduced welfare. These included: (a) self-inflicted skin lesions as a result of redirected behaviours, (b) gut damage, including lesions, ulceration and damaged mucosal tissue as a consequence of a wind-sucking and crib-biting [27,49] and (c) increased frequency of laminitis as a result of stereotypic movements such as weaving and box walking [27].
This literature review aims to provide an overview of studies that have evaluated the effects of changes in factors relating to horses' environmental requirements (specifically social contact, social companionship, free movement, and access to roughage) on behavioural or physiological parameters (Table 1). We asked (a) whether a meta-analysis of the relevant literature supports the claim that social contact, social companionship, free movement and unlimited access to roughage are basic needs in horses, and (b) whether certain measurements can be considered reliable indicators for the analysis of animal welfare and basic need restrictions. Table 1. Literature on restrictions in "proposed basic needs" in horses. Background colours indicate studies on the restricted basic needs: grey = no restriction, light orange = social contact, light blue = social companionship, light yellow = movement, dark grey = feed, orange = social contact and feed, green = movement and social contact, yellow = movement, social contact and feed. The columns list the test condition or changes in test conditions, the response measurements and the horses' responses that were observed. For detailed information on the horses and their management conditions please see Supplementary Materials (Table S1).

Materials and Methods
From August 2020 to February 2021, we searched the research platforms Research Gate, PubMed, Web of Science, Science Direct and Google Scholar for studies on social contact, social companionship, free movement and unlimited access to roughage in horses. We identified 38 studies (Table 1 and Table S1) on behavioural and physiological responses to management conditions in which one or more of the four proposed basic needs were restricted, and this will be the basis of this literature review. Information on the horses observed in the studies and the management conditions they lived under are given in the Supplementary Materials (Table S1). Of the studies we identified, 17 evaluated horses' responses to the given management situation, and 21 studied horses' responses to changes in management conditions. One study is cited, but not included in the analysis, as it describes the horses' responses to particular management conditions but does not analyse responses to management restrictions.
As horse housing is a complex setting, isolating only one of the aspects is very difficult. Therefore, only a few studies have analysed changes in only one of the proposed basic needs, and most consider two or more needs simultaneously (see Table 1). These generally compared two or more horse groups under housing conditions that differed in one or more needs (see Table 1), but a few examined changes in housing conditions of one group.

Data Processing
Four steps were applied in the evaluation of the literature (Table 1 and Figure 1). Firstly, the literature was assigned to the proposed basic needs that were restricted. Secondly, the studies were categorized according to the type of response shown by the horse, and thirdly, we assessed whether changes in the behavioural and physiological reactions indicated that the restrictions were compromising the horses' welfare. Finally, the frequencies of studies reporting responses were compared with the frequency of those reporting no responses to the restrictions. Figure 1. Number of reports of responses (dark colours, i.e., black and blue) versus number of reports of no response (light colours: skin and yellow) to restrictions in proposed "basic needs" described for response types across all studies. Abnormal Behaviour responses were significantly shown across all studies (Binomial Test: p = 0.02). Responses were also shown when social contact, free movement and access to roughage were mutually restricted (Binomial Test: p < 0.001). More detailed information on the responses, the test conditions and the horses included in the studies are given in Table 1 and in the Supplementary Materials (Table S1).
The studies were grouped according to the proposed basic needs that were restricted (Table 1): 1.
Social Contact and Access to Roughage restricted, n = 1 7.
Free Movement and Social Contact restricted, n = 16 8.
Free Movement and Access to Roughage restricted, n = 1 9.

Data Analysis
The R-Project statistical software (R Development Core Team 2021, https://www.rproject.org/) was used for the statistical analysis and Excel for creating the figures when analysing the frequencies of studies on basic need restrictions. Some of the data were not normally distributed (Shapiro-Wilk Test). Therefore, Generalized Linear Models (GLMs) for multivariate testing with fixed factors were applied. For a general approach, we analysed the frequency of studies reporting reactions or no reaction under the particular restrictions of basic needs and the different responses shown. The GLM (formula = number of manuscripts evaluating response~response versus no response + response type, family = poisson (identity), data = Dataset) was used. We continued by applying a nested Generalized Linear Model (GLM) to analyse whether manuscripts reported differences for showing or not showing responses nested within the different types of response. Therefore, the GLM (formula = number of manuscripts. evaluating response~response versus no response % in % response type, family = poisson (identity), data = Dataset) was applied. The full statistical data are given in the Supplementary Materials (File S1). Binomial Tests were applied to compare the frequencies of showing or not showing certain types of responses under certain management restrictions. All tests were two-tailed, and the significance level was set at 0.05.
Of the studies on changes in behavioural and physiological measurements of 'Stress Responses', n = 13 studies reported that horses showed responses and n = 7 did not (Table 1, Figure 1). The studies did not clearly indicate stress responses for single restrictions, as when studies found several stress measurements changed, i.e., indicating stress, the same study or other studies also found other stress measurements remained unchanged, i.e., indicating no stress. When the reactions to all the different restrictions were compared, the number of reports showing stress responses correlated with the number of reports showing no response (GLM: n = 37, z = 2.49, p = 0.01; Figure 1, Table 1, File S1).
The same was true for the 'Active Responses', with n = 14 studies reporting that active responses were shown and n = 8 reporting they were not ( Figure 1, Table 1). Again, while many parameters changed almost as many did not change when analysing a particular restriction and there was a general correlation between active responses and no active response over all the basic need restrictions (GLM: n = 37, z = 2.63; p = 0.008, Figure 1, Table 1, File S1).
'Passive Responses' were reported in n = 14 studies and no 'Passive Responses' for n = 8 studies (Figure 1, Table 1). 'Passive Responses' to particular restrictions indicated more clearly than 'Stress Responses' and 'Active Responses' that horses responded with changes in behaviour to certain restrictions. The literature revealed only a trend in correlation between the horses showing passive responses to those showing no passive response when comparing all the different restrictions in the basic needs (GLM: n = 37, z = 1.85, p = 0.06; Figure 1, Table 1, File S1).
Finally, a significant number of horses demonstrated the response 'Abnormal Behaviour' in response to any restriction in basic needs ('Abnormal Behaviour': n = 12, no 'Abnormal Behaviour': n =3, Binomial test: p = 0.02, Figure 1). Horses reacted more clearly by either showing abnormal behaviour or not showing abnormal behaviour to particular restrictions, as there was again only a trend for a correlation between reports of abnormal behaviour and reports of no abnormal behaviour across the studies on all the different restrictions examined (GLM: n = 37, z = 1.74, p = 0.08; Figure 1, Table 1, File S1).
The literature provides no clear conclusion as to whether horses generally show responses to restrictions in just one of the parameters of social contact, social companionship, free movement and access to roughage in isolation (Figure 1). For most of the comparisons the sample sizes were too small for applying statistical tests, the few that allowed statistical testing revealed no difference between measurements that show and those that do not show a response to the restrictions (Binomial tests: all p > 0.05; Figure 1). When social contact, free movement and access to roughage were mutually restricted, horses showed behavioural and/or physiological responses (Binomial test: p < 0.001; Figure 1).

Discussion
The aim of this study was to evaluate whether certain measurements can be considered reliable indicators for the analysis of animal welfare under basic need restrictions by analysing the results published in the literature. The interpretation of the studies calls for caution as positive reporting biases may result in more studies that prove a certain measurement to be effective than those reporting no effect [56]. However, the literature also contains several studies that actually report missing responses of horses to the measurements under observation. We therefore found the results of the meta analysis of these studies worth discussing.
Especially the development of abnormal behaviour and stress responses under the long-term stress of compromised environmental conditions has been considered a maladaptive strategy which does not provide the animals with options to deal with such conditions [8,57,58]. When animals develop these responses, they are said to be clearly suffering [58,59]. The present literature review provides evidence that this claim is justified in horses, especially in relation to abnormal behaviour. Horses develop significant levels of abnormal behaviour when social contact, social companionship, free movement and access to roughage are compromised. Therefore, it appears to be justified to use the display of abnormal behaviour as a behavioural animal welfare indicator when analysing the quality of horse housing and training, as has been established for animal welfare protocols [60].
However, it remains debatable whether a long-term display of abnormal behaviour provides strong evidence of a horse suffering under its present management conditions. Stereotypic behaviour may have developed under previous conditions and persist, even when management and training return to favourable conditions [45,46,57]. A recent development in EEG wave pattern analysis [26] offers promising new insights into this debate. Horses that had consistently displayed stereotypic behaviour for at least one year and lived under restricted management conditions showed EEG wave patterns comparable to those indicating negative emotional states in humans. However, caution should be exercised when evaluating welfare in horses displaying stereotypic behaviour as horses secrete dopamine when engaging in such behaviour and this elicits a positive emotional state in the animal (see for review: [61]).
Similarly, the measurement of behavioural and physiological stress parameters did not necessarily indicate that all horses experienced stress in compromised management conditions (Figure 1). Some studies found that horses did respond with behavioural or physiological stress parameters, and others did not (Table 1). This discrepancy may be due to difficulties in assessing long-term stress. When animals suffer stress for longer periods, such as when their environmental requirements are restricted for a long time, several physiological parameters, such as stress hormones and cardiovascular functions may return to base levels or below [13,25,62,63].
There are promising stress parameters that may allow long-term stress to be evaluated, such as immune cell suppression, changes in motor laterality [25,64], increased hemispheric laterality and EEG wave patterns analogous to those measured in humans with negative emotional arousal [26]. However, there is not yet sufficient evidence for these to be included in the present literature survey because they have only been studied in a few pioneering papers [25,26,64,65].
There may also be individual differences between horses in their stress resistance and the importance an individual attaches to any particular stressor [63]. In addition, previous experience with the restrictions in basic needs may be a factor. For example, horses that were born and raised in a stabled environment may be less stressed by restricted movement than horses raised at grass and then moved to a stabled environment [25]. Moreover, horses that had previously experienced individual housing did not display any significant differences in physiological stress responses between individual housing with semi-contact to conspecifics and group housing [13,24], whereas horses that were naïve to individual housing showed significant physiological stress responses when moved from group to individual housing [17,19,25]. A combination of several physiological and behavioural stress parameters may provide the strongest evidence for stress in horses as some studies found conflicting results when comparing a limited spectrum of physiological and behavioural data [63].
However, the proportion of horses showing passive responses supports the claim that many horses suffer long term stress under the investigated management restrictions [22,52,66]. Passive responses, such as reductions in activity, feeding, behaviour displays, contact to persons or other conspecifics and reactions to the environment indicate that horses withdraw from external stimuli and may show a depressive-like state [22,66]. These responses are maladaptive for animals such as horses, which are both flight animals that rely on fast responses to acute challenges and social animals that rely on fast responses to social challenges [9].
Interestingly, the analysis of 'Active Responses' did not provide a clear conclusion. Some studies reported that horses showed active responses when the animals were faced with restricted basic needs and others did not (Table 1 and Table S1). It may be difficult to clearly distinguish the level of activity that constitutes a stress response, as elevated aggression and movement may counteract mild stress [34,37].
As only very few studies succeeded in isolating the effects of restrictions in just one of the four proposed basic needs of social contact, social companionship, free movement and access to roughage, a clear statement on whether animals can generally cope with the particular restrictions remains elusive. However, there were sufficient studies on combined restrictions in social contact, free movement and access to roughage, and when the horses' environments were restricted in these three conditions, they appeared to suffer, as 'Abnormal Behaviour' and 'Passive Response' had clearly developed. This is consistent with the finding that abnormal behaviour display may be caused by a variety factors (e.g., genetics, rearing conditions, housing and feeding [48]).

Conclusions
We conclude that under combined restrictions of social contact, social companionship, free movement and access to roughage horses display signs of suffering. The literature available on individual so called "basic needs", does not allow us to isolate the effect of each of them, especially in the case of social companionship. However, the development of abnormal behaviour and passive coping strategies, can be considered signs of suffering, and these were displayed under separate restrictions in social contact, free movement and free access to roughage, as well as under combined restrictions of two or more of the proposed basic needs.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/ani11061798/s1, File S1: Statistical Data. Complete Generalized Linear Models (GLMs), Table  S1: Summary table of literature content. More detailed information on the restrictions, the horses, their management and their responses.