Feed Distribution in Large Scale Sea Cage Aquaculture: Experiments, modelling and simulation

Abstract

The objective of this thesis was to gain new insight into the process of feeding in large scale sea cage aquaculture and investigate novel methods of feeding in order to increase profitability, welfare and minimize environmental impact. Compared to livestock farming, Atlantic salmon (Salmo salar ) farming is a young industry which has experienced an almost exponential growth rate and the product continues to be in high demand. A single cage in Norway may contain more than a 1000 tonnes of fish in the form of 200.000 individuals. Such a figure is difficult to comprehend, but one may draw a parallel to the equivalent of 1600 cows inside a single cage. Feeding of fish kept in sea cages is a complicated endeavour compared to land based farming for a number of reasons. Thousands of individuals co-exist in a single three-dimensional dynamic space, observation is restricted to surface inspections or a submerged camera with limited field of view, feed can not be given to a specific fish and the location of feed is difficult to predict as a consequence of currents and fish induced turbulence. In addition, feed which is not consumed from the time it is distributed over the surface to it passes through the cage represents a direct economic loss and acts as an unnecessary nutrient discharge to the environment. Over 10.000 kg of feed may be administered to a single cage towards the end of a production cycle and is the single largest cost in Norwegian salmon farming. Even though the process of feeding is a complicated one, the systems used to distribute feed are simple. Significant effort has been made in determining the ration size, meal frequency and at what time of day Atlantic salmon should be fed. This thesis looks into the temporal feed availability on a meal to meal basis and goes into depth with respect to the spatial availability of feed within the sea cage. Many studies on a smaller scale indicate that spatially and temporally restrictive feeding may lead to unequal feed accessibility, loss of growth potential and elevated levels of aggression. With respect to controlling the spatial distribution of feed, it has been shown that current methods cover a small area of the cage surface. In addition, existing methods have limited ability to increase the feed distribution without exhibiting other detrimental effects such as increased pellet breakage. There is also no way of controlling where feed is placed as a consequence of wind or currents. Experimental results are presented to better understand the dynamics of a feed spreader, a model has been developed and the performance of different spreader designs investigated. Further experimental results for settling rate and diffusion of pellets are presented and have been used to parametrize a full sea cage model. This model enables simulation of environmental factors, feeding methods and fish to predict the effect on central production parameters. Finally, using these two models, different feeding regimes are simulated and the consequent effects on spatiotemporal feed distribution, feed intake and feed loss are commented upon. It is likely that by increased use of environmental measurements run through feed distribution models and having more adaptable methods of feed placement, one can in the future minimize the environmental impact whilst maintaining high growth rates and good fish welfare.