Butler et al.: Biology and population dynamics of Sebastes levis in the southern California Bight 



273 



0.07 



06 - 



005 - 

 o 

 S 004 - 



Q. 



^ 0,03 - 

 0.02 

 01 



-o- - Data 

 Model fits 



I960 



1970 



1980 

 Year 



1990 



2000 



Figure 10 



Area-weighted catch per unit of effort (CPUE) for cowcod in the commercial 

 passenger fishing vessel fleet during 1964-96. "Observed" values are habitat- 

 area-weighted means. "Model fits" are predicted values from the stock 

 assessment model fitted to the data. 



erations, we used average estimated recruitment during 

 the 1951-80 seasons as the assumed level of constant 

 recruitment during the virgin and historical periods in the 

 final model run used to estimate biomass, recruitment, and 

 fishing mortality for management purposes. 



Trends in historical estimates indicate that cowcod 

 biomass was near the virgin level in 1916 and remained 

 relatively stable throughout the historical period (prior 

 to 1951), whereas catches were low and recruitment was 

 assumed relatively high (Table 1 and Fig. 11). Biomass 

 began to decline slowly in the 1950s, and the decline 

 accelerated through the 1970s because catch increased 

 while recruitment remained relatively constant. Recruit- 

 ment, biomass, and catches all declined dramatically after 

 the early 1980s. Reduced catches from the mid-1980s to 

 1990s were not enough to offset continuous reductions in 

 biomass and recruitment; fishing mortality rates increased 

 and usually exceeded 0.1/yr. SCB cowcod biomass during 

 the 1998 season (about 238 t, CV 337r) was 7.4% (CV 28%) 

 of the level in the 1951 season (3198 t) and 6.5% of the vir- 

 gin level (3472 t). During the most recent decade ( 1989-98 

 seasons), recruitment biomass averaged 8 t/yr and catches 

 averaged 52 t/ yr. 



Discussion 



We used the best available information to estimate total 

 cowcod catch but the data and our estimates were impre- 

 cise. It is possible, for example, that CPFV captains over- 

 report cowcod catches as a means of attracting business, 

 although cowcod are a small part of the catch on most 

 CPFV vessels, but we are not aware of such a practice. Un- 



certainty about catch affects the magnitude of F and bio- 

 mass estimates, but had little effect on estimated biomass 

 trends after 1951 or the ratio of current and virgin biomass 

 (Butler etal., 1999). 



Our estimates of habitat area for cowcod (Fig. 1) were 

 crude because depth preferences are uncertain and because 

 we were not able to distinguish rocky areas that are pre- 

 ferred by adult cowcod. Fortunately, our analysis of CPUE 

 in the CPFV fishery was robust to errors in estimating 

 total potential habitat because habitat areas were used 

 as relative (rather than absolute) weights in computing 

 average CPUE. 



Biomass, recruitment, and fishing mortality estimates 

 from the stock assessment model for cowcod were less 

 precise for seasons prior to 1951 because historical esti- 

 mates were based on catch data and no abundance data 

 and because they involved assumptions about virgin bio- 

 mass and an educated guess about average recruitment 

 during 1916-51 (Kimura and Tagart, 1982; Kimura et al., 

 1984; Kimura, 1985). Despite these caveats, the estimates 

 for seasons prior to 1951 were plausible and based on all 

 available data. 



Rare-event data 



Our experience with cowcod suggests that rare-event data 

 for fish stock assessment work is an important topic for 

 future research. Rare events may be particularly important 

 in understanding the population dynamics of naturally 

 rare, severely overexploited or difficult to sample organ- 

 isms, particularly if long-term data from intensive sam- 

 pling programs are unavailable. The underlying theory is 

 understood (Mangel and Smith, 1990), statistical tools are 



