- Email: [email protected]
Avoidance of tape-recorded milking facility noise by dairy heifers in a Y maze choice task
Applied Animal Behaviour Science 109 (2008) 201–210 www.elsevier.com/locate/applanim
Avoidance of tape-recorded milking facility noise by dairy heifers in a Y maze choice task Naomi A. Arnold a,*, Kim T. Ng b, Ellen C. Jongman c, Paul H. Hemsworth a a
Animal Welfare Science Centre, The University of Melbourne, Victoria 3010, Australia b School of Psychology, Psychiatry and Psychological Medicine, Monash University, Clayton Campus, Wellington Road, Clayton, Victoria 3800, Australia c Department of Primary Industries, Victoria, Werribee Centre, 600 Sneydes Rd, Werribee 3030, Australia Accepted 5 February 2007 Available online 6 March 2007
Abstract The effect of noise pre-recorded from a commercial milking facility on the choice behaviour of dairy heifers in a Y maze was examined. Sixteen animals were individually trained to associate noise onset with a particular maze arm and no noise (quiet) with the other maze arm. After a maze familiarisation period, each heifer was exposed to a mixture of training trials, where access to only one maze arm was made available, and choice trials, where both maze arms were available. In order to reduce possible interference from animals’ original side preferences or response patterns under two-alternative choice situations, training and choice trials were interspersed and the noise stimulus was presented in the maze arm first chosen after familiarisation. Over 11 exposures to the maze across 3 days, choice of maze arm (during choice trials), avoidance behaviour (time taken to enter maze arm, number of stops, handler intervention required) and heart rate (HR) were measured. Results showed that the percentage of heifers that chose the quiet arm changed significantly ( p < 0.01) from 31.3% to 81.3% over the course of the experiment. In addition, during training trials, animals took longer ( p < 0.05) to enter the maze arm, stopped more, required more handler intervention when entering and were more restless in the noise compared to the quiet arm. During training trials there was also a trend ( p = 0.06) indicating some increase in HR during noise trials compared to quiet trials. Overall, the results of this experiment indicate that milking facility noise is fear-provoking for dairy heifers and that they will learn to avoid this noise when given the opportunity. The experiment has also demonstrated successful use of this Y maze choice methodology for assessment of environmental stimuli in dairy heifers (previously Y maze methods have only been used to assess handling and husbandry practices in cattle). # 2007 Elsevier B.V. All rights reserved. Keywords: Fear; Heart rate; Habituation; Y maze; Noise; Choice; Dairy cows
* Corresponding author. Tel.: +61 3 8344 8383; fax: +61 3 8344 5037. E-mail address: [email protected] (N.A. Arnold). 0168-1591/$ – see front matter # 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.applanim.2007.02.002
202
N.A. Arnold et al. / Applied Animal Behaviour Science 109 (2008) 201–210
1. Introduction Arnold et al. (2007) investigated the effect of unavoidable exposure to noise recorded from a milking facility on the behavioural and physiological responses of dairy cows. The findings indicate some activation of fear-related responses in the presence of this noise based on behavioural responses and changes in HR. The current experiment was aimed at assessing the impact of the same noise on heifers’ choice behaviour. A number of studies have investigated the aversiveness of noise for animals (but not cattle) using choice behaviour (e.g., McAdie et al., 1993; Talling et al., 1998). Use of both direct aversion and preference testing methods has been advocated by Pajor et al. (2003), who assessed responses of dairy cows to various handling treatments using both methods (Pajor et al., 2000, 2003). The way in which fear-related responses of cows to various stimuli, such as noise, manifest in preference behaviour (under volitional control) is an important consideration in commercial milking situations, as voluntary entry by cows into milking facilities is desirable. Thus, the present study was aimed at developing a methodology to assess the effect of tape-recorded milking shed noise on the choice behaviour (i.e., under animal’s volitional control) of dairy heifers. The Y maze as a measure of choice behaviour has been used to assess restraint procedures for sheep (Grandin et al., 1986; Rushen, 1986), deer (Pollard et al., 1994) and beef cattle (Grandin et al., 1994). In dairy cows, choices relating to various treatments, including feeding, shouting, electric shock, hitting (Pajor et al., 2003) and being milked (Prescott et al., 1998) have been assessed using Y maze methodology. Most of these studies did not include measurement of any additional indicators of fear, such as avoidance behaviour and/or physiological indices. Of those that did, Pollard et al. (1994) found that, in addition to a preference for no restraint, deer hesitated more and required more force to enter a crush restraint option than a no restraint option, providing additional validation for interpretation of an aversion to restraint. Because choice behaviour may be affected by different attributes of the stimulus or the animal that lead to attraction or avoidance responses (e.g., Prescott et al., 1998), the inclusion of alternative measurement techniques will assist with more accurate interpretation of choice outcomes. Furthermore, many Y maze experiments fail to account for a number of other factors which may affect overt choice behaviour; for example, preferred response patterns, original location preferences (e.g., Grandin et al., 1986), and the impact of other situational variables on ability to learn associations between maze location and stimulus presentation. For example, in the study by Pollard et al. (1994), animals that were trained using forced (i.e., only one option available) exposures to the two maze options more readily learnt location–stimulus associations than animals that were trained under conditions of free access to either option. Thus, forced training trials appear to be an important inclusion in Y maze procedures for use with farm animals. In addition to incorporating forced training trials (i.e., trials where only one option is made available in the Y maze), the methodology in the present experiment was also designed to minimise impact of location and response pattern preferences on choice behaviour. There are two common response patterns reported in the literature that can naturally occur under twoalternative choice conditions such as in a Y maze. The first is spontaneous alternation (Dember and Richman, 1989), where the animal oscillates between the two maze arms over a number of choice trials. The second is perseverance (Rodriguez et al., 1992), where the animal persists with a particular maze arm choice over repeated trials. In order to minimise potential interference of these response patterns in the current experiment, training trials were arranged in an alternating
N.A. Arnold et al. / Applied Animal Behaviour Science 109 (2008) 201–210
203
Table 1 Example sequence of trials for one session of Y maze procedure (forced trials 2–4 are alternating), based on an initial left choice at trial 1 Trial
Type
Maze side
1 2 3 4 5
Choice Forced Forced Forced Choice
Lefta Right Left Right (a) will go lefta if following original preference (b) will go left if following an alternation response pattern (c) will go right if avoiding noise onset
a
Assumed to be representative of any original side preference.
pattern (i.e., left side, right side, left side). This enabled a dissociation of preferences for particular response patterns from preferences for avoiding the test stimulus (see Table 1). Furthermore, choice trials (i.e., where both options are available) were interspersed with training trials in order to reduce repetition of consecutive visits to the same maze arm, which may occur if choice trials are grouped together. Repetition of visits to one maze arm may lead to formation of a location preference (based on familiarity) that could interfere with the effect of the test stimulus on choice behaviour. In addition, the noise stimulus was paired with the maze arm initially chosen on the first day of training as a method for controlling effects of any original side preference of individual animals (based on the assumption that the first arm chosen will be consistent with any side preference). Measurements of heart rate (HR), restlessness, difficulty moving the animal into the maze arms, and time taken to make a choice were included as supplementary indicators of aversion. If noise is an aversive stimulus, as indicated in Arnold et al. (2007), then choice behaviour should be consistent with avoidance of noise in the Y maze paradigm. 2. Method 2.1. Subjects Sixteen Holstein Friesian heifers were used in this experiment. The animals were 1.5 years old and five were 7 months pregnant at the time of this experiment. 2.2. Materials The experimental facility consisted of a Y-shaped maze built with 2 m high steel panels covered with Hessian sacking (see Fig. 1). There was a 2.5 m gap between the two inside panels of each maze arm. The noise stimulus was transmitted through a speaker positioned, at cow head height, at the halfway point between the two arms, facing away from the maze junction. At the junction of the maze arms, a gate (gate 2) was located that could be in one of three positions—open, right arm closed, or left arm closed. The noise used in the experiment was a playback of sounds present in a commercial milking facility during milking while the animals passed through the raceway. The noise consisted of several different components, such as milking machinery, animal (cow), human, stall gate hydraulics, and radio music. A sample of each of these sounds was combined into a 2 min segment and a random portion of this segment was played at each trial. The noise was played at a set volume that ensured heifers were exposed to 85 dB of intensity through the majority of the raceway. This was the average decibel level recorded in the milking facility where the tape recording was created. A standard tape recorder was used to record and play the noise.
204
N.A. Arnold et al. / Applied Animal Behaviour Science 109 (2008) 201–210
Fig. 1. Y maze apparatus used to determine if dairy heifers avoid noise (grey lines indicate point at which shoulder must pass for a choice to be recorded).
The entry to the Y maze was linked to a concrete yard and curved race leading to a standard cattle crush. The crush was used to restrain the heifers during attachment of HR monitors before the beginning of each animal’s experimental session. A Polar HorseTrainerTM transmitter (distributed by Pursuit Performance Pty Ltd., Adelaide, Australia), consisting of two electrodes connected to a transmitter and a Polar Accurex PlusTM Heart Rate wrist monitor were used to receive and record heart-rate signals telemetrically (within a range of 1 m). Positive electrodes were placed 10 cm below the top of the wither, just behind the shoulder blade. Negative electrodes were placed behind the left elbow. Obstetrical gel was applied to the electrodes to enhance conductivity of the connection between the electrodes and the animal’s skin. The electrodes were held in position with an elastic belt, to which the transmitter and wrist monitor were attached. 2.3. Procedure The 16 heifers were housed in two groups of 8 with one group completing the experimental task in the mornings and the other in the afternoons. At the beginning of each experimental session, the group was moved into a holding yard adjacent to the experimental facility. Each heifer was then individually moved into the crush and fitted with a heart rate monitor before completing the following procedures for each day of the experiment. Day 1: Four trials per heifer were completed on day 1 to familiarise animals with the maze apparatus. No noise stimulus was present in either maze arm for any of the trials on this day. In the first trial, each heifer was individually introduced to the maze in the open position. The animal was allowed 10 s to move into an arm. If no movement occurred, a handler moved behind the heifer and tapped her with a length of flexible plastic pipe on the middle of the top of her tail once every 5 s until a choice was made. Choices were defined as having taken place when the animal’s shoulders moved past a previously designated ‘choice line’ (see Fig. 1). Once a choice had been made, the animal was moved to the end of the chosen arm and held for 30 s before the end door was opened and the animal released to move through to the end of the building where a feed reward of calf meal was available. Trial 2 was carried out in a similar way, but gate 2 was positioned to enforce movement into the opposite arm to that chosen in trial 1. Trials 3 and 4 were also enforced runs, with alternating gate positions. The last trial of the session was a choice trial, with gate 2 in the open position. Trials were run consecutively for each heifer, with all four trials being completed for each animal before the next was run. Days 2 and 3: There was a total of five trials per heifer on each of these days, with choices on trials 1 and 5, and alternating enforced runs on trials 2–4 (again with gate positions on trial 2 enforcing movement into the opposite arm to that chosen in trial 1). A description of the trials is provided in Table 2. The same method of moving heifers through the maze was used as described for day 1 with the same feed reward available at the end of the building, and again, all five trials were run consecutively for each heifer before the next animal was run. On days 2 and 3, the tape recording of milking facility noise was played at 85 dB during the 30 s holding period in one arm (but not the other). For each heifer, the arm in which the noise was played was the arm chosen in trial 1 on day 2. The noise was switched on as soon as the animal crossed the ‘choice line’ of this arm and, for days 2–4, the noise was played only when the animal entered this ‘noise’ arm.
N.A. Arnold et al. / Applied Animal Behaviour Science 109 (2008) 201–210
205
Table 2 Trial descriptions (identical for day 2 and day 3) T1 T2 T3 T4 T5
Choice trial Animal forced in opposite arm to that chosen in T1 Forced trial—alternate arm to T2 Forced trial—alternate arm T3 Choice trial
Day 4: Each animal completed a single choice trial on day 4, with the maze in the open position. All trials were video-recorded and these records were used to measure time taken to make a choice (or move through the junction), force required (i.e., number of handler interventions required to encourage movement), and level of restlessness (i.e., proportion of time moving or lifting the hooves) during the 30 s holding period in the maze arms. In addition, HR in the maze junction area and during the holding period in each trial was recorded. 2.4. Statistical analysis Change in choice behaviour in relation to the noise and quiet maze arms was analysed using Cochran’s test for repeated observations (Cochran, 1950), where quiet choices were treated as ‘successes’ and noise choices as ‘failures’. Time to enter the maze arm, restlessness in the arm, HR, and ease of movement variables were analysed separately with ANOVA with repeated measures on treatment (noise vs. quiet) and day (day 2 versus day 3) using the Statistical Package for the Social Sciences (SPSS Inc., 1989–2004). These analyses were conducted on the third and fourth trial of each day (day 2 and 3) only. These trials were chosen because they include both a noise and a quiet trial (the second exposure to each for that day) and do not include a choice trial. Choice trials could not be included in this comparison because not all animals were represented in both noise and quiet options (as only one choice could be made per trial) and, depending on the choice made, animals were not consistently represented in both options. Type I error rate for all analyses was set at 0.05.
3. Results The choice of arm on all choice trials throughout the 4 days was recorded for each animal and is presented in Table 2. In the first trial on day 2, the majority of heifers (81.3%) chose the left arm. Consequently, all but three of the animals received the noise treatment in the left arm, because treatment to arm allocation was based on the first side entered after day 1. After completion of enforced trials on day 2, 5 of the 16 animals (31.3%) chose the quiet arm in the last choice trial of the day. The following day (day 3) 11 of the 16 (68.8%) chose the quiet arm in both trial 1 and trial 5. On day 4, 13 (81.3%) of heifers chose the quiet arm. 3.1. Description of choice pattern As shown in Fig. 2, tracking the distribution pattern of noise and quiet arm choices after the noise stimulus had been experienced (i.e., after trial 1 on day 2) provided a valuable perspective on choice behaviour in this experiment. Half (eight) of the heifers chose the quiet arm on every choice after the initial choice trials. Of the other eight animals, five chose quiet on the last choice and/or the second to last choice. The three remaining heifers still chose noise on the last trial and were also the only animals who made a choice switch from quiet back to noise at some point during the experiment.
206
N.A. Arnold et al. / Applied Animal Behaviour Science 109 (2008) 201–210
Fig. 2. Dendrite diagram showing the pattern of noise (N) and quiet (Q) side choices and the overall percentage of quiet arm choices made by the 16 heifers across each of the four choice trials after introduction of the noise stimulus in the Y maze task.
3.2. Preference for noise/quiet Analysis of choice outcomes for each of the four choice trials in the experiment was conducted to examine whether there was a difference over time in the likelihood of animals choosing the quiet arm (representing a learning effect). Cochran’s test for repeated observations was used to compare the proportion of animals making quiet arm choices across all the four trials. No overall difference in proportions was found (x2ð3;13Þ = 1.10, p = 0.594). However, when the proportion of noise and quiet choices for each of choice trials 2–4 with those of choice trial 1, Cochran’s test revealed a significant change in the proportion of heifers making quiet arm choices from choice 1 to choice 2 (x2ð1;15Þ = 7.20, p < 0.01), choice 1 to choice 3 (x2ð1;15Þ = 4.50, p < 0.01) and choice 1 to choice 4 (x2ð1;15Þ = 5.00, p < 0.01). 3.3. Analysis of measured variables The effects of the noise treatment on measurements of time to enter the maze arm, HR in the junction, HR in the arm, restlessness in the arm and ease of movement in the arm were all examined over days 2 and 3 using trials 3 and 4 on each day. The effects of treatment (noise versus quiet) and day (day 2 versus day 3) were examined using repeated measures ANOVA. Means for each measurement on days 2 and 3, are displayed in Fig. 3. Heifers took significantly longer to enter the maze arm during noise trials than quiet trials (F (1,15) = 15.691, p < 0.01) and were significantly more restless in the maze arm during noise trials than during quiet trials (F (1,15) = 12.785, p < 0.01). There was no interaction effect and no day effect on either of these measurements. There was a trend towards higher HR in the arm during noise trials than in quiet trials (F (1,15) = 4.289, p = 0.06). Further, HR in the maze arm was greater on day 2 than on day 3 (F (1,15) = 13.249, p < 0.05). There was no significant effect of noise on HR in the junction (F (1,15) = 1.929, p = 0.192). However, there was a significant decrease in HR in the junction from day 2 to day 3 (F (1,15) = 0.045, p = 0.016). There were no interaction effects on HR. Animals stopped significantly more in the junction during noise trials than during quiet trials (F (1,15) = 6.454, p < 0.023) and required more handler intervention during noise trials than quiet
N.A. Arnold et al. / Applied Animal Behaviour Science 109 (2008) 201–210
207
Fig. 3. Mean time to enter, movement in arm, HR variables and ease of movement variables for the 16 heifers during noise (black) and quiet (grey) training trials in the Y maze task (i.e., trials 3 and 4 on each day).
trials (F (1,15) = 13.492, p < 0.01). There were no interaction or day effects on either of these variables. 4. Discussion The analysis of choice behaviour in this study indicated a tendency for animals to switch from a noise arm choice to a quiet arm choice over the course of the experiment. This result suggests that, when given the opportunity, the majority of heifers developed a preference to avoid noise recorded from a commercial milking facility and that this preference developed after only two to three exposures to noise. Most heifers succeeded in learning the association between treatment and maze arm location. Similar successful learning performances have been reported in cows, including learning to associate Y maze arms with particular events (e.g., Pajor et al., 2003) and learning the spatial location of reward in radial arm mazes (e.g., Bailey et al., 1989). In addition to changes in choice behaviour, heifers were also more restless and showed a tendency toward increased HR while in the arm during noise than during quiet trials. Further, during noise trials, animals took longer to enter the maze arm, stopped more and required more handler intervention to successfully move through the apparatus compared to quiet trials. These findings are all indicative of a fear response to noise including avoidance of noise exposure and
208
N.A. Arnold et al. / Applied Animal Behaviour Science 109 (2008) 201–210
possible increased physiological (HR) and behavioural (restlessness) reactivity consistent with a flight response when exposed to the noise stimulus. This interpretation is supported by Waynert et al. (1999, p. 37) who stated that ‘‘Increased HR and escape movements are the normal components of the fear response in cattle’’. However, it should be noted that the observed trend of increases in heart rate and increased restlessness during noise trials may have been at least partially a result of the greater handler intervention that was required during noise trials than quiet trials. Nevertheless, increases in handler intervention were a direct result of avoidance behaviour (stops) during noise trials and therefore any increase in HR or restlessness was caused by the noise stimulus, either directly or indirectly. Heifers were exposed to noise in the maze a minimum of four and maximum of seven times (depending on choice outcomes) over days 2–4 and there was no statistical evidence of habituation of behavioural responses to the noise during this time. This finding is consistent with Waynert et al. (1999) who reported an acute stress response in heifers exposed to noise, with habituation of physiological but not behavioural responses after 5 days (one exposure per day). Similarly, Arnold et al. (2007), reported that although heifers showed habituation of HR response after the first exposure to noise recordings identical to those used in the current experiment, there was no evidence of habituation of escape-related (avoidance) behaviour during noise exposure found after 12 exposures. It is possible that habituation of these responses would have been observed if animals had been exposed to the noise over more trials. In the current experiment, analysis of variables during the first exposures on day 2 and 3 was not possible as they occurred over a mixture of choice and training trials which may have confounded results due to the qualitative differences between training and choice trials. Therefore it is possible that by trial 3 on day 2 (i.e., from the point that analyses were conducted) heifers’ HR responses had already partially habituated to the noise; supported by the finding of only a trend indicating higher HR during noise trials beyond this point. The decision to use only training trials to compare responses to noise and quiet exposure was made in order to remove a number of potentially confounding effects arising from the qualitative differences between having a choice (i.e., high controllability of outcome) and having only one alternative available (i.e., no controllability of outcome). For example, it has been suggested that a perceived lack of controllability can contribute to increases in stress (Wiepkema, 1987) or influence emotional responses (De´sire´ et al., 2002) and therefore heifers’ responses during training trials may be more pronounced than those during choice trials. As the specific designation of noise and quiet exposures to either training or choice trials was not experimentally controlled but was determined by the heifers’ own choice behaviour during initial trials on days 2 and 3, controllability (i.e., whether exposure was voluntary or forced) and order of presentation were not equally balanced amongst noise and quiet trials. Therefore, it was prudent to use only one type of trial for analyses of heifers’ behavioural and HR responses. This experiment also provides some information about choice behaviour patterns in dairy heifers. The noise stimulus was presented in the first maze arm that each individual animal entered on the first trial after familiarisation (i.e., trial 1 on day 2). This was based on the assumption that the first arm entered was likely to be the preferred side of the maze and, therefore, it would be prudent to present the hypothesised aversive stimulus in this preferred location in order to ensure a more conservative test of preference for the quiet arm. Some of the choice behaviour results are consistent with this assumption of an original preference. For example, on the first choice trial (i.e., trial 1 on day 2), just over 80% of the animals chose the left arm, which is considerably greater than the 50% that would have been expected if there had been an equal likelihood of animals choosing either arm of the maze. This evidence of a side
N.A. Arnold et al. / Applied Animal Behaviour Science 109 (2008) 201–210
209
preference in dairy heifers is supported by findings of a preferred side of the milking shed in dairy cows (Hopster et al., 1998; Paranhos da Costa and Broom, 2001). The majority of heifers altered their choice behaviour over the course of the experiment so as to avoid the noise stimulus. Interestingly, this preference expression in some animals appeared to change overnight, from choice trial 1 at the end of day 2 to choice trial 2 at the beginning of day 3. A spontaneous improvement in memory consolidation (process of forming associations; Gordon, 1989), known as an incubation effect (a result of consolidation processes occurring while the organism is at rest, through reactivation of the neural activity patterns corresponding to recent event Marr, 1971) may explain the observed change in choice behaviour in the absence of intervening training trials. Regardless of some anomalies in expression of preference through choice behaviour in the Y maze, the results of this study provide some evidence that dairy heifers learn to avoid noise recorded from a conventional dairy when given the opportunity. The animals were able to successfully learn an association between location in the Y maze and the noise stimulus in a limited amount of training, suggesting that cows can readily learn to avoid the location of an aversive stimulus. In addition, exposure to noise increased avoidance behaviour, as indicated by increases in stopping and amount of required handler intervention, as well as increasing restlessness, suggesting that noise in the milking facility has direct implications for on-farm efficiency related to improving cow flow and human–animal interactions during the milk harvesting process. Acknowledgements These experiments were funded by Dairy Australia. We wish to thank the technical staff at the Department of Primary Industries, Werribee Centre for their assistance with constructing the Y maze, setting up video recording equipment and handling animals during experimental sessions. References Arnold, N.A., Ng, K.T., Jongman, E.C., Hemsworth, P.H., 2007. The behavioural and physiological responses of dairy heifers to tape-recorded milking facility noise with and without a pre-treatment adaptation phase. Appl. Anim. Behav. Sci. 106, 12–25. Bailey, D.W., Rittenhouse, L.R., Hart, R.H., Richards, R.W., 1989. Characteristics of spatial memory in cattle. Appl. Anim. Behav. Sci. 23, 331–340. Cochran, W.G., 1950. The comparison of percentages in matched samples. Biometrika 37, 256–266. Dember, W.N., Richman, C.L., 1989. Spontaneous Alternation Behaviour. Springer-Verlag, New York. De´sire´, L., Boissy, A., Veissier, I., 2002. Emotions in farm animals: a new approach to animal welfare in applied ethology. Behav. Proc. 60, 165–180. Gordon, W.C., 1989. Learning and Memory. Pacific Grove, Brooks-Cole. Grandin, T., Curtis, S.E., Widowski, T.M., Thurmon, J.C., 1986. Electro-immobilisation versus mechanical restraint in an avoid–avoid choice test for ewes. J. Anim. Sci. 62, 1469–1480. Grandin, T., Odde, K.G., Schultz, D.N., Behrns, L.M., 1994. The reluctance of cattle to change a learned choice may confound preference tests. Appl. Anim. Behav. Sci. 39, 21–28. Hopster, H., van der Werf, J.T., Blokhuis, N.H.J., 1998. Side preference of dairy cows in the milking parlour and its effects on behaviour and heart rate during milking. Appl. Anim. Behav. Sci. 55, 213–229. Marr, D., 1971. Simple memory: a theory for archicortex. Phil. Trans. R. Soc. B 262, 23–81. McAdie, T.M., Foster, T.M., Temple, W., Matthews, L.R., 1993. A method for measuring the aversiveness of sounds to domestic hens. Appl. Anim. Behav. Sci. 37, 223–238. Pajor, E.A., Rushen, J., de Passille, A.M.B., 2000. Aversion learning techniques to evaluate dairy cattle handling practices. Appl. Anim. Behav. Sci. 69, 89–102.
210
N.A. Arnold et al. / Applied Animal Behaviour Science 109 (2008) 201–210
Pajor, E.A., Rushen, J., de Passille, A.M.B., 2003. Dairy cattle’s choice of handling treatments in a Y-maze. Appl. Anim. Behav. Sci. 80, 93–107. Paranhos da Costa, M.J.R., Broom, D.M., 2001. Consistency of side choice in the milking parlour by Holstein–Friesian cows and its relationship with their reactivity and milk yield. Appl. Anim. Behav. Sci. 70, 177–186. Pollard, J.C., Littlejohn, R.P., Suttie, J.M., 1994. Responses of red deer to restraint in a Y maze preference test. Appl. Anim. Behav. Sci. 39, 63–71. Prescott, N.B., Mottram, T.T., Webster, A.J.F., 1998. Relative motivations of dairy cows to be milked or fed in a Y-maze and an automatic milking system. Appl. Anim. Behav. Sci. 57, 23–33. Rodriguez, M., Gomez, C., Alonso, J., Afonso, D., 1992. Laterality, alternation and perseveration relationships on the Tmaze test. Behav. Neurosci. 106, 974–980. Rushen, J., 1986. Aversion of sheep for handling treatments: paired-choice studies. Appl. Anim. Behav. Sci. 16, 363–370. Talling, J.C., Waran, N.K., Wathes, C.M., Lines, J.A., 1998. Sound avoidance by domestic pigs depends upon characteristics of the signal. Appl. Anim. Behav. Sci. 58, 255–266. Waynert, D.F., Stookey, J.M., Schwartzkopf-Genswein, K.S., Webster, A.J.F., 1999. The response of beef cattle to noise during handling. Appl. Anim. Behav. Sci. 62, 27–42. Wiepkema, P.R., 1987. Behavioural aspects of stress. In: Wiepkema, P.R., van Adrichem, P.W.M. (Eds.), Biology of Stress in Farm Animals: An Integrative Approach. Martinus Nijhoff Publishers, Boston, pp. 111–133.
Avoidance of tape-recorded milking facility noise by dairy heifers in a Y maze choice task Naomi A. Arnold a,*, Kim T. Ng b, Ellen C. Jongman c, Paul H. Hemsworth a a
Animal Welfare Science Centre, The University of Melbourne, Victoria 3010, Australia b School of Psychology, Psychiatry and Psychological Medicine, Monash University, Clayton Campus, Wellington Road, Clayton, Victoria 3800, Australia c Department of Primary Industries, Victoria, Werribee Centre, 600 Sneydes Rd, Werribee 3030, Australia Accepted 5 February 2007 Available online 6 March 2007
Abstract The effect of noise pre-recorded from a commercial milking facility on the choice behaviour of dairy heifers in a Y maze was examined. Sixteen animals were individually trained to associate noise onset with a particular maze arm and no noise (quiet) with the other maze arm. After a maze familiarisation period, each heifer was exposed to a mixture of training trials, where access to only one maze arm was made available, and choice trials, where both maze arms were available. In order to reduce possible interference from animals’ original side preferences or response patterns under two-alternative choice situations, training and choice trials were interspersed and the noise stimulus was presented in the maze arm first chosen after familiarisation. Over 11 exposures to the maze across 3 days, choice of maze arm (during choice trials), avoidance behaviour (time taken to enter maze arm, number of stops, handler intervention required) and heart rate (HR) were measured. Results showed that the percentage of heifers that chose the quiet arm changed significantly ( p < 0.01) from 31.3% to 81.3% over the course of the experiment. In addition, during training trials, animals took longer ( p < 0.05) to enter the maze arm, stopped more, required more handler intervention when entering and were more restless in the noise compared to the quiet arm. During training trials there was also a trend ( p = 0.06) indicating some increase in HR during noise trials compared to quiet trials. Overall, the results of this experiment indicate that milking facility noise is fear-provoking for dairy heifers and that they will learn to avoid this noise when given the opportunity. The experiment has also demonstrated successful use of this Y maze choice methodology for assessment of environmental stimuli in dairy heifers (previously Y maze methods have only been used to assess handling and husbandry practices in cattle). # 2007 Elsevier B.V. All rights reserved. Keywords: Fear; Heart rate; Habituation; Y maze; Noise; Choice; Dairy cows
* Corresponding author. Tel.: +61 3 8344 8383; fax: +61 3 8344 5037. E-mail address: [email protected] (N.A. Arnold). 0168-1591/$ – see front matter # 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.applanim.2007.02.002
202
N.A. Arnold et al. / Applied Animal Behaviour Science 109 (2008) 201–210
1. Introduction Arnold et al. (2007) investigated the effect of unavoidable exposure to noise recorded from a milking facility on the behavioural and physiological responses of dairy cows. The findings indicate some activation of fear-related responses in the presence of this noise based on behavioural responses and changes in HR. The current experiment was aimed at assessing the impact of the same noise on heifers’ choice behaviour. A number of studies have investigated the aversiveness of noise for animals (but not cattle) using choice behaviour (e.g., McAdie et al., 1993; Talling et al., 1998). Use of both direct aversion and preference testing methods has been advocated by Pajor et al. (2003), who assessed responses of dairy cows to various handling treatments using both methods (Pajor et al., 2000, 2003). The way in which fear-related responses of cows to various stimuli, such as noise, manifest in preference behaviour (under volitional control) is an important consideration in commercial milking situations, as voluntary entry by cows into milking facilities is desirable. Thus, the present study was aimed at developing a methodology to assess the effect of tape-recorded milking shed noise on the choice behaviour (i.e., under animal’s volitional control) of dairy heifers. The Y maze as a measure of choice behaviour has been used to assess restraint procedures for sheep (Grandin et al., 1986; Rushen, 1986), deer (Pollard et al., 1994) and beef cattle (Grandin et al., 1994). In dairy cows, choices relating to various treatments, including feeding, shouting, electric shock, hitting (Pajor et al., 2003) and being milked (Prescott et al., 1998) have been assessed using Y maze methodology. Most of these studies did not include measurement of any additional indicators of fear, such as avoidance behaviour and/or physiological indices. Of those that did, Pollard et al. (1994) found that, in addition to a preference for no restraint, deer hesitated more and required more force to enter a crush restraint option than a no restraint option, providing additional validation for interpretation of an aversion to restraint. Because choice behaviour may be affected by different attributes of the stimulus or the animal that lead to attraction or avoidance responses (e.g., Prescott et al., 1998), the inclusion of alternative measurement techniques will assist with more accurate interpretation of choice outcomes. Furthermore, many Y maze experiments fail to account for a number of other factors which may affect overt choice behaviour; for example, preferred response patterns, original location preferences (e.g., Grandin et al., 1986), and the impact of other situational variables on ability to learn associations between maze location and stimulus presentation. For example, in the study by Pollard et al. (1994), animals that were trained using forced (i.e., only one option available) exposures to the two maze options more readily learnt location–stimulus associations than animals that were trained under conditions of free access to either option. Thus, forced training trials appear to be an important inclusion in Y maze procedures for use with farm animals. In addition to incorporating forced training trials (i.e., trials where only one option is made available in the Y maze), the methodology in the present experiment was also designed to minimise impact of location and response pattern preferences on choice behaviour. There are two common response patterns reported in the literature that can naturally occur under twoalternative choice conditions such as in a Y maze. The first is spontaneous alternation (Dember and Richman, 1989), where the animal oscillates between the two maze arms over a number of choice trials. The second is perseverance (Rodriguez et al., 1992), where the animal persists with a particular maze arm choice over repeated trials. In order to minimise potential interference of these response patterns in the current experiment, training trials were arranged in an alternating
N.A. Arnold et al. / Applied Animal Behaviour Science 109 (2008) 201–210
203
Table 1 Example sequence of trials for one session of Y maze procedure (forced trials 2–4 are alternating), based on an initial left choice at trial 1 Trial
Type
Maze side
1 2 3 4 5
Choice Forced Forced Forced Choice
Lefta Right Left Right (a) will go lefta if following original preference (b) will go left if following an alternation response pattern (c) will go right if avoiding noise onset
a
Assumed to be representative of any original side preference.
pattern (i.e., left side, right side, left side). This enabled a dissociation of preferences for particular response patterns from preferences for avoiding the test stimulus (see Table 1). Furthermore, choice trials (i.e., where both options are available) were interspersed with training trials in order to reduce repetition of consecutive visits to the same maze arm, which may occur if choice trials are grouped together. Repetition of visits to one maze arm may lead to formation of a location preference (based on familiarity) that could interfere with the effect of the test stimulus on choice behaviour. In addition, the noise stimulus was paired with the maze arm initially chosen on the first day of training as a method for controlling effects of any original side preference of individual animals (based on the assumption that the first arm chosen will be consistent with any side preference). Measurements of heart rate (HR), restlessness, difficulty moving the animal into the maze arms, and time taken to make a choice were included as supplementary indicators of aversion. If noise is an aversive stimulus, as indicated in Arnold et al. (2007), then choice behaviour should be consistent with avoidance of noise in the Y maze paradigm. 2. Method 2.1. Subjects Sixteen Holstein Friesian heifers were used in this experiment. The animals were 1.5 years old and five were 7 months pregnant at the time of this experiment. 2.2. Materials The experimental facility consisted of a Y-shaped maze built with 2 m high steel panels covered with Hessian sacking (see Fig. 1). There was a 2.5 m gap between the two inside panels of each maze arm. The noise stimulus was transmitted through a speaker positioned, at cow head height, at the halfway point between the two arms, facing away from the maze junction. At the junction of the maze arms, a gate (gate 2) was located that could be in one of three positions—open, right arm closed, or left arm closed. The noise used in the experiment was a playback of sounds present in a commercial milking facility during milking while the animals passed through the raceway. The noise consisted of several different components, such as milking machinery, animal (cow), human, stall gate hydraulics, and radio music. A sample of each of these sounds was combined into a 2 min segment and a random portion of this segment was played at each trial. The noise was played at a set volume that ensured heifers were exposed to 85 dB of intensity through the majority of the raceway. This was the average decibel level recorded in the milking facility where the tape recording was created. A standard tape recorder was used to record and play the noise.
204
N.A. Arnold et al. / Applied Animal Behaviour Science 109 (2008) 201–210
Fig. 1. Y maze apparatus used to determine if dairy heifers avoid noise (grey lines indicate point at which shoulder must pass for a choice to be recorded).
The entry to the Y maze was linked to a concrete yard and curved race leading to a standard cattle crush. The crush was used to restrain the heifers during attachment of HR monitors before the beginning of each animal’s experimental session. A Polar HorseTrainerTM transmitter (distributed by Pursuit Performance Pty Ltd., Adelaide, Australia), consisting of two electrodes connected to a transmitter and a Polar Accurex PlusTM Heart Rate wrist monitor were used to receive and record heart-rate signals telemetrically (within a range of 1 m). Positive electrodes were placed 10 cm below the top of the wither, just behind the shoulder blade. Negative electrodes were placed behind the left elbow. Obstetrical gel was applied to the electrodes to enhance conductivity of the connection between the electrodes and the animal’s skin. The electrodes were held in position with an elastic belt, to which the transmitter and wrist monitor were attached. 2.3. Procedure The 16 heifers were housed in two groups of 8 with one group completing the experimental task in the mornings and the other in the afternoons. At the beginning of each experimental session, the group was moved into a holding yard adjacent to the experimental facility. Each heifer was then individually moved into the crush and fitted with a heart rate monitor before completing the following procedures for each day of the experiment. Day 1: Four trials per heifer were completed on day 1 to familiarise animals with the maze apparatus. No noise stimulus was present in either maze arm for any of the trials on this day. In the first trial, each heifer was individually introduced to the maze in the open position. The animal was allowed 10 s to move into an arm. If no movement occurred, a handler moved behind the heifer and tapped her with a length of flexible plastic pipe on the middle of the top of her tail once every 5 s until a choice was made. Choices were defined as having taken place when the animal’s shoulders moved past a previously designated ‘choice line’ (see Fig. 1). Once a choice had been made, the animal was moved to the end of the chosen arm and held for 30 s before the end door was opened and the animal released to move through to the end of the building where a feed reward of calf meal was available. Trial 2 was carried out in a similar way, but gate 2 was positioned to enforce movement into the opposite arm to that chosen in trial 1. Trials 3 and 4 were also enforced runs, with alternating gate positions. The last trial of the session was a choice trial, with gate 2 in the open position. Trials were run consecutively for each heifer, with all four trials being completed for each animal before the next was run. Days 2 and 3: There was a total of five trials per heifer on each of these days, with choices on trials 1 and 5, and alternating enforced runs on trials 2–4 (again with gate positions on trial 2 enforcing movement into the opposite arm to that chosen in trial 1). A description of the trials is provided in Table 2. The same method of moving heifers through the maze was used as described for day 1 with the same feed reward available at the end of the building, and again, all five trials were run consecutively for each heifer before the next animal was run. On days 2 and 3, the tape recording of milking facility noise was played at 85 dB during the 30 s holding period in one arm (but not the other). For each heifer, the arm in which the noise was played was the arm chosen in trial 1 on day 2. The noise was switched on as soon as the animal crossed the ‘choice line’ of this arm and, for days 2–4, the noise was played only when the animal entered this ‘noise’ arm.
N.A. Arnold et al. / Applied Animal Behaviour Science 109 (2008) 201–210
205
Table 2 Trial descriptions (identical for day 2 and day 3) T1 T2 T3 T4 T5
Choice trial Animal forced in opposite arm to that chosen in T1 Forced trial—alternate arm to T2 Forced trial—alternate arm T3 Choice trial
Day 4: Each animal completed a single choice trial on day 4, with the maze in the open position. All trials were video-recorded and these records were used to measure time taken to make a choice (or move through the junction), force required (i.e., number of handler interventions required to encourage movement), and level of restlessness (i.e., proportion of time moving or lifting the hooves) during the 30 s holding period in the maze arms. In addition, HR in the maze junction area and during the holding period in each trial was recorded. 2.4. Statistical analysis Change in choice behaviour in relation to the noise and quiet maze arms was analysed using Cochran’s test for repeated observations (Cochran, 1950), where quiet choices were treated as ‘successes’ and noise choices as ‘failures’. Time to enter the maze arm, restlessness in the arm, HR, and ease of movement variables were analysed separately with ANOVA with repeated measures on treatment (noise vs. quiet) and day (day 2 versus day 3) using the Statistical Package for the Social Sciences (SPSS Inc., 1989–2004). These analyses were conducted on the third and fourth trial of each day (day 2 and 3) only. These trials were chosen because they include both a noise and a quiet trial (the second exposure to each for that day) and do not include a choice trial. Choice trials could not be included in this comparison because not all animals were represented in both noise and quiet options (as only one choice could be made per trial) and, depending on the choice made, animals were not consistently represented in both options. Type I error rate for all analyses was set at 0.05.
3. Results The choice of arm on all choice trials throughout the 4 days was recorded for each animal and is presented in Table 2. In the first trial on day 2, the majority of heifers (81.3%) chose the left arm. Consequently, all but three of the animals received the noise treatment in the left arm, because treatment to arm allocation was based on the first side entered after day 1. After completion of enforced trials on day 2, 5 of the 16 animals (31.3%) chose the quiet arm in the last choice trial of the day. The following day (day 3) 11 of the 16 (68.8%) chose the quiet arm in both trial 1 and trial 5. On day 4, 13 (81.3%) of heifers chose the quiet arm. 3.1. Description of choice pattern As shown in Fig. 2, tracking the distribution pattern of noise and quiet arm choices after the noise stimulus had been experienced (i.e., after trial 1 on day 2) provided a valuable perspective on choice behaviour in this experiment. Half (eight) of the heifers chose the quiet arm on every choice after the initial choice trials. Of the other eight animals, five chose quiet on the last choice and/or the second to last choice. The three remaining heifers still chose noise on the last trial and were also the only animals who made a choice switch from quiet back to noise at some point during the experiment.
206
N.A. Arnold et al. / Applied Animal Behaviour Science 109 (2008) 201–210
Fig. 2. Dendrite diagram showing the pattern of noise (N) and quiet (Q) side choices and the overall percentage of quiet arm choices made by the 16 heifers across each of the four choice trials after introduction of the noise stimulus in the Y maze task.
3.2. Preference for noise/quiet Analysis of choice outcomes for each of the four choice trials in the experiment was conducted to examine whether there was a difference over time in the likelihood of animals choosing the quiet arm (representing a learning effect). Cochran’s test for repeated observations was used to compare the proportion of animals making quiet arm choices across all the four trials. No overall difference in proportions was found (x2ð3;13Þ = 1.10, p = 0.594). However, when the proportion of noise and quiet choices for each of choice trials 2–4 with those of choice trial 1, Cochran’s test revealed a significant change in the proportion of heifers making quiet arm choices from choice 1 to choice 2 (x2ð1;15Þ = 7.20, p < 0.01), choice 1 to choice 3 (x2ð1;15Þ = 4.50, p < 0.01) and choice 1 to choice 4 (x2ð1;15Þ = 5.00, p < 0.01). 3.3. Analysis of measured variables The effects of the noise treatment on measurements of time to enter the maze arm, HR in the junction, HR in the arm, restlessness in the arm and ease of movement in the arm were all examined over days 2 and 3 using trials 3 and 4 on each day. The effects of treatment (noise versus quiet) and day (day 2 versus day 3) were examined using repeated measures ANOVA. Means for each measurement on days 2 and 3, are displayed in Fig. 3. Heifers took significantly longer to enter the maze arm during noise trials than quiet trials (F (1,15) = 15.691, p < 0.01) and were significantly more restless in the maze arm during noise trials than during quiet trials (F (1,15) = 12.785, p < 0.01). There was no interaction effect and no day effect on either of these measurements. There was a trend towards higher HR in the arm during noise trials than in quiet trials (F (1,15) = 4.289, p = 0.06). Further, HR in the maze arm was greater on day 2 than on day 3 (F (1,15) = 13.249, p < 0.05). There was no significant effect of noise on HR in the junction (F (1,15) = 1.929, p = 0.192). However, there was a significant decrease in HR in the junction from day 2 to day 3 (F (1,15) = 0.045, p = 0.016). There were no interaction effects on HR. Animals stopped significantly more in the junction during noise trials than during quiet trials (F (1,15) = 6.454, p < 0.023) and required more handler intervention during noise trials than quiet
N.A. Arnold et al. / Applied Animal Behaviour Science 109 (2008) 201–210
207
Fig. 3. Mean time to enter, movement in arm, HR variables and ease of movement variables for the 16 heifers during noise (black) and quiet (grey) training trials in the Y maze task (i.e., trials 3 and 4 on each day).
trials (F (1,15) = 13.492, p < 0.01). There were no interaction or day effects on either of these variables. 4. Discussion The analysis of choice behaviour in this study indicated a tendency for animals to switch from a noise arm choice to a quiet arm choice over the course of the experiment. This result suggests that, when given the opportunity, the majority of heifers developed a preference to avoid noise recorded from a commercial milking facility and that this preference developed after only two to three exposures to noise. Most heifers succeeded in learning the association between treatment and maze arm location. Similar successful learning performances have been reported in cows, including learning to associate Y maze arms with particular events (e.g., Pajor et al., 2003) and learning the spatial location of reward in radial arm mazes (e.g., Bailey et al., 1989). In addition to changes in choice behaviour, heifers were also more restless and showed a tendency toward increased HR while in the arm during noise than during quiet trials. Further, during noise trials, animals took longer to enter the maze arm, stopped more and required more handler intervention to successfully move through the apparatus compared to quiet trials. These findings are all indicative of a fear response to noise including avoidance of noise exposure and
208
N.A. Arnold et al. / Applied Animal Behaviour Science 109 (2008) 201–210
possible increased physiological (HR) and behavioural (restlessness) reactivity consistent with a flight response when exposed to the noise stimulus. This interpretation is supported by Waynert et al. (1999, p. 37) who stated that ‘‘Increased HR and escape movements are the normal components of the fear response in cattle’’. However, it should be noted that the observed trend of increases in heart rate and increased restlessness during noise trials may have been at least partially a result of the greater handler intervention that was required during noise trials than quiet trials. Nevertheless, increases in handler intervention were a direct result of avoidance behaviour (stops) during noise trials and therefore any increase in HR or restlessness was caused by the noise stimulus, either directly or indirectly. Heifers were exposed to noise in the maze a minimum of four and maximum of seven times (depending on choice outcomes) over days 2–4 and there was no statistical evidence of habituation of behavioural responses to the noise during this time. This finding is consistent with Waynert et al. (1999) who reported an acute stress response in heifers exposed to noise, with habituation of physiological but not behavioural responses after 5 days (one exposure per day). Similarly, Arnold et al. (2007), reported that although heifers showed habituation of HR response after the first exposure to noise recordings identical to those used in the current experiment, there was no evidence of habituation of escape-related (avoidance) behaviour during noise exposure found after 12 exposures. It is possible that habituation of these responses would have been observed if animals had been exposed to the noise over more trials. In the current experiment, analysis of variables during the first exposures on day 2 and 3 was not possible as they occurred over a mixture of choice and training trials which may have confounded results due to the qualitative differences between training and choice trials. Therefore it is possible that by trial 3 on day 2 (i.e., from the point that analyses were conducted) heifers’ HR responses had already partially habituated to the noise; supported by the finding of only a trend indicating higher HR during noise trials beyond this point. The decision to use only training trials to compare responses to noise and quiet exposure was made in order to remove a number of potentially confounding effects arising from the qualitative differences between having a choice (i.e., high controllability of outcome) and having only one alternative available (i.e., no controllability of outcome). For example, it has been suggested that a perceived lack of controllability can contribute to increases in stress (Wiepkema, 1987) or influence emotional responses (De´sire´ et al., 2002) and therefore heifers’ responses during training trials may be more pronounced than those during choice trials. As the specific designation of noise and quiet exposures to either training or choice trials was not experimentally controlled but was determined by the heifers’ own choice behaviour during initial trials on days 2 and 3, controllability (i.e., whether exposure was voluntary or forced) and order of presentation were not equally balanced amongst noise and quiet trials. Therefore, it was prudent to use only one type of trial for analyses of heifers’ behavioural and HR responses. This experiment also provides some information about choice behaviour patterns in dairy heifers. The noise stimulus was presented in the first maze arm that each individual animal entered on the first trial after familiarisation (i.e., trial 1 on day 2). This was based on the assumption that the first arm entered was likely to be the preferred side of the maze and, therefore, it would be prudent to present the hypothesised aversive stimulus in this preferred location in order to ensure a more conservative test of preference for the quiet arm. Some of the choice behaviour results are consistent with this assumption of an original preference. For example, on the first choice trial (i.e., trial 1 on day 2), just over 80% of the animals chose the left arm, which is considerably greater than the 50% that would have been expected if there had been an equal likelihood of animals choosing either arm of the maze. This evidence of a side
N.A. Arnold et al. / Applied Animal Behaviour Science 109 (2008) 201–210
209
preference in dairy heifers is supported by findings of a preferred side of the milking shed in dairy cows (Hopster et al., 1998; Paranhos da Costa and Broom, 2001). The majority of heifers altered their choice behaviour over the course of the experiment so as to avoid the noise stimulus. Interestingly, this preference expression in some animals appeared to change overnight, from choice trial 1 at the end of day 2 to choice trial 2 at the beginning of day 3. A spontaneous improvement in memory consolidation (process of forming associations; Gordon, 1989), known as an incubation effect (a result of consolidation processes occurring while the organism is at rest, through reactivation of the neural activity patterns corresponding to recent event Marr, 1971) may explain the observed change in choice behaviour in the absence of intervening training trials. Regardless of some anomalies in expression of preference through choice behaviour in the Y maze, the results of this study provide some evidence that dairy heifers learn to avoid noise recorded from a conventional dairy when given the opportunity. The animals were able to successfully learn an association between location in the Y maze and the noise stimulus in a limited amount of training, suggesting that cows can readily learn to avoid the location of an aversive stimulus. In addition, exposure to noise increased avoidance behaviour, as indicated by increases in stopping and amount of required handler intervention, as well as increasing restlessness, suggesting that noise in the milking facility has direct implications for on-farm efficiency related to improving cow flow and human–animal interactions during the milk harvesting process. Acknowledgements These experiments were funded by Dairy Australia. We wish to thank the technical staff at the Department of Primary Industries, Werribee Centre for their assistance with constructing the Y maze, setting up video recording equipment and handling animals during experimental sessions. References Arnold, N.A., Ng, K.T., Jongman, E.C., Hemsworth, P.H., 2007. The behavioural and physiological responses of dairy heifers to tape-recorded milking facility noise with and without a pre-treatment adaptation phase. Appl. Anim. Behav. Sci. 106, 12–25. Bailey, D.W., Rittenhouse, L.R., Hart, R.H., Richards, R.W., 1989. Characteristics of spatial memory in cattle. Appl. Anim. Behav. Sci. 23, 331–340. Cochran, W.G., 1950. The comparison of percentages in matched samples. Biometrika 37, 256–266. Dember, W.N., Richman, C.L., 1989. Spontaneous Alternation Behaviour. Springer-Verlag, New York. De´sire´, L., Boissy, A., Veissier, I., 2002. Emotions in farm animals: a new approach to animal welfare in applied ethology. Behav. Proc. 60, 165–180. Gordon, W.C., 1989. Learning and Memory. Pacific Grove, Brooks-Cole. Grandin, T., Curtis, S.E., Widowski, T.M., Thurmon, J.C., 1986. Electro-immobilisation versus mechanical restraint in an avoid–avoid choice test for ewes. J. Anim. Sci. 62, 1469–1480. Grandin, T., Odde, K.G., Schultz, D.N., Behrns, L.M., 1994. The reluctance of cattle to change a learned choice may confound preference tests. Appl. Anim. Behav. Sci. 39, 21–28. Hopster, H., van der Werf, J.T., Blokhuis, N.H.J., 1998. Side preference of dairy cows in the milking parlour and its effects on behaviour and heart rate during milking. Appl. Anim. Behav. Sci. 55, 213–229. Marr, D., 1971. Simple memory: a theory for archicortex. Phil. Trans. R. Soc. B 262, 23–81. McAdie, T.M., Foster, T.M., Temple, W., Matthews, L.R., 1993. A method for measuring the aversiveness of sounds to domestic hens. Appl. Anim. Behav. Sci. 37, 223–238. Pajor, E.A., Rushen, J., de Passille, A.M.B., 2000. Aversion learning techniques to evaluate dairy cattle handling practices. Appl. Anim. Behav. Sci. 69, 89–102.
210
N.A. Arnold et al. / Applied Animal Behaviour Science 109 (2008) 201–210
Pajor, E.A., Rushen, J., de Passille, A.M.B., 2003. Dairy cattle’s choice of handling treatments in a Y-maze. Appl. Anim. Behav. Sci. 80, 93–107. Paranhos da Costa, M.J.R., Broom, D.M., 2001. Consistency of side choice in the milking parlour by Holstein–Friesian cows and its relationship with their reactivity and milk yield. Appl. Anim. Behav. Sci. 70, 177–186. Pollard, J.C., Littlejohn, R.P., Suttie, J.M., 1994. Responses of red deer to restraint in a Y maze preference test. Appl. Anim. Behav. Sci. 39, 63–71. Prescott, N.B., Mottram, T.T., Webster, A.J.F., 1998. Relative motivations of dairy cows to be milked or fed in a Y-maze and an automatic milking system. Appl. Anim. Behav. Sci. 57, 23–33. Rodriguez, M., Gomez, C., Alonso, J., Afonso, D., 1992. Laterality, alternation and perseveration relationships on the Tmaze test. Behav. Neurosci. 106, 974–980. Rushen, J., 1986. Aversion of sheep for handling treatments: paired-choice studies. Appl. Anim. Behav. Sci. 16, 363–370. Talling, J.C., Waran, N.K., Wathes, C.M., Lines, J.A., 1998. Sound avoidance by domestic pigs depends upon characteristics of the signal. Appl. Anim. Behav. Sci. 58, 255–266. Waynert, D.F., Stookey, J.M., Schwartzkopf-Genswein, K.S., Webster, A.J.F., 1999. The response of beef cattle to noise during handling. Appl. Anim. Behav. Sci. 62, 27–42. Wiepkema, P.R., 1987. Behavioural aspects of stress. In: Wiepkema, P.R., van Adrichem, P.W.M. (Eds.), Biology of Stress in Farm Animals: An Integrative Approach. Martinus Nijhoff Publishers, Boston, pp. 111–133.