Yochum et al. A comparison of methods for evaluating mortality of discarded Cancer magister 
131 
in that time frame because of a captivity effect. For 
the analysis, “condition 1” was assigned to crab with 
a reflex impairment score equal to 0. Owing to small 
sample sizes, scores 1-6 were combined and assigned 
as “condition 2.” Sample size was reduced because of 
limited numbers of crab with scores greater than 0 for 
these fisheries and because the comparison had to be 
done by fishery, sex, and shell hardness given that the 
laboratory study determined different mortality rates 
for these variables. 
Relative survival rates between conditions 1 (score 
0) and 2 (score greater than 0) after the “recovery pe¬ 
riod” (i.e., 5 d in holding for the laboratory study, and 
5 d at large for the tag-return study) were estimated 
for both studies and compared. This was done with all 
data combined for the laboratory study, and by release 
event (i.e., sampling trip) for the tag study to control 
for the influence on return rate of days-at-large, natu¬ 
ral and fishing-induced mortality, and temporal vari¬ 
ability in catchability, fishing effort, tag loss, and re¬ 
porting rate. 
_ C 2 ! C x 
T 2 /Ti’ 
( 1 ) 
where C x - Number of ‘surviving’ crab after the re¬ 
covery period (tag study: recaptured tags; 
laboratory study: surviving crab) for the ith 
condition 
Tj = Total number of experimental crab (tag 
study: tagged; laboratory study: held) for 
the ith condition. 
Two-sided confidence intervals (CIs) for the survival 
rates were calculated to compare the 2 studies. They 
were calculated as (Hueter et ah, 2006): 
{Re~ Zl - a/2 ^,Re Zl - al2 ^ ) j, ( 2 ) 
where Z 1 _ ot/2 is the 100 (l-oc/2)th percentile of the stan¬ 
dard normal distribution and 
v = T 1~ C 1 + T 2~ C 2 ( 3 ) 
T X C X T 2 C 2 ' 
Relative long-term survival 
If there was a more chronic, sublethal effect from the 
capture, handling, and discard process (e.g., nonlethal 
physiological impact or change in behavior), a differ¬ 
ence in survival rates between conditions could con¬ 
tinue beyond a short recovery period (here, 5 d). This 
difference in survival would indicate that laboratory 
holding under-estimates mortality by evaluating only 
over the short-term. To evaluate the potential change 
in relative survival, by condition over time, we used 
logistic regression in R software, vers. 3.1.1 (R Core 
Team, 2014) to model the probability that a recaptured 
tag was from a condition 2 (score greater than 0) crab, 
including a variable for days-at-large, and 3 indicators 
of sex and shell hardness. This analysis allowed us to 
determine whether the odds of return between the 2 
conditions change over time. A nonzero time-dependent 
slope coefficient indicates different long-term survival 
between conditions. The intercept would be 0 if the 
same number of tagged crabs were released in the 2 
conditions and they suffered the same rate of short¬ 
term mortality after 5 d at large (t= 0). If the 2 condi¬ 
tions had the same short-term survival, a nonzero in¬ 
tercept would reflect the ratio of the number of tagged 
animals in the 2 conditions at t-0. We compared crab 
within a release event with the assumption that rela¬ 
tive natural mortality, catchability, and reporting prob¬ 
ability are the same for a given condition and release 
event. 
Evaluating the return to water 
In addition to the laboratory holding and tag-return 
studies, 2 experiments were conducted, one in Novem¬ 
ber 2013 and a second in April 2014, to evaluate the 
contribution to discard mortality by dropping crab into 
water. This is a potential cause of different mortality 
rates between the tag and laboratory studies, and a 
possible source of error in laboratory-determined mor¬ 
tality rates. For these drop studies, score-0 crab were 
collected by using recreational fishing gear, tagged, and 
were held for 2 weeks before experimentation. Then, 
for each drop height of each experiment, 20 crab at a 
time were taken out of holding and, for transport to 
the drop location, were placed in a large ice chest filled 
with wet burlap (to provide barriers among the crab). 
For the second study conducted in April 2014 only, the 
ice chest used for transport was filled with sea water 
in addition to the burlap so that the crab were in wa¬ 
ter until they were dropped. At the drop location (ap¬ 
proximately 200 m away), 3 crab at a time were taken 
out of the ice chest, lifted to one of 3 drop heights, and 
released (dropped) one-by-one into a tank of sea water. 
Drop heights for both experiments were 8 m (“high”) 
and 3 m (“medium”). Crab were also dropped from 1 m 
(“low”) for the first study and from 6 m (‘high’) for the 
second study. The drop distances reflected an attempt 
to mimic the distance a crab would typically fall during 
discard from recreational and commercial vessels (“low” 
and “medium” distances, respectively). The “high” dis¬ 
tances were approximations of the maximum distance 
that a crab would be thrown from a pier or dock during 
shoreside recreational fishing at low tide (8 m), and a 
height similar to the distance from the rail of the New¬ 
port Pier to the water at mean lower low water (6 m). 
For the first experiment, we attempted to drop half of 
the crab from each height such that they would land 
dorsally and the other half ventrally. For the second 
experiment, the side on which the crab landed was not 
forced (merely noted). After 3 crab were dropped, they 
were removed from the sea water tank and placed in 
an ice chest filled with sea water and burlap. For each 
height treatment, once all 20 crab were dropped, they 
were taken back to the holding tanks and placed in in¬ 
dividual compartments, and mortality was determined 
after 5 d. 
