Towards better survival: Modeling drivers for daily mortality in Norwegian Atlantic salmon farming

Mortality in the production of farmed salmonids is a major constraint to the sustainability of this form of animal husbandry. We have developed a model for the daily mortality in salmon farming over a full production cycle from stocking to harvest, considering different environmental and production factors. These factors included sea temperature, salinity, day within year, fish weight at stocking, stocking day, four types of lice treatments and the possible occurrence of pancreas disease (PD). We considered a generalized additive model following full pro-duction cycles, allowing for non-linear descriptions of how relevant factors relate to the daily mortality. We saw a high overall mortality rate immediately after stocking, which decreased the first three months in the cycle and thereafter increased. We found that the total mortality could be reduced by 21% if avoiding all lice treatments, and similarly reduced by 20% if no PD infections occurred. If avoiding jointly PD and all lice treatments, the accumulated mortality could be reduced by 34%. A single thermal or hydrogen peroxide treatment was asso-ciated with a mortality of around 1.6% and 1.3%, respectively. This modeling approach gave a unique oppor-tunity to model how different factors interact on the overall global mortality and can easily be extended by other factors, such as additional fish diseases.

Automated summary

Mortality in farmed salmon production is a significant challenge to sustainability. A model was developed to predict daily mortality based on various environmental and production factors. Factors such as sea temperature, lice treatments, and presence of pancreas disease were considered. The model revealed a high mortality rate after stocking, gradually decreasing for the first three months and then increasing. Avoiding lice treatments and pancreas disease could significantly reduce total mortality.