This is an entry in the 2019 Australian Dairy Conference's Young Dairy Scientist Award.
Virtual fencing (VF) technology has great potential to alleviate the costs, time and labour devoted to the establishment and maintenance of traditional physical fences. Virtual fence technology as collar-mounted devices (eShepherd, Agersens Pty Ltd, Melbourne, Vic, Australia), are now available for commercial use.
Virtual fencing works through ‘associative learning’, whereby a cow associates an audio cue with an electrical pulse. Animals receive a warning audio tone when they reach a VF boundary, with continued movement forward resulting in the delivery of a short, sharp electrical pulse.
Over time the animal learns to associate the audio tone with the electrical pulse to remain within a VF boundary. An online application allows the farmer to remotely create or change the virtual fence boundary, allowing for real-time control and flexibility of cow location and movement.
Currently, there is limited knowledge on how individual cows learn and respond to the technology, and how this is influenced by herd mates. Cattle are gregarious animals; therefore, complex herd interactions may influence or alter response to VF cues. An understanding of these relationships is essential for the effective adoption of VF into pasture-based livestock systems.
We conducted an experiment using multiparous Holstein-Friesian dry cows at the University of Sydney’s Mayfarm. The cows were fitted with an experimental prototype automated virtual fencing collar (eShepherd, Agersens Pty Ltd, Melbourne, VIC, Australia). We assessed animal learning and the individual variability of dairy cow response to VF cues, and whether the response changes when that individual is exposed to a VF in a group.
For the experiment, a large laneway of 100 metre x 20m of mown ryegrass pasture was used. A VF boundary was set at 50m down the laneway with a feed attractant of lucerne cubes at the end.
Twenty-four cows were trained to access the feed attractant before the VF was activated. The VF was then activated, and cows were divided into two treatments for training to VF cues, either as an individual or in a group of six cows.
Cows were tested four times in their treatments, after which time they were crossed over to assess the retention of learning across contexts; whereby individuals were tested in groups and groups as individuals for a final two tests.
The behavioural response of cows to VF cues, the number of cues each cow received (audio tone and electrical pulse) and whether animals crossed the VF to eat the feed attractant were recorded.
Our results show that cows can learn a VF relatively quickly and can be successfully contained within a grazing allocation. Within four interactions with a VF, the number of animals crossing the VF was below 10 per cent (see Figure 1). There was a small difference in learning between animals trained to the VF as individuals versus those trained in groups.
After the crossover, animals that were trained in groups were more likely to cross the VF to reach the feed when tested individually (20pc), in comparison to those trained as individuals when moved to a group (4pc).
These results have important implications for training of cows to a VF. It is clear that in the context of our experiment, not all animals in the group interacted with the VF equally.
We noted some animals responded to the reactions of their herd mates to the VF, and therefore did not interact directly with the VF to learn the association of the cues. Therefore, future training protocols should allow for more time and less grazing pressure so that individual animals have the opportunity to equally interact with the VF and establish adequate learning.
These findings indicate that dairy cows can learn a VF to remain within a boundary, however, highlights the requirement for appropriate training for successful implementation. Future work by the Dairy Science Group at the University of Sydney will focus on how the technology can be used to control cow movement to and from the dairy, and to control sub-groups of animals within a herd. This program is supported by the Australian Government of Agriculture and Water Resources as part of its Rural R&D for Profit program. Ultimately, experiments like this will help direct future research and incorporation of the eShepherd collars into the pasture-based dairy systems.
The six finalists were: Caelie Richardson, Latrobe University (Vic).
Read her article on Breeding burpless bovines Teanna Cahill, Murdoch University (WA).
Read her article on What mastitis bugs grow on your farm? Patricia Colusso, University of Western Sydney (NSW).
Read her article on Learning of a virtual fence by cows Juan Guragilo, University of Sydney (NSW).
Read his article on Larger and tech-savvy farms Chaya Smith, Latrobe University (Vic).
Read her article on Improving perennial ryegrass quality Felicity Searle, Murdoch University (WA).
Read her article on Getting more bucks per bang