Modelling the Effectiveness of Gene-Edited Salmon at Sea Lice Control and the Use of Refugia to Mitigate...

Advances in gene-editing technologies offer opportunities to improve disease management in aquaculture. Gene-editing applications for farmed Atlantic salmon (Salmo salar) include harnessing innate parasite resistance to protect against salmon lice (Lepeophtheirus salmonis). The potential for salmon lice to counter-adapt to changes in the host should be considered. However, salmon farms are highly connected through louse transmission, and so it is important to gauge the impact of new technologies over large scales. Exploring the epidemiology and evolution of lice across a farm network is possible using metapopulation models. Here, we expand upon an eco-evolutionary model to simulate the stocking of theoretical gene-edited Atlantic salmon that rejected lice to a similar degree as the more resistant coho salmon (Oncorhynchus kisutch). Model outputs suggested that such louse resistance would be highly effective at controlling outbreaks and reducing the need for additional delousing treatments. Lice were controlled more efficiently when gene edits were prioritized at key farms in the louse dispersal network. In scenarios where gene edits selected for adaptive traits in the louse population, however, lice rapidly evolved counter-resistance, leading to a significant reduction in treatment efficacy. When highly connected farms were left as refugia (not stocked with edited salmon), the rate of adaptation was slowed, thus extending the effectiveness of gene edits through time. The refuge effect was further enhanced if there were fitness trade-offs to counter-resistance in lice. We note that the long-term benefits of the refugia approach-to individual farms and to the wider industry-must be balanced with the costs in the short term, especially for the refuge farms. Careful planning of how to distribute new technologies can maximize efficiency and help safeguard them against parasite evolution. Spatial eco-evolutionary models are powerful tools for scenario testing that assist with decision making.