Field validation of growth models used in Atlantic salmon farming

Several models for description of fish growth are commonly used in Atlantic salmon farming, including the specific growth rate (SGR), the thermal growth coefficient (TGC), the Ewos growth index (EGI), and average daily weight gain (ADG). In the present study, a subset of a commercial database containing information from 827 fish groups from the year-classes 2000–2005, produced along the Norwegian coastline, was used to validate these four growth models. A number of biotic and abiotic factors were fitted to the models of interest in order to evaluate model strengths and weaknesses and to identify additional factors whose inclusion may improve model performance. Preliminary analysis indicated non-linear relations; to account for this we applied generalized additive models (GAM) in regression analysis.Our findings indicate that ADG was strongly associated with harvest weight and was thus deemed unsuitable for describing growth in Atlantic salmon. SGR was also associated with fish size and biased towards small fish when fish of uneven stocking size were compared. The TGC, SGR and the EGI models were all moderately associated with harvest weight, and the same three models were more strongly associated with mean temperature and mean day-length. These models might therefore present bias when used to compare growth in varying environmental conditions. The EGI was considered the most robust model overall for predicting growth at different sizes exposed to variable abiotic exposure such as temperature and light. Finally, the study suggests that the robustness of growth models can be improved by accounting for non-linear effects on growth and including abiotic factors such as temperature, light, and latitude.