Abstract .—We studied phenotypic 
variation in larval and juvenile growth 
and development, using laboratory- 
reared winter flounder, Pleuronectes 
americanus. Larvae were reared indi- 
vidually to metamorphosis and beyond 
and were measured at weekly intervals. 
Growth in length was rapid until 30 d 
but slowed thereafter until metamor- 
phosis. Standard length peaked and 
often declined as metamorphosis ap- 
proached, and notochord length de- 
creased during flexion. Length at 30 d 
(an index of larval growth rate) was 
inversely related to age at metamorpho- 
sis, confirming previous assertions that 
larvae that grow rapidly also develop 
most rapidly. The relation between 
growth rate and larval-period duration, 
however, was not straightforward. The 
time from the day of peak larval length 
until metamorphosis (7-35 d) appeared 
to be inversely related to larval growth 
rate. Juvenile growth rates during the 
first 3 weeks following metamorphosis 
were unrelated to length at 30 d. Addi- 
tional juveniles, reared in groups as 
larvae and tracked as individuals fol- 
lowing metamorphosis, showed no 
change in growth rates during the first 
4 weeks of the juvenile period in rela- 
tion to increasing age at metamorpho- 
sis or larval growth rates. These results 
are consistent with earlier findings that 
size at age does not diverge continually 
throughout the larval and juvenile pe- 
riods. Compensatory juvenile growth 
among fish that grew slowly as larvae 
was observed but not to the same ex- 
tent as previously reported. We empha- 
size the utility of the individual-based 
approach for identifying patterns of 
phenotypic variability in growth and 
development during the early life 
stages in fishes. 
Manuscript accepted 13 August 1996. 
Fishery Bulletin 95:1-10 (1997). 
Individual variation in growth and 
development during the early life 
stages of winter flounder, 
Pleuronectes americanus 
Douglas F. Bertram* * 
Thomas J. Miller * 
William C. Leggett*** 
Department of Biology, McGill University 
I 205 Avenue Docteur Penfield, 
Montreal, QC, Canada H3A 1B1 
Mechanisms controlling survival 
and recruitment of fishes operate at 
the level of the individual (Crowder 
et al., 1992). Further, small initial 
differences among individual larvae 
and juveniles within fish popula- 
tions may have disproportionate 
effects on the probability of their 
survival (Crowder et al., 1992; Rice 
et al., 1993). Consequently, research 
programs in fisheries have begun to 
focus on phenotypic variability 
within cohorts in an effort to iden- 
tify particular traits that may be 
unique to the small minority of sur- 
vivors (Fritz et al., 1990; Taggart 
and Frank, 1990). If survivors are 
not random subsets of the original 
cohort, interpretations of recruit- 
ment processes based upon analy- 
ses of population averages are likely 
to be misleading (Pepin and Miller, 
1993). Consequently, individual- 
based approaches are increasingly 
favored (Crowder et al., 1992). How- 
ever, there have been few quantita- 
tive measurements of either indi- 
vidual variation in early life history 
traits of fishes or in their survival 
consequences (but see Rosenberg 
and Haugen, 1982; Rice et al., 1987; 
Chambers et al., 1989; Chambers 
and Leggett, 1992; D’Amours, 1992; 
Bertram and Leggett, 1994; Loch- 
mann et al., 1995; Miller et al., 
1995). In theory, longitudinal data 
can be obtained from sequential 
measurements of individuals or 
from back calculations of size at age 
from otolith microstructure. 
Variation in larval growth rates 
is widely believed to be a central 
feature in year-class formation in 
fishes (Leggett and Deblois, 1994). 
Traditionally, growth parameters 
are estimated from a restricted 
number of samples of the popula- 
tion. Each sample includes a range 
of fish lengths and ages. Impor- 
tantly, each fish provides only a 
single estimate of length at age. 
Such data are termed cross-sec- 
tioned. The calculated growth pa- 
rameters represent composite pic- 
tures and cannot reveal variability 
among the growth patterns of indi- 
viduals simply because they aggre- 
gate data at a level higher than the 
individual. Chambers and Miller 
(1995) have discussed the effects of 
the level of aggregation of data on 
the inferences that can be made. In 
addition, composite growth curves 
Present address: Department of Biologi- 
cal Sciences, Simon Fraser University, 
Burnaby, British Columbia, Canada 156 
V5A. E-mail address: dbertram@sfu.ca 
* Present address: The University of Mary- 
land System, Chesapeake Biological 
Laboratory, P.O. Box 38, Solomons, Mary- 
land 20688-0038. 
Present address: Queen’s University, 206 
Richardson Hall, Kingston, Ontario, 
Canada K7L 3N6. 
