Salmon lice from fish farms in open net pens pose a threat to the survival of wild salmon post-smolts migrating through areas with high farm and lice densities. Reliable estimation of this mortality is fundamental for the sustainable management of aquaculture in such areas but is challenged by considerable uncertainty about several of the processes that link reported lice numbers in fish farms to post-smolt mortality. Utilising recent access to lice data from post-smolt trawling, we revised a virtual post-smolt model that estimates salmon lice-induced mortality of seaward-migrating wild salmon post-smolts. We also assessed the sensitivity of model results to model assumptions that differ among the virtual post-smolt models currently being used as management support in Norway. The spatiotemporal variation in infestation pressure along the entire Norwegian coast was calculated based on monitoring data for salmon lice, temperature and fish abundance in all Norwegian salmonid farm locations. The revised model integrates lice data from sentinel cage experiments and post-smolt trawling to quantify the spatiotemporal variation in infestation rate of wild post-smolts from the spatiotemporal infestation pressure, while also estimating the effect of calibration data type on infestation rate. Results show 10 times higher infestation rate in trawl-caught post-smolts than in post-smolts from sentinel cage experiments. Mortality estimates calibrated to trawl data are more sensitive to assumptions about possible density dependence in lice infestation rate than to assumptions about migration speed. These findings contribute to explaining why different virtual post-smolt models provide different estimates of salmon lice-induced mortality.
Read full abstract