Should Dairy Cattle Be Trained to a Virtual Fence System as Individuals or in Groups?

SIMPLE SUMMARY: A virtual fence (VF) system is being evaluated for commercial implementation in the Australian livestock industries. For this to work in dairy systems, cows will require training to learn the association between paired stimuli for livestock containment. We aimed to understand if cow learning and response to VF stimuli would differ when trained as individuals or in groups in a controlled experimental environment. Twenty-three dairy cows were trained to a VF as individuals or in groups of 5–6, and then moved to the alternate context to test the retention of learning. Cows trained in groups were more likely to interact with the VF when tested as individuals, indicating they might rely on the response of their conspecifics rather than directly receiving stimuli themselves. It is important that all individuals learn the association between stimuli to ensure they remain within a boundary, and to minimise potential welfare implications on animals that do not learn. However, training individual cattle is impractical, therefore, further work should evaluate effective group training protocols that provide the time and space for all individuals to learn the VF. ABSTRACT: Pre-commercial virtual fence (VF) neckbands (eShepherd(®), Agersens, Melbourne, Vic, Australia) can contain cows within a designated area without the need for physical fencing, through associative learning of a paired audio tone and electrical pulse. Cattle are gregarious, so there may be an impact of herd mates on the learning process. To evaluate this, a VF was set 30 m down one of three test paddocks with a feed attractant 70 m past the VF. Twenty-three Holstein-Friesian cows were all fitted with VF neckbands and trained as individuals or in groups (5–6) for four 10 min tests; then, cows were crossed over to the alternate context for two more 10 min tests. The number of cows breaking through the VF and the number of paired stimuli reduced across time (from 82% to 26% and 45% to 14%, respectively, p < 0.01). Cows trained in a group (88%) were more likely to interact with the VF in the crossover compared to those trained as individuals (36%) (p < 0.01), indicating an influence of group members on individual cow response. Individual training is impractical, therefore, future research should evaluate group training protocols ensuring all cows learn the VF to avoid any adverse impacts on animal welfare.