Accuracy and precision of harvest stock estimation in Atlantic salmon farming
2013
Aunsmo, Arnfinn | Skjerve, Eystein | Midtlyng, Paul Johan
Accuracy and precision of number, mean weight and biomass estimation at time of harvest of farmed Atlantic salmon were investigated using harvest-time data from 240 pens belonging to three large salmon producers in mid-Norway. The same database was used for statistical analysis of potential factors associated with estimation errors of the three outcome variables in a generalized linear model using a sandwich estimator for handling clustering and dependencies in the dataset. The results show high accuracy and poor precision of estimates where, at the pen level, mean errors are close to zero, but approximately 50% of estimates fall outside 3% errors and approximately 10% remain outside ±10% of the true values. The mean absolute biomass estimation error on pen level was 5.1%. Errors were reduced when figures were aggregated at the site and company level due to opposite directed errors cancelling each other. The symmetric distribution around zero suggests that the poor precision is an effect of random errors. Mortality above median was found to be statistically associated with underestimation of the fish numbers for harvest. Error in number estimation was found to be inversely associated with errors in mean weight estimation. Norwegian salmon production is regulated by maximum allowed biomass per license where random errors in stock estimation may contribute to overstocking and subsequent regulatory actions. The poor precision in stock estimation is currently an obstacle to further development and profitability in the industry. Correcting the factors identified in this study may give minor improvements in stock estimation while major improvements will require new methods and/or equipment.
Afficher plus [+] Moins [-]Mots clés AGROVOC
Informations bibliographiques
Cette notice bibliographique a été fournie par National Agricultural Library
Découvrez la collection de ce fournisseur de données dans AGRIS