FISHERY BULLETIN: VOL. 85, NO. 1 



area, has implications for management. The higher 

 representation of the CowHtz-Kalama subgroup in 

 the sport fisheries than in the troll fisheries of com- 

 mon times and areas suggests a greater suscep- 

 tibility of this subgroup to sport harvests. In addi- 

 tion, the relative abundance of the Cowlitz-Kalama 

 subgroup compared with the Spring Creek subgroup 

 was higher in more southern areas for both commer- 

 cial and sport fisheries. If these trends continue to 

 be observed, different management strategies could 

 be applied for these groups when warranted. 



The low estimates of the Priest Rapids subgroup 

 of upriver brights relative to the two less abundant 

 subgroups suggest different oceanic distributions of 

 these subgroups. However, the coded wire tagging 

 data (Table 5) indicate that at least the Snake River 

 and Priest Rapids subgroups are harvested much 

 more intensely in areas to the north of those sam- 

 pled in this study (no tagging data were available 

 for the Deschutes subgroup). Any attempts to iden- 

 tify and protect the weaker subgroups within the 

 sampling areas of this study would be futile unless 

 similar efforts could be applied to these much larger 

 catches in more northern areas. 



A general occurrence of larger proportions of 

 Puget Sound and Canadian fish in the northern sam- 

 pling areas is suggested by the similar observations 

 for 2 consecutive years and by the particularly high 

 estimates for these fish in area 4B. Since 1983, more 

 detailed GSI estimates from area 4B have, in fact, 

 been used by the WDF to monitor and regulate 

 Chinook salmon fisheries in the Strait of Juan de 

 Fuca and Puget Sound areas. 



Preliminary results from the September gill net 

 fishery in the lower Columbia River (based on a sub- 

 sampling of 500 fish) were available on the day 

 following the collection of the samples. This poten- 

 tial for rapid turnaround time increases the value 

 of the GSI as a management tool by permitting in- 

 season regulatory adjustments. Such information 

 would allow greater harvest of a healthy stock while 

 continuing to provide for maximum protection of a 

 depressed stock. For example, in years when bright 

 fish are expected to return in great abundance and 

 tules in low abundance, the GSI method could be 

 used to monitor extended fall gill net fisheries to 

 time the entry of tules. When ratios of tules to 

 brights became unfavorable, fisheries could be 

 curtailed. 



It is important to emphasize the arbitrary nature 

 of many of this study's groupings, which were neces- 

 sary to provide a manageable basis for reporting. 

 A focus on the tule and upriver bright contributions 

 was appropriate because of the extensive baseline 



data from the Columbia River drainage, the domi- 

 nance of the tule runs in ocean fisheries, and the 

 distinct oceanic distributions of the tule and upriver 

 bright groups. However, a similar focus on other 

 groupings (e.g., Columbia River spring runs or wild 

 and hatchery stocks of the Oregon coast) is equally 

 feasible, and could easily provide a basis for more 

 detailed information on the distributions of in- 

 dividual populations within such groups. 



The completeness and the reliability of the sets 

 of baseline data that are used affects the accuracy 

 of GSI estimates. This study's focus on the contribu- 

 tion of Columbia River populations to stock mixtures 

 in ocean areas adjacent to the mouth of the Colum- 

 bia River was appropriate for the sets of baseline 

 data that were used. Most estimates were obtained 

 through a data base that included most of the major 

 contributing groups within the Columbia River and 

 allele frequency data from 17 polymorphic loci. 

 These same baseline data can be used over succes- 

 sive years, providing the allele frequencies remain 

 stable among year classes and over succeeding 

 generations. Such stability has been observed for 

 some loci and populations of anadromous salmonids 

 (e.g., Utter et al. 1980; Grant et al. 1980; Altukhov 

 1981). 



This temptation to regard the present baseline 

 data as a static entity should nevertheless be re- 

 sisted for a number of reasons. Gene flow, genetic 

 drift, and selection could modify allelic frequencies 

 over extended time periods; thus, periodic updating 

 of previously sampled populations is desirable. Tem- 

 poral changes in allele frequencies of chinook salm- 

 on have been reported (Carl and Healey 1984; Kris- 

 tiansson and Mclntyre 1976). The extensive stock 

 transplantations of chinook salmon within the 

 Columbia River make the possibility of gene flow 

 particularly likely for the focal populations of this 

 study. Hatchery populations perpetuated by limited 

 numbers of breeders are particularly susceptible to 

 allele frequency changes through genetic drift 

 (Allendorf and Ryman 1987). Previously unsampled 

 baseline populations should be added, particularly 

 in areas where limited sampling has occurred, to in- 

 crease the accuracy and broaden the usable range 

 of analyses. The discriminatory powers of GSI 

 analyses are substantially increased as new variable 

 loci are added (see Milner et al. fn. 4). The con- 

 tinuing search for additional markers requires col- 

 lection of electrophoretic data from previously 

 sampled populations for each new variable locus that 

 is found. 



Increasing application of procedures used in this 

 study seems virtually inevitable in view of the per- 



22 



