600 



Fishery Bulletin 97(3), 1999 



quently resulted in sharp declines in fishery catch 

 rate (Fujioka, 1986; McDevitt, 1986). 



An estimate of the area occupied by adult sable- 

 fish was used to calculate the depletion and hook- 



Asymptotic selectivity, age 

 and length data 



B Asymptotic selectivity, 

 age data 



(0 



E 

 o 

 .n 



a 

 ra 



o 



Q. 

 X 



LU 



79 81 83 85 87 89 91 93 95 



C Asymptotic selectivity, 

 length data 



79 81 83 85 87 89 91 93 95 



D Dome-shaped selectivity, 

 age data 



600 

 500 

 400 

 300 

 200 

 100 

 



-4- 



-M- 



79 81 83 85 87 89 91 93 95 



79 81 83 85 87 89 91 93 95 



Year 



Figure 10 



Exploitable bioma.ss (thousand.s metric tons) versus year for simulated data. 



The "true" values ( ) are compared with the average values ( ) for the 



simulations. 



spacing estimates of absolute biomass. The estimated 

 area is inexact and, correspondingly, so are these 

 biomass estimates. However this area would have to 

 be an order of magnitude higher or lower to be con- 

 sistent with the observed catch effects on the abun- 

 dance indices, a level of error unlikely given the 

 extensive longline surveys for sablefish. 



Data and model requirements 

 necessary to estimate abundance 



Estimates of absolute abundance for sablefish were 

 unbiased (Fig. 10, A-C) and relatively precise (Fig. 

 11) when selectivity was asymptotic. Abundance 

 estimates were relatively precise regardless of 

 whether age data alone, length data alone, or age 

 and length data together were used. Even though 

 length data alone is less desirable than age data 

 for estimating annual recruitment, the precision 

 of the abundance estimates was similar, support- 

 ing the notion that length data is useful for abun- 

 dance estimation with age-structured analyses. 

 Dome-shaped selectivity affects abundance es- 

 timation. Sablefish abundance esti- 

 mates were positively biased (Fig. 

 lOD) and less precise (Fig. 11) for a 

 survey with dome-shaped selectiv- 

 ity. Abundance estimates for a simu- 

 lated widow rockfish, Sebastes 

 entomelas, fishery off the western 

 coast of the continental United 

 States also were less accurate and 

 less precise for a survey with dome- 

 shaped selectivity than one with 

 asymptotic selectivity (Bence et al., 

 1993). Thus, surveys with asymp- 

 totic selectivity appear better for 

 abundance estimation than surveys 

 with dome-shaped selectivity. 



The sablefish age-structured 

 model was fitted with 17 years of 

 survey length data, but only five 

 years of survey age data and no fish- 

 ery age or length data. Two assump- 

 tions probably were the keys to how 

 it was possible to fit an age-struc- 

 tured model with only limited age 

 data (Table 1 ). One key assumption 

 is that selectivity is equal for the 

 fishery and survey, so that no fish- 

 ery age or length data were required 

 to independently estimate fishery 

 selectivity. Another key assumption 

 is that the age-length transition 

 matrix is constant over time, so that 



