Methodology:

Trawl surveys

Data on the abundance and distribution of Champsocephalus gunnari were collected during 28 demersal trawl surveys conducted primarily during the austral summer (Nov-Feb) at South Georgia and Shag Rocks, between 1987 and 2023. The survey design has remained constant throughout the time series despite changes in survey vessel. From the 1991 survey onwards, gear was standardized (commercial sized otter trawl FP-120, wingspread 18-20 m, headline height 3-6 m, cod end mesh 40 mm). Before this, broadly similar gear was used on each of the early surveys (Belchier et al., 2015).

The area was surveyed using a random-stratified design, subdividing into five sectors: Shag Rocks plus the four quadrants of the South Georgia shelf; northwest, northeast, southwest, and southeast. Trawls were conducted for 30 minutes at a speed of 3-4 knots, and at two depth zones: 100-200 m and 201-350 m. The trawl dataset also contains data from winter surveys, performed using the standard survey design but in April/May (2008 and 2021) or September (1997 and 2007), and dedicated deepwater surveys (350-1000 m)

During each survey, the total catch of C. gunnari in each trawl was recorded along with information on trawl location, time, depth, and trawl characteristics (trawl duration, distance trawled, horizontal and vertical net opening).

Diet data

Since 2004, stomach contents of mackerel icefish have been collected on a routine basis using a standard methodology developed by Main et al. (2009). Samples of C. gunnari (typically less than or equal to 30 per trawl) were collected from all trawls in which the species was caught. Sampled fish were measured and weighed, stomachs were removed and immediately frozen at -20 degrees centigrade, for later processing and macroscopic identification of stomach contents. Stomach contents were thawed and sorted in the laboratory at King Edward Point, South Georgia. Once the stomach contents were sorted, individual prey items were counted and weighed to the nearest 0.01g.

Diet composition per fish sampled were aggregated at two levels. Gross survey averages of diet composition were produced by first assigning prey items into one of 11 broad taxonomic groups (Euphausia superba, Other Euphausia spp., Thysanoessa sp., Themisto gaudichaudii, Antarctomysis sp., Other/unknown crustacean, Nototheniidae, Channichthyidae, Myctophidae, Other/unknown fish, Other/unknown), then finding the summed mass for each group across all fish sampled from the Shag Rocks and South Georgia shelves per survey. These are then presented as proportional composition by dividing by the net total across groups per survey and shelf area. The proportion of krill in stomach contents per trawl was calculated by taking the summed mass of krill and dividing by the summed mass across all prey taxa, where the sums are performed across fish stomach contents from each trawl with diet information.

Data collection:

All data processing and cleaning were performed in R version 4.3.3 (R Core Team 2024).

Data quality:

Raw trawl data were cleaned to remove trawls hauled early due to unsuitable ground, and those that were deemed unrepresentative due to gear failure.