liCARNECCHlA and WAC.NER CONTRIBUTION OF COHO SALMON To OCEAN SPORT FISHERY 



ITaBLE 3. — Description of scale characters measured or counted 

 in this study from known hatchery and wild coho salmon- All 

 measurements are made at angles ventral to the longest axis. 



) Character Description 

 1 Radius of preocean zone at 20 



2 Number o( circuli in the preocean zone at 20 



3 Distance between circuli 1 and 5 ol the preocean 



zone at 90 



4 Distance between circuli 1 and 10 of the preocean 



zone at 90 



5 Distance between circuli 1 and 15 of the preocean 



zone at 90 

 ji 6 Radius of preocean zone at 90 



 7 Number of circuli in preocean zone at 90" 



7 3 Number of broken or branched circuli within precise- 



ly defined zone (see fvlethods) 



released by hatcheries of the Oregon Department 

 of Fish and Wildlife. Most hatchery-reared smolts 

 currently being released by Oregon's hatcheries 

 are larger than wild smolts (Oregon Department 

 of Fish and Wildlife"). Because radii of scales and 

 number of circuli appear to be well correlated to 

 length of Pacific salmon smolts (Clutter and 

 Whitesel 1956), they are logical selections for 

 scale characters to use in separating hatchery and 

 wild fish. 



.Some scales had no "plus" (Anas and Murai 

 1969) or estuarine growth whereas others had 

 substantial amounts. We chose to measure total 

 freshwater growth plus any spring and estuarine 

 growth and to call that distance the "preocean" 

 zone. 



We chose three spacing characters, 3 through 5 

 (Table 3), to determine whether the plentiful food 

 supply of hatchery coho salmon would yield differ- 

 ent spacing of circuli than that observed for wild 

 fish. We measured these characters at 90 to the 

 longest axis of the scale because breaking and 

 branching of circuli is less at that angle than at 

 lower angles to the longest axis. 



The number of broken or branched circuli 

 (character 8, Table 3) was used to determine if 

 circuli of hatchery fish were more or less branched 

 than circuli of wild fish. It was postulated that 

 regular feeding by hatchery fish would result in 

 less breakmg and branching of circuli. For this 

 character an acetate sheet with thin parallel lines 

 1 cm apart and a dotted line parallel to and mid- 

 way between these lines was used as a guide. A 

 small point at the end of the dotted line was placed 

 at the center of the focus of the scale and a dotted 

 line extended outward at 90" ventral to the longest 



axis. The two solid outer lines then enclosed a 

 rectangular area. Within this area, we counted 

 circuli 5 through 12 inclusive in the preocean zone 

 of the scale and recorded the number of these cir- 

 culi that were broken or branched. 



The eight characters in Table 3 were measured 

 and counted from scales of known hatchery fish 

 (Table 1) and known wild fish (Table 2). These 

 measurements were subjected to discriminant 

 function analysis, which reduced all characters for 

 each scale to a single value and then, through a 

 linear model, classified the scales as hatchery or 

 wild (Nie et al. 1975). Assumptions in this 

 analysis were that data were multivariate normal 

 and had common variance-covariance matrices. 

 Plotting the data for each of the eight characters 

 individually showed that only character 8 de- 

 viated somewhat from normal. Although normal- 

 ity of individual characters does not imply joint 

 normality, it indicates that the data conform fairly 

 well with the assumption of multivariate normal- 

 ity- 



From the discriminant function analyses, it was 

 concluded that preocean radius at 20° ( character 1 ) 

 was the most efficient individual character for 

 separating hatchery and wild fish. Preocean 

 radius at 20° is generally larger in hatchery fish 

 than in wild fish. By using the character, we reli- 

 ably separated S'2f7c of the hatchery fish and 89'/f oi 

 the wild fish. 



While characters 1, 3, and 8 in combination 

 would do as well, there would be no benefit to their 

 use except that a slightly higher percentage (1.1) 

 of hatchery fish would be correctly classified at the 

 expense of a lower percentage (1.2) of wild fish 

 correctly classified (Table 4). 



Since preocean radius at 20° was the most useful 

 character for discriminating between adult hatch- 

 ery and wild coho salmon, this character was mea- 



T.ABLE 4. — Combinations of scale characters to which discrimin- 

 ant function analysis was applied and effectiveness at classify- 

 ing coho salmon as to wild or hatchery origin. 



Percentage correctly classified 



'Oregon Department of Fish and Wildlife. Unpubl. slat, of the 

 Fish Culture Division. 17330 SE Evelyn Street. Clackamas. OR 

 97015. 



1 , 8, and 3 

 1 and 2 

 1 through 8 

 1. 8. 3.2, and 5 



619 



