Abstract
This study describes the patterns of mortality and investigates the sources of variation in mortality during the marine phase of commercial salmon farming. The study included daily mortality records from stocking to harvest of 21 million salmon from ten hatcheries in 136 fish‐groups (fish in the same cage from the same hatchery). The fish was stocked in 2017–2018 at 21 marine farms within two Norwegian companies. The sources of variation in mortality were investigated using multilevel linear regression models with ‘fish‐group’ nested within ‘farm’ as a random effect, cross‐classified with ‘hatchery’. In the final model, ‘fish‐group’ was the source of most variation (70%). Furthermore, the mortality categories ‘smolt‐related mortality’, ‘infectious diseases’ and ‘handling and treatment’ were responsible for 10%, 17% and 29% of the total number of dead fish respectively. Overall, the study shows that smolt‐related mortality is one of the major causes of death in the first part of the production, while handling and treatment was the dominating cause of mortality in total. Mortality varied by fish‐group to a large extent. This means that targeted preventive strategies to decrease mortality for individual fish‐groups might be more effective than overall measures at farm or hatchery level.