Fear responses to novelty in testing environments are related to day-to-day activity in the home environment in dairy cattle

Behavioural tests for cattle take time to perform and can be stressful for the animals but are currently the only way of assessing behavioural reactions to fear-causing stimuli in a standardised manner. It may be possible to use behavioural data collected remotely in the home pen environment through the use of activity monitors and robotic milkers to identify fearful cattle without testing. In this study eighty five dairy cows were given a novel arena novel object (NANO) test and 79 of these were also human approach (HAP) tested, both thought to reflect fear. All animals had their activity recorded for 40 days prior to the testing period using a tri-axial accelerometer activity monitor. High numbers of novel object contacts in the test was associated with younger animals with fewer lying bouts per day and were less variable in their lying bout duration (View the MathML source, F3,75 = 4.65, P = 0.005). Cows with a higher tolerance for human approach had fewer lying bouts per day, a shorter average standing bout duration and presented themselves to the robot milker more often (View the MathML source, F3,69 = 3.12, P = 0.032). Personality traits constructed from a principle components analysis of the observed NANO behaviours were also associated with home pen activity. Cows which scored highly on the first component termed ‘neophobia’ were older, had more lying bouts and a greater variation in the duration of their average lying bout (View the MathML source, F3,75 = 5.32, P = 0.002) while cows which scored highly on the second component termed ‘boldness’ were older cows with less variation in their average lying bout duration (View the MathML source, F2,75 = 5.63, P = 0.005). To conclude, significant relationships exist between behaviours in short-term personality tests and home pen activity recorded over several weeks. As fearfulness is reflected in spontaneous home pen behaviours, activity databases could be incorporated into models predicting fearfulness and welfare assessment protocols.