Carls and O'Clair: Responses of Chionoecetes bairdi to cold air 



47 



cal treatment ranges (0-32 minutes, -20 to +5°C; see 

 Results section) and could be described by the same 

 types of simple linear or nonlinear models, we used 

 the same technique here. 



Regression techniques and logit analysis were used 

 to relate response variables to exposure (Berkson, 

 1957; BMDP, 1983). We compared median lethal re- 

 sponses with log-likelihood ratio tests (Fujioka, 

 1986). Multiple regression was used to test for dif- 

 ferences in the slopes of regression lines and to ad- 

 just for covariates (Kleinbaum and Kupper, 1983). 

 The relation of selected response variables to one 

 another was tested with parametric correlation. Af- 

 ter one-way analysis of variance, comparisons of 

 treatment means were made with Tukey's or 

 Dunnett's a posteriori multiple comparison tests and 

 judged significantly different if P<0.05. Proportional 

 data were arcsine transformed. Reported error 

 ranges are ±95% confidence limits. 



Results 



Mortality 



Below -1 to -3 degree hours, exposure to cold air 

 killed crabs. Almost all mortality occurred 1-2 days 

 after exposure; in groups where more than half the 

 crabs died, mortality always reached 50% within 

 2 days. Mortality was inversely related to exposure 

 and increased rapidly below -l°h for juveniles and 

 below -3°h for adults (logistic regressions [large P- 

 values indicate good fits], P Juvenile =0.959, P adul =0.882; 



Fig. 2). Nearly all deaths occurred within 8 days af- 

 ter exposure; no crabs died after day 16. For juve- 

 niles, calculated median lethal exposures rose from 

 -7.7 ± 3.4°h 1 day after exposure to -3.3 ± 0.8°h 16 

 days after exposure, and for adults from -7.2 ± 1.6°h 

 to -4.3 ± 0.5°h over the same time period (Table 2). 



Righting response 



The speed with which crabs righted themselves when 

 placed on their backs was inversely related to expo- 

 sure (Fig. 3A). The response was most clearly de- 

 scribed by the percentage of crabs not righting within 

 two minutes (logistic regressions, P=0.799 [n=6] for 

 juveniles; P=0.978 [ra=22] for adults; Fig. 3B). Per- 

 centages of crabs not righting increased sharply be- 

 low — 1.0°h for juveniles and below -2.2°h for adults, 

 and crabs ceased righting entirely after exposure to 

 <-4.0°h for juveniles and <-6.9°h for adults (Fig. 3B). 

 Median exposures causing one-half the crabs to cease 

 righting (EC50) were -1.2 ± 0.3°h for juveniles and 

 -2.1 ± 0.3°h for adults, measured immediately after 

 exposure; values declined to —1.6 ± 0.3°h for juve- 

 niles and -3.8 ± 0.5°h for adults measured 32 days 

 after exposure (Table 3). The percentage of crabs 

 unable to right themselves immediately after expo- 

 sure was significantly correlated with cumulative 



mortality (P Juvenile =0.003, r' 



venile 



= 0.91, n=6; 



P . „<0.001,r^7,=0.67,/i=22) and, therefore, could 



adult ' adult > 



serve as a predictor of death. 



Righting times tended to improve (decrease) dur- 

 ing the first eight days after exposure, but this re- 

 covery was generally not statistically significant. 



