808 



Fishery Bulletin 92(4). 1994 



tag recaptures with useful information were reported 

 to the AFSC through March 1992 for a recovery rate 

 of 3.0%. Although tag returns occurred over a broad 

 area and time period, this rate is much lower than 

 the 24-26% recovery rate reported by Canadian and 

 U.S. researchers working off British Columbia, 

 Canada, and in Puget Sound, Washington 

 (Westrheim, 1984; Karp, 1982). In our study, research 

 trawl-caught fish accounted for the majority of tag 

 releases (97%), and commercially fished bottom 

 trawls and longlines accounted for the majority of 

 tag recoveries (Table 1). Thompson ( 1994) estimated 

 that the exploitation rate for Pacific cod in the east- 

 ern Bering Sea during 1981-1992 averaged 9-11% 

 annually. This suggests additional tag loss, tagging 

 mortality, or under-reporting of rates which sum to 

 about 2/3 in some combination. 



Fish lengths at release were between 25 and 118 

 cm, representing fish as young as age 2 yr as well as 

 very large, mature fish. The distribution of lengths 

 at recovery corresponds well with the overall 

 tag-release size frequency but is shifted to the right 

 due to growth and gear selectivity (Fig. 2). 



In the commercial fisheries, Pacific cod are first 

 recruited at about 40 cm or age 3 yr. They become 

 available to different gear types (i.e. initially to bot- 

 tom trawls, then to longline gear) at progressively 

 older ages and larger mean size (Shimada, 1985). 

 Most tag recaptures were of commercially recruited, 

 sexually mature fish, older than age 5 yr, and larger 

 than about 60 cm, as defined by Teshima ( 1985). 



More than 75% of all tag releases were from U.S. 

 survey vessels in the eastern Bering Sea (Table 1; 

 Fig. 1 ). Cooperative foreign research vessels operat- 

 ing in the Aleutian Islands and Gulf of Alaska were 

 responsible for the remaining 25%>. Twenty-four per- 

 cent of recoveries were made over the inner shelf, 

 and 29% over the outer shelf and upper slope (Table 

 1; Fig. 1 ). Of particular note was the high concentra- 

 tion of tag recoveries (>42%) in Unimak Pass and its 

 surrounding waters (including the adjacent western 

 Gulf of Alaska) during the winter months (Fig. 1). 

 Only 17 tags (<5%) were recovered from outside the 

 three primary study areas (i.e. from the outer Aleu- 

 tian Islands and the central Gulf of Alaska). 



Mapping seasonal movements 



Tagged Pacific cod exhibited marked spatial and tem- 

 poral displacement from their point of initial release 

 (Fig. 4). Individual movements generally conformed 

 to seasonal shifts in centers of Pacific cod abundance 

 (Ketchen, 1961; Bakkala, 1984) and to the corre- 

 sponding movement of fishing effort ( Shimada, 1985 ). 



We attribute the observed pattern in tagged fish 

 movements to hypothesized migratory shifts between 

 perennial summer (feeding) and winter (spawning) 

 areas (Moiseev, 1952, 1953; Ketchen, 1961). This is 

 most easily seen in the vector movements of indi- 

 vidual fish into and out of the main spawning area, 

 Area 3. These data were grouped in two ways to show 

 1) the origin offish released in all areas and recov- 

 ered within Area 3 [Fig. 4A]; and 2) the outward re- 

 coveries of fish tagged within Area 3 [Fig. 4B]. Al- 

 though the movement into and out of Area 3 is clear, 

 the movement into the spawning areas seems to oc- 

 cur in two stages: 1) movement off the inner shelf 

 [Area 1| into slope areas [Area 2] [Fig. 4C|; and 2) 

 subsequent movement into spawning areas in Areas 

 2 and 3 [Fig. 4D]. This shift is counterbalanced by 

 spring and summer recaptures on the inner shelf 

 [Fig. 4B]. 



The annual cycle of Pacific cod migration appears 

 to begin in late September, when tagged fish move 

 off the eastern Bering Sea shelf and seaward to the 

 200 m shelf break. By fall, tags were recovered, pri- 

 marily along the outer shelf edge. In winter, Pacific 

 cod converged in large spawning masses over rela- 

 tively small areas. Major aggregations were usually 

 encountered between Unalaska and Unimak islands 

 on the Bering Sea side of Unimak Pass. Other recur- 

 ring centers of abundance were located southwest of 

 the Pribilof Islands along the shelf edge and near 

 the Shumagin Island group in the western Gulf of 

 Alaska (Fig. 1). 



Following the spawning season, tagged Pacific cod 

 dispersed from these overwintering areas and were 

 recaptured farther inshore in concert with seasonal 

 warming of the inner shelf environment. For ex- 

 ample, fish tagged in areas of deep off-shelf waters 

 adjoining Unimak Pass, close to the time of winter 

 spawning, were recaptured progressively over the 

 shelf (and especially north of the Alaska Peninsula 

 following the 30-m isobath ) beginning in late spring. 

 Tagged Pacific cod also moved to the northwest outer 

 shelf (100-200 m) during the spring quarter. By sum- 

 mer, the feeding range was well established back in 

 central Bristol Bay (30-50 m) and the outer shelf. 

 This distribution persisted until late fall and the 

 beginning of the next yearly cycle. 



The seasonal nature of Pacific cod movement is 

 most easily seen in the average vector movement of 

 fish tagged within a particular 2° latitude and 5 lon- 

 gitude rectangle and recovered during a specific sea- 

 son (Fig. 5). From these maps, the off-shelf move- 

 ment is clearly visible in fall, and movement to the 

 spawning ground is clearly visible in winter. How- 

 ever, spring and summer maps show relatively little 

 directed movement, because during these time peri- 



