McGurk Allometry of marine mortality of Pacific salmon 



85 



The most important source of error in this analy- 

 sis was associated with estimating mean body 

 weights of smolts and adults. The methods of esti- 

 mating mean adult weights were less than optimal 

 mainly because of a lack of information on this sub- 

 ject in the primary literature. Part of the problem in 

 obtaining adult weights is that salmon fisheries are 

 rarely located at the ocean terminus of the salmon's 

 spawning migration. Rather, fishing pressure on a 

 single stock is often distributed over thousands of 

 square kilometers and over several separate man- 

 agement jurisdictions. For those studies that did not 

 report mean adult weight, it was necessary to as- 

 sume that all fisheries were terminal in order to es- 

 timate adult weight from catch records. The assump- 

 tion was undoubtedly violated for many stocks that 

 are subjected to fishing mortality for a considerable 

 portion of their last year of sea life. For those popu- 

 lations, mean weight at sea will tend to underesti- 

 mate weight at re-entry. 



A second source of error was the inability to adjust 

 mean adult weight for the presence of jacks, those 



male salmon that mature at an early age and small 

 size. The effect of a significant number of jacks in a 

 population is to underestimate adult weight at re- 

 turn, thereby increasing the bias introduced by as- 

 suming that adult weight can be estimated from com- 

 mercial catches. However, sensitivity analyses 

 showed that adult weight would have to be consis- 

 tently over- or under-estimated by more than 25% 

 before these two sources of error would seriously af- 

 fect the conclusions of this study. 



A third source of error was the use of weights pre- 

 dicted from length rather than observed weights, 

 because weight-length relationships may vary among 

 populations and among brood years within popula- 

 tions. However, Koenings and Burkett (1987) re- 

 ported that known weight-length relationships for 

 sockeye smolts in Alaska were virtually identical. 

 Further, this source of error would not bias the analy- 

 sis in a specific direction. 



The lack of smolt and adult weights derived from 

 the same brood years resulted in the within-popula- 

 tion variation in survival being confounded with 



