Discussion

As anticipated in the first hypothesis, individual sheep did stay further away from the group sheep when a live stimulus was presented than with the less biologically significant novel object. This suggests that the distance maintained from the group sheep in the arena test can be taken to indicate the aversiveness of the stimulus to the sheep. Erhard (2003) also found that individual sheep stayed further away from their companions when a more aversive stimulus was presented (forward-facing human versus human facing away).

Sheep also stayed further away from the group sheep when the Dog was present than when the Goat was present, suggesting that they found the Dog more aversive. This result agrees with the traditional beliefs about the relationships between sheep and dogs, and sheep and goats. Domestic dogs are widely documented to be predators of sheep (e.g. ColI, 1 922; Roy and Dorrance, 1 976; Boggess et al., 1 978; Robel et al., 1 98 1 ; Schaefer et aI., 1 98 1). Experimentally, their presence has been shown to elicit

physiological stress responses, including increases in heart rate (MacArthur et aI., 1 979; Harlow et aI., 1 987; Baldock and Sibly, 1 990) and in plasma concentrations of cortisol, ACTH, adrenaline and noradrenaline (Harlow et aI., 1 987; Komesaroff et aI. , 1 998). Goats are not predators and neurophysiological evidence suggests that they are recognized, along with conspecifics, as non-threatening by sheep (Kendrick and Baldwin, 1 987).

It was also hypothesized that differences in other behaviours would be observed when different stimuli were presented. Differences in vigilance behaviours, exploration and other behaviours such as foot stamping, vocalization, urination, lip-licking and departures were observed. Individual sheep were highly vigilant and spent little time exploring the arena when the Dog was present. Baldock and Sibly ( 1 990) found that sheep spent more time alert in the presence of both a dog and human, than with a human alone, and dogs have previously been reported to inhibit investigatory behaviour in sheep (Torres-Hemandez and Hohenboken, 1 979). There is generally a trade off between exploratory and vigilance behaviour in prey animals. This is because these behaviours reflect competing motivations for resource acquisition and predator avoidance (Illius and FitzGibbon, 1 994).

Sheep also glanced at the group sheep more often when the dog was present. This result agrees with evidence that Bonnet Macaques tend to look at nearby troop members when presented with leopard models, possibly to capitalize on their companions' assessment

of any possible threat (Co ss and Ramakrishnan, 2000). Griffin et al. (2000) speculated that social cues become more salient in a 'risky' environment.

S heep foot stamped almost exclusively when the Dog was present, but its presence seemed to inhibit urination and departures. This pattern of behaviour may have evolved specifically in response to predators (Dwyer, 2004). For example, it may be dangerous to turn your back on a predator in order to walk away. Indeed, a sheep departing from a predator is reported to stimulate attack (Jansen, 1 974; Connolly et aI. , 1 976). Likewise, urination would provide olfactory cues to a potential predator in the vicinity.

While the presence of the human or dog elicited significantly more lip-licking than the presence of the box, the significance of this sheep behaviour in terms of fear or aversion

is unknown. However, taken with the other results of this study, a higher frequency of lip-licking may reflect more fear or aversion in sheep in the arena.

Vocalization did not appear to be suppressed by the Dog or Human, as has been

previously reported (Torres-Hernandez and Hohenboken, 1 979; Romeyer and Bouissou, 1 992) . Rather, the Goat appeared to elicit more vocalization than the other stimuli. In

addition, sheep sniffed the Goat more often than the Human or Dog. It is contended that the Goat was recognized as a strange conspecific, and the test sheep may have been contact calling. This idea is supported by the separation, along the second dimension of the canonical discriminant analysis, of the Goat from the three other stimuli, suggesting that the Goat was perceived in a different category. Neurophysiological evidence shows that the image of a homed goat stimulated the same cells involved in facial recognition as a homed sheep of a different breed to the test sheep, implying that goats may be recognized in a similar emotional category as strange conspecifics (Kendrick, 1 99 1 ). It

would have been interesting to present a ' Stranger Sheep' stimulus as well, to compare behavioural responses between a strange conspecific and the Goat.

It was hypothesized that observed differences in the distance maintained from the group sheep and other behaviours would indicate that the Human was less aversive than the

Dog, but more aversive than the Goat. Sheep presented with the Goat or Human came and stayed closer to the group sheep, were less vigilant and explored more, and showed less fear-related behaviour (Fearfulness Score) than when the Dog was present.

Moreover, the distance maintained and almost all other behaviours were similar in response to the Goat and Human. These results suggest that the Human was less aversive than the Dog, and similarly aversive to the Goat. However, while sheep often sniffed the goat, they rarely sniffed the human. The integrated Fearfulness Scores also indicate that the sheep exhibited more fear-related behaviour in the presence of the Human than with the Goat, suggesting that the Human may have been more aversive than the Goat.

Whether dogs or humans cause more fear in sheep is not known, and the nature of the relationship between humans and sheep is uncertain (Rushen, 1 990). Some postulate that humans are viewed as dominant conspecifics, while others argue that humans are seen as predators (Rushen et aI. , 1 999). B ased on neurophysiological and behavioural evidence, Kendrick and B aldwin ( 1 987) suggested that dogs and humans are recognized with similar emotional significance. However, the results of this study suggest that this particular Dog was more aversive to individual sheep than the Human presented.

The use of live non-human animals in this experiment may be problematic, as

inconsistencies in stimulus behaviour can influence the responses of the test sheep. For example, the Goat was somewhat aggressive towards test sheep, which may account for the fact that exploration of the area near the Goat was inhibited. In addition, while I could ask the human to maintain a non-threatening demeanour, I could not control the dog's behaviour, beyond keeping it in square 2 . However, in order to determine the behavioural responses of sheep to a range of 'real' stimuli, the use of live animals was deemed most appropriate.

There may also be confounding effects of differences in experience with the stimuli on the behavioural responses of the sheep (Cockram, 2004). However, these test animals had been born on the same farm, and had run as one mob since weaning, and their experiences could be expected to be as consistent as any group of normal farm animals. These sheep were raised outdoors, with some regular contact with humans and dogs. Most of this contact would be considered aversive e.g. herding, yarding, docking,

drenching. However, this contact represents the normal experience of intensively managed sheep in New Zealand, the behavioural responses of which I wished to study in the arena test. B ecause experience with a specific human, dog or goat will influence a sheep ' s subsequent behavioural response to that stimulus, all stimuli used were

unfamiliar to the test sheep. The sheep had had little, if any, experience with goats. Therefore, the goat may have been considered by the sheep to be somewhat novel, and subsequently, some of their behaviour towards the goat could have reflected this novelty e.g. increased vocalization.

2.6 Conclusions

The results of this experiment suggest that the arena test is an appropriate tool to test the relative aversiveness of different stimuli to individual sheep, as suggested by Erhard (2003). Distance from the group sheep, and levels of vigilance and exploratory

behaviour differed between sheep presented with different stimuli. In addition, the Dog elicited a pattern of behaviours that may have specifically evolved in response to a predator. In this arena test, individual sheep found the Dog most aversive, and the Box least. The Goat and Human appeared similarly aversive, except that the sheep rarely sniffed the human. The integrated Fearfulness Score indicates that, overall, this Human elicited more fear-related behaviour than this Goat, suggesting that the Human may have been slightly more aversive. Future work should correlate physiological reactions to different stimuli with behavioural responses in the arena test.