424 



Fishery Bulletin 97(3), 1999 



2,000 



1,500 



S 1,000 



6 500 - 



[ n 1 1 I Pf 

 35 40 44 48 52 56 60 64 68 72 76 80 84 

 Fork length (cm) 



n 1 1 1 1 1 

 92 96 



Figure 4 



Size distribution of North Pacific albacore tag releases, 1971-89. 



Nuisance parameters 



We assumed that tag shedding was the only source 

 of type- 1 and type-2 tag loss and therefore used val- 

 ues for a and t// of 0.12 and 0.0245 per quarter, re- 

 spectively, based on analysis of double-tagging data 

 (Laurs et al., 1976). There are no independent esti- 

 mates of nonreporting of North Pacific albacore tags, 

 although Kleiber and Baker (1987) stated that 

 "nonreporting losses are small for the major fisher- 

 ies that recovered the tags"; they assumed p = 0.9. 

 We also made this assumption but tested the sensi- 

 tivity of our results to values of /3in the range 0.1 to 

 1.0. The value of /( 0.87) was estimated directly from 

 the tag-return data. 



Reparameterization of Fishing mortality 



To create sufficient degrees of freedom to enable a 

 statistical estimation of parameters, the number of 

 fishing mortality parameters to be estimated, F„y 

 needs to be reduced. If we assume for the moment 

 that F,jf is independent of release group /, F if may be 

 reparameterized as a function of fishing effort, Ejf, 

 and catchability, g,/ : 



'jf 



(2) 



We may then propose models that constrain q . in 

 some sensible way. Let S^ denote a matrix that con- 

 tains, for each f, a season for each time period in- 

 dexed hy j. We tested two seasonal schemes: no sea- 

 sonal variation for any fleet and seasonal variation 

 (Ql=Jan-Mar, Q2=Apr^un, Q3=Jul-Sep, Q4=Oct- 

 Dec) for all fleets. 



