N 



N 



i 



LAKE 

 MICHIGAN 



Figure 1 . — Map of area. Water from Lake Michigan ( L) was 

 collected upwind of river mouth. 



used in the later analysis. A control series with 

 the same stimulus in each of the four holders 

 demonstrated that delivery through any one of 

 the four sample holders gave comparable results. 



Water samples for testing responses of coho 

 salmon were collected from the two home 

 streams (a tributary to the Ahnapee River (A) 

 and a tributary to the Kewaunee River (K)), 

 the Ahnapee River (R) 4 km above the entrance 

 of the home-stream tributary, and Lake Mich- 

 igan (L) (Figure 1). Samples from the home 

 streams were taken above (Aa, Ka) and below 



(Ab, Kb) barrier fish traps so that Aa and Ka 

 were home-stream waters while Ab and Kb 

 were home-stream water plus the scents from 

 the returning migrants. Both home streams 

 (A, K) had similar physical features and coho 

 salmon stocking histories; but the tributary to 

 the Kewaunee River (K) contained a greater 

 proportion of well water than stream water 

 compared with the tributary to the Ahnapee 

 River (A). 



All sample presentations were made in 

 groups of four to each fish, and responses were 

 quantified as a ratio to the response to water 

 from station Ab. This water sample was in- 

 cluded as one of the four presentations for all 

 fish and was chosen as a base because it evoked 

 intense responses. Nineteen Ahnapee fish were 

 tested: 11 with water samples from Aa, Ab, R, 

 and L; 3 with Aa, Ab, R, and water from an un- 

 related source; 1 with Aa, Ab, and 2 samples 

 from unrelated sources; and 4 with water from 

 Aa, Ab, Ka, and Kb. Six fish from Kewaunee 

 were tested with water from Aa, Ab, Ka, and 

 Kb. (The water samples from other sources 

 showed no consistent patterns and will not be 

 discussed.) 



Results 



For the Ahnapee fish, samples from the ter- 

 minus of the return migration evoked greater 

 responses than the sample from the origin (L). 

 But this result is probably not an example of 

 Hara's (1970) "specific electric" response to the 

 home stream: (1) One-half of the "stimulus 

 strength" was added at the trap, although water 

 collected above the trap (Aa) must be considered 

 home stream (Table 1). (2) Agreement on order 

 of responses to each water (Ab>Aa>R>L) was 

 good among fish [W^ = 0.87, k = 11, N = 4, 

 P<0.01 Kendal coefficient of concordance test 

 (Siegel, 1956)], but the absolute difference in 

 responses between presumably nonhome water 

 from the Ahnapee River (R) and home water 

 (Aa) was small. (3) Six fish collected from the 

 Kewaunee River and four controls from the 

 Ahnapee site were tested with water collected 

 at Ab and Aa and from the corresponding sta- 

 tions from Kewaunee, Kb, and Ka. Kewaunee 

 fish responded to their home-stream water more 



894 



