874 
Fishery Bulletin 96(4), 1998 
king crab, which are not retained by the fishery, were 
sampled aboard commercial vessels by occasionally 
collecting all individuals from one randomly chosen 
trap. Sampled crabs were first measured for cara- 
pace length, then examined for the presence of 
snailfish eggs and larvae, without regard to species, 
by removing their carapaces. Legal-size male crab 
were sampled from the commercial catches when 
they were landed at the processing plant. Catches 
were first separated into groups of live and dead 
crabs, then counted and subsampled by examining 
up to 100 individuals from each group for the pres- 
ence of snailfish eggs. Port samples were collected 
for crabs caught either in the Bering Sea (three land- 
ings) or the Aleutian Islands (six landings; Fig. 2), 
but at-sea samples were collected only in the Aleu- 
tian Islands. 
Thirteen golden king crabs found, during commer- 
cial sampling, to contain snailfish eggs or larvae were 
frozen whole, with the eggs or larvae left in place for 
later processing. Eighteen of 23 egg and larvae 
samples were nonhatched egg masses that were 
placed in water overnight to fully rehydrate, then 
weighed to the nearest 1 mg. Displacement volume 
of the egg masses (mL) was measured in a gradu- 
ated cylinder. Egg masses were then teased apart 
and the total number of eggs counted. Diameters of 
approximately 200 eggs randomly chosen from each 
egg mass were measured to the nearest 0.03 mm with 
an optical micrometer. Volumes of the gills on both 
the affected and unaffected side of the crab were 
measured by displacement after they were excised 
with a scalpel at their bases. Volume of the empty 
branchial chamber on one side of the crab was then 
measured by securing the carapace in its proper place 
on the crab and then injecting Polycel insulating foam 
into the branchial cavity through a small hole. After 
any excess was removed, the volumes of the foam 
casts were then measured by displacement to the 
nearest milliliter. The 23 egg and larvae samples 
were then subjected to restriction fragment length 
polymorphism (RFLP) analysis (Dowling et al., 1996) 
by using six restriction enzymes to determine spe- 
cies identity. 2 Species identification patterns were 
established by using the RFLP analysis on six 
Careproctus rastrinus, eight C. furcellus, and one 
C. cypselurus collected in crab pots during commer- 
cial sampling. 
The incidence (e.g. probability of occurrence) of 
snailfish eggs and larvae as a function of several pre- 
dictor variables was described with a generalized lin- 
ear model, with a logit link function and a binomial 
2 Lopez, A. 1996. Marine Molecular Biological Laboratory, Uni- 
versity of Washington, Seattle, WA. Personal commun. 
variance function (Venables and Ripley, 1994). For 
each crab, model inputs included a binary response 
variable (presence or absence of eggs or larvae). For 
the 1982 survey data, predictors included two con- 
tinuous variables (carapace length and depth) and a 
discrete variable (sex). For the 1996 port sampling 
data, predictors included two discrete variables (area 
and whether the crab was alive or dead). Model se- 
lection was accomplished iteratively by first testing 
the significance of each interaction term with a like- 
lihood ratio test, then by discarding the least signifi- 
cant term. This process was then repeated on each 
main effect and any associated interaction terms. 
Differences in the depth distributions of male and 
female crab and of spawning and nonspawning 
snailfish were tested by first calculating a Cramer- 
von Mises statistic from catch-per-tow and depth 
data, then by determining the significance of the sta- 
tistic with randomization (Syrjala, 1996). 
Association between the developmental stage of 
eggs carried by a female golden king crab and the 
developmental stage of snailfish eggs within a crab’s 
branchial chambers was tested with a Spearman’s 
rank order correlation coefficient. Because king crabs 
extrude eggs onto their pleopods soon after molting 
(Powell and Nickerson, 1965; Sloan, 1985), the stage 
of embryonic development is a crude measure of the 
relative time since molting. Positive correlation be- 
tween the developmental stages of crab and snailfish 
embryos would indicate that snailfish preferentially 
choose newly molted crabs as hosts. To simplify in- 
terpretation, 3 of 26 females were not used because 
they contained more than a single egg mass or lar- 
vae cluster. Embryonic developmental stages for 
snailfish (uneyed, eyed, and larval) and crabs 
(uneyed, eyed, and hatched) were considered as or- 
dered variables and were assigned numeric codes (i.e. 
1, 2, 3). Mature female crabs without any obvious 
egg remnants attached to the pleopods were grouped 
with females carrying hatched eggs because egg rem- 
nants may have been missed in the macroscopic ex- 
amination performed at-sea. 
The mortality rate of male crabs in the holding 
tanks of commercial vessels was estimated separately 
for crabs with and without snailfish eggs and larvae 
on the basis of 1996 port sampling data for commer- 
cial crab landings. For each landing, the total num- 
ber of live crabs that were infested was estimated by 
multiplying the number of live crabs landed by the 
infested proportion in the catch subsample. The to- 
tal number of infested dead crabs was estimated simi- 
larly. The mortality rate of infested and uninfested 
crabs was then estimated as the number dying di- 
vided by the total in each category. Average mortality 
rate was then estimated as the mean over all landings. 
