Table 4. — Statistics on variability studies on mean annua) catch per 

 unit effort <CPUE) of walleye pollock by pair trawlers of the 120-219 

 gross registered tons class in the Japanese surimi (a minced fish 

 product) mothership fishery in the eastern Bering Sea, 1964-76. 



significant increase occurred from 1971 to 1972. There- 

 after, CPUE declined significantly from 1972 to 1975. 



data. These calculations were based on data summarized 

 by major statistical blocks (MBLs, Fig. 1) and months 

 (MBL-months) from five vessel class-gear categories 

 known to fish mainly for pollock in the surimi fishery 

 (Table 2). These vessel-gear types were: 



Type 1— Pair trawlers, 120-219 GRT from the mother- 

 ship fishery. 



Type 2 — All pair trawlers combined from the mother- 

 ship fishery. 



Type 3 — Stem trawlers smaller than 300 GRT from 

 the mothership fishery. 



Type 4— Factory stern trawlers of the 1,500-3,500 GRT 

 class. 



Type 5 — Factory stern trawlers larger than 3,500 

 GRT. 



Our calculations showed that differences in ADIs by 

 the two procedures were generally small when they oc- 

 curred and that ADIs were higher for pollock-majority- 

 data (Table 5). Comparison of monthly ADIs (Table 6) 

 shows that: 1) in over 80' h of the observed number of 



COMPARISON OF ADIs FROM 



ALL-DATA VERSUS 

 POLLOCK-MAJORITY-DATA 



The manner in which data were selected for ADI 

 computations by the Japanese scientists and for CPUE 

 computations by the U.S. scientists was different. 

 Japanese scientists used data from those trawls that pre- 

 sumably targeted on walleye pollock, which were deter- 

 mined as those in which the catch of pollock was greater 

 than that of £iny other species (pollock-majority-data). 

 U.S. scientists used data from all trawls because pollock 

 wets the target species almost all the time. 



Either procedure can produce biased indices of abun- 

 dance. Abundance may be overestimated by the exclu- 

 sion of catch-effort data when the proportion of pollock 

 in catches was low. On the other hand, it may be under- 

 estimated when data were included from trawls that 

 were not fishing pollock. 



In order to compare the two procedures, we calculated 

 ADIs for 1973-76 using all-data and poUock-majority- 



Table 5.— Average density indices, in metric tons per hour trawled, 

 for all-data versus walleye pollock-majority-data by vessel-gear 

 categories in the eastern Bering Sea, 1973-76. 



'Type 1 — Surimi (a minced fish product) mothership. pair trawlers, 



120-219 gross registered tons (GRT). 

 Type 2 — Surimi mothership. all pair trawlers. 

 Type 3 — Surimi mothership. stem trawlers. <300 GRT. 

 Type 4— Surimi factory stern trawlers, 1.500-3,500 GRT. 

 Type 5 — Surimi factory stem trawlers. >3.500 GRT. 



Table 6.— Number of major statistical block months in which average density indices computed from all-data ver- 

 sus walleye pollock-majority data in the eastern Bering Sea were the same, differed by 0.1, or differed by more 

 than 0.1 metric ton per hour trawled. 



'Type 1 — Surimi la minced fish product) mothership, pair trawlers, 120-219 gross registered tons (GRT). 

 T>"pe 2 — Surimi mothership, all pair trawlers. 

 Type 3 — Surimi mothership. stern trawlers, <300 GRT. 

 Type 4 — Surimi factory stern trawlers. 1.500-3,500 GRT. 

 Type 5 — Surimi factory stern trawlers, >3,500 GRT. 



