591 



Abstract.— Relative abundance of 

 sablefish, Anoplopoma fimbria, in the 

 waters off Alaska has been measured 

 annually since 1979 with longline sur- 

 veys. These extensive surveys provide 

 precise measures of relative abun- 

 dance. An age-structured model was 

 fitted to the longline survey data to es- 

 timate absolute abundance. Estimates 

 of recent exploitation rates for fully se- 

 lected ages averaged IC/r. Monte Carlo 

 simulations of the age-structured model 

 indicated that absolute abundance of 

 Alaskan sablefish can be estimated re- 

 liably if age selectivity is asymptotic 

 and not dome-shaped. Abundance esti- 

 mates were reliable even when only 

 length data and no age data were avail- 

 able. Dome-shaped selectivity gave bi- 

 ased and less precise estimates, prob- 

 ably owing to a parameter interaction 

 between catchability and the shape of 

 the selectivity function. 



Estimation of sablefish, Anoplopoma fimbria, 

 abundance off Alaska with an age-structured 

 population model 



Michael F. Sigler 



Auke Bay Laboralory, Alaska Fisheries Science Center 

 11305 Glacier Hwy, Juneau, Alaska 99801-8626 

 E-mail address Mike Sigler@noa3.gov 



Manuscript accepted 18 August 1998. 

 Fish. Bull. 97:591-603 (19991. 



Sablefish, Anoplopoma fimbria, is 

 a long-lived species that inhabits 

 the northeast Pacific Ocean and 

 Bering Sea. This species supports a 

 fishery in Alaskan waters, with 

 catches ranging from about 10,000 

 to 35,000 metric tons (t) during the 

 last two decades (Fig.l). The fish- 

 ery mostly uses longline gear and 

 primarily occurs on the upper con- 

 tinental slope, which is inhabited by 

 adult sablefish (Fig. 2). Previously 

 the fishery became compressed 

 year-round during the mid-1980s to 

 10 days in some areas, until 1995, 

 when management switched to indi- 

 vidual fishing quotas and an eight- 

 month season. The delay of the start 

 of the season from 1 January to 1 

 April and later 15 May accompanied 

 the compressed fishery season. 



In Alaska, relative abundance of 

 sablefish has been best measured by 

 annual longline surveys since 1979. 

 Longline surveys are preferred over 

 trawl surveys for assessing sable- 

 fish because the longline surveys 

 generally cover the areas that adult 

 sablefish inhabit, namely the upper 

 continental slope. The longline sur- 

 veys have occurred between early 

 May and late September. The sur- 

 vey and fishery generally take place 

 in the same area, the upper conti- 

 nental slope but the fishery gener- 

 ally takes place over a narrower 

 depth range. Trawl surveys have 

 also been conducted in Alaska, but 

 compared with the longline surveys, 

 they have not sampled as deeply 

 and have covered fewer years. In 

 stock assessments for Alaskan 



sablefish prior to 1996 (Fujioka, 

 1995; Lowe, 1995), the limited trawl 

 data were used to convert relative 

 abundance from the longline survey 

 to absolute abundance by calibra- 

 tion to the trawl (Rose, 1986), and 

 the population was modeled by us- 

 ing a delay-difference analysis 

 (Kimura, 1985). The formulation of 

 the delay-difference analysis ap- 

 plied to Alaskan sablefish was modi- 

 fied to provide annual recruitment 

 estimates, with the assumption that 

 trawl surveys measure absolute 

 abundance (Fujioka, 1995; Lowe, 

 1995). The Alaskan sablefish stock 

 assessment was like other Alaskan 

 groundfish stock assessments, 

 which historically have assumed 

 that trawl surveys measure abso- 

 lute abundance (Alverson and 

 Pereyra, 1969). This assumption 

 probably is wrong because fish can 

 escape under the net (Engas and 

 Godo, 1989a) and be herded by the 

 bridles (Engas and Godo, 1989b). 

 The motivation for the current 

 study was to assess Alaskan sable- 

 fish without this assumption, rely- 

 ing only on longline survey esti- 

 mates of relative abundance. 



In the current study, I estimated 

 absolute abundance for Alaskan 

 sablefish with an age-structured 

 population model, evaluated the 

 estimation approach with Monte 

 Carlo simulations, and investigated 

 interactions of model parameters. 

 Delay-difference analyses use an- 

 nual abundance estimates and 

 catches, whereas age-structured 

 analyses additionally use annual 



