Sigler: Estimation of abundance of Anop/opoma fimbria off Alaska 



597 



0.3 1 



23456789 11 16 



3456789 11 16 



2 3 4 5 6 7 



9 11 16 



23456789 11 16 



23456789 11 16 



Age class 



Figure 5 



Observed ( i and expected ( i age compositions. Data were available only from 



the 1983. 1987. 1989. 1991. and 1993 surveys. See Table 1 for age-class definitions. 



age and length data were used (Fig. lOA). Estimated 

 error (CE [exploitable bioniass]) was about 0.08, in- 

 creasing for the later years to about 0.12 (Fig. Ill, 

 values similar to the assumed CV for the abundance 

 index of 0.10. These values all imply that sablefish 

 abundance could be estimated given the type and 

 quality of data assumed in these simulations. These 

 simulations match my evaluation of the actual data. 

 Biomass estimates also were unbiased for asymptotic 

 selectivity, no matter whether age or "length" data 

 were used (Fig. 10, B and C); CE (exploitable bio- 

 mass) ranged from 0.10 to 0.15 (Fig. 11). In contrast, 

 biomass estimates were positively biased by 149^ to 

 179^ for dome-shaped selectivity (Fig. lOD) and CE 

 ( exploitable biomass ) ranged from 0.40 to 0.50 ( Fig. 11). 

 Thus, estimating abundance from a survey with as- 

 ymptotic selectivity appears reasonable; biomass 



estimates derived from a survey with dome-shaped 

 selectivity can be estimated, but may be biased and 

 less precise. 



Discussion 



Comparison of age-structured and 

 delay-difference analyses 



The estimates of sablefish exploitable biomass from 

 age-structured analysis and delay-difference analy- 

 sis are similar (Fig. 12). Both analyses show exploit- 

 able biomass rising from about 170,000 t in 1979, 

 peaking at about 420,000 t around 1986, and falling 

 since then to about 200,000 t in 1995. The trend from 

 the sablefish delay-difference analysis is less smooth 



