tions for immature fish is scanty, but the circum- 

 stantial evidence certainly indicates tliat the Bris- 

 tol Bay sockeye salmon undertake substantial 

 feeding migrations during their second and third 

 summers at sea licfore they return to tlie home 

 stream. 



Fourth, much of the migration is not to and 

 fro, but circular. The circuit which is closed only 

 upon return home appears to consist of a single 

 loop in the pink salmon from southeastern Alaska- 

 I'ritish Cohunbia and Kamchatka and multiple 

 loops in tlic Bristol Bay sockeye salmon. In none 

 of these cases does it appeal' to be possible for the 

 salmon to use memorized -stimuli tliat could be 

 followed back in reverse order. 



During tliese long migrations the races tliat liavc 

 been studied extensively occupy a dis^^-inctive but 

 \ery large j)art of the ocean. Thus, tlie ranges of the 

 various stocks of the sevei-al species overlap to an 

 enormous extent. For example, tagged salmon 

 have returned to Bristol Bay and also to the 

 Fraser River from a group caught in a single set 

 of the net in the northwestern part of the Gulf of 

 Alaska. The salmon of the several species tagged 

 near Adak Island, Alaska, have been recovered 

 from neai'ly the western, northern, and eastern 

 extremities of the range of Pacific salmon. 



We believe that particular stocks of salmon have 

 no tendency to school as a group in the oc*aii. 

 We reinforce this assertion by the observation 

 that usually we capture salmon of different species, 

 age groups, and sizes on single sets of gear excejit 

 when close to a destination of maturing salmon 

 where a single stock may predominate. Even in 

 such areas with a dominating single stock, the 

 numbei-s caug'ht per seine set are much more uni- 

 form over considerable areas and on successive 

 days than would he. true if the salmon were in 

 separate, compact schools as are lierring. for 

 example. 



Perhaps the most startling evidence of the in- 

 dividual behavior of salmon is indicated liy the 

 distribution of the mature Bristol Bay sockeye 

 salmon in ]\Iay and early June, 4 to 10 weeks liefore 

 they arrive in Bristol Bay (fig. 1.5). At this time 

 they are .spread over some 2,000 miles (3,704 km.) 

 of ocean in an east-west direction and some must 

 migi-ate to Bristol Bay from as far as 1,200 miles 

 (2,222 km.) away, either directly or in a dogleg to 

 circunmavigate the Alaskan Peninsula. Within 



this range no evidence has appeared of segregation 

 of the runs to the individual Bristol Bay rivers. 



The salmon appear to be nearly continuous 

 travelers. Many of them average about 10 miles 

 (18.5 km.) per day while immature. AYhen matur- 

 ing, they commonly travel an average of 25 to 30 

 miles (46.3-55.6 km.) a day and occasionally may 

 average more than 45 miles (83.3 km.) daily over 

 long distances. 



These speeds are clearly faster than most ocean 

 currents that may carry the fish. The directional 

 catciies of the purse seine in many parts of the 

 ocean for both mature and inimaiture fish indicate 

 clearly that the salmon exceed the^ speed of the 

 current. The migration is i)ositive, not a passive 

 drifting. 



These long migrations terminate on a remark- 

 ably consistent schedule. The migration of the 

 Bristol Bay sockej-e salmon past Adak Island 

 (al)out 1»(>0 miles [1,607 km.] from home) is com- 

 pleted in about the same length of time and in 

 about the same sequence of individuals as the 

 migration through the fishing areas in Bristol Bay 

 (Hartt, U»66). The runs in Bristol Bay in 1956-65 

 peaked on the average date of July 5, and the 

 earliest and latest j^eaks were July 2 and July 9 

 (Royce, 1965). Eighty percent of the run in the 

 same years arrived in 9 to 22 days. Other runs of 

 salmon in more southerly latitudes commonly oc- 

 cur over greater periods of time, but we believe 

 this spread is usually due to different schedules of 

 I lie several populations that make up these runs. 

 It appears to be common for a single interbreed- 

 ing population to keep a schedule that varies from 

 the average by only a few days. 



The arrival of the salmon is less variable than 

 the seasonal change in the weather. For example, 

 average water temperatures at Weather Station 

 P, located at lat. 50° X., long. 145° W. in the central 

 (iulf of Alaska, show an average monthly increase 

 of 2.2° C. from May to July (Bureau of Commer- 

 cial Fisheries, 1957-65), but the i-ange of attained 

 mean monthly temperatures for 1957-65 shows 

 that the May mean varied from 6.1° to 7.8° C, 

 tiie June mean from 7.8° to 10.6° C, and the July 

 mean from 10.6° to 13.3° C. If the timing of sahn- 

 iin migrations were governed by certain critical 

 temperatures in tiie waters through which they are 

 distributed, then the arrival date would vary by 

 about 2 weeks around a mean. 



456 



U.S. FISH AND WILDLIFE SERVICE 



