FISHERY BULLETIN: VOL. 80. NO. 3 



chinus neumayeri, Odontaster validus). After a 

 suitable acclimation period (24-48 h), the fish or 

 fishes were released and allowed to feed. Obser- 

 vations were made from above and provided in- 

 formation on feeding method. Foraging strategy 

 is inferred from the feeding method, morphology, 

 and diet of the fish and microhabitat in which it 

 was captured or observed. 



Percentage frequency of occurrence of each 

 prey taxon and percentage composition of diet by 

 number and point volume were calculated for 

 each species. Hoffman (1978) suggested a meth- 

 od for determining an adequate sample size in 

 feeding studies. Using this method, three sam- 

 ples containing 35 individuals of each of four 

 species were examined. No additional informa- 

 tion was obtained in sample sizes >9-23 individ- 

 uals (x = 17). Where possible, at least 20 individ- 

 uals were examined. When a sufficient number 

 of specimens was collected, the sample was seg- 

 regated by size of fish, season, or area of capture. 

 These subsamples were then compared using a 

 X test for association (Remington and Schork 

 1970). Fullness indices, measures of feeding in- 

 tensity (Windell 1971), were calculated for each 

 subsample, and significance was determined by 

 Wilcoxon sum of ranks test (paired samples) or 

 Kruskal-Wallis x 2 test (3 or more samples) 

 (Langley 1970). Mean prey size was calculated 

 by dividing total volume per taxon by total num- 

 ber of prey items consumed. Percentage diet 

 similarity by number and volume was deter- 

 mined using: 



S= 100(1 -% X\p„- p w \) 



where p„ and p yi are the proportions of the diets 

 of species x and y respectively of prey item i (Lin- 

 ton et al. 1981; Abrams 1980; Schoener 1970). 

 Diet diversity was examined by the number of 

 taxa found in the diet of each species (P) and 

 diversity index H = — Sp,ln(p), where p, — the 

 proportion of the ith species in the sample (Shan- 

 non and Weaver 1949). 



RESULTS 



Feeding Behaviors 



Fishes were observed using variations of four 

 basic behaviors: ambush feeding, bottom slurp- 

 ing, water column feeding, and grazing. Ambush 

 feeding was observed most frequently in the 

 field. Harpagifer bispinis, Notothenia neglecta, 



Trematomus bernacchii usually, and N. gibber- 

 ifrons, on occasion, perched among rocks or 

 partially buried themselves in soft mud and 

 waited for a prey organism to approach. As the 

 prey neared, the fish lunged and then engulfed 

 the item. Treatment of the prey after capture de- 

 pended upon its relative size and morphology. 

 If possible, the item was swallowed whole. Ex- 

 ceptions to this were scaleworms. In the labora- 

 tory, I observed H. bispinis capture a scaleworm 

 on 16 occasions, pull it into its mouth, spit it out, 

 immediately pull it into its mouth again, and re- 

 peat the process several more times (x = 6, range 

 = 2-16). This procedure successfully removed all 

 scales from the worm before the fish actually 

 consumed it. This process apparently occurred 

 in the wild since scales are rarely found in H. 

 bispinis stomachs although scaleworms are an 

 important part of its diet (below). If the prey item 

 was too large to engulf whole, it was eaten in 

 parts. I observed N. neglecta capturing fish one- 

 third to one-half as long as itself on five occasions 

 in laboratory tanks. The predatory N. neglecta 

 pulled the prey fish T. bernacchii or N. nudifrons 

 into its mouth, usually head first, retreated to a 

 protected area, and began to digest that part of 

 the fish in its mouth and stomach. This process 

 took up to 12 h during which time the predator 

 was quiescent. Large prey items taken from the 

 stomachs of N. neglecta commonly showed signs 

 of differential digestion, which indicates that 

 this method of feeding occurs in the wild. Fish 

 using ambush feeding tended to be largely car- 

 nivorous and preyed upon relatively large, motile 

 organisms. Fishes were observed to take only 

 moving organisms. On many occasions in the lab- 

 oratory, H. bispinis ignored stationary amphi- 

 pods close to its mouth and readily visible. When 

 the amphipod moved, it was consumed. Often 

 movement consisted only of a twitching antenna. 



The slurp feeding method was observed most 

 frequently in N. gibberifrons which swam over 

 mud bottoms, sucked up and sifted through large 

 quantities of mud, and consumed the organisms 

 encountered (Daniels and Lipps 1978). Mud and 

 small rocks were also swallowed and passed 

 through the gut. Fish using this method usually 

 fed upon sedentary or slow-moving invertebrates 

 and rarely consumed plant matter. Bacteria ad- 

 herent to mud may also have been an important 

 part of their diet. 



Water column feeding was characteristic of 

 the pelagic P. antarcticum, juvenile T. newnesi, 

 and, on occasion, demersal forms like N. neglecta. 



578 



