Smart submergence of sea cages to improve profitability, minimise environmental impact, and ensure welfare of salmon .


Is it possible to maintain neutral buoyancy in submerged farmed salmon? The submersion of salmon farms aims to permit increased production by creating opportunities to occupy alternative aquaculture sites, in addition to avoiding unsuitable surface conditions. After several attempts to hold salmon submerged at 3-10 m depth, for 3-6 week periods, this method has proven difficult to execute due to the salmon’s swimbladder anatomy; salmon need to re-fill the swimbladder with air at the surface in order to stay neutrally buoyant and not sink. Sinking can, for shorter periods, be compensated by faster swimming and more structured schooling formation, but for longer periods this will lead to tilted swimming during night time, slower growth and consequently poorer welfare.


In a small submerged pilot-cage (5 x 5 x 7 m), we observed how 15 salmon learned to refill the swim bladder in an air-filled dome, hence maintaining their neutral buoyancy. To investigate how a large group of salmon would behave and utilise an “artificial surface”, we introduced air to 5000 salmon submerged to 10 m depth in two different ways; 1) in a small air-filled dome and 2) from ascending air-bubbles.

Four commercial-scale cages (approx. 2000 m3) were used; two submerged and two control treatments were used. A 1 × 1 m clear plastic dome, containing ~120 L of air, was attached beneath the roof netting in the submersible cages. The salmon was submerged for 49 days from March 15 to May 02, 2012. Nine days after the completion of these treatments, one cage was submerged and an air-hose with 20 holes (Ø = 2 mm) was lowered to the bottom of the cage to release air bubbles. Four days after submergence, air was supplied daily between 15:00 and 22:00, and fish monitored by the echosounder system for 7 days.


As seen in previous experiments, the swimming speed for submerged salmon increased 1.6 times compared to that of the salmon in control cages. Additionally, the submerged fish maintained a structured school and paid little attention to the small air-dome. After around 28 days, random re-fill events were observed in one of the submerged cages, and a group of 200-300 salmon could be seen near the dome, separated from the deeper main school. This small group of fish seemed to be neutrally buoyant, as they swam slowly or glided with their head slightly downwards.

This fact is supported from the echosounder signals, as the decline of average swim bladder gas stopped at the same time as the refill behaviour started. The appetite for the submerged fish seemed to fall during the experiment, and at the end the average growth rates was lower for the submerged fish compared to the control fish. Nevertheless, one of the submerged groups grew almost at similar rate as the lowest control group (SGR at 0.39 and 0.44 % BW day-1, respectively). During feeding, the schooling structure could be observed as a tall cylinder in the submerged groups, contrasting to the more normal donut-shaped school in the control cages. Further evidence for poor air-dome utilisation was the high rate of snout wounds in the submerged compared to control individuals.

The other method tested was to offer salmon air in a submerged cage by introducing air-bubbles at the bottom of a cage, to determine whether salmon were able to re-fill their swim bladder from bubbles ascending in the water column. The fish were not scared by the bubbles, as they swam unaffected through the “wall” of bubbles, however re-filling of the swim bladder was not evident in the echosounder signals. After re-surfacing the submerged cages, intense surface activity was observed, which strongly indicate that the fish lost air from their swim bladders during the 7 days of submergence. 

Overall, the results show that most fish did not utilise the air-dome to re-fill their swim bladders during submergence. It is possible that the surface area of the air-dome was too small at 0.7% (1 m2) of the total surface area in the cage. In addition, the roof was angled at 40° and the air-dome was placed in the centre, the combination of which might have made it less available for the schooling fish. Therefore, the design of submersible cages must be developed to provide easy access to the artificial surface. We believe farming of salmon can be achieved in submersible cages if a large artificial area (one large or several small) is placed in a flat roof to ease access for the schooling fish. The feed should also be introduced in the air-dome to attract and teach the fish to find the available air.

Published September 20, 2013

Tore S Kristiansen (leader), Øyvind Korsøen (post doc), Jan Erik Fosseidengen, Frode Oppdal, Ørjan Karlsen (Institute of Marine Research)
Tim Dempster, Samantha Bui (SINTEF Fisheries and Aquaculture and Melbourne University).

Egersund Net
Lerøy Seafood Group
Marine Harvest