Development of feeding behaviour

The young ruminant relies initially on milk and begins sampling solid feed within the first few weeks after birth (Nicol and Sharafeldin, Reference Nicol1975). The timing of the transition from nursing to a solid diet is highly variable among individuals; for example, natural weaning in domestic cattle was reported to be between 7 and 14 months after birth (Reinhardt and Reinhardt, Reference Rice, Jongman, Borg, Butler and Hemsworth1981). Young ruminants will begin to graze by learning from social models such as the mother and conspecifics or learning by trial and error, leading to individual preferences and aversions to plants, and individual differences in ability to forage efficiently (Provenza and Balph, Reference Provenza and Burritt1987; Kyriazakis et al., Reference Kurvers, Hamblin and Giraldeau1999). Even at an older age, social models may be useful when introducing naïve animals to new feeding systems; when dairy heifers were turned out to pasture for the first time without an experienced companion, some individuals took over 3 h to begin to graze compared with just 1 h for those that were pastured with an experienced grazer (Costa et al., Reference Costa, Adderley, Weary and von Keyserlingk2016a).

In most dairy cattle production systems, calves are raised apart from their mother. Dairy farms vary in how much and how often milk is delivered to the calves and in opportunities for social learning (Vasseur et al., Reference Veissier and Stefanova2010; Hötzel et al., Reference Hewitson, Gordon and Dumont2014; USDA, Reference Val-Laillet, Passillé, Rushen and von Keyserlingk2014); both factors may influence the development of feeding behaviours (reviewed by Miller-Cushon and DeVries, Reference Miller-Cushon and DeVries2015). For example, calves reared individually must learn on their own where, how and what to eat. The lack of a social model may be particularly important during the transition from milk onto a solid diet, especially given that weaning occurs much earlier than in nature (Enríquez et al., Reference Dumont, Boissy, Achard, Sibbald and Erhard2011; reviewed by Khan et al., Reference Jørgensen, Andersen and Bøe2016).

Young ruminants vary in the amount of milk that they choose to consume when milk is provided ad libitum. This variability can contribute to differences in growth rates during the pre-weaning period. For example, de Passillé et al. (Reference de Passillé and Rushen2016) reported a large range in milk intakes during the first 2 to 4 days of age in Holstein dairy calves, ranging from 2.4 l to 12 l/day (7 to 30% of BW), resulting in differences in BW gains (ranging from 0.07 to 1.2 kg/day in the first month of age). Similar variability in milk feeding patterns were reported for artificially-reared lambs, ranging from 0.3 to 2.9 l/day milk consumption (David et al., Reference David, Bouvier, Ricard, Ruesche and Weisbecker2014).

Dairy calves raised indoors are typically introduced to a concentrate diet soon after birth. Considerable variability in concentrate consumption has been reported by several authors. Neave et al. (2018) showed that calves first found and began to consume grain at between 4 to 41 days of age when fed either 6, 8, 10 or 12 l/day of milk. Calves fed 12 l/day of milk first consumed 200 g/day of grain at between 23 to 82 days of age and first reached a daily grain consumption of 1400 g/day at between 58 to 94 days of age (de Passillé and Rushen, Reference Patison, Swain, Bishop-Hurley, Pattison and Robins2016). Calves fed less milk typically begin to consume more grain at an earlier age; however, Roth et al. (Reference Rushen, Taylor and de Passillé2009) showed that even when calves were fed just 6 l of milk/day, the age range when they first consumed 2000 g/day of grain was between 45 and 98 days of age. Furthermore, when group-housed calves were offered free choice of milk replacer, concentrate, maize silage, hay and straw, there was large individual variability in intake of each component, suggesting that calves develop diet preferences from a young age (Webb et al., Reference Webb, Engel, Berends, van Reenen, Gerrits, de Boer and Bokkers2014). These preferences may also arise from associations between sensory properties and nutritional value of the diet (Forbes and Kyriazakis, Reference Escós, Alados and Boza1995), the experiences associated with the first encounter with the feedstuff, or the physical properties of the diet that are important for stimulating ruminal development (Baumont, Reference Baumont1996).

This evidence highlights the variation in feeding behaviour among individuals from a young age. It is well known that reduced milk intake or reluctance to transition to solid feed can result in impaired growth during the pre-weaning and weaning periods in young ruminants (e.g. dairy calves: de Passillé et al., Reference de Passillé and Rushen2016; goat kids: Warmington and Kirton, Reference Warmington and Kirton1990; lambs: Greenwood et al., Reference Giraldeau and Caraco1998). Furthermore, early-life nutritional disadvantages have been shown to affect future foraging behaviour in other species (e.g. Andrews et al., Reference Andrews, Viviani, Egan, Bedford, Brilot, Nettle and Bateson2015). Understanding how and why some individuals develop feeding patterns that result in better or worse performance is important if we are to help all animals thrive.

Expression of feeding behaviour

Characteristic feeding patterns that develop from an early age in ruminants appear to also be present in adulthood. The rearing environment, as well as morphological and physiological differences, will have a profound impact on how individuals express their feeding preferences and patterns as adults (Provenza and Balph, Reference Provenza and Burritt1987). Individuals within a herd can show feeding patterns that are widely variable between individuals but remain relatively consistent over time within individuals (Melin et al., Reference Meagher, Weary and von Keyserlingk2005). This is not to say that feeding behaviour is inflexible, but rather that the degree of flexibility in feeding patterns generally remains consistent within individuals over time.

The selection of, and preference for, plants to graze or browse will depend on the individual’s nutritional needs and on prior experience with these food sources and ability to cope with toxins (Provenza et al., Reference Ralphs, Graham and James2003). Ruminants are known to make trade-offs in selecting diets that meet the requirements of their internal state (e.g. hunger, and physiological state such as pregnancy) while reducing costs in the selection of the diet (e.g. environmental or social pressures) (Kyriazakis et al., Reference Kurvers, Hamblin and Giraldeau1999; Arsenos et al., Reference Arsenos, Emmans and Kyriazakis2000). Thus an individual’s diet selection is flexible with changing internal state, with changes in diet dependent upon any nutritional deficiency and post-ingestive feedback from ingested foods (e.g. Tolkamp et al., 1998). Day et al. (Reference Day, Kyriazakis and Rogers1998) proposed that food selection is also influenced by a motivation to explore the feeding environment, which functions to identify new food items and to monitor and update information on existing food sources; individuals are able to modify their feeding behaviour if needed.

Foraging animals must choose between continuing to exploit an existing site or searching for a superior foraging site (see Giraldeau and Caraco, Reference Galindo and Broom2000). One approach to understanding this dichotomy is the producer-scrounger model, originally developed to describe the feeding strategies of sparrows (Barnard and Sibly, Reference Barnard and Sibly1981). ‘Producers’ take the role of finding higher food quality food patches. These individuals benefit from first access to the new patch, but pay the cost of lost foraging opportunities (and perhaps increased predation) while searching for new patches (Giraldeau and Caraco, Reference Galindo and Broom2000). Other individuals adopt a ‘scrounger’ strategy of following ‘producers’ to exploit their findings rather than searching for food themselves. This framework has been applied to foraging strategies in goats (Stears et al., Reference Thouless2014) and sheep (Hewitson, Reference Haskell, Simm and Turner2002). Individuals within a herd may also adopt ‘leader’ and ‘follower’ roles in making decisions when to move between feeding locations including cattle (Dumont et al., Reference Dingemanse, Kazem, Réale and Wright2005), sheep (Squires and Daws, Reference Stears, Kerley and Shrader1975), goats (Escós et al., Reference Eitam, Vaya, Brosh, Orlov, Khatib, Izhaki and Shabtay1993) and buffalo (Prins, Reference Proudfoot and Habing1996). ’Producers’ may be more likely to be ‘leaders’ but this line of research has yet to be explored.

Given the evidence of individual variability in diet selection of foraging animals (Kyriazakis et al., Reference Kurvers, Hamblin and Giraldeau1999; Arsenos et al., Reference Arsenos, Emmans and Kyriazakis2000), it follows that there would be similar variability in the feeding patterns of confined ruminants. For instance, Melin et al. (Reference Meagher, Weary and von Keyserlingk2005) found that as much as 84% to 98% of the variation in feeding patterns could be attributed to individual differences between dairy cows. Several studies have shown substantial between-cow variability for meal frequency (e.g. ranging from 5 to 9 meals/day) and feeding time (dairy cattle: 250 to 450 min/day; beef cattle: 8 to 90 min/day or 86 to 120 min/day) (Schwartzkopf-Genswein et al., Reference Sibbald, Erhard, Hooper, Dumont and Boissy2002; DeVries et al., Reference de Passillé, Rabeyrin and Rushen2003). Such feeding patterns have been attributed more to phenotypic than to genetic variation among individuals (Løvendahl and Munksgaard, Reference Loerch and Fluharty2016).

Studies have also shown that some individuals will adjust their feeding behaviour in response to social or environmental changes. Crossley et al. (Reference Crossley, Harlander-Matauschek and DeVries2017) reported that when dairy cows competed for access to a feeding area, there was an increase in variability in feeding time, feeding rate and meal duration. In a companion study, increasing feeding frequency did not reduce variability in feeding time, feeding rate or dry matter intake (Crossley et al., Reference Crossley, Harlander-Matauschek and DeVries2018). This variability is likely due to individual motivations to access the feed bunk; some animals reduce feeding time and others strive to maintain feeding times even under high levels of competition (dairy cattle: Val-Laillet et al., Reference Van Reenen, O’Connell, Van Der Werf, Korte, Hopster, Jones and Blokhuis2008b; goats: Jørgensen et al., Reference Huzzey, DeVries, Valois and von Keyserlingk2007). Some individuals will also respond more negatively than others when experiencing environmental changes. For instance, Rice et al. (Reference Rice, Jongman, Butler and Hemsworth2016a) found that 18% of lambs entering a feedlot spent less than 30 min/day feeding, lost weight during the 1^st week, and were more likely to visit the feeder when no other lambs were present. This evidence suggests that the feeding behaviour of some individuals will change in response to social or environmental pressure.

In summary, there is wide individual variability in feeding behaviour from an early age. Understanding why feeding behaviour is variable among individuals will be the focus of the remainder of this review, particularly how personality traits such as exploration, fear or reactivity and sociability affect how ruminants interact with their feeding environment.

Foraging strategies

Studies on the personality characteristics of individuals adopting producer-scrounger or leader-follower foraging strategies often profile animals along an exploration-avoidance or boldness-shyness axis, and individuals that are more exploratory or bold are thought to be more likely be leaders or producers by searching for food rather than relying on others (Kurvers et al., Reference Krohn and Konggaard2012). For example, sheep that were more exploratory in an unfamiliar arena with novel objects were also more likely to move away from conspecifics while grazing, enabling them to explore more of the pasture area (Sibbald et al., Reference Sibbald and Hooper2009). Sheep that were more exploratory were also more likely to split into smaller subgroups, indicating these animals made the trade-off to explore their feeding environment rather than to remain together as a cohesive group (Michelena et al., Reference Melin, Wiktorsson and Norell2009). Domestic deer that spent more time close to or investigating novel objects made a similar trade-off, spending less time engaged in vigilant behaviours and more time investigating a novel food (Bergvall et al., Reference Bergvall, Schäpers, Kjellander and Weiss2011). In addition, beef heifers that spent more time interacting with a novel object tended to be positioned at the front of the herd (Ramseyer et al., Reference Rault, Lay and Marchant-Forde2009). These studies suggest that more exploratory individuals (as indicated by greater investigation of novel objects or food) adopt riskier foraging behaviour and that this results in increased opportunities to forage.

In confined farming systems, ruminants are typically provided uniform diets at specific times of the day, reducing the need for deciding when and where to forage. Indoor-housed animals will still perform exploratory behaviour, particularly when the feed quality is variable (Huzzey et al., Reference Hötzel, Longo, Balcão, Cardoso and Costa2013). Meagher et al. (Reference Lynch, Hinch and Adams2017) offered feed bins with different forage varieties or flavours along a feed bunk and recorded the number of bin switches as a measure of exploratory feed sampling. Heifers that spent more time in contact with a novel object in a previous test also spent more time exploring and eating the varied feed. Moreover, those that were quicker to reach a novel food in the individual test also spent more time at the varied feed and switched between bins more often. Most recently, dairy calves that were most exploratory in a novel environment began to eat grain earlier, consumed more grain and gained more weight (Neave et al., Reference Nawroth, von Borell and Langbein2018). Collectively this evidence indicates that some individuals are more proficient in exploring and sampling varied or novel feeds.

Feed sampling and sorting behaviour

Dairy cows are known to preferentially sort for concentrate and against long forage components in a mixed ration but this type of behaviour is highly variable among individuals, with some even sorting against the typically preferred finer particles (Leonardi and Armentano, Reference Lensink, Fernandez, Boivin, Pradel, Le Neindre and Veissier2003). Interestingly, sorting behaviour did not decrease when cows were fed in a competitive feeding environment (Hosseinkhani et al., Reference Hewitson2008), suggesting that individuals engaging in this behaviour are motivated to do so even when access to feed is limited. A possible explanation for this finding is that individuals that continued to sort in a competitive environment were also higher in social rank and thus could maintain their position at the feed bunk (see Favati et al., Reference Enríquez, Hötzel and Ungerfeld2014).

Sorting is also likely a learned behaviour, related to post-ingestive feedback mechanisms (Provenza, Reference Provenza and Balph1995), and familiarity of feed from an early age (Miller-Cushon and DeVries, Reference Michelena, Sibbald, Erhard and McLeod2011). Consequently, young dairy calves are able to sort a mixed ration and will adjust this behaviour in response to the availability of grain (Costa et al., Reference Costa, Costa, Weary, Machado Filho and von Keyserlingk2016b). Feed sorting is seen as a risk factor for ruminal acidosis in adult cows (Cook et al., Reference Cook, Nordlund and Oetzel2004), so farms often strive to prevent this type of behaviour. If indeed the motivation behind sorting behaviour stems from a desire to explore the feeding environment, management practices may be able to redirect this behaviour by offering other opportunities for environmental exploration or manipulation. To our knowledge, no work to date has explored such opportunities.

Overall, individuals differ in how much they explore their feeding environment. Some individuals are producers or scroungers, and some are leaders or followers when it comes to deciding how and where to find food. In confined housing, these foraging differences have yet to be documented, but individuals that sample their feeding environment have been shown to be more exploratory and bold in novel situations. Individuals that are more reactive in response to novelty (rather than exploratory or bold) also show differences in their feeding behaviour; this will be the focus of the next section.

Reactivity to novel environments

Dairy heifers are known to differ in their responses when first introduced to the milking parlour; Van Reenen et al. (Reference Vasseur, Borderas, Cue, Lefebvre, Pellerin, Rushen, Wade and de Passillé2002) showed that some individuals had consistently higher physiological (i.e. cortisol) and behavioural reactivity (i.e. stepping and kicking) during milking preparation and teat cup attachment. Heifers habituated to the milking parlour for several weeks before calving had improved feed intake compared with those that were not habituated (Daniels et al., Reference Daniels, Donkin, Eicher, Pajor and Schutz2007). Thus changes in feeding behaviour may be an indication of individuals that are especially affected by novel processes like milking.

Several authors have reported individual variability in physiological stress responses in cattle (e.g. Loerch and Fluharty, Reference Llonch, Somarriba, Duthie, Haskell, Rooke, Troy, Roehe and Turner1999; Eitam et al., Reference Dumont and Boissy2010) and lambs (Rice et al., Reference Rioja-Lang, Roberts, Healy, Lawrence and Haskell2016b) when introduced to a feedlot. Lambs with a high cortisol response during the 1^st week in this new environment were also more reactive during isolation and restraint tests and had a greater number of feeding bouts likely due to a high number of displacements.

Some personality traits may drive consistent behavioural responses across a range of situations, whereas other traits may only manifest under specific contexts (Sloan Wilson et al., Reference Squires and Daws1994; Beausoleil et al., Reference Beausoleil, Blache, Stafford, Mellor and Noble2012). For instance, lambs that spent < 30 min/day feeding in the 1^st week after arriving at the feedlot (termed ‘shy-feeders’) had no relationship with behavioural responses during isolation and restraint (Rice et al., Reference Rice, Jongman, Butler and Hemsworth2016a). However, this study used a standardized test that elicited a fear response specific to isolation which may be unrelated to responses to stressors encountered in a highly social feeding environment. Future work should aim to identify personality traits that are specific to individuals that show changes in their feeding behaviour when introduced to new environments and how modifications to that environment may be beneficial.

Food neophobia

Food neophobia, defined as a reluctance to eat unfamiliar foods (Chapple and Lynch, Reference Chapple and Lynch1986), is well-known in ruminants and is thought to help animals avoid toxic plants (Provenza and Balph, Reference Provenza and Burritt1987). This fear of novel diets must be overcome for livestock to transition to different feed types (Launchbaugh et al., Reference Launchbaugh and Howery1997). When this transition is coupled with a change in environment, food neophobia is greater compared with when animals remain in familiar environments (Burritt and Provenza, Reference Burritt and Provenza1997). Individuals will even consume familiar foods containing toxins over novel feeds when in an unfamiliar environment (Burritt and Provenza, Reference Burritt and Provenza1997). However, early exposure to a diversity of foods can increase acceptance of novel foods especially in unfamiliar locations (Villalba et al., Reference Villalba, Manteca and Provenza2012).

Variability in feeding behaviour may be due in part to differences in food neophobia. For instance, Rice et al. (Reference Rice, Jongman, Butler and Hemsworth2016a) suggested that highly reduced feeding times in some lambs (‘shy-feeders’) may be due to individual differences in food neophobia. This reluctance to sample novel feeds can be consistent over time and in different contexts. For example, Costa et al. (Reference Costa, von Keyserlingk and Weary2014) demonstrated that dairy calves were consistent across days in their willingness to sample two types of novel foods (carrots and hay), and heifers that were quick to find and eat more of a novel food in an arena also spent more time eating flavoured and varied forages offered at the feed bunk (Meagher et al., Reference Lynch, Hinch and Adams2017). In lambs, individuals that were more food neophobic were also more fearful in a novel arena and exhibited more stress-induced hyperthermia (Villalba et al., Reference von Keyserlingk and Weary2009), suggesting that the test of food neophobia reflects fearfulness. These experimental findings were recently supported in a study of pair-housed calves on a commercial farm that consumed nearly three times as much novel food compared with isolated calves (Whalin et al., Reference Whalin, Weary and von Keyserlingk2018). No studies have investigated how food neophobia is related to other personality traits, such as exploration and sociability.

Reactivity to handling

Individuals differ in their reactivity to handling and restraint. Many such interactions occur in farm animal production including vaccinations, dehorning, branding or castrating. With the advances of automated management technologies, opportunities for positive interactions and habituation to humans may be limited (Rushen et al., Reference Rutter1999; Butler et al., Reference Butler, Holloway and Bear2012). Poor handling and fear of humans are expected to alter the behaviour of the animals, including changes in feeding behaviour. For instance, beef cattle that are especially reactive when in the chute or isolated in a pen with a handler (i.e. nervous, vigorous or violent movement, or attempts to escape), and have high flight speeds exiting the chute, also have reduced feed intake (Black et al., 2013; Llonch et al., 2018), reduced feeding times (Cafe et al., 2011) and increased feeding bouts of shorter duration (Llonch et al., 2018). Similar effects on feeding behaviour have been shown in physiologically more reactive cattle (e.g. high cortisol response when in the chute; Llonch et al., Reference Leonardi and Armentano2016). However, other studies have not found a relationship between reactivity and feed intake (Petherick et al., Reference Prins2002; Francisco et al., Reference Forbes and Kyriazakis2015) or feeding time (Nkrumah et al., Reference Nicol and Sharafeldin2007). These inconsistent results may be related to how personality measures were analysed (e.g. categorical, such as ‘adequate’ v. ‘excitable’ in Francisco et al., Reference Forbes and Kyriazakis2015, or continuous, such as scoring personality from 1 to 5 in Cafe et al., Reference Cafe, Robinson, Ferguson, Mcintyre, Geesink and Greenwood2011) or may be related to the degree of negative experiences associated with humans (e.g. blood samples were taken before personality measures were scored in the chute; Cafe et al., Reference Cafe, Robinson, Ferguson, Mcintyre, Geesink and Greenwood2011).

Apart from feed intake and feeding time, no studies have examined how reactivity to handling influences other aspects of feeding behaviour in ruminants. However, work on other farm animals shows an association. For example, reactivity during weighing was associated with more visits to the feeder and less intake per visit in pigs (Ros-Freixedes et al., Reference Roth, Keil, Gygax and Hillmann2014). Rohrer et al. (Reference Ros-Freixedes, Sadler, Onteru, Smith, Young, Johnson, Lonergan, Huff-Lonergan, Dekkers and Rothschild2013) found that pigs that struggled more during restraint in the supine position (i.e. ‘reactive’ pigs) tended to have fewer daily meals and these were of longer duration compared with ‘proactive’ pigs. ‘Reactive’ pigs also preferred to eat at times when the feeder was less occupied. These studies suggest that active movement during restraint may be related to active avoidance of social conflict at the feeder.

Stress from prolonged or repeated unpleasant handling can lead to impaired growth and productivity (e.g. Lensink et al., Reference Launchbaugh, Provenza and Werkmeister2000). Whether growth is reduced due to changes in feeding behaviour that limit feed intake, or due to poorer feed efficiency, remains unknown. Recent research has shown that human presence can have a profound impact on goat behaviour; even changes in a human’s head position can alter the behaviour of goats (Nawroth et al., Reference Miranda-de la Lama, Sepúlveda, Montaldo, María and Galindo2015). Thus, the behaviour of stockpersons during management practices may elicit stress responses that in turn affect feeding behaviour. Together this evidence suggests that reactive individuals may be more prone to changes in feeding behaviour.

In summary, individuals differ in the way they respond to stressful events on farms. Feeding behaviour and feed intake can be impaired in individuals that are particularly reactive to a change in environment (feedlot, milking parlour), change in diet (food neophobia) and handling by humans (e.g. restraint in a chute). Aspects of the social environment may also be stressful for some individuals, and the way in which individuals respond to stressors may be related to the social relationships within the herd. We turn to this topic in the following section.

Dominant-subordinate relationships

Domestic ruminants are gregarious and will organize themselves into social hierarchies with dominant and subordinate individuals (e.g. cattle: Bouissou et al., Reference Bouissou, Boissy, Le Neindre and Veissier2001; goats: Miranda-de la Lama and Mattiello, Reference Miller and Wood-Gush2010). An individual’s position in the hierarchy is often expressed through agonistic interactions when gaining or maintaining access to resources (Miller and Wood-Gush, Reference Miller-Cushon and DeVries1991; Barroso et al., Reference Barroso, Alados and Boza2000). For grazing ruminants on rangeland, where space is less limited, this hierarchy is related to priority of access to high-quality grazing areas (Barroso et al., Reference Barroso, Alados and Boza2000), which can be expressed as an ‘avoidance order’ whereby, subordinate animals avoid conflict with dominant ones (Sárová et al., Reference Schwartzkopf-Genswein, Atwood and McAllister2010).

Feeding behaviour is related to the social rank of the individual with dominant individuals typically having priority access to food. This is especially evident when forage is limited and low in quality. For example, higher-ranking goats will out-compete subordinates to consume preferred shrubs (Barroso et al., Reference Barroso, Alados and Boza2000) and dominant individuals are often more efficient foragers (e.g. Thouless, Reference Tolkamp, Kyriazakis, Oldham, Lewis, Dewhurst and Newbold1990), likely due to less selection required when given priority access to feed resources. To achieve these foraging advantages, dominant animals position themselves toward the front of the herd and cover shorter distances relative to subordinates during periods of foraging (Sárová et al., Reference Schwartzkopf-Genswein, Atwood and McAllister2010). Consequently, subordinates are often forced to graze areas of lower quality; these individuals could move away from the group in search of improved grazing opportunity but risk increased exposure to predators (Thouless, Reference Tolkamp, Kyriazakis, Oldham, Lewis, Dewhurst and Newbold1990; Barroso et al., Reference Barroso, Alados and Boza2000). Indeed, the slower bite rate of subordinates is thought to reflect the trade-off between grazing and maintaining vigilance for predators and dominant animals, in addition to the increased necessity to select forage (Thouless, Reference Tolkamp, Kyriazakis, Oldham, Lewis, Dewhurst and Newbold1990). Interestingly, subordinates also reduce bite rate when dominant individuals are nearby, and cease grazing altogether to avoid neighbouring dominant animals (Thouless, Reference Tolkamp, Kyriazakis, Oldham, Lewis, Dewhurst and Newbold1990). Together this evidence indicates that dominant-subordinate relationships are important drivers of foraging behaviour in grazing systems.

Dominant–subordinate relationships in confined housing systems also influence feeding behaviour. Subordinate individuals may fail to gain access to the feed bunk and eat at times that are less preferred (Huzzey et al., Reference Hosseinkhani, DeVries, Proudfoot, Valizadeh, Veira and von Keyserlingk2006), and even sacrifice higher quality feed to avoid feeding near a dominant (Rioja-Lang et al., Reference Rohrer, Brown-Brandl, Rempel, Schneider and Holl2009). These effects may be exacerbated when competition for feed increases (e.g. Jørgensen et al., Reference Huzzey, DeVries, Valois and von Keyserlingk2007). Although dominance rank is often scored as the number and outcome of agonistic interactions between dyads (Galindo and Broom, Reference Francisco, Resende, Benatti, Castilhos, Cooke and Jorge2000), displacements at the feeder by cattle has been reported to be bi- or tri-directional and nonlinear with subordinate cows occasionally displacing dominants (Val-Laillet et al., Reference Val-Laillet, Veira and von Keyserlingk2008a). Of most interest is that high-ranking cows do not necessarily have the longest feeding times; dominance at the feed bunk may be related to individual motivation to gain access to feed or to defending the resource (Val-Laillet et al., Reference Val-Laillet, Veira and von Keyserlingk2008a; Reference Van Reenen, O’Connell, Van Der Werf, Korte, Hopster, Jones and Blokhuisb).

Variability in social behaviour at the feeder cannot be explained entirely by dominance. For example, Miranda-de la Lama et al. (Reference Miranda-de la Lama and Mattiello2011) described four social strategies or ‘identity profiles’ in goats: ‘passive’ goats (submissive but made no attempt to avoid or engage in agonistic behaviour) spent the least time at the feeder, whereas ‘avoider’ goats (submissive and avoided both agonistic and even non-agonistic behaviour) spent the most time feeding. Goats that were ‘aggressive’ (highly dominant and mediated other social conflicts) and ‘affiliative’ (average dominance and engaged in socio-positive behaviours) were similar in time spent at the feeder but intermediate to ‘avoider’ and ‘passive’ goats. Thus individuals can share similar dominance ranks yet adopt different social strategies that impact feeding behaviour.

Temporal feeding patterns and other measures of feeding behaviour such as feeding rate can further our understanding of how individuals are able to maintain or adjust feeding time under competitive conditions when adopting different social strategies. Zobel et al. (Reference Zobel, Schwartzkopf-Genswein, Genswein and von Keyserlingk2011) noted that beef heifers varied in how they responded to a highly competitive feeding environment; some individuals actively engaged in competition while maintaining a relatively stable feeding rate, but others appeared to actively avoid competition and shifted their feeding until after peak feeding time. Still, others appeared to adopt a ‘get-in-and-get-out’ strategy characterized by increased feeding rate and avoidance of agonistic interactions. Indeed, Nielsen (1999) noted that changes in feeding rate could be a valuable indicator of social pressure experienced by individuals in group feeding environments.

These results suggest avenues for future research. Given that individuals differ in their responsiveness to environmental change (i.e. behavioural plasticity; reviewed in Dingemanse et al., Reference de Passillé and Rushen2010), we might expect that some individuals will adjust their social behaviour at the feed bunk in response to a change in social environment (e.g. reduce aggression when there is low stocking density or when there are many other dominant individuals), whereas others will remain consistent across different social environments (e.g. maintain aggression despite plenty of space to access feed). This research would help to advance our understanding of, and opportunities for management of, the social grouping of ruminants in both grazing and confined feeding environments.

Affiliative relationships and sociability

Herd-living animals also have affiliative relationships, and these are also likely to influence social behaviour in the feeding environment. The formation of affiliative relationships among individuals has been reported among domestic ruminants, including cattle (Bouissou et al., Reference Bouissou, Boissy, Le Neindre and Veissier2001), goats (Miranda-de la Lama and Mattiello, Reference Miller and Wood-Gush2010) and sheep (Lynch et al., Reference Løvendahl and Munksgaard1992). Dairy cows form and maintain preferred partnerships from an early age (Raussi et al., Reference Reinhardt and Reinhardt2010) with large within- and between-cow variability in the frequency of social interactions and time spent in close proximity to partners (Gutmann et al., Reference Goetsch, Gipson, Askar and Puchala2015). This variability in sociability has been proposed as a distinct personality trait underlying how individuals respond to environmental challenges (Koolhaas and Van Reenen, Reference Koolhaas, Korte, De Boer, Van Der Vegt, Van Reenen, Hopster, De Jong, Ruis and Blokhuis2016). Sociability can be defined as the motivation to remain close to conspecifics (Sibbald et al., Reference Sibbald, Erhard, McLeod and Hooper2005), and appears to be linked with feeding patterns of grazing animals that must make trade-offs between social and feeding motivations. For example, when high-quality forage is distributed in patches, the intake and diet composition of each individual depends on their willingness to move away from the herd to graze preferred patches (Sibbald and Hooper, Reference Sibbald and Hooper2003). In a subsequent paper, these authors (Sibbald and Hooper, Reference Sih and Bell2004) demonstrated that more sociable sheep (i.e. those that spent the most time in close proximity to other sheep) were less likely to move away from the group to access a preferred grazing patch.

These foraging trade-offs appear to be influenced by the strength of relationships with herd mates. Dumont and Boissy (Reference DeVries, von Keyserlingk, Weary and Beauchemin2000) reported that ewes penned with familiar companions chose to graze away from the group for longer periods, vocalized less and were less vigilant than those grazing with unfamiliar companions. Brahman steers were also more willing to leave a familiar companion to approach a food bowl but did not do so when housed with an unfamiliar companion (Patison et al., Reference Petherick, Holroyd, Doogan and Venus2010). These authors suggested that a combination of a lack of social support and fear of isolation motivated individuals to remain close to unfamiliar companions. Sociability is also related to group movements between foraging sites. Ramseyer et al. (Reference Rault, Lay and Marchant-Forde2009) demonstrated that heifers with limited affiliative partnerships and ewes that often grazed away from the group were those that most often initiated group movement.

In confined housing systems there is also evidence that individual differences in sociability influence feeding behaviour. Gibbons et al. (Reference Galef and Laland2010) found that dairy cows who took longer to return to the group following separation at the end of a passageway had more limited partner associations, were less synchronized with the group, and did not feed during peak feeding times. The strength and type of partner associations appears to affect feeding behaviour. Individuals housed together for longer periods were more likely to be preferential partners during feeding (dairy cattle: Gutmann et al., Reference Goetsch, Gipson, Askar and Puchala2015; goats: Aschwanden et al., Reference Aschwanden, Gygax, Wechsler and Keil2008), and these feeding partners showed more positive social contact such as allogrooming (Val-Laillet et al., Reference Val-Laillet, Veira and von Keyserlingk2008a). These studies suggest that preferential associations among individuals can influence feeding behaviour in both grazing and confined herds. Future research should aim to understand how the quality and quantity of relationships between individuals influences how individuals make foraging decisions and affect feeding patterns. For instance, individuals that develop close social bonds may become more reliant on a social partner to find high-quality food in a grazing system. Furthermore, individuals with strong social bonds or many social partnerships may be better able to cope with stressful feeding environments given that social partners often mediate stress responses in farm animals (Rault et al., Reference Raussi, Niskanen, Siivonen, Hã, Hepola, Jauhiainen and Veissier2011).

Social facilitation and social learning

Social companions are important sources of information for making foraging decisions in both grazing and confined housing systems. Social facilitation and social learning play important roles in the development of foraging behaviour in neonatal ruminants on pasture (Launchbaugh and Howery, Reference Laland2005). Social facilitation is the phenomenon where the stimulus of another animal eating, approaching or manipulating feed may increase attention toward the feed, and subsequently encourage consumption of feed by others, whereas social learning describes the mechanism of learning through observation of others (Zentall and Galef, Reference Zentall and Galef1988).

Young ruminants learn from their mother and conspecifics the location of food, water and shelter, and consequently develop diet and habitat use patterns that resemble those of social models (Provenza and Balph, Reference Provenza and Burritt1987). For instance, individuals that were conditioned to avoid a particular plant when grazing alone began to consume this plant when grazing with others that had not learned to avoid this plant (cattle: Ralphs et al., Reference Ramseyer, Boissy, Thierry and Dumont1994; lambs: Provenza and Burritt, Reference Provenza, Villalba, Dziba, Atwood and Banner1991). Social models can be influential in learning how to graze; naïve beef calves had increased grazing activity when turned out to pasture with experienced companions compared with those without an experienced social model (Hessle, Reference Gutmann, Špinka and Winckler2009).

For young ruminants that are raised in confinement, social facilitation and social learning may influence how quickly individuals begin to use feeding equipment. Lambs were quicker to learn to drink milk from a teat when grouped with lambs that had previously learned this behaviour (Veissier and Stefanova, Reference Villalba, Catanese, Provenza and Distel1993). Dairy calves raised with a social partner consumed more starter during the milk feeding period (Costa et al., Reference Costa, Meagher, von Keyserlingk and Weary2015), were quicker to first visit a new concentrate feeder when regrouped after weaning (de Paula Vieira et al., 2010) and had more frequent concentrate meals before and during weaning (Miller-Cushon and DeVries, 2016) compared with calves raised alone. Consequently, socially housed calves have improved feed intake and weight gains (reviewed by Costa et al., Reference Costa, Daros, von Keyserlingk and Weary2016c). Together the evidence in both grazed and housed ruminant species suggests that learning about the feeding environment is influenced by the presence of social models.

Social foraging theory suggests that obtaining information by observing others is less costly than gaining the same information personally (Galef and Laland, Reference Forkman, Boissy, Meunier-Salaün, Canali and Jones2005). An exciting area of research is how individuals differ in their use of social information, such that some may be more likely to learn from social information in a feeding context. Sih and Bell (Reference Sloan Wilson, Clark, Coleman and Dearstyne2008) suggest that more sociable individuals may be more likely to acquire social information, given their close proximity and increased interactions among group mates compared with less sociable individuals. Indeed, the development and extent of social relationships among individuals in a grazing herd of cattle have been shown to be critical in the efficiency of information transfer (Launchbaugh and Howery, Reference Laland2005). Similarly, the development of solid feeding behaviour in young ruminants requiring the discovery and exploitation of a novel food resource may be influenced by the sociability of the animal such that strong affiliations may lead to increased attention to social cues. Naïve observers may pay more attention to experienced demonstrators if they share a particular relationship (see review by Nicol, Reference Neisen, Wechsler and Gygax1995).

The personality of the demonstrator may also suggest to observers within the herd that they are a reliable source of information. For instance, observers may watch and learn from the foraging behaviour of more exploratory or dominant individuals (Nicol, Reference Neisen, Wechsler and Gygax1995). Indeed, there is evidence suggesting that the feeding behaviour of dominant individuals is copied by subordinates (Laland, Reference Kyriazakis, Tolkamp and Emmans2004). This line of research has been tested in some ruminants. In sheep, dominance status of both the observer and demonstrator affected decisions to leave a food patch for a potentially better foraging opportunity (Hewitson et al., Reference Hessle2007); subordinate sheep would only follow a dominant when the expectation for discovering high-quality food was high. In contrast, Baciadonna et al. (Reference Baciadonna, McElligott and Briefer2013) found that the use of social information by goats to locate a food patch was not dependent upon the dominance rank of the demonstrator. However, these authors also found that goats favoured the use of personal rather than social information to locate food. Reliance upon personal information may be more important when foraging in a variable or patchy environment.

These studies illustrate how social status affects the use of information when making foraging decisions. However, no work in ruminants has explored the effects of affiliative relationships on the use of social information. In addition, efforts could be made to understand if some personality types utilize or convey social information more than others.

In summary, the social environment influences feeding behaviour in grazing and confined housing systems. Dominant–subordinate and affiliative relationships affect how individuals forage, gain access to feed and adjust feeding patterns when the social environment changes.