were maintained in holding aquaria without tor- 

 rent sculpins for 24 h before being subjected to 

 predation in test conditions. Two types of test 

 groups, each consisting of 20 coho salmon (per 

 stream aquarium), were used. In the naive group, 

 all coho salmon were naive; in the naive and con- 

 ditioned group, 10 naive and 10 conditioned fish 

 were tested together. 



The procedure for testing naive coho salmon was 

 to collect torrent sculpins and place them in hold- 

 ing aquaria without food; on the second day, coho 

 salmon were collected and 20 individuals, 37 to 42 

 mm FL, were placed in each stream aquarium; on 

 the third day, 10 torrent sculpins, 83 to 127 mm 

 TL, where lengths averaged about 100 mm per 

 test group, were transferred from the holding 

 aquarium to each stream aquarium. Forty-eight 

 hours later, the surviving coho salmon were 

 counted and experimental fish were discarded. 



The procedure for testing the naive and con- 

 ditioned group of coho salmon was similar to the 

 foregoing test procedure except that on the second 

 day, 10 naive coho salmon were collected and 

 placed in each stream aquarium with 10 con- 

 ditioned coho salmon. The tip of a ventral fin of the 

 conditioned coho salmon was clipped at the time 

 they were introduced into the stream aquarium to 

 allow them to be recognized at the end of the test. 

 Thus, if there was an adverse effect from clipping, 

 it would be on the group with the greater expected 

 survival. Eight replicate tests were made on each 

 of the two conditions. 



Length Relation 



The experimental procedure placed the coho 

 salmon in close proximity to torrent sculpins for a 

 prolonged period to enhance the possibility of pre- 

 dation. Torrent sculpins responded to this oppor- 

 tunity by preying on larger coho salmon than has 

 been observed in nature (Table 1). The maximum 

 size of coho salmon a torrent sculpin is capable of 

 preying upon is probably limited by the physical 

 size of a coho salmon that a torrent sculpin can 

 catch, subdue, and swallow. While the swimming 

 ability is probably greater for larger coho salmon, 

 this may not be too important because predation 

 by torrent sculpins is accomplished by ambush 

 rather than by pursuit. Torrent sculpins under 

 natural conditions rarely eat coho salmon 40 to 80 

 mm FL, indicating that some factor of coho salmon 

 behavior must decrease their susceptibility .to 

 predation. 



Predator Avoidance Response 



The average survival of the naive group consist- 

 ing only of naive fish was 45.5%; within the naive 

 and conditioned group, consisting of conditioned 

 and naive coho salmon tested together, the naive 

 fish had a 71% survival, and the conditioned coho 

 salmon had a 75% survival. Cumulative chi- 

 square tests of homogeneity showed no significant 

 differences within the naive test group or within 

 the naive and conditioned group (Table 2). TheZ 

 test showed no significant difference between the 

 conditioned and naive coho salmon that were 

 tested together (Z P0 05 = + 0. 53 < 1.645). There was, 

 however, a significant difference between the 

 group consisting of naive coho salmon only and the 

 group consisting of naive plus conditioned coho 

 salmon (Z Pom = +5.29>1.645). 



Mortalities of coho salmon were significantly 

 reduced by conditioning; also, naive fish tested 

 with conditioned fish behaved as conditioned fish. 

 The results of these tests are probably due to rapid 

 conditioning of the coho salmon and a transferable 

 predator avoidance reaction. Rapid conditioning 

 was evident because conditioning of fish to a 

 stimulus other than predators is usually ac- 

 complished only after many trials. Conditioning 

 coho salmon to evade predation by exposing them 

 to torrent sculpins probably reinforces a strong 

 innate avoidance behavior. In another case, rapid 

 conditioning of sockeye salmon, O. nerka, to evade 

 predation by rainbow trout, Salmo gairdneri, has 

 been demonstrated by Ginetz and Larkin (1976). 

 Experiments by Russians have shown that certain 

 fishes, including the chum salmon, O. keta, in- 

 creased their ability to evade predation after a 2- 

 to 4-day training period with predators (Kanid'yev 

 et al. 1970). 



TABLE 2. — Comparative survival of two groups of coho salmon 

 that were exposed to predation by the torrent sculpin. One group 

 consisted of naive fish only and the other consisted of naive and 

 conditioned coho salmon combined. The initial number of coho 

 salmon per group per stream aquarium was 20. 



458 



