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



733 



no discoloration or scarring. The number of legs with 

 each of these three types of injuries was recorded. 



(D) Survival or delayed mortality A subset of the 

 injury-assessed crabs (Step C) was placed into one of 

 four tanks of flowing seawater, ~ 1.0 x 1.0 x 0.7 m, and 

 held for 48 hours, after which they were reexamined 

 for delayed mortality, using the same procedures 

 outlined above for vitality. The subset usually consisted 

 of all crabs coded for injures from one haul per day, 

 although towards the end of the experiment several 

 hauls per day were treated in this manner. Crabs which 

 were determined to be dead before placement into the 

 tanks were retained for at least 6 hours for confirma- 

 tion before removal. 



Crab processing time was recorded for each tow as 

 the interval between the arrival of the codend on deck 

 and the time the crabs were examined and measured. 

 Although examination usually required anywhere from 

 0.5 to 1.5 hours for an entire sample, the midpoint of 

 that period was used as the mean endpoint of process- 

 ing time for all crabs in a haul. Towing time, i.e., the 

 difference between the time the net was determined 

 to be on bottom and the time haulback started, was 

 obtained from each catcher vessel and divided by 2, in 

 order to approximate the average time each crab spent 

 in the net before delivery to the Svlak. Total captur- 

 ing and processing time (CAPTIME) was calculated as 

 the sum of (towing time)/2 and processing time. 



Total weight of the catch in each haul was estimated 

 by measuring the height and width of the codend at 

 several intervals along its length, and calculating the 

 volume based on the shape of an oblate cylinder, with 

 density determined by weighing a known volume of 

 fish. 



Data analysis 



In order to improve the precision of the survival esti- 

 mates, overall survival was calculated for each species 

 in a stratified manner (Cochran 1963). Vitality codes 

 were used as strata because they provided additional 

 information on the probability of survival, i.e., crabs 

 coded as alive and active were deemed more likely to 

 survive than crabs coded as moribund. For immediate 

 survival estimates only, each crab was weighted by the 

 sampling factor for the tow from which it originated. 

 Estimates of delayed mortality were not weighted, 

 because subsampling was accounted for at the level of 

 immediate mortality. The formulae used were: 



S„ = 1/N I N,p, 



(1) 



where Sst = stratified estimate of population sur- 

 vival rate. 

 V(Sst) = variance of stratified estimate, 



Nv = number of crabs in stratum (vitality 

 group) V, 



nv = number sampled from stratum v (placed 

 in live tanks), 



Pv = proportion of subsample surviving in 

 live tanks, 



Cjv = 1 - Pv , 



N = weighted total number of crabs given 

 vitality codes. 



The effect of CAPTIME on immediate mortality was 

 determined by fitting the data to a logistic curve. Crabs 

 were grouped into hourly intervals of CAPTIME (as 

 described above), and mortality data (weighted by 

 sampling factors) for each interval were fitted using 

 an iterative procedure (FSAS FISHPARM, Saila et al. 

 1988). The equation fitted was: 



M = l/{l + exp(-r[X-X5„])} 



(3) 



and 



V(S,t) = 1/N- I N,(N, - n,)p,q,/(n, - 1) (2) 



where M is expected mortality and 0<M< 1, X is the 

 value of the predictor variable (CAPTIME) in hours, 

 and r and X-,,, are parameters of the equation, the 

 latter representing the time required for 50% mortal- 

 ity to occur. Both king and Tanner crabs were treated 

 separately. Delayed and overall survival were not fitted 

 because there were too few observations. The effect 

 of total catch weight on immediate survival was also 

 determined in the same manner, with crab data 

 grouped by catch weight intervals of 1.0 t. 



To determine the effect of size, crabs were grouped 

 by 15-mm intervals of CL (king crab) or CW (Tanner 

 crab). For king crabs, immediate, delayed, and overall 

 mortality were calculated for each size group. Only im- 

 mediate mortality was calculated for each size group 

 of Tanner crabs, due to limited data. Immediate mor- 

 tality was also determined for each gear type, by 

 species. 



Procedures used by National Marine Fisheries Ser- 

 vice (NMFS) observers aboard fishing vessels to code 

 crab vitality differ from those used in this study. In this 

 study, vitality codes were based on activity level of 

 crabs regardless of injuries, and injuries were coded 

 separately, whereas the NMFS observer method is to 

 assign a vitality code of 1 for crabs which are active 

 and uninjured, and a vitality code of 2 for crabs which 

 are either inactive or injured. Contingency table and 

 loglinear analysis was performed in order to assess the 

 difference between these two methods and determine 

 which provided the best estimate of delayed mortality, 

 and to determine the relative contribution of vitality 

 and injuries to subsequent delayed mortality. Three 



