through the race, or the mean force required to move the animal through the race. Nor did these variables differ at any one specific trial. Of the 13 heifers on the food trial,
Ž .
only 8 started to eat consistently defined as eating on two consecutive trials by the fourth trial. Since this may have effected the results, the data were re-analysed using
only these 8 animals. However, the results of the analysis were the same as when the entire data set was used.
6. Discussion
The results demonstrate the effectiveness and limitations of the aversion race to evaluate handling practices for dairy cattle. Experiment 1 demonstrated the empirical
validity of the aversion race by comparing treatments known to differ in aversiveness. Adult cows clearly demonstrated which treatments were aversive and which were
rewarding; cows that were hit and shouted at took longer to enter the race took more time and required more force to move down the race than cows that were given food.
These results are in agreement with other studies showing that cows avoid handlers that
Ž .
hit them Breuer et al., 1997; Munksgaard et al., 1997; Rushen et al., 1999 and Ž
. approach handlers that feed them Jago et al., 1999, Rushen et al., 1999 . The use of the
aversion race to clearly differentiate between a known positive and a known negative handling treatment allows us to conclude that it is a valid method of evaluating the
aversiveness of various handling practices.
Experiment 1 also examined human–animal contact that did not involve feeding Ž
. brushing and petting . Non-aggressive forms of tactile contact between humans and
Ž .
animals gentling, petting, brushing have often been reported to decrease fear of Ž
humans and make handling easier in cattle Boissy and Bouissou, 1988; Boivin et al., .
1992; de Passille et al., 1996; Munksgaard et al., 1997 . However, these studies combine
´
contact with providing food. The few studies that have looked only at gentle contact Ž
. report conflicting results Boissy and Bouissou, 1988; Hemsworth et al., 1996 . Our
results suggest that for dairy cows, brushing was more aversive than being given food or receiving no treatment at all, at least during initial treatments. These results support the
Ž .
conclusion of Jago et al. 1999 who found that gentling was less effective than giving food in the development of human–animal interactions. However, the differences
between cows on the brushing and feeding treatments decreased with time suggesting Ž
. that cows may habituate to the brushing. Similarly, Boivin et al. 1998 reported that
acceptance of contact appears to be more the result of a habituation process than one of positive reinforcement.
In the second experiment, we compared techniques commonly used to move cattle. Cows treated with an electric prod took more time and required more force to walk
down the race than control cows, and this was the most aversive treatment on the final trials. Both cattle and swine have been found to avoid their handlers that use electric
Ž .
prods Gonyou et al., 1986; de Passille et al., 1996 . Cows that were shouted at in our
´
experiment also took more time and required more force to move through the race than control cow, and on the earlier trials this appeared to be as aversive as the electric prod.
Our conclusion that animals find a human shouting to be aversive is supported by
Ž .
Ž Waynert et al. 1999 who found that beef cattle responded more higher heart rates and
. more movements when played the sounds of a shouting human voice than metal
Ž .
clanging. In contrast, Hemsworth et al. 1986 found no differences in the approach behaviours of young pigs to humans using a loud voice or a soft quiet voice.
The race did not find that the other handling treatments, tail twist and hitting, to be significantly more aversive than the control. Twisting the tail of the cow results in the
Ž .
animal moving forward and is thought to be aversive Grandin, 1994 . Although our results suggest that tail twisting is only slightly aversive, it must be done with great care
as tail breakage may result. Hitting the cow on the rump with an open hand was not found to be significantly more aversive than control. Hitting has been shown to increase
Ž flight distance in both cows Breuer et al., 1997; Munksgaard et al., 1997; Rushen et al.,
. Ž
. 1999 and pigs Gonyou et al., 1986; Hemsworth and Barnett 1991 and cows being hit
Ž .
are reported to defecate and urinate more Munksgaard et al., 1997 . However, hitting Ž
. the rump of the animal as done in experiment 2 may be relatively less aversive to the
cow then hitting the muzzle of the animal. Finally, the inability of the aversion race to discriminate significantly between hitting and control indicates the limits on the sensitiv-
ity of the test. There are numerous procedural and animal factors that may explain the failure to clearly distinguish between some handling treatments. The inability to detect
differences between handling treatments in experiment 2 is largely due to the large variation among individual cows. This was substantially less in experiment 1 than in
experiment 2, which may be due to differing experience that the animals had with human contact. The non-lactating cows on experiment 2 had little daily contact with
people, while the lactating cows on the first experiment were handled at least twice a day.
Heifers are generally more difficult to handle then adult cows; the refusal of dairy heifers to enter the milking parlour is a well-known problem in the dairy industry
Ž .
Ž .
Albright and Arave, 1997; Bremner, 1997 . Hutson 1985 suggested that providing a positive reward immediately after handling may make animals easier to move in the
future. Since providing animals with food was found to be a positive treatment in Ž
experiment 1 and gentling is thought to be positive for calves Boissy and Bouissou, .
1988 , we hypothesised that heifers would move through a race more easily if these treatments were provided at the end of the race. This was not the case, as heifers that
received these treatments did not differ in the amount of time or force required to walk down the race in comparison with heifers that did not receive any treatment. Heifers that
had little experience with people were offered food from a person’s hand. Consequently, they may have found both gentling and taking food from the hand of a relatively
unfamiliar person aversive.
One limitation with the use of aversion learning techniques is that the animals must learn to associate moving to the end of the race with the treatment they receive at the
end of the race, which requires that they be exposed to the treatment a number of times. This repeated exposure may change the aversiveness of the treatment as the animals get
used to the treatment. For example, in experiment 2, animals on the shout treatment took a great deal of time to walk through the race in the middle of the experiment but
appeared to habituate to the treatment and the time to walk through the race declined over the final three trials. Habituation to sound has been previously reported in studies
Ž .
Ž .
with pigs Talling et al., 1996 and beef cattle Waynert et al. 1999 . Repeated exposure is also important to minimise other sources of fear associated with using the race. The
novelty of the procedure and the handling involved with moving the animal to and from Ž
the race are potential sources of fear Boissy, 1995; Schwartzkopf-Genswein et al., .
Ž .
1997 which repeated exposure should minimise Grandin, 1980 . In all of our experi- ments, the amount of time taken by animals on the control treatment decreased over the
course of the experiment. Of the measures taken, we conclude that the latency taken to move the animal into
the race is least useful. Although our measure of latency to enter the race differed Ž
between treatments on experiment 1, it was generally a less sensitive measure more .
variable than the amount of time and force required to move through the race. This may have been because it was the most difficult measure to standardise. Although animals
were moved to the race entrance in a standardised manner, occasionally animals would balk at the entrance to the race, turn and run back toward the pre-test holding pen. As a
result the overall latency to enter the pen could be a reflection of the skills of the handler, the distance the animal ran back towards the holding pen, the size of the animal
or other factors not associated with the treatment. The force to move through the race is difficult to standardise, and this also affects the total time taken to move through the
race. Although the handler always used the same pattern of force application, the intensity could have varied. Furthermore, if the animal required a great deal of force to
move through the race then animals latency to enter and time to move through the race may be associated with the application of force rather than the treatment received at the
end. Automating the moving of the animal down the race, in a manner where constant pressure is applied to the cow and resistance to this pressure is quantified would
improve the quality of the measures taken in the race.
7. Conclusion