Stevens Survival of king and tanner crabs captured incidentally in the Bering Sea 



739 



ner crabs were typically broken or crushed carapaces 

 which were associated with lower survival. In general, 

 injuries were present in a greater proportion of dead 

 crabs than live crabs. Combined body and leg injuries 

 were most fatal for king crabs, whereas leg-only in- 

 juries accounted for the greatest number of dead Tan- 

 ner crabs. The low survival odds for uninjured king 

 crabs suggests that suffocation in the net or bunkers 

 may have contributed strongly to their mortality. 



Contingency table analysis (Table 3) showed that 

 significant interactions occurred between vitality (VIT) 

 and delayed mortality (DMORT; X- = 138.6), indicat- 

 ing that VIT was an excellent predictor of future sur- 

 vival or mortality for those king crabs held in survival 

 tanks. Similarly, interaction between presence or 

 absence of injuries (INJ) and DMORT (X^ = 16.46) in- 

 dicated that significantly more injured crabs died than 

 were expected to, and that between VIT and INJ 

 (X- = 9.77) suggests that injuries were partly respon- 

 sible for observed vitality levels. 



Loglinear analysis with DMORT as the dependent 

 variable, and the interactions of DMORT with both VIT 

 and INJ as independent variables (Table 4) indicated 

 that the overall odds of survival were 2.087:1. These 

 odds were increased by a factor of 4.713 for crabs coded 

 as active (VIT = 1), and by a factor of 1.395 for crabs 

 without injuries (INJ = 1). The total delayed survival 



Table 4 



Parameter estimates and survival odds for each factor in the 

 iogit model for king crabs. The dependent factor, DMORT, 

 represents the mean survival odds (ratio of survivors to mor- 

 talities) and has no alternative outcome. Each interaction term 

 has two alternative outcomes; alternative 1 represents cells 

 with equal indices, i.e., values for Vitality and Injuries both 

 equal 1 or both equal 2 (Table 5. lines 1 and 4). and alternative 

 2 represents cells with unequal indices (Table 5, lines 2 and 

 3). The survival odds for active crabs with no injuries (Vital- 

 ity and Injuries = 1) is the product of antilogs for DMORT, 

 DMORT X VIT,, and DMORT x INJ,, or 13.73. Survival 

 odds are equal to the antilogs of 2 (coefficient). 



Factor 



Coefficient 



Survival odds 

 Alt. 1 Alt. 2 



DMORT 

 DMORT X VIT 

 DMORT X IN,J 



Sum 



0..368 

 0.775 

 0.167 



0.736 

 1.550 

 0.333 



2.087 

 4.713 

 1.395 



0.212 

 0.717 



:.619 



13.73 



0.317 



odds for active, uninjured crabs eciualed 13.73, the 

 product of all three values. Similarly, survival odds for 

 inactive, injured crabs were 0.317, the product of the 

 mean odds (2.087), the odds for inactive crabs (0.212), 



