McKenna: Trophic relationships of Antarctic demersal fish of South Georgia Island 



645 



greater than zero were selected arbitrarily from the 

 catch for gut content analysis. The sex of each fish was 

 also noted. These data were recorded and the stomach 

 was assigned an identification number for correlation 

 with station information. 



Stomachs were removed, taking care to prevent loss 

 of any contents. Each stomach was preserved by in- 

 jection with 10% formalin and wrapped in gauze or 

 paper towels. Large stomachs were soaked in a 10% 

 formalin solution for at least 24 hours. Stomachs were 

 sealed in plastic ziplock bags and stored until they could 

 be examined. 



Examination of stomach contents 



In the laboratory, the weights of all fish and stomach 

 contents were measured to within 0.001 g on an elec- 

 tronic balance. Objects weighing more than 160g were 

 measured on a triple-beam balance. Before contents 

 were removed from the stomach, the total formalin- 

 preserved wet weight was measured. The stomach was 

 then opened with a longitudinal incision through the 

 stomach lining from esophagus to intestine. The con- 

 tents were removed and sieved through a 0.5 mm mesh 

 screen. Wet weight of the empty stomach lining was 

 then measured. This allowed back-calculation of the wet 

 weight of unidentifiable material (GORP) by the dif- 

 ference between total wet weight and the sum of the 

 weights of the separate, identifiable items plus stomach 



lining. Empty stomachs and those containing only 

 unidentifiable material were not used in the analysis 

 of similarity. 



Items remaining on the sieve screen were sorted 

 into general taxonomic groups (e.g., fish, amphipod, 

 isopod) and then identified to family or species where 

 possible. Each group was enumerated and weighed. 

 Specimens were then placed in a drying oven at 60 °C 

 and dried to constant weight. 



Analysis of the diet data 



Frequency of occurrence and dietary coefficients 

 (Linkowski et al. 1983) were determined as simple 

 measures of the importance of each potential prey item 

 in the diet of each species (Table 2). To remove the 

 biases due to varying stomach size and total number 

 of items contained in each stomach, all diet data were 

 converted to percent composition. Percent composition 

 of the diet was determined for each stomach based on 

 numerical abundance, wet weight, and dry weight of 

 the prey items. Results refer to the use of percent com- 

 position by dry weight unless stated otherwise. Aver- 

 age percent composition of the diet was then calculated 

 for each of the 15 species of fish for which stomachs 

 had been collected (Table 3). Only stomachs which con- 

 tained identifiable prey items were included in the diet 

 analysis. 



