25 Effects of diet on behavior – normal and abnormal
Abnormal oral behavior associated with diet and feeding
Time budgets and feeding patterns in the wild and in modern husbandry
Digestive processes and links with behavior
Further evidence linking stereotypy with digestion and potential confounding factors
Effects of dietary carbohydrate, fiber and oil on behavior
Evidence for diet-mediated physiological changes
Glucoregulation and the serotonergic system
Behavior modification by feed supplements
Abnormal oral behaviors like crib-biting are undesirable for many reasons. They are widely thought to indicate sub-optimal welfare and high proportions of owners perceive such behaviors to lower the economic value of a horse and adversely affect performance (McBride & Long 2001). A combination of welfare concerns and pragmatic economic considerations has inspired a considerable body of research into the role of diet and feeding in the etiology of abnormal oral behavior. In contrast, diet’s influence on normal behavior, reactivity and response to stressful stimuli has been little studied, despite the potential benefits to performance, the human–horse relationship and safety. The results of the small number of studies in this area converge to suggest that diet may affect behavior through the alteration of digestive function. Where the underlying processes have been revealed, they are generally consistent with the mechanisms that have been hypothesized to influence stereotypy development. Dietary influences on normal and abnormal behavior may therefore be governed by the same processes. These findings may help both to improve understanding of the development of stereotypies and to design dietary regimens that are beneficial in terms of horses’ physical and mental well-being.
This chapter therefore considers evidence that modern management and feeding practices influence aspects of normal behavior such as foraging, and may contribute to the development of redirected or stereotypic behavior. It then describes mechanisms by which digestive processes might exert an influence on behavior and examples of where this has been observed. Recent studies are reviewed that may provide ways to alter diet to optimize behavior or performance. Where possible, suggestions are also made for how these might be applied to reducing the risks of stereotypies. Finally, examination of other potential methods of modifying behavior through dietary supplements or foraging enrichment are considered.
Abnormal oral behavior associated with diet and feeding
Crib-biting, wind-sucking and wood-chewing are abnormal, apparently functionless behaviors whose initiation is thought by many to be associated with diet and feeding management. Crib-biting occurs when a fixed object is grasped by the incisor teeth, the lower neck muscles contract to retract the larynx caudally and air is drawn into the cranial esophagus, producing a characteristic grunt (McGreevy et al 1995). Wind-sucking involves the same suite of movements, except that the horse does not actually grasp a fixed object. Once developed by an individual horse, crib-biting (used here to include wind-sucking) tends to become increasingly fixed and rigid in form and orientation over time and is usually performed repetitively. Wood-chewing is sometimes called crib-biting by horse owners but is a more varied and flexible behavior pattern lacking the invariant, repetitive movements that define a true stereotypy (Mason, 1991) so is perhaps best described as a redirected behavior. Wood-chewing is a behavior worthy of study, however, as it often precedes or is associated with crib-biting. In a four-year prospective study, Waters et al (2002) recorded the onset of stereotypic behavior in young horses. Some 30.3% of young horses showed wood-chewing and 10.5% showed crib-biting, but notably, 74% of the horses that developed crib-biting had previously shown wood-chewing.
Locomotor stereotypies such as weaving, box walking and nodding will not be considered here as there is little evidence to suggest they are associated directly with diet. For example, weaving (a stereotyped lateral movement of the head and neck) peaks in frequency just prior to feeding (Clegg et al 2008, Ninomiya 2007) and appears to represent anticipation of an exciting or stressful event combined with frustration at the horse’s inability to leave the confined area. Factors that reduce the general aversiveness of the confined environment, particularly providing increased visual access to other horses, can significantly reduce the incidence of weaving (Cooper et al 2000, McAfee et al 2002). In contrast, such enrichment of the stabled environment appears to have little effect on the incidence of crib-biting where it is an established habit (Nicol 1999a), but foraging enrichment appears to have at least short term benefits (see ‘Foraging Enrichment’ page 451)
Cross-sectional surveys estimating the prevalence of abnormal oral behaviors have spanned various countries, breeds and management conditions. Not surprisingly, their estimates have varied considerably but summarizing early surveys from the UK, Italy, Sweden, and Canada on a range of breeds and uses, Nicol (1999a) described an overall mean prevalence of 4.13% crib-biting and 11.78% wood-chewing. Recent reports have reported similar values. A survey of 312 nonracing horses in Prince Edward Island yielded an owner-declared prevalence of 3.8% crib-biting and 3.8% wood chewing (Christie et al 2006). Albright et al (2009) reported an overall prevalence of 4.4% crib-biting in a study of 3574 American horses, though only 3.5% of a Swiss population of horses showed either locomotor or oral stereotypies (Bachmann et al 2003). In contrast, a 4-year longitudinal study of 225 foals revealed substantially higher rates: crib-biting was initiated by 10.5% of horses at a median age of 20 weeks, and wood-chewing by 30.3% of horses at a median age of 30 weeks (Waters et al 2002). It is not clear whether some young horses cease wood-chewing and crib-biting as they mature, or whether the lower levels reported in adults reflect a differential loss of these individuals to the adult population. Even if the former is the case, stereotypic oral behaviors affect a considerable number of horses and can come to dominate the behavioral repertoire of some individuals. Nicol et al (2002) reported one foal that performed crib-biting for nearly 50% of the time. Using a sample of 5 Thoroughbred geldings with an established crib-biting habit, Clegg et al (2008) recorded all incidences of crib-biting for 30 s of every 5 min period for 22 h per day. By extrapolation, they estimated that the horses in their study performed an average of 1470 crib-bites during each 22-h stabled period.
• Abnormal oral behaviors such as crib-biting and wood-chewing appear to be associated with diet and feeding management whereas locomotor stereotypies are more closely linked with frustration, anticipation and aversive confined environments
• Crib-biting and wood-chewing habits are often initiated early in life
• Estimates vary but these behaviors are observed in a few percent of horses across various breeds and uses (and can come to dominate the behavioral repertoire of some individuals)
Time budgets and feeding patterns in the wild and in modern husbandry
Grazing is the predominant activity of the free-living horse’s day. Studies of feral horses and primitive Przewalski’s horses under near-natural conditions generally yield observations that at least 50% of daylight is spent feeding (van Dierendonck et al (1994, Przewalski’s horse): 30–68%; Lamoot and Hoffman (2004): 68%, Mayes and Duncan (1986): 57–75% of daylight, 49–53% of nocturnal hours; Boyd (1988): 48–59%). Berger et al (1999) expressed time grazing as 52% of all activity.
Przewalski’s horses reintroduced into the wild were mostly active during daylight (Berger et al 1999) tending to graze at dawn and in the afternoon and rest during the morning (van Dierendonck et al 1994). Mayes and Duncan (1986) also describe predominant daytime grazing by Camargue horses but with significant periods also occurring before dawn and after dusk. Feeding activity too varies with season. It constituted a smaller proportion of the time budget in summer of Przewalski’s horses (Berger et al 1999) and of feral Haflinger mares (Lamoot & Hoffman 2004) and Camargue mares (Mayes & Duncan 1986). This probably reflects the lower nutritional quality of forage during the colder months, when body condition (Berger et al 1999, Kuntz et al 2006) and activity levels have both been observed to drop (Berger et al 1999) in Przewalski’s horses. Ransom et al (2010) similarly found that feral female horses in low body condition spent more time feeding and less time resting and in maintenance and social behavior than high condition females.
Studies of equine time budgets have used different observation methods but some estimates of other activities have been made. Lying time has been estimated at between 0.4% (Boyd 1988, Przewalski’s horses) and 5% (Salter & Hudson 1979, feral horses) of daytime hours and time standing resting between 11% and 23% in feral horses (Duncan 1980, 1985, Salter & Hudson 1979). Locomotion occurs at a low level while grazing and otherwise seems to occur between around 5% and 15% of observed time (Boyd 1988, Duncan 1980, 1985).
The eating habits of stabled horses tend to be more limited by the timing of meals and forage provision, but very little information is available on feeding behavior or indeed time budgets of such animals. Stabled mares fed hay ad libitum spent an average of 60% of a 24-hour period eating (Elia et al 2010). In a group of stabled horses fed concentrate and forage meals in the morning and evening, with additional forage at midday, scan sampling showed more foraging occurred at night than during the day (though this may have been influenced by providing most of the forage ration in the evening). Forage eating took up over 30% of the horses’ time across the 24-hour period. Lying behavior took up 16.1% of scans and occurred almost entirely during the night (Cooper et al 2007). When seven Welsh mountain pony mares were offered ad lib access to a complete diet, they spent approximately 50% of the day feeding in summer, and 44% in winter and consumed feed in discrete meals of around 40 minutes. The rate of food consumption was inversely associated with body condition score (Dugdale et al 2011).
Crib-biting has not been reported in studies of feral or free-living domestic horses, but it has been observed when Przewalski horses are kept in captivity (Boyd 1986), suggesting that domestic management practices are a necessary cause of the behavior. Modern horse management is associated with substantial changes in the nature, quantity and frequency of food consumption in horses that are often confounded with other aspects of management. For example, horses receiving a substantial proportion of their energy requirements as discrete meals of concentrate feeds are likely to be those that are regularly stabled. Stabling curtails opportunities for a range of behaviors in the horse’s normal repertoire such as foraging, locomotion and social contact. Separating the behavioral consequences of diet from those of frustration, excess energy reserves or lack of stimulation is therefore likely to be difficult. Controlled experiments comparing groups of horses, or the same individuals over varying treatments, suggest that even if the provision of concentrate feeds removes the physiological need to forage throughout the day, the motivation may remain. Through a series of experimental manipulations using fistulae and injection of nutrients, Ralston and colleagues concluded that oropharyngeal stimuli (taste, chewing, smell), nutrient feedback, and changes in energy availability (measured as glucose and insulin levels) all exert some degree of control over feeding behavior in horses (reviewed in Ralston 1984). Horses given ad libitum access to concentrate feed ate around 10 meals per day and engaged in multiple “nibbling” bouts (less than 150 g) in between meals, which were rarely separated by more than 3 hours (Ralston et al 1979). More recently, Dugdale et al (2010) reported that ponies provided with ad libitum access to a short chopped forage based meal ate an average of 16 meals per day and had a slower rate of intake than when food was limited. A patch choice experiment indicated that when grazing, horses forage so as to maximize energy intake rate (Edouard et al 2009). This will have been an adaptive strategy for a species that evolved to survive on grasses, and may contribute to horses’ tendency to consume even large, energy-dense meals quickly.
Behaviors such as coprophagy (ingestion of feces), wood-chewing and bed-eating are often considered to be aberrant but may simply reflect motivation to feed outside of mealtimes due to cues from gut fill, time since the last meal, or a drop in blood glucose. These behaviors may specifically represent attempts to ingest fiber as they are often ameliorated by greater provision of roughage. Horses fed an all-concentrate diet spent significantly more time engaged in wood-chewing and coprophagy than horses fed only hay (Willard et al 1977) and Redbo et al (1998) reported that the risk of wood-chewing correlated negatively with the amount of roughage available. Bed-eating is most common in horses bedded on straw and those lacking access to fibrous feed (Mills et al 2000). When Zeyner et al (2004) varied the amount of hay provided in addition to oats, they observed restless and nervous behavior in the horses with the lowest intake. Behavior was “quieter” and both aggressive behavior at feeding time and coprophagy were eliminated when the diet contained more hay. In this particular study, total caloric intake was lower in horses fed less hay so hunger may also have been an important determinant of behavior. Mares offered ad-lib orchard grass hay or a complete pelleted feed in a counterbalanced order had a higher motivation for hay – measured by an operant response – when fed the pelleted diet. Time spent in food searching behavior was also an order of magnitude lower when fed hay (Elia et al 2010).
Crib-biting is similarly increased by a low forage or high-grain diet (Bachmann et al 2003, McGreevy et al 1995, Redbo et al 1998, Waters et al 2002) and decreased by the use of straw bedding, possibly because this may function as an additional source of dietary fiber (McGreevy et al 1995, Christie et al 2006). Most of these studies have been cross-sectional in nature so conclusions about cause and effect must be drawn with great care. However, a longitudinal study has shown that feeding practices have a significant effect on the relative rate of development of oral stereotypies. Waters et al (2002) found that feeding grain-based feeds immediately after weaning resulted in a four-fold increase in risk. A prospective study of foal behavior suggested that individual differences in feeding patterns or feeding motivation of foals may also have an influence. Foals that developed abnormal oral behavior (wood chewing, crib-biting or both) after weaning had, prior to weaning, spent more time suckling and twice as much time teat nuzzling as other foals (Nicol & Badnell-Waters 2005). It is possible that foals with the greatest suckling motivation may be most affected by sudden weaning methods that instantly prevent suckling.
• Crib-biting has only been observed in domesticated or captive equids
• Free-living horses evolved to spend a large proportion of their time grazing on fibrous food
• The diets of many domesticated horses include infrequent, large and energy dense meals with limited forage.
• Undesirable oral behaviors are likely to reflect a motivation to graze and/or ingest additional fiber
Digestive processes and links with behavior
A closer look at digestive processes reveals plausible mechanisms by which both abnormal and undesirable normal behavior might result from concentrate-heavy feeding practices. There is a common perception that excess energy from concentrate feeds causes “fizzy” or unwanted excitable behavior (Jansson et al 2002). Large, high starch meals result in large fluctuations in plasma glucose and insulin after feeding (Harris 2005, Stull & Rodiek 1988, Pagan et al 1999), which are indeed likely to cause peaks and troughs of energy. Low fiber, high starch diets are also associated with a number of digestive and metabolic disorders (Hoffman 2003, Kronfeld & Harris 2003). Many problems seem to stem from the horse being unable to regulate gut acidity during digestion and absorption (Mills & Clarke 2002) and excitable, irritable or abnormal oral behavior may plausibly stem from visceral discomfort in either the fore- or hind-gut, or from its sequalae.
After ingesting large cereal-based meals, the higher proportion of dry matter in the stomach contents slows the mixing of feed and gastric juice, leading to the potential for dysfermentation in the stomach (Harris et al 2006). Wild horses would have historically eaten wild cereals that they encountered and so ingestion of cereals is not an “unnatural” modern phenomenon; it is more that their proportions in modern diets lead to a vastly decreased time spent chewing compared with a grazing lifestyle. For example, the average horse eating a kilo of hay would chew 3400 times, taking 40 minutes while the same weight of oats could be consumed in 10 minutes with only 850 jaw movements (see Harris & Arkell 2005). Elia et al (2010) found that horses chewed over 43,000 times a day when eating ad-lib hay, and only around 10 000 times a day when fed a pelleted diet. This exacerbates matters by reducing opportunities to moisten food with alkaline saliva. As a consequence, large starchy meals may result in discomfort and even gastric colic (Harris & Arkell 2005). Ralston (2007) states that feeding large grain meals is strongly correlated with ulceration but notes that careful verification through controlled studies is still needed. Estimated starch intakes of more than 2 g/kg bodyweight per day or 1 g/kg per meal were both found to be risk factors for the development of equine gastric ulceration syndrome (EGUS) in a study of Danish horses not in active race training (Luthersson et al 2009). Stall housing was previously reported as a risk factor for ulceration of the squamous epithelium (Murray & Eichorn 1996). Luthersson et al (2009) suggested that the free movement associated with grazing may help to move stomach contents through the gastrointestinal tract, but gut pH did not decrease over a 24-hour period when horses were stall-housed rather than grazing in paddocks (Husted et al 2008). As subjects had unrestricted access to hay in that study, perhaps stall housing as a risk factor reflects its tendency to co-occur with intermittent feeding rather than a direct effect of confinement per se. Fecal pH was observed to be lower in horses fed a pelleted diet compared with when the same horses ate ad-lib hay (Elia et al 2010).
The fact that crib-biting horses are similarly at increased risk of digestive upset including certain types of colic (Archer et al 2004, 2008, Hillyer et al 2002), gastric ulceration (Nicol et al 2002) and altered gut-transit time (McGreevy & Nicol 1998, McGreevy et al 2001) hints at an involvement of digestive processes in stereotypy. Interestingly, Waters et al (2002) found that sudden weaning methods that involve the isolation of foals also significantly increase risk of developing crib-biting. They suggest that this may reflect upset eating patterns in the young foal, stressed by the sudden removal of its mother and access to milk. Even short periods of not eating lead to stomach acidity, early onset of ulceration (Murray & Eichorn 1996, Nieto et al 2009) and an increased risk of some colics (Archer et al 2008). The newly isolated foal will therefore be particularly at risk of gut-related problems. Any of these digestive imbalances are likely to impair the welfare of affected horses but evidence is emerging that they may have economic consequences as well. Experimentally-induced gastric ulceration adversely affected physiologic indices of performance (Nieto et al 2009) and a case study reported that treatment of existing EGUS with the proton pump inhibitor omeprazole resulted in improved race earnings in 4 racehorses (Franklin et al 2008). This presents financial corroboration of previous findings that moderate to severe gastric ulceration in racehorses improved significantly faster with omeprazole versus treatment using the same compound minus the drug component (Murray et al 1997).
Because horses only produce saliva when chewing (Alexander 1966) but secrete acid into the stomach continuously, it has been proposed that horses suffering from foregut acidity problems may wood-chew or crib-bite in an attempt to stimulate additional saliva production (Nicol 1999b). In support of this hypothesis, crib-biting foals showed significantly greater evidence of stomach inflammation and early ulceration than normal foals (Nicol et al 2002). Feeding antacids reduces these clinical signs, and also tends to result in a reduction in crib-biting behavior (Nicol et al 2002, Mills & Macleod 2002). Measurement of saliva production from the submandibular gland indicated that crib-biting horses produced smaller quantities of saliva than non-stereotypic horses but that this difference was compensated for by the action of crib-biting (Moeller et al 2008). This study measured behavior over a short period and it would be interesting to explore how substantially crib-biting might mitigate the effects of low saliva production in the long term. Associations between crib-biting and gastrointestinal disturbance would seem to indicate that it is not wholly successful, but it is equally possible that such disturbance would be worse in the absence of crib-biting. This highlights the potential welfare consequences of simply treating the outward symptoms of stereotypies (such as the use of anti-cribbing collars) without understanding their underlying causes. If crib-biting becomes an anticipatory or habitual behavior, the initiating cause may also not be immediately apparent and this further complicates both diagnosis and treatment.
Large meals can overwhelm the capacity of the stomach and small intestine and increase transit rates of digesta. Potter et al (1992, cited in Delobel & Cuvelier 2008) concluded that more than 3.5–4 g/kg body weight starch per meal (about 3 kg of barley for a 500 kg horse) exceeded capacity for enzymatic digestion and increased the amount of starch reaching the posterior ileum. Luthersson et al (2009) recently advised a lower limit, 2 g starch/kg body weight per day or 1 g/kg per meal, as they found that higher starch levels were a risk factor for EGUS. Intense episodes of hydrolyzable carbohydrate fermentation in the hindgut and a resulting drop in pH can occur, as measured in the caecum (Willard et al 1977) or feces (Rowe et al 1994, dos Santos et al 2009). Correspondingly, feeding more than 4 kg grains (usually oats in this study) was recently identified as a risk factor for low fecal pH in New Zealand racehorses (Williamson et al 2007). This finding was not general to commercial pre-mixed feeds, possibly because cereal grains within such mixes tend to undergo a greater degree of processing, which in turn facilitates fermentation earlier in digestion (Harris & Arkell 2005). Tinker et al (1997) did report an increased colic risk associated with levels of concentrate feeding greater than 2.5 kg per day. Interestingly, in Williamson’s study, fecal pH was slightly but significantly higher in the very small sample of only six horses recorded as displaying stereotpyies. However, interpretation of this finding is difficult as these included both oral and locomotor stereotypies.
Discomfort caused by the overflow of readily fermentable carbohydrates into the hindgut has been linked with increased anxiety and aggression in rats (Hanstock et al 2004) and with undesirable oral behaviors including bed eating and wood chewing in horses. These were reduced by dietary supplementation with virginiamycin, an antibiotic which altered fecal pH (Johnson et al 1998). Presumably selective bacterial proliferation was halted, preventing rapid fermentation of sugar and starch. The study did not include any crib-biting horses. Willard et al (1977) also reported that cecal pH was lower in three horses fed a sweet concentrate feed compared with when the same individuals were fed only hay. Relative proportions of specific volatile fatty acids (VFAs) in the cecum were also altered, and infusions of disodium carbonate both increased cecal pH in concentrate-fed horses and reversed the pattern of VFA proportions. It was also noted that the infusions reduced the time spent wood chewing and engaging in coprophagy but the finding did not reach statistical significance in this very small sample.
Increased total gut transit time has been recorded in crib-biters (McGreevy et al 2001), and when both crib-biting and eating hay were prevented, orocecal transit time was significantly longer than in non-stereotypic horses (McGreevy & Nicol 1998). Freire et al (2008) later suggested that crib-biting and weaving are not influenced by hindgut acidosis. This conclusion was based on the failure of virginiamycin supplementation to reduce performance of these stereotypic behaviors but seems speculative in light of interpretational difficulties presented by the study design. For example, no mention was made of an increase in the amount of grain in the subjects’ diet partway through the study, revealed by examination of a companion paper (Freire et al 2009). Virginiamycin reduced exploratory behavior and altered water intake but similar results were attributed to the change in diet in the other study. Measurements were averaged across two time periods and this nominally accounted for the horses consuming different diets in each period, but the completely separate analyses permit no consideration of interactions or carry-over between periods.
Virginiamycin supplementation was apparently successful in mitigating the moderate to severe lameness due to laminitis that was otherwise induced by feeding 8 kg of maize-based pellets per day (Rowe et al 1994). It has been suggested that hindgut acidosis in racehorses may commonly lead to low-level laminitis that can limit the horse’s performance potential (Linford et al 1993