Informing krill fishery management through modelling spatial distribution and diet of a major krill predator, mackerel icefish ( <i>Champsocephalus gunnari</i> ), at South Georgia

Abstract Demersal fish are a critical component of Antarctic marine ecosystems and may consume large quantities of Antarctic krill (Euphausia superba), placing them at risk from competition with commercial krill fishing. However, estimating relative overlap between krill fishing and consumption by demersal fish is beset by uncertainty regarding fish distribution and diet. In this study, we develop models of mackerel icefish (Champsocephalus gunnari) distribution and diet around the subantarctic island of South Georgia to predict the distribution of krill consumption attributable to mackerel icefish to aid in krill fishery spatial management. We use trawl survey data (1986–2023) to construct distribution models for mackerel icefish biomass across the South Georgia and Shag Rocks shelf. Using gut content data, we also developed spatial models for the proportion of krill in mackerel icefish diets, allowing for a novel evaluation of spatial variation in potential krill availability. Models indicated that mackerel icefish are distributed across the South Georgia and Shag Rocks shelves to 400 m depth, with high density areas towards the shelf edge associated with seafloor topography. Spatial variation in diet was evident between South Georgia, where krill predominated, and Shag Rocks, where diets were more piscivorous. Higher krill diets along the South Georgia shelf edge were also coincident with elevated icefish density, suggesting that icefish distribution and diet may be associated with krill availability. Spatial variation in krill consumption by mackerel icefish suggests that overlap with the krill fishery is currently low despite mackerel icefish being a major krill consumer. However, estimates of mackerel icefish biomass and total krill consumption were uncertain due to uncertainties surrounding icefish catch and feeding rates, and were likely negatively biased due to the inability to account for the unknown proportion of fish in the water column that are unavailable to trawl sampling. Despite those concerns, our results provide the foundation for including demersal fish information in spatial management of the krill fishery.