Threat assessment by domestic ducklings using visual signals: implications for animal–machine interactions
Introduction
Information about the responses of animals to approaching stimuli is of practical use when considering the design of machines to work amongst animals. Autonomous robots have much potential in animal agriculture, where they could relieve humans of dangerous, difficult or monotonous tasks (Wathes, 1994), e.g. for inspection and herding. The dynamic responses of animals to machines and robots are unknown but are important if animal welfare is not to be compromised through inappropriate design or operation. Animals have not encountered machines in their evolutionary history, and have therefore not developed behaviours that are necessarily adaptive in their presence: machines do not intentionally represent historical enemies and may not evoke any fear responses. Milking robots, for example, do not form part of the evolutionary precepts of a cow, and the limited studies to date have shown rapid adaptation within a few days (Prescott et al., 1998). Mechanised broiler catchers can reduce injury and may induce a lower stress response than manual collection Duncan et al., 1986, Lacy and Czarick, 1998. Animals may adapt more readily to predictable machines than to an often unpredictable and inconsistent stockman: if a stimulus is highly predictable, its aversiveness is substantially diminished (Hemsworth et al., 1993).
Effective herding robots have to pose a sufficient threat on approach to interrupt ongoing animal behaviour without causing panic or flight reactions. In our experiments, different stimuli were evaluated for their effectiveness in interrupting feeding behaviour as they approached pairs of ducklings at a constant velocity. Domestic ducklings were used as a model species in these experiments because they have been used successfully in pilot studies of a herding robot (Vaughan et al., 2000). Three stimuli were compared in the first experiment. A human was selected as the stimulus that domestic birds are most likely to encounter in current commercial practice and also for comparison with related work (Henderson, 1999). The second stimulus was a vertical cylinder, which had a similar appearance to the herding robot developed by Vaughan et al. (2000). A model fox (Vulpes vulpes) was also chosen as it represented a more naturalistic and predatory stimulus with features that could potentially be added to a robot for more effective herding. Experiments 2 and 3 were specific investigations of the relevance of different features of the model fox in interrupting duckling feeding behaviour.
Section snippets
Animals
Three batches each of 36 female ducklings (Pekin×Aylesbury) were used. Day-old birds were supplied by a commercial producer and randomly allocated to one of three flocks upon arrival. Each flock of 12 ducklings was housed in an identical adjacent pen. Visual contact was not possible between flocks. All flocks were subject to identical husbandry procedures; each pen was cleaned out daily in the morning during training/experimental sessions.
All flocks were fed once a day at the same time relative
Results
The data for the latency to resume feeding were skewed and were transformed using a log10 transformation. As many of the values were zero, which represented cases when birds continued to feed as the stimulus approached, a period of 0.5 s was added to all values prior to transformation. The results of the analysis of variance supported the regression analysis (see Henderson, 1999).
Discussion
The overall pattern of duckling response in the three experiments was similar, indicating that the protocol developed was robust and that similar behavioural responses were measured. In the first experiment, the approaching human caused the greatest interruption of feeding behaviour at all distances. The human was the only live stimulus and differed from the other stimuli in features such as height, realistic walking movement and facial features. Birds may also have responded to information
Acknowledgements
We are grateful to James Francis and Len Burgess for technical assistance. Birds were provided by Cherry Valley Farms, UK. JVH was in receipt of a studentship from the Biotechnology and Biological Sciences Research Council.
References (20)
- et al.
The future of applied ethology
Appl. Anim. Behav. Sci.
(1993) - et al.
Fear reactions of domestic sheep confronted with either a human or a human-like model
Behav. Proc.
(1995) - et al.
Slides of conspecifics as representative of real animals in laying hens (Gallus domesticus)
Behav. Proc.
(1993) The functional organisation of anti-predator behaviour in the Pied Flycatcher. A study of avian visual perception
Anim. Behav.
(1975)- et al.
The assessment of fear in domestic hens exposed to a looming human stimulus
Behav. Proc.
(1981) - et al.
Are faces special for sheep? Evidence from facial and object discrimination learning tests showing effects of inversion and social familiarity
Behav. Proc.
(1996) - et al.
Mechanical harvesting of broilers
Poult. Sci.
(1998) Total panoramic vision in the mallard duck (Anas platyrhnchos)
Vision Res.
(1986)- et al.
Fear reactions of ewes to photographic images
Behav. Proc.
(1994) - et al.
Mimicry and the eye of the beholder
Proc. R. Soc. London, Ser. B
(1993)
Cited by (7)
Genomic signatures of selection, local adaptation and production type characterisation of East Adriatic sheep breeds
2023, Journal of Animal Science and BiotechnologyWelfare of ducks, geese and quail on farm
2023, EFSA Journal