HJORT and SCHRECK: PHKNOTYIMC DIFFERENCES AMONG COHO SALMON 



study all should play a role in choosing stocks for 

 transfer to other stream systems. 



The third trend mentioned (that of hatchery 

 and wild stocks diverging toward different 

 phenotypes) presents a problem to managers 

 who must choose the best stock for transfer to 

 other stream systems. The separate clustering 

 of hatchery and wild stocks suggests that hatch- 

 ery stocks have become dissimilar to wild 

 stocks — even those that inhabit the same drain- 

 age. Studies with steelhead trout indicated that 

 hatchery fish survived better in hatchery ponds, 

 whereas wild fish had higher survival in streams 

 (Reisenbichler and Mclntyre 1977). The dis- 

 similarity between hatchery and wild stocks 

 may play a role in reducing the survival of 

 hatchery-reared coho salmon when they are re- 

 leased into a stream system. 



CONCLUSIONS 



Individual characters of the stocks examined 

 by us showed a variety of responses to stream 

 characters. Time of peak spawning was strongly 

 correlated with latitude, whereas other charac- 

 ters were significantly correlated with several 

 environmental gradients, suggesting that 

 interactions determining stock phenotypes are 

 complex. The variability of the stock character 

 may also change along environmental gradients, 

 as demonstrated by the transferrin genotype 

 (Figs. 4, 5). 



The results of the cluster analysis indicate that 

 stocks that are geographically close are similar, 

 that stocks from large stream systems are simi- 

 lar to each other, that stocks from coastal stream 

 systems are similar to each other, and that hatch- 



Similarity of Stream Systems and 

 Wild Stocks 



Because coho salmon appear to have similar 

 phenotypes in similar environments, one could 

 possibly relate phenotypes of stocks with de- 

 scriptions of their stream basins (Tables 5, 9). 

 However, comparisons of an agglomerative 

 cluster analysis of wild stocks (Fig. 7) with a 

 cluster analysis of stream characters (Fig. 8) in- 

 dicated that they were less similar than we had 

 anticipated — although we expected some differ- 

 ences because the stream characters were not 

 necessarily related to taxonomic characters used 

 in this study. 



Table 9.— Fish species and myxosporidan parasite. Cerata- 

 myxa skasta, present in the Oregon stream systems. X = 

 present. 



Stream systems 



CO 



o 



°>.c 



CD 



CD 



3 



a 



W 



II 



u> — 



cr 



3 



5 

 o 



CO 



o a 



CO "> 



T3 



a 



it* 



in 



CD m 

 Q-ra 



T3 



CO 



CD 



33 



on 



c/> CD 



3 

 O 



E m 



oro 



TO 



X 



co co 



2-c 



CO 



Nehalem River 

 Trask River 

 Nestucca River 

 Salmon River 

 Siletz River 

 Beaver Creek 

 Alsea River 

 Smith River 

 Umpqua River 

 Tenmile Lake 

 Coquille River 

 Rogue River 



X 

 X 

 X 



X 

 X 

 X 



•I 



in 



4.0 

 3.5 

 3.0. 

 2.5 



z.o 



1.5 



1.0. 



.5 



>- 

 l- 



a. 



in 



HI 



5.0. 

 4.5 

 4.0. 

 3.51 

 3.0 

 2.51 

 Z.O 

 1.5 

 1.0 

 .5 



STOCK 



STREAM SYSTEM 



Figure 7.— Dendrogram of the agglomerative cluster analy- 

 sis for wild coho salmon stocks with a Euclidean distance 

 dissimilarity measure and group average clustering strategy. 

 Location codes are as in Figure 1. 



Figure 8.— Dendrogram of the agglomerative cluster analy- 

 sis for stream systems with a Euclidean distance measure and 

 group average clustering strategy. Location codes are as in 

 Figure 1. 



117 



