584 
Fishery Bulletin 99(4) 
9 mm, and 17 mm are consistent with sizes at age 0, 1, 
and 2, respectively, within Bristol Bay. This may be due, 
in part, to reduced molt frequency in Bristol Bay: the da- 
ta obtained from Weber (1967) indicate that mid-winter 
molting was not uncommon during his study, whereas our 
data suggest that this may not be typical in Bristol Bay. It 
is difficult to determine whether the discrepancies repre- 
sent actual regional differences, or simply differences be- 
tween different studies conducted at different sites and at 
different times, but our results indicate that it is probably 
inappropriate to apply the early juvenile growth rate ob- 
tained by Weber (1967) to the Bristol Bay stock; use of the 
faster growth rate likely results in lower estimates of age- 
at-recruitment than the population displays. 
Variable growth of older juveniles (age 3+) may further 
delay age-at-recruitment. This variability is evident when 
comparing growth of the 1976 year class to that of the 
1990 year class; age-at-recruitment was 1-2 years greater 
in the former, due to slow growth of prerecruits. In par- 
ticular, note that females of the 1976 year class, averaging 
83.5 mm CL in 1983, displayed a mean increase in cara- 
pace length during molting (molt increment [MI]) of ~9 
mm during the 1984 spring molt, whereas females of the 
1990 year class, also averaging 83.5 mm CL in 1996, ex- 
hibited a mean MI of ~14 mm during the 1997 spring molt. 
As a result, females from the 1976 year class required 
four years to grow from a mean CL = ~50 mm to a mean 
CL = ~92 mm CL and at ~8 years after settlement were 
still slightly smaller than the estimated size for full repro- 
ductive recruitment (Zheng et ah, 1995a, 1995b). Females 
from the 1990 year class were able to accomplish slightly 
more mean growth, from ~53 mm to ~97 mm CL, in only 
three years. 
The large difference in growth rate between the 1976 
and 1990 year classes may have been caused by water 
temperature differences during the two time periods. Molt 
schedules and growth rates can be strongly influenced 
by ambient temperature (Kurata, 1960, 1961; Nakanishi, 
1985), and considerable variability in size at maturity has 
been observed over the species’ geographic range in both 
males (Paul et al., 1991) and females (see review in Blau, 
1990; Otto et al., 1990). Though a number of factors may 
contribute to the observed variability, reduced growth as- 
sociated with colder bottom temperatures has been in- 
voked to explain the smaller size-at-maturity observed in 
the Norton Sound population as compared with other Ber- 
ing Sea stocks (Blau, 1990; Otto et al., 1990), and modeling 
suggests that regional and temporal variation in temper- 
ature can have broad effects on age-at-recruitment (Ste- 
vens, 1990; Stevens and Munk, 1991). June bottom tem- 
perature profiles in Bristol Bay suggest that the 1976 year 
class was subjected to lower temperatures than the 1990 
year class, primarily at early juvenile ages. Although a 
detailed analysis of temperature-dependent growth would 
require year-round temperature records, which are not 
available for this region, June temperatures in Bristol Bay 
may serve as a proxy for thermal conditions throughout 
the year. Bottom temperatures in Bristol Bay are linked to 
seasonal sea ice, that in some years covers much of Bristol 
Bay (NIC, 1994; Wyllie-Echeverria, 1995; Neibauer, 1998), 
and the development of sea ice can have strong effects on 
bottom temperature conditions throughout the year. “Cold 
pool” bottom waters (<1°C) produced in the winter during 
ice formation may persist well into the summer, and po- 
tentially into the following winter, once insulated from 
surface heating by the development of the summer ther- 
mocline (Azumaya and Ohtani, 1995). 
In many Crustacea, temperature primarily affects the 
molt schedule and has little influence on the magnitude of 
the MI (Hartnoll, 1982; Wainright and Armstrong, 1993), 
but laboratory studies conducted with red king crab indi- 
cate substantial variability in Ml-at-age, across ranges of 
temperatures, as well as under stable environmental con- 
ditions. Rearing crabs ~6.3 mm CL under constant tem- 
peratures of ~10°C, Molyneaux and Shirley (1988) report- 
ed changes in CL at molt that ranged from -4.4% to 52.2%; 
similarly, for juvenile premolt crabs 33-36 mm CL, reared 
at ~5.0°C, Gharrett (1986) observed Mis ranging from 3 
to 8 mm. At reproductive age, Weber and Miyahara ( 1962) 
observed that MI varied between 5 and 23 mm CL per 
molt in adult males, and large variability in MI associated 
with water temperatures between 0° and 12°C has been 
demonstrated for ovigerous females (Shirley et al., 1990). 
Because the changes in mean CL the we observed for the 
Bristol Bay stock were determined through modal analysis 
of the entire population, it is reasonable to suspect that ap- 
parent differences in growth between years and cohorts do 
not represent MI variability but may be explained as vari- 
ability in the number of molted versus unmolted crab with- 
in particular survey years. This is reasonable to assume, 
considering that red king crab may skip molting so that the 
annual molt schedule is replaced by a biennial or triennial 
cycle (Weber and Miyahara, 1962; McCaughran and Pow- 
ell, 1977; Balsiger, 1974). However, closer examination of 
the trawl survey data indicates that, of the 23 size modes of 
crab observed, none comprised less than 94% new-shelled 
crabs that had recently molted (senior author, unpubl. da- 
ta). A high proportion of newly molted crabs was character- 
istic of nearly all the identifiable size modes (senior author, 
unpubl. data): of the 53 modes identified, 35 comprised en- 
tirely new-shelled crabs, 17 comprised 94-99% new-shelled 
crabs, and only one comprised >10% old-shell individuals 
(the female cohort with mean CL=~94mm CL in 1982; pop- 
ulation^ 1.4% old-shell). Thus, the difference in growth 
rates observed between the two year classes of late juve- 
niles cannot be explained by variations in molt frequency; 
the trawl survey data support the conclusion that variabil- 
ity in MI is a characteristic shared by both sexes across a 
range of ages. 
Substantial variability in MI is an important life his- 
tory characteristic that confounds attempts to assign dis- 
crete size-at-age categories to a population. The greater 
the variability in MI among individuals and over time, the 
greater will be the range of sizes associated with crabs in 
a given year class, making modes more difficult to resolve 
from one another. Size-at-age values presently used for 
Bristol Bay red king crabs were derived from studies 
consisting of 1-4 years of data (Weber and Miyahara, 
1962; Weber, 1967; McCaughran and Powell, 1977; Incze 
et al., 1986), but our analyses show that a strong tenden- 
