Graver: Feeding ecology of Theragra chalcogramma in the Gulf of Alaska 



465 



a single species. This index was chosen because it does 

 not require resource-availabihty data (Wallace 1981), 

 it is independent of sample size (Kohn and Riggs 1982), 

 and it estimates overlap accurately for real overlaps 



between 7 and 85%, whereas three other indices per- 

 formed poorly (Linton et al. 1981). PS values greater 

 than 60% are generally considered to demonstrate 

 significant overlap (Mathur 1977, Wallace and Ramsey 

 1983). In order to minimize size-related bias in the 

 measurement of diet overlap, proportional similarity 

 indices were calculated for only the 15.0-19.9 mm SL 

 size class. 



Results 



Standard length tended to increase with date of col- 

 lection (Table 1, correlation coefficient r = 0.678, P 

 <0.05). The dominant size class for all stations was 

 15.0-19.9 mm SL, accounting for 62% of the larvae, 

 with a minimum of 21 larvae of this size collected at 

 each station. The mean lengths of larvae from five sta- 

 tions located along a transect parallel to the Alaska 

 Peninsula were significantly different (P< 0.001) 

 (ANOVA, Zar 1974), while of the two transects perpen- 

 dicular to the Alaska Peninsula, larval lengths were 

 significantly different (P<0.001) along only one of 

 these transects (stations 44, 45, and 48). 



