2 
Fishery Bulletin 95(1 ), 1997 
are subject to several potential biases (Chambers and 
Miller, 1995). For example, composite curves are not 
accurate in cases where mortality of individual age 
classes are biased towards small or large individu- 
als (Litvak and Leggett, 1992; Pepin et al., 1992). 
Hence, if the survival consequences of variability in 
growth rates are to be evaluated adequately, indi- 
vidual phenotypic variability in growth must be 
quantified (Lynch and Arnold, 1988; Chambers, 
1993). This requires that longitudinal data based 
upon repeated measures of individuals be collected. 
These problems are illustrated in the following ex- 
ample. Consider a cohort of 220 larvae with an aver- 
age size of 5.5 mm and a uniform size distribution of 
20 larvae in each of ten 0.1-mm size classes from 5 
to 6 mm. In a hypothetical 7-d interval, no larval 
growth occurs, but a gape-limited predator consumes 
all larvae less than 5.5 mm, leaving 120 larvae with 
an average size of 5.75 mm. If only cross-sectional 
data were available, the larval growth rate was esti- 
mated as 0.11 mm/d for the 7-d interval. However, if 
longitudinal data were available (i.e. measurements 
of survivors at the beginning and end of the week), it 
would be clear that no growth had occurred. 
Bertram et al. ( 1993) have argued that the dynam- 
ics of larval and juvenile growth rates should be ex- 
amined in unison, rather than separately. Using labo- 
ratory-reared winter flounder, Pleuronectes ameri- 
canus, they have shown that size-at-age trajectories 
do not diverge continually during the larval and ju- 
venile periods. In fact, juvenile size-at-age trajecto- 
ries converge because fish that grew slowly as lar- 
vae compensated for their slow growth by growing 
rapidly as juveniles. However, Bertram et al. (1993) 
assumed that larval growth was linear; therefore 
juvenile fish used in their experiments were pooled 
into groups. This approach precluded the study of 
individual phenotypic variability. The objectives of 
the present study were 1 ) to provide estimates of the 
individual variability in growth rate in fish during 
early life stages because it is upon this individual 
variability that phenotypic selection acts and 2) to 
evaluate the validity of previous estimates of larval 
growth rate. Also, we explore how individual varia- 
tion in larval growth affects growth during the sub- 
sequent juvenile period. 
Methods 
Rearing protocoS 
The research in this study was conducted at the 
Huntsman Marine Science Centre, St. Andrews, New 
Brunswick, during summer 1991. In May, adult win- 
ter flounder were collected from Passamaquoddy Bay 
with a bottom trawl and held at ambient seawater 
temperature (7-8°C). When ripe, eggs from individual 
females were combined with sperm pooled from three 
males to create half-sibling families. Families were 
maintained separately throughout the study. Fertil- 
ized eggs were placed in a slurry of diatomateous 
earth for 12 h following fertilization to prevent clump- 
ing. Incubation temperature was 7 (±0.5)°C (mean ± 
SD). At approximately 24-h after fertilization, the 
eggs were immersed in solutions of penicillin (0.0158 
g/L) and streptomycin (0.02 g/L) for 24 h. Filtered, 
UV-sterilized seawater was replaced every 2-3 d until 
hatching commenced at approximately 14 d after fer- 
tilization. 
Upon hatching, 118 larvae from two families (fami- 
lies 1 and 2) were individually stocked into black 
cylindrical 0.4-L containers (15 cm diameter x 6.5 
cm high) with clear plexiglass bottoms. Water tem- 
perature was maintained at 10 (±0.5)°C in a tem- 
perature controlled room with a 16:8 day:night pho- 
toperiod. At weekly intervals, 75% of the water was 
removed from each container and replaced with UV- 
sterilized filtered seawater. Additional “reserve” lar- 
vae from the same families were reared in groups 
under identical conditions in 18-L cylindrical, black 
plastic containers. Individual larvae that died within 
the first 3 weeks were replaced with siblings from 
the appropriate reserve group. 
Larvae were also reared in groups in 38-L aquaria 
covered externally with black plastic. Five aquaria 
were each stocked with four-hundred 1-d-old larvae 
drawn from another half-sibling family (family 3). 
Temperatures in the aquaria were maintained at 10 
(±0.5)°C. At weekly intervals, 3 liters of water were 
removed from each aquarium and replaced with UV- 
sterilized filtered seawater. Dead larvae were si- 
phoned regularly from the tank. 
All larvae were fed Brachionus sp. at 2 /(mL-d) until 
the end of week 7. Rotifers were cultured by using 
Isochrysis sp. and Chaetoceros sp. Twenty-four hours 
prior to being fed to larvae, rotifers were provided 
with Microfeast artificial plankton (Provesta Corpo- 
ration) to enhance their nutritional quality. From the 
end of week 5 onwards, larvae were also offered 
Artemia nauplii (0.25/[mL-d]). Nauplii were enriched 
with Microfeast 24 h prior to use. 
At metamorphosis, larvae from family 3, which had 
been reared in groups, were individually stocked into 
0.4-L rearing containers (see above) to examine ju- 
venile growth. To standardize the developmental 
stage of individuals used in this study, we used only 
fish whose left eye had just crossed the midline on 
its migration to the right side of the body (stage H of 
Seikai et al., 1986). All fish that metamorphosed on 
