FISHERY BULLETIN: VOL. 86, NO. 4 



val of 1,912 to 9,339. Based on these figures, it is 

 estimated that coho salmon consumed a total of 162 

 (9%) of the tagged pink salmon released on 6 June, 

 with a 95% confidence interval ranging from 76 (4%) 

 to 375 (21%). 



DISCUSSION 



Determining the prey preference of predators in 

 the wild is an important but difficult problem. The 

 two experiments reported here illustrate a new ap- 

 proach to determining the selectivity of predators 

 in the wild. Aside from logistic problems, the suc- 

 cess of this method depends on the validity of four 

 main assumptions: 1) predators that are captured 

 and examined, and their stomach contents, are truly 

 representative of the total predator population of 

 interest; 2) tagging or marking the prey does not 

 result in abnormal behavior of either the prey or 

 predators; 3) ingestion and partial digestion of the 

 prey by the predators does not significantly alter 

 the characteristics of the prey that are of primary 

 interest; 4) all of the tagged prey remain equally 

 "available" to the predators for the duration of the 

 experiment. 



The first assumption should be valid if the sam- 

 pling program is appropriately designed, consider- 

 ing the statistical tests that will be used to analyze 

 the data. This is a complex topic and an in-depth 

 discussion is beyond the scope of this paper. How- 

 ever, many extensive references are available (e.g., 

 Anderson and McLean 1974; Cochran 1977; Mont- 

 gomery 1976). 



The two experiments reported here were designed 

 only to demonstrate the utility of this approach and 

 do not clearly show how generally applicable the 

 results are. Only a small number of tagged fish were 

 released and samples of predators were collected 

 with a beach seine, which undoubtedly is biased to 

 some degree in terms of the species and sizes of fish 

 that were captured. In addition, the samples were 

 collected at only one location in the first experiment 

 and over a relatively small area in the second experi- 

 ment. However, extensive sampling and examina- 

 tion of the stomach contents of fish predators over 

 a 3-yr period indicates that yearling coho salmon are 

 the major predators of juvenile pink salmon through- 

 out Masset Inlet and Masset Sound (Hargreaves in 

 press). The results of these two experiments are also 

 consistent with those obtained from enclosure ex- 

 periments, which indicate that yearling coho salmon 

 are size selective when feeding on juvenile pink or 

 chum salmon (Parker 1971; Hargreaves and LeBras- 



seur 1986). Thus, although quite limited in scope, 

 these two experiments provided results that are con- 

 sistent with those obtained by two other indepen- 

 dent, but considerably more expensive and labori- 

 ous, methods. 



The assumption that tagging and marking the 

 prey does not affect the behavior of predators or 

 prey can be assessed either by direct observation 

 or by conducting additional experiments. In some 

 cases it may be possible to design the experiment 

 to allow observation of both the predators and prey 

 throughout the experiment and directly observe any 

 unusual behavior. However, in many cases, addi- 

 tional experiments will probably be required. For 

 the two experiments reported here, the pink salmon 

 used in the first experiment were tagged and fin- 

 clipped; the fish released in the second experiment 

 were tagged but were not fin-clipped. The results 

 of the first experiment indicate that tagging was 

 more effective than fin-clipping for recognizing fish 

 recovered from the stomachs of predators. In terms 

 of behavioral changes, previous work indicated that 

 the mortality of tagged and untagged juvenile 

 salmon was not significantly different when exposed 

 to predators and that tagging juvenile salmon had 

 no noticeable affect on the behavior of either the 

 predators or prey (Hargreaves and LeBrasseur 

 1986). However, the tagged fish used in these en- 

 closure experiments were not fin-clipped. 



Earlier studies have indicated that amputation of 

 fins from juvenile salmon typically results in lower 

 survival rates (Ricker 1949). Marked pink salmon 

 fry also suffer higher mortality than unmarked fry, 

 possibly due to a bias on the part of predators for 

 marked prey (Parker et al. 1963). This is not a major 

 concern in the two experiments reported here be- 

 cause the intent was to determine the size selectivity 

 of predators, rather than any selectivity for marked 

 or unmarked prey. In addition, fin-clipped pink 

 salmon were used only in the first experiment but 

 predators consumed significantly more of the 

 smaller prey in both experiments. This supports the 

 assertion that fin-clipping the pink salmon did not 

 substantially affect the prey size selectivity of the 

 predators. In general, the possibility that the results 

 of these types of experiments may not apply to un- 

 tagged or unmarked prey can be eliminated by using 

 only tags or marks that are known to have negli- 

 gible effects on the behavior of both the prey and 

 predators. 



The third assumption, that ingestion and partial 

 digestion of the prey by the predators does not sig- 

 nificantly alter the important characteristics of the 



768 



