FISHERY BULLETIN: VOL. 82, NO. 2 



lighter smolts and increased curvilinearly with 

 longer char (Fig. 3B, C). The maximum increase in 

 char consumption of smolts due to changes in 

 smolt weight and char length was about 1.4 and 

 0.8 smolts/char per 24 h, respectively. 



The predicting power of this model is weak at 

 high consumption rates by char. This problem 

 arises from increasing residual variability as the 

 predicted value increases. Increasing residual 

 variability is usually corrected by using a weight- 

 ing factor; however, when applied, the only data 

 points carrying weight were those near the origin 

 and any relationship between the variables was 

 lost. Linear models were attempted, but did not 

 approach the fit of the nonlinear model. Thus, at 

 high rates of consumption by char, the model is 

 best used for descriptive purposes. 



1976 



1979 



1980 



Comparison of consumption rates by char be- 

 tween years demonstrates the large variability 

 that may be explained by smolt abundance, smolt 

 weight, and char length (Fig. 5). During 1980 

 there were more than three times the number of 

 migrating smolts, the weight of smolts was 

 30-50% less, and length of char was 24-68 mm 

 greater than in any of the previous years. The 

 combined effects of these variables resulted in a 

 relatively large number of consumed smolts per 

 char, which was also predicted by the model. 



Percent Smolt Mortality 



Two different percent smolt mortality curve 

 types were produced from the two hypothetical 

 numerical responses (number feeding) and the es- 

 timated consumption rates of char. A Type II curve 

 (Fig. 2B) exhibiting an inverse relationship be- 

 tween percent smolt mortality and smolt abun- 

 dance was produced from the assumption that all 

 1,100 char fed each day (Fig. 6A). Smolt mor- 

 tality ranged from to 100% when the number of 

 migrating smolts was < 6,750 smolts/24 h and 

 <15% when the number of migrating smolts ex- 

 ceeded 20,000 smolts/24 h. A Type III percent mor- 

 tality curve was produced from the assumption 

 that the number of char feeding varied with smolt 

 abundance (Fig. 6B). Although variability exists, 

 percent mortality increased at low smolt abun- 

 dances (< 20,000 smolts/24 h), then decreased 

 after char became overwhelmed and/or satiated 4 

 by smolts. 



Char Consumption of Smolts by Length 



The comparison of mean lengths of smolt con- 

 sumed by char with mean length in the migration 

 indicates that less full char consumed larger than 

 average smolts (d = 1.7-2.9 mm, a = 0.05; Fig. 7). 

 Char with full stomachs consumed smolts that 

 were not different than the average length in the 

 migration (d = -0.1-0.6 mm, a = 0.05). The length 

 of char did not have a significant effect on the 

 length of smolt consumed. 



The comparison of length of smolts consumed 

 in each stomach fullness category with the length 



FIGURE 5. — Comparison of observed and predicted smolt 

 consumption/Arctic char per 24 h with smolt abundance, average 

 smolt weight, and char length during each sampling year. Data 

 grouped into 3- to 5-d sampling periods. Dash line indicates 

 mean value for all years Log smolt abundance calculated from 

 hundreds of smolts migrating. Arrows indicate entry of adult 

 sockeye salmon in to Little Togiak River. 



4 During days of large smolt migrations, the number of smolts 

 observed in individual char ranged from to 45 smolts (not 

 corrected by digestion period). Because of this variability in 

 consumption, it is difficult to determine whether the char were 

 overwhelmed by smolt abundance or satiated. This observed 

 variability in consumption may be due to individual char mi- 

 grating from the local lake area to the river at different times, 

 thereby causing variable feeding durations. 



406 



