Article Text


Behavioural and physiological effects of virginiamycin in the diets of horses with stereotypies
  1. R. Freire, BSc, PhD1,
  2. H. A. Clegg, BAppSc, MScVetSc1,
  3. P. Buckley, BVSc, DipVetClinStud, MVetClinStud, MACVSc1,
  4. M. A. Friend, BSc, PhD1 and
  5. P. D. McGreevy, BVSc, PhD, MACVSc, MRCVS2
  1. 1 School of Animal and Veterinary Sciences, Charles Sturt University, Wagga Wagga, NSW 2650, Australia
  2. 2 Faculty of Veterinary Science, University of Sydney, Sydney, NSW 2006, Australia


The effects of dietary supplements of virginiamycin on the behaviour and physiology of 17 thoroughbred geldings (five cribbers, six weavers and six control horses) were compared with the effects of a placebo over a period of 16 weeks. Virginiamycin had no effect on the horses' stereotypic behaviour, but it reduced their explorative behaviour, possibly owing to a reduction in feeding motivation. Virginiamycin increased the water intake of the cribbers and decreased the water intake of the control horses, but it was not possible to eliminate possible confounding factors for this effect. Virginiamycin had no other significant effects on the behaviour or physiology of the horses, and had no effect on the digestibility of their diets.

Statistics from

STEREOTYPIC behaviours, defined by being repetitive, relatively unchanging and apparently functionless, commonly occur in intensively managed horses (Nicol 1999a). Grain-based, low-roughage diets have been suggested as a possible cause for the development and performance of stereotypies, possibly as a result of increased gastric acidity (Nicol 1999a, Nicol and others 2002) or increased hindgut fermentation (Willard and others 1977, Johnson and others 1998, Cuddeford 1999, Rowe and others 2001, Zeyner and others 2002, Zeyner and others 2004).

A horse's stomach is at its most acidic when it is empty for long periods, as happens when the horse is fed small, concentrated feeds (Nicol 1999a, Nicol and others 2002). It has been suggested that oral stereotypies may be due either to an attempt to buffer this acidity by increasing the production of saliva (Nicol 1999b), or to gastric ulceration related to the acidity (Nadeau and others 2000). Nicol and others (2001) and Mills and Macleod (2002) reduced cribbing behaviour in young and adult horses, respectively, by feeding an antacid diet, suggesting that gastric acidity may play a significant role in oral stereotypy. The reduction was particularly evident in the period after feeding (Mills and Macleod 2002), a period that has been particularly associated with intense stereotypic behaviour by cribbers (Kennedy and others 1993, Gillham and others 1994, McGreevy and others 1995).

It has also been suggested that the development and expression of oral stereotypies may be due to the reduction in hindgut pH caused by the fermentation of grain in the hindgut (Willard and others 1977, Johnson and others 1998, Cuddeford 1999, Rowe and others 2001, Zeyner and others 2002, Zeyner and others 2004). However, the mechanism by which a reduction in hindgut pH may influence a horse's behaviour is not fully understood. Studies by Willard and others (1977), Johnson and others (1998) and Zeyner and others (2004) found that concentrate diets low in roughage increased the occurrence of a variety of undesirable behaviours, including wood-chewing, coprophagy, bed-eating, grasping, nervousness, restlessness and aggressive behaviour. The increase in these behaviours was accompanied by a decrease in faecal pH, suggesting that hindgut acidosis might be implicated in abnormal behaviours when horses are fed highly concentrated diets. It has been suggested that hindgut acidosis may cause visceral discomfort, resulting in the expression of anxiety-related oral behaviours (Johnson and others 1998).

The addition of virginiamycin to the diet of various species has been shown to reduce the accumulation of lactic acid in the rumen or hindgut that results from the fermentation of grain (Nagaraja and others 1987, 1995, Godfrey and others 1993, 1995, Rowe and others 1994, Thorniley and others 1996, 1998, Marounek and others 1998, Al Jassim and Rowe 1999). However, it has also been shown to cause a reduction in feed intake when first included in the diet of sheep and cattle (Thorniley and others 1996) and horses (Rowe and others 1994), possibly because it reduces palatability. Johnson and others (1998) observed that horses supplemented with virginiamycin ate more slowly, possibly owing to a reduction in palatability, although their observations were not quantified or tested for significance. Thorniley and others (1996) recorded a significant reduction in the digestibility of feed when virginiamycin was included in the diet of sheep and cattle. Its effects on the digestibility of horses' diets have not been determined.

The addition of virginiamycin to the diet of horses has been shown to decrease bed-eating and grasping significantly, and tended to decrease wood-chewing (Johnson and others 1998); the faecal pH of horses fed virginiamycin was significantly higher than the faecal pH of horses fed concentrate diets with no additional virginiamycin. When the faecal pH and oral behaviour during virginiamycin supplementation were compared with control data, there was significant correlation between them, with a lower pH being associated with greater expression of behavioural abnormalities. The study did not involve any crib-biting or wind-sucking horses, but it was proposed that there might be a similar relationship between caecal/faecal pH and these stereotypies (Johnson and others 1998).

Waters and others (2002) showed that foals weaned on to a concentrate diet were four times as likely to develop crib-biting as foals weaned on to a forage diet, suggesting that crib-biting may be strongly related to diet. The link between diet and weaving is less well established, but Thorne and others (2005) showed that increasing the amount of foraging that horses can express, for a period as short as seven days, is sufficient to reduce or eliminate weaving.

The hypothesis for this study was that the addition of virginiamycin to the diet of horses with established stereotypies would reduce the expression of their oral behaviours by reducing fermentative acidosis in the hindgut. The virginiamycin was also expected to reduce stereotypic behaviour in weaving horses, possibly to a smaller extent than cribbing, because weaving is a locomotory stereotypy thought to be more closely related to frustration at confinement (McAfee and others 2002, Mills and Davenport 2002) than to digestive effects.



Seventeen adult thoroughbred geldings, ranging in age from three to 18 years, were studied. Their case histories and initial observations of their stereotypic behaviour were used to classify them as five cribbers, six weavers and six control horses. Their initial bodyweights ranged from 464 to 630 kg, with mean (se) initial weights for the cribbers, weavers and nonstereotypic controls of 541 (27) kg, 503 (16) kg and 539 (19) kg, respectively


The horses were housed in stables (4 × 4 m) that allowed olfactory, visual and tactile contact with neighbouring horses on three sides. The walls consisted of wooden panelling 1·2 m high, topped with vertical steel bars (15 mm diameter, 10 cm apart) extending a further 1·2 m. Vertical bars above the stable doors prevented the horses putting their heads over the doors. Sawdust bedding was used on the concrete floors and tap water was provided ad libitum in 60 litre plastic bins.

All the horizontal surfaces in the stables were made of metal. As an alternative to cribbing on metal, a wooden bar (approximately 60 cm long × 10 cm × 8 cm) was fixed to a side wall of each stable to facilitate and focus crib-biting. A standard video surveillance camera was mounted above each stable and linked to a sequential switcher and time-lapse video recorder. Constant fluorescent lighting allowed for behavioural recording throughout the night, but the horses were returned to natural photoperiods after the experiment.

Experimental design

Experimental work was carried out over two 17-week periods, August to December 1999 and February to June 2001, with 10 horses tested in the first replicate and seven in the second replicate. In the first week the horses were familiarised with their stables and this was followed by 16 weeks of treatment. The horses spent 22 hours a day in their stables and were given free exercise in a paddock for two hours each afternoon, starting at 13.30, during which period no data were collected.

The horses were fed a diet of whole oats, wheaten chaff, lucerne chaff and lucerne hay, in accordance with nrc (1989) guidelines, at an energy concentration of 10 mj digestible energy/kg dry matter. They were given two equal, light-work feeds of 2·5 per cent bodyweight at 10.00 and 17.00 each day.

Virginiamycin was fed as Founderguard (Virbac), a dietary supplement designed to prevent laminitis, which inhibits the development of lactic acidosis associated with the onset of this disease (Rowe and others 1994). The placebo was obtained from the same manufacturer. Each additive was mixed with the oat portions of the evening feed at a rate of 5 g/100 kg bodyweight, as prescribed by the manufacturer. The placebo was fed during weeks 1 to 4 and 9 to 12, and the virginiamycin product was fed during weeks 5 to 8 and 13 to 16. The changes in diet between the treatments were made gradually over the first week of each treatment period to allow the digestive systems of the horses to adapt.

Sampling for behaviour, stress physiology and digestibility were carried out on all the horses throughout the 16-week trials. Blood glucose concentration, orocaecal transit time (octt) and the digestibility of the diets were measured in the five cribbers and six control horses at the end of weeks 4, 8, 12 and 16.

Routine measurements


The horses were under constant video surveillance during the 22 hours per day that they were in the stables. A sequential switcher was attached to the camera system, so that each horse was filmed for 30 seconds every five minutes. The video recordings were later viewed and the presence or absence of nine behaviours (Table 1) was recorded for each 30-second period (Martin and Bateson 1993). The video recordings were marked only with the date, so that when they were viewed the observer was blind to the dietary treatment. It was, however, obvious from the behaviour of the horses which were cribbers, weavers or control horses.

Stress physiology

The concentration of cortisol was measured in venous blood taken from each horse by venepuncture every second morning at 09.00, by means of a radioimmunoassay (ria) kit (Active Cortisol ria dsl — 2100; Diagnostic Systems Laboratories).

The horses wore heart rate monitors (Accurex Plus; Polar Electro) attached to lunge rollers, which recorded their average daily heart rate (excluding turnout time).

Orocaecal transit time

A method developed for use in horses by McGreevy and Nicol (1998) was used to determine the octt. After a baseline prefeeding blood sample had been taken, the horses were dosed orally with a sulfasalazine paste (Salazopyrin; Kabi Pharmacia) at a rate of 50 mg/kg, followed immediately by the morning feed. Venous blood samples were collected from each horse every 30 minutes for the following seven hours. Sulfapyridine was detected by high-performance liquid chromatography, as described by McGreevy and Nicol (1998). The threshold of its first appearance was set at the point where the ratio of sulfapyridine to the internal standard (sulfamethazine) reached 0·02.


N-alkane concentrations in feed and faeces were measured to determine feed digestibility, using the method described by Mayes and others (1986) and modified by Dove (1992). The horses wore harnesses throughout the trial (Equisan) to facilitate the collection of faeces.

Grab-samples of faeces were taken from the faecal collection harnesses every six hours for a period of 48 hours during the final week of each treatment, that is, weeks 4, 8, 12 and 16. Feed and faecal samples were stored at −4°C, before being dried to a constant weight at 60°C and ground through a 1 mm sieve using a Cyclotec 1093 sample mill (Tecator).

Alkane concentrations were measured using the alkane extraction method described by Dove (1992). Values for C31 were chosen for use in the analysis, because its concentration is high in feed and faeces and its concentrations had the highest correlation between feed and faecal values. The raw data were corrected for faecal recovery, using a mean (se) recovery value of 94·0 (2·8) per cent (O'Keefe and McMeniman 1998).

Feed/faecal pH

The pH of the feed samples containing virginiamycin was compared with the pH of samples containing the placebo. Five replicate samples of 10 g of each feed sample were ground and mixed with 50 ml distilled water. The pH of the resultant suspension was measured after 30 minutes. Unfortunately, the pH of the samples of faeces could not be measured because the samples were lost owing to the breakdown of a freezer.

Blood glucose

Venous blood samples were collected through an indwelling jugular catheter at the same 30-minute intervals as those for the octt trial. They were analysed immediately for blood glucose, using a blood glucose monitor and glucose testing strips (Accu-Chek Advantage; Roche Diagnostics). Blood glucose levels were not measured in the 10 horses in the first replicate, but were measured in the seven horses in the second replicate.

Food intake

Before the horses were given their morning feed, any uneaten food was removed and weighed to determine the daily food intake.

Water intake

Each afternoon, the volume of water required to refill each 60 litre water bucket was recorded to measure the volume of water the horse had consumed.


The horses were weighed once a week on a walk-on electronic weighbridge just before they were fed at 17.00.

Statistical analyses

Because the placebo and drug treatments were given in two alternating periods (that is placebo, drug, placebo, drug), an average for each treatment was calculated for the two periods to reduce any possible time effects. The horses' behaviour, water intake, cortisol concentration and heart rate were analysed in an analysis of variance (anova) with group (weavers, cribbers and control horses) as a between-subject factor and drug treatment (virginiamycin and placebo) as a within-subject factor.

Blood glucose measurements were obtained from only seven horses (one cribber, two weavers and four control horses) and analysed in a repeated measures anova. Measurements of octt and digestibility were obtained from only 10 horses (five cribbers and five control horses) and they were analysed in a repeated measures 2 × 2 anova.

All the variables except the observations of stereotypic behaviour met the criteria for parametric analysis; the measures of stereotypic behaviour were (ln + 1) transformed to normalise their distribution. Rubbing behaviour was observed infrequently and was not analysed. SPSS version 14.0 was used to perform the analyses.


Virginiamycin did not reduce cribbing or weaving and did not affect these two types of stereotypic activities differently (Fig 1). There were no significant effects on the time the horses spent drinking each day (Table 2), but virginiamycin seemed to increase the water intake of the cribbers but decrease the water intake of the control horses, with no effect on the weavers (drug/horse group interaction, P=0·02) (Fig 2).


Mean (se) numbers of stereotypic behaviours expressed per day by five cribbers and six weavers when they were fed diets containing virginiamycin or a placebo


Daily water consumption of five cribbers, six weavers and six control horses while they were fed diets containing virginiamycin or a placebo

As expected, the frequency and form of stereotypic behaviour differed between the three groups of horses, with the control horses showing no cribbing or weaving. However, there were no significant differences between the three groups in their frequency of eating, exploring, lying down, urinating or defecating, or in the interactions between the groups and their drug treatment (Table 2). Virginiamycin did reduce the amount of exploratory behaviour by the horses (P=0·01), but had no significant effects on the frequency of eating, lying down, urinating and defecating (Table 2).

Virginiamycin had no significant effect on the horses' plasma cortisol concentrations (P=0·06) (Table 3), and there was no difference between the groups of horses and no significant drug treatment/horse group interaction. Virginiamycin had no significant effect on the horses' heart rate (Table 3).

Virginiamycin had no significant effect on the horses' peak blood glucose response, octt or the digestibility of their diet (Table 3). The mean (sd) pH of the diet containing virginiamycin was 5·6 (0·09), the same as the pH of the diet containing the placebo 5·6 (0·1).


The different stereotypic behaviours of the cribbers and weavers were not significantly changed by the virginiamycin treatment. Virginiamycin had different effects on the horses' drinking behaviour, increasing the water intake of the cribbers but decreasing the water intake of the weavers, and having no effect on the water intake of the control horses. Virginiamycin reduced the explorative behaviour of all the horses but had no other effect on their behaviour. It increased the horses' plasma cortisol concentrations, but not significantly (P=0·06) and had no other effects on their physiology or on the digestibility of their diet.

Rowe and others (1994), Johnson and others (1998) and Al Jassim and Rowe (1999) all found that supplementing the diet of horses with virginiamycin reduced the acidity of their faeces, so it seems reasonable to assume that in this study virginiamycin reduced hindgut fermentation. Since dietary supplementation with virginiamycin did not result in any change in the horses' stereotypies, it is suggested that cribbing and weaving may not be influenced by hindgut acidosis. However, if postprandial cribbing is related more to foregut pathology (Nicol 1999a, Nicol and others 2001, 2002, Mills and Macleod 2002) than to hindgut acidosis (Willard and others 1977, Johnson and others 1998, Cuddeford 1999), the lack of effect of virginiamycin on the frequency of cribbing would not be not surprising.

The increase in drinking by the cribbers when they were fed virginiamycin suggests that they may have been trying to flush an aversive flavour from their mouths, but if that were the case it is not clear why the control horses and weavers did not also drink more. Caution is required in interpreting this finding, because it may have been related to the ambient temperature and humidity, which were not recorded. Exploratory behaviour in stabled horses has been interpreted to indicate that they may be searching for forage (Goodwin and other 2002), and the reduction in exploration by all the horses fed virginiamycin may indicate a reduction in their motivation to find food. If so, the findings are consistent with studies indicating that virginiamycin increases the time taken by horses to eat a feed (Johnson and others 1998) and reduces the total amount of feed they consume (Rowe and others 1994). However, any effect on the feeding motivation of the horses appears to have been minor, because neither their frequency of eating nor the time they took to consume the meal completely were affected by virginiamycin.

The results of this study suggest that if virginiamycin reduces hindgut fermentation, as has been reported by other researchers, the change in caecal pH it causes has no effect on the signs of physiological distress or stereotypic behaviour of adult horses with established stereotypies, and affects only their exploration of their stables and possibly their drinking behaviour. Virginiamycin also had no effect on the horses' octt or blood glucose concentrations, and did not affect the digestibility of their diets


This project was completed with the financial assistance of the New South Wales Racing Research Fund and the in-kind contribution of Virbac International. Thanks are due to Professor Martin Sillence, Kellie Munn, Dr Hugh Dove and Terry Flynn for their technical assistance with sample and data analysis. Thanks are also due to the owners for the loan of the horses studied.


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.