4 
Fishery Bulletin 95 ( 1 ), 1997 
lationship between juvenile growth rates and aver- 
age larval growth rate. For fish that had metamor- 
phosed early, measurements of size at age were avail- 
able until week 7 of the juvenile period. For these 
individuals we compared growth rates during weeks 
1-4 with those during weeks 5-7, using a paired £-test. 
Results 
Individually reared larvae 
Thirty-two individually reared fish, 31 of which were 
from a single family (family 1), survived until meta- 
morphosis. For 5 of these fish, weekly measures were 
available from hatching to metamorphosis. For the 
remaining 27 larvae weekly measures began at day 
22. We could not detect size-dependent mortality in 
the weeks following: hatching (£=1.83, df=116), week 
1 (£=0.18, df=87), week 2 (£=1.4, df=32), or week 3 
(£=0.21, df=ll). Data from the 32 fish provided esti- 
mates of individual larval growth trajectories (Fig. 
1A). The trajectories exhibited considerable pheno- 
typic variation in size at age, maximum larval size, 
size at metamorphosis, latency period, and the du- 
ration of the larval period (Fig. 1A). To demonstrate 
the loss of information introduced by basing growth 
parameters on cross-sectional data, we treated the 
original individual-level longitudinal data as cross- 
sectional. When individual larval sizes were depicted 
in this composite fashion (Fig. IB), mean larval 
length at age increased rapidly from day 1 until day 
30 and then leveled off. Important information, how- 
ever, can be obtained only from cross-sectional data. 
For example, coefficients of variation (CV ) for length 
at age increased from 0.07 on day 1 to 0. 12 on day 30 
but declined subsequently and leveled off at approxi- 
mately 0.08. 
The largest larvae at 30 d were also largest at 22 d 
(r=0.66, ti= 18, P=0.03) and at 36 d (r=0.5, n=18, 
P=0.034), indicating positive covariance in size at age 
for individually reared larvae that were alive at each 
of those ages. There was a significant positive rela- 
tionship between larval length at 30 d and maximum 
larval size (r=0.6, ti= 27, P=0.001). However, larval 
length at 30 d and age at metamorphosis were nega- 
tively correlated (n=27, r=-0.589, P=0.001; Fig. 2A). 
The negative correlation coefficient between size at 
30 d and age at metamorphosis was larger than corre- 
lation coefficients calculated for all other age classes. 
Average larval growth rate and length at 30 d were 
positively correlated (n=25, r=0.49, P=0.01; Fig. 2B). 
The age of maximum larval size (log-transformed) 
was negatively correlated to length at 30 d (r=-0.71, 
n= 27, P<0.001; Fig. 3A). The latency period (range:7- 
2 1 1 1 1 1 ; I 
0 10 20 30 40 50 60 
Figure I 
(A) Growth trajectories for individual laboratory- 
reared winter flounder, Pleuronectes americanus, 
larvae (n= 32) based upon spline functions fitted to 
repeated weekly measures of standard length. Lar- 
val standard length (mean ±SD) versus age. (B) The 
sample size and CV for each age class are given above 
and below the error bar (respectively). Also shown 
is the Gompertz curve based upon the equation re- 
ported by Jearld et al. (1993) for winter flounder. 
35 d; 14.4 ±7.5 d, 7i=31)(log-transformed) was in- 
versely related to age of maximum larval size (r= 
-0.58, re=31, P=0. 001; Fig. 3B). In contrast, the latency 
period showed a positive trend with increasing length 
at 30 d, but the relationship was not significant (r= 0.2, 
n=26, P=0.38). Age at metamorphosis ranged from 44 
d to 71 d (55.2 ±7.9 d). Length at metamorphosis ranged 
from 6.1 mm to 7.5 mm (6.6 ±0.3 mm). Length and age 
at metamorphosis were positively correlated for indi- 
vidually reared larvae (i'=0.46, n=29, P=0.012). 
Among individually reared larvae, subsequent ju- 
venile growth rate during the first 3 weeks of the 
juvenile stage bore no relation to age at metamor- 
