Barnentos el al Differences In tfie feeding patterns of Sebastes capensis 



491 



eralist versus specialist) of the predator. The verti- 

 cal axis of a hypothetical index represents the feeding 

 strategy of the predator in terms of specialization or 

 generalization. Prey points located at the upper left 

 of the diagram would be indicative of specialization 

 of individual predators, and those in the upper right 

 would represent specialization of the predator popu- 

 lation. Data seen in the upper right of the diagram 

 (population specialization) necessarily restricted to a 

 single point or to a few points, would reflect a predator 

 population with a narrow niche diet. If all prey points 

 were located along or below the diagonal from the upper 

 left to the lower right, the predator population would 

 have a broad niche width. Prey points located in either 

 the upper left or the lower right corner would represent 

 prey that make the same overall contributions to the 

 population diet, but they would be indicative of totally 

 different feeding strategies for the individual predators. 

 Prey with high specific abundance and low occurrence 

 (upper left) would be consumed by a few individuals 

 displaying specialization, whereas prey with a low spe- 

 cific abundance and a high occurrence (lower right) 

 would be eaten occasionally by most individuals. These 

 differences in feeding strategy are related to the be- 

 tween- and within-phenotypic contributions to the niche 

 width. In a population with a high between-phenotypic 

 component, individuals specialize on different resource 

 types. In a population with a high within-phenotype 

 component, most of the individuals use many resource 

 types simultaneously. 



Statistical analysis 



The Mann Whitney test was used to compare the sizes 

 of fish between sexes, and the Kruskal-Wallis test was 

 performed to compare the sizes of fish from different 

 locations (Zar, 1999). The x" contingency tables were 

 used to evaluate the possible differences in the prey 

 items clustered by phylum (Arthropoda, Chordata, and 

 Mollusca) among locations. In addition, the G-statistic 

 was used to evaluate differences in the occurrences 



of different prey types between sexes (Zar, 1999) and 

 among the different size classes offish defined arbitrary 

 (class 1: fish between 18 and 25 cm, and class 2: fish 

 between 26 and .38 cm) (Zar, 1999). 



A correspondence analysis was used to evaluate the 

 association between occurrences of prey items and the 

 different locations, allowing the ordination of samples 

 in a reduced space (two dimensions) (Digby and Kemp- 

 ton, 1987). 



Results 



Fish characteristics 



A total of 613 specimens were analyzed and of those 

 296 stomachs contained some prey, 189 had empty stom- 

 achs, and 128 had everted stomachs (48.29%, 30.83% 

 and 20.88%, respectively). Table 1 summarizes the 

 characteristics of the fish analyzed by location. There 

 was no significant difference in body size between males 

 and females ({7=9854.5. P=0.26). The size of fish dif- 

 fered significantly among locations: fish from Talca- 

 huano and Valdivia did not differ significantly in size 

 but were significantly larger than those from the others 

 locations (Table 1). 



Diet analysis 



Across the geographical range, a total of 60 prey items 

 were identified (Table 2): 40 crustaceans, 9 fishes, 3 

 cephalopods, 7 snails, and 1 polychetes. Crustaceans 

 were the predominant prey in the diet of S. capensis. 

 The low occurrence of other prey items allowed us to des- 

 ignate this species as carcinophagous (=feeding mainly 

 on crustaceans). 



Overall, the most important prey item was Mysida- 

 cea unident. (75.06% IRI), followed by Osteichthyes 

 unident. (6.29% IRI), the rock shrimp Rhynchocinetes 

 typiis (6.03% IRI), deep red crab (Petrolisthes desrnar- 

 estii) (4.67 % IRI), the rock crab Cancer setosus (2.32% 



