As global aquaculture expands, intensifies and becomes increasingly automated, the monitoring of health and welfare status of fish becomes increasingly challenging, particularly given the animals are out of sight underwater, and the population sizes are many times higher than in most terrestrial farming operations (Huntingford et al. 2018, 2023).
There is thus an increasing need to constantly monitor the welfare status of whole populations within farms, predict disease outbreaks and enable early interventions. Disease prediction of this nature was demonstrated by Maloy (2020) who used hydroacoustics to retrospectively detect behavioural indicators of pancreatic disease in Atlantic salmon, 4 weeks prior to its diagnosis by conventional methods.
The potential for such an application has led to development of a hydroacoustic system which can monitor the behavioural patterns of most of the fish population in large cages 24/7, independently of water quality and light levels. We have used this system to collect datasets covering extended periods on several commercial salmon farms, and are analysing them for behavioural signatures related to specific stressors or health events. The initial findings of these analyses will be presented. The work is part of WelfareShield, a project to develop a health & welfare monitoring system using hydroacoustic and optical sensors, appetite monitoring, AI and deep learning to provide fish farmers with important information on the status of their stocks in real time (See figure).
Huntingford, F.A., Kadri, S & Saraiva, J.L. (2023) Welfare of Cage Cultured Fish under Climate Change In: Woo & Subasinghe (eds) Climate Change on Diseases and Disorders of Finfish in Cage Culture, 3rd edition Wallingford, Oxfordshire, UK. CABI pp. 462-498
Huntingford, F.A., Turnbull J.F. & Kadri, S (2018) Methods to Increase Fish Production: Welfare and Sustainability Implications. In: Grandin & Whiting (eds) Are Pushing Animals to their Biological Limits? Welfare and Ethical Implications. Wallingford, Oxfordshire, UK. CABI
Maloy, H. (2020) EchoBERT: A Transformer-Based Approach for Behavior Detection in Echograms
(2020) IEEE Access 8, pp. 218372 - 218385