FISHERY BULLETIN: VOL. 79, NO. 4 



scaphopods, were the least often selected food at 

 both locations. Aplacophoran molluscs, however, 

 were neither significantly selected nor rejected. At 

 SG29, the numerical proportions of crustaceans 

 (amphipods and cumaceans) in stomachs and box 

 core samples were nearly equal. At SGIO, however, 

 these crustaceans appeared highly selected. 



Among polychaetes at SG29, the most highly 

 selected prey species were Glycinde picta and 

 Lumbrineris latreilli (in fish of all sizes, i.e., >11 

 cm), Sigalionidae (especially in fish >21 cm), 

 Aricidea ramosa (in fish of all sizes, but especially 

 those >21 cm), Cossura sp. (in fish of all sizes, but 

 especially those >29 cm) , Pectinaria californiensis 

 (in fish >34 cm), and Pista cristata (in fish >35 

 cm) (Table 3). No principal mollusc taxa were 

 selected, with the exception of aplacophorans, 

 which appeared to be neither significantly 

 selected nor ignored. The cumacean Eudorella 

 pacifica was negatively selected by fish >34 cm. 

 The phoxocephalid amphipod species Harpiniop- 

 sis excavata and H. fulgens were not present in box 

 cores, indicating high selectivity by fish or in- 

 adequate box core sampling. If values for all 

 phoxocephalids were pooled, it appeared that the 

 family was selected positively by fish <36 cm. 



Changes in electivity related to fish size and 



prey taxa are shown in Figure 3. Values of £ were 

 calculated for each principal prey taxon found at 

 SG29 within three fish size intervals (11-21, 22-31, 

 and 32-42 cm). The patterns of changes in electiv- 

 ity with fish size are similar to the patterns found 

 for changes in frequency of occurrence of prey with 

 fish size in Figure 2. In one pattern, electivity of a 

 prey taxon was low by small fish, higher by inter- 

 mediate fish, and remained constant by large fish 

 relative to intermediate values (Figure 3A). Taxa 

 belonging to this category included the poly- 

 chaetes Anaitides groenlandica (21), Sternaspis 

 fossor (26), Prionospio spp. (23), Spiophanes spp. 

 (24), and Terebellides stroemii (28); and the Sca- 

 phopoda (32). In the second pattern (Figure 3B), 

 electivity of prey taxa decreased with fish size. 

 Taxa in this category included the polychaetes 

 Myriochele oculata (15), M. heeri (14), and Am- 

 pharetidae (2); phoxocephalid amphipods; the 

 cumacean Eudorella pacifica (34); and the com- 

 bined pelecypod group Adontorhina cyclia- 

 Axinopsida serricata (30). In the third category 

 (Figure 3C), electivity was highest by 

 intermediate-sized fish, and lower for larger and 

 smaller fish. Taxa following this pattern included 

 the polychaetes Cossura sp. (6), Tharyx sp. (5), 

 and Apistobranchus ornatus (3). In the fourth 



> 



UJ 



O.SOr- 



0.60- 



0.40- 



>• 

 > 



.- 0.20 



UJ 



_l 



UJ 



o 



X 



UJ 

 Q 



0.20 



-0.40- 



- -0.60 



■0 80- 



-1.00 



PHOX 



1 1 -*2I '22-31 I32-42I ' 11-21 I 22-31 '32-42' 



FISH LENGTH INTERVAL (cm) 



I 00 



0.80- 



11-21 '22-31 '32-42' 



1-21 I22-31I32-42' 



FISH LENGTH INTERVAL (cm) 



Figure 3. — Changes in Ivlev indices of electivity with fish length at SG29. Numbers designate taxa of prey in Table 3. Phox = 

 Phoxocephalidae. Taxa with similar patterns of change are grouped as Figures A, B, C, and D. 



756 



