Aquaculture Europe 2023

September 18 - 21, 2023

Vienna,Austria

Add To Calendar 19/09/2023 15:45:0019/09/2023 16:00:00Europe/ViennaAquaculture Europe 2023INVESTIGATING SALMON BEHAVIOURAL CHANGES TO A STRESSOR FOR AUTOMATED FEEDING AND IMPROVED WELFARE IN AQUACULTURE FARMSStrauss 3The European Aquaculture Societywebmaster@aquaeas.orgfalseDD/MM/YYYYaaVZHLXMfzTRLzDrHmAi181982

INVESTIGATING SALMON BEHAVIOURAL CHANGES TO A STRESSOR FOR AUTOMATED FEEDING AND IMPROVED WELFARE IN AQUACULTURE FARMS

Meredith Burke* , Dragana Nikolic, Pieter Fabry, Hemang Rishi, Trevor Telfer,  and  Sonia Rey Planellas

 

Institute of Aquaculture , University of Stirling , Stirling, UK, FK9 4LA

meredith.burke@stir.ac.uk

 



Introduction

 Aquaculture is expanding globally, valued at  USD 281.5 billion in 2020 , with Atlantic salmon dominating finfish production at 2.7 million tonnes annually (FAO, 2022). As the industry grows, more sophisticated technology is needed to monitor farms and ensure their sustainability. 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 (Yang et al., 2020) . The development of algorithms to analyse fish behaviour related to feeding may be used to fully automate the feeding process and reduce environmental and economical waste. The goal of this study is to identify changes to Atlantic salmon ( Salmo salar ) behaviour related to responses to stressors such as the  well boat  treatment for gill disease (freshwater)  and sea lice (FLS).

Materials and Methods

For this study, 5 cameras were deployed at a Scottish Atlantic salmon farm consisting of 10 cages, each 100 m in circumference and ~15 meters in depth. The cameras were deployed in one cage in the following orientation: 3 down the centre (4 m, 9 m, 14 m), 2 at 9 m on the inner and outer areas of the cage, respectively. An algorithm was created by Observe Technologies to process video footage from these cameras and transform it into behavioural data useful for farmers ( e.g., activity, speed, schooling, shoaling).  For this project, daily internal validation occurs whereby experts compare the videos to  the  output from the algorithm. The analysis for this study used activity as a proxy for distribution, as an increase in activity corresponds to an increase in the number of fish observed in the videos. Activity was recorded f or 5 days surrounding the mechanical treatment  and statistical differences were determined with two-sided Kolmogorov-Smirnov tests. Additionally, latency to feed was documented as the time between when feeding was turned on to when activity increased at camera 2.

Results

 The  fish were distributed predominantly in two areas of the cage, at mid-depth towards the inner farm and at the bottom of the cage, which was observed 2 days prior to treatment (T-2; Fig. 1 ).  The day prior to treatment  (T-1), the fish were starved, and started to move towards the middle of the cage,  indicated by the higher activity at cameras 2 and 4. On the day of treatment (T), the cameras were pulled from the water , so  the  day post-treatment (T+1) was used to analyse fish stress response.  On  T+1, there was significantly higher activity in cameras at mid-depth , cameras 2 (30.8% to 51.9% ; D=0.96, p<0.001), 4 (25.2% to 46.3%; D=0.92, p<0.001) , and 5 (43.9% to 51.3%; D=0.69, p<0.001) , compared to  on T-2 .  This coincided with a  significant  decrease in activity in the bottom of the cage (52.3% to  40.5%; D=0.90, p<0.001 ).  Two days post-treatment (T+2), the fish moved back down to the bottom of the cage, however they appear to swim to the surface sporadically .  Finally, three days post treatment (T+3), the fish resume the distribution observed at T-2 .  It should be noted that while the latency to feed was highest on T+1 (39.5 min), it was similar on days T-2, T+2, and T+3 (14.75, 12, 8.75 mins, respectively).

Discussion

 Behaviour is a useful  welfare indicator as it is the first change to occur after exposure to a stressor (Sadoul et al., 2014).  The distribution of fish during this  study period is largely governed by temperature. Temperatures ranged from  8 ºC at the surface to 9 ºC at the bottom of the cage. As salmon show a temperature preference between 16-18ºC , within the lower range of temperatures available they will congregate in the warmest depth, which wa s the bottom of the cage (Johansson et al., 2006).  There appears to be a subsection of “shy” fish within the inner cage ,  which  prefer to stay away from the bottom of the cage and  the potential for predation.  After treatment,  the f ish stress response  was to move towards the centre of the cage, likely due to their innate defence behaviour mechanism of  shoaling  away from  areas of higher potential for predation (Sadoul et al., 2014).  On day T+2, while there we re  sporadic increases in activity at the surface, the latency to feed was not significantly different from  days T-2 and T+3, indicating that changes to feeding behaviour only occurs 1-day post-treatment . Understanding these timelines can provide insight to the farmers  on how to structure their feeding post- treatment to ultimately reduce feed waste.     

References

FAO. 2022. The State of World Fisheries and Aquaculture 2022. Towards Blue Transformation. Rome, FAO. https://doi.org/10.4060/cc0461en

 Johansson D., Ruohonen K., Kiessling A., Oppedal F., Stiansen J.-E., Kelly M., Juell J.-E., 2006. Effect of environmental factors on swimming depth preferences of Atlantic salmon ( Salmo salar L.) and temporal and spatial variations in oxygen levels in sea cages at a fjord site. Aquaculture 254 ,  594–605. doi: 10.1016/j.aquaculture.2005.10.029.

Sadoul , B., Evouna Mengues , P., Friggens , N.C., Prunet , P., Colson, V., 2014. A new method for measuring group behaviours of fish shoals from recorded videos taken in near aquaculture conditions. Aquaculture 430, 179–187. https://doi.org/10.1016/j.aquaculture.2014.04.008

 Yang, L., Liu, Y., Yu, H., Fang, X., Song, L., Li, D., Chen, Y., 2021. Computer Vision Models in Intelligent Aquaculture with Emphasis on Fish Detection and Behavior Analysis: A Review. Archives of Computational Methods in Engineering 28, 2785–2816. https://doi.org/10.1007/s11831-020-09486-2