Precision Farming in Aquaculture: Use of a Non-Invasive, Ai-Powered Real-Time Automated Behavioural Monitoring Approach to Predict Gill Health and Improve Welfare in Atlantic Salmon (Salmo Salar) Aquaculture Farms

As the aquaculture industry is growing, more sophisticated technology is required to monitor farms and ensure sustainability and good fish welfare, in line with the precision livestock farming concept. Using behaviour as a non-invasive form of monitoring, in combination with artificial intelligence and machine learning, can allow for higher control over farm management. The goal of this study was to use a novel machine learning algorithm to quantify and assess changes to farmed Atlantic salmon (Salmo salar) behaviour related to fish health and welfare status. Main behaviours recorded were shoal-like cohesion, feeding, swimming activity and fish distribution in the cage. Video cameras were deployed within all cages in two Scottish Atlantic salmon marine farms. Furthermore, one cage in each farm was equipped with additional cameras (5 and 4 for site 1 and 2, respectively), for higher spatial coverage of fish behaviour and distribution throughout the cage. The algorithm processed video footage from these cameras and outputted behavioural data termed ‘activity’, which encompasses fish abundance, speed, and shoal cohesion. Additionally, gill health, Operational Welfare Indicators (OWI), mortality, and Specific Feeding Rate (SFR) were monitored, recorded and scored weekly at both sites. During the summer 2023, gill health issues arose at both farms, leading to fish stress which was reflected in the behavioural data. Prior to the onset of poor gill health, the average (± standard deviation) activity levels of the fish across all cages were 28.3 ± 10.5% and 24.9 ± 7.0% for site 1 and 2, respectively. Post-onset of poor gill health, the activity rose significantly in all cages with a mean of 43.3 ± 13.5% and 32.6 ± 9.6%, respectively. A generalised linear mixed model revealed that PGD was the main driver of this increase in activity. This increase in activity coincided with a fish migration to the centre of the cage, meaning the fish were shoaling tighter, which is a normal stress response often seen in relation to predators and other environmental or health stressors. Additionally, mortality increased and SFR decreased throughout the cages after the onset of poor gill health. Individual level OWIs did not change significantly, indicating that group-level OWIs, specifically changes to behaviour, are important for predicting and observing present impacts to fish health and welfare, while individual-based OWIs tend to emerge when fish are already compromised. The use of behaviour as a non-invasive welfare indicator and the potential to use artificial intelligence and machine learning to automatise the process of behavioural identification takes fish farming to a higher level of control, allowing farmers to offer their animals a ‘life worth living’, according to the best welfare standards.