128 



Fishery Bulletin 105(1) 



al., 1990; Pearcy, 1992; Sogard, 1997; Mortensen et al., 

 2000; Beamish and Mahnken, 2001; Moss et al, 2005). 

 This result is also in accord with the critical-size and 

 critical-period hypothesis in which brood-year survival 

 is determined by the number of juvenile salmon that 

 have reached a critical size by the end of their first ma- 

 rine summer (Beamish and Mahnken, 2001; Beamish 

 et al., 2004). The assumption with this hypothesis is 

 that fish that do not reach a critical size after their first 

 summer at sea will die because they are unable to meet 

 minimum metabolic requirements during late fall and 

 winter (Beamish and Mahnken, 2001). Although our re- 

 sults indicate that larger juvenile sockeye salmon have 

 higher relative marine-stage survival rate after their 

 first year at sea, it is difficult to directly address when 

 the mortality would occur because sockeye salmon can 

 spend an average of 2 to 3 years at sea. However, the 

 overwhelming evidence from field and laboratory studies 

 of juvenile stages of teleost fishes seems to indicate that 

 size-selective mortality occurs during winter because 

 larger members of a cohort are better than smaller 

 members at tolerating physical extremes and enduring 

 longer periods without food (Sogard, 1997). 



One other test of the critical-size and critical-pe- 

 riod hypothesis is that mortality rates after this period 

 should be large in relation to other sources of early ma- 

 rine mortality (Beamish et al., 2004). To interpret our 

 indices of marine-stage survival rate as the actual post- 

 survey marine survival rate requires making a variety 

 of questionable assumptions (e.g., that the vulnerability 

 of juvenile salmon to our gear is known). However, if 

 our estimates are close to correct, they would indicate 

 that marine-stage mortality rates of juvenile sockeye 

 salmon may be greater than 70% (Table 4) after our 

 late-summer-early-fall surveys. These marine-stage 

 mortality rates are substantial and approach late fall 

 and winter mortality rates of greater than 90% found 

 for other Pacific salmon (Beamish et al., 2004). 



Lengths of juvenile sockeye salmon differed signifi- 

 cantly among years if we assumed daily growth rates 

 of 0.3 mm and greater. Differences in fork length of 

 juvenile sockeye salmon could reflect annual differ- 

 ences in early marine growth rates or may also reflect 

 annual differences in the size of smolt leaving Bristol 

 Bay lake systems. However, limited surveys of sockeye 

 salmon smolt from the Kvichak and Ugashik Rivers 

 during 2000 through 2002 (Egegik River sampling was 

 not undertaken in 2002) by ADF&G indicate that dif- 

 ferences in smolt length among years and within age 

 classes and river systems were less than 9%. In ad- 

 dition, the smallest average smolt size among these 

 three years was seen during 2002, the year with the 

 largest juvenile sockeye salmon size. Thus, it is likely 

 that annual differences in length observed during our 

 survey were due to differences in marine growth rates 

 between years. 



The annual variability in juvenile sockeye salmon 

 size and in indices of marine-stage survival rates may 

 be linked to the early marine migration of these salmon 

 along the eastern Bering Sea shelf. Although we had 



only three years of data, size and survival indices of 

 Bristol Bay sockeye salmon were lowest when juvenile 

 sockeye salmon were distributed nearshore along the 

 Alaska Peninsula (i.e., the coastal migration pathway) 

 and highest when they were distributed farther north 

 and offshore. In support of this theory, the coastal mi- 

 gration pathway of juvenile Bristol Bay sockeye salmon 

 observed by Straty (1981) during the late 1960s and 

 early 1970s coincided with a significantly lower produc- 

 tion of Bristol Bay sockeye salmon that occurred before 

 the mid 1970s (Adkison et al., 1996). 



The annual variability in seaward migration path- 

 ways is likely related to ocean conditions on the shelf 

 during spring and summer. Recent studies indicate that 

 sea surface temperatures along the eastern Bering Sea 

 in summer, the period when juvenile sockeye salmon are 

 present on the shelf, is positively correlated with Bristol 

 Bay sockeye salmon survival rates (Mueter et al., 2002). 

 It is possible that the effect of sea surface temperatures 

 on survival rates of juvenile Bristol Bay sockeye salmon 

 is a result of its influence on early marine distribution 

 of juvenile sockeye salmon. For example, during the 

 late 1960s and early 1970s, the nearshore migration 

 of juvenile Bristol Bay sockeye salmon was thought to 

 be a result of sockeye salmon using the warmer near- 

 shore waters rather than the colder sea surface tem- 

 peratures offshore in order to maximize their growth 

 (Straty, 1981). Depth-averaged sea temperatures from 

 an oceanographic mooring along the eastern Bering Sea 

 middle shelf domain from mid-July to mid-September 

 were consistently warmer during 2001 through 2002 

 than during 1995 through 1997 (Overland and Sta- 

 beno, 2004). Presumably, the warmer sea temperatures 

 during 2001 would have been conducive to offshore 

 migration of juvenile sockeye salmon during that year. 

 Although sea temperatures were warmer during 2001 

 through 2002, sea temperatures along the shelf were 1° 

 to 2°C cooler from late June to September during 2001 

 than during 2002 (Overland and Stabeno, 2004). Thus, 

 it may be that warmer sea temperatures during the 

 time juvenile sockeye salmon first are present over the 

 eastern Bering Sea shelf (beginning in June) provide 

 a conduit for rapid offshore migration (and possibly 

 higher survival) and that cooler sea temperatures delay 

 offshore migration. 



Our results indicate that after the first summer in 

 the Bering Sea, larger juvenile sockeye salmon may 

 gain a survival advantage over smaller individuals. 

 This result, coupled with previous findings of reduced 

 juvenile-to-adult survival for pink (Moss et al., 2005) 

 and coho (Beamish et al., 2004) salmon that spend 

 their first summer in the coastal waters of the Gulf of 

 Alaska and Strait of Georgia, indicates that reduced 

 growth of Pacific salmon during their first year at sea 

 may lead to substantial salmon mortality, presumably 

 during their first winter at sea. This phenomenon may 

 not be seen if size of the salmon after their first year 

 at sea is inferred from the scale growth increments of 

 returning adults, because these individuals could be a 

 biased sample from the faster-growing portion of the 



