Kupchik and Shaw: Age, growth, and recruitment of larval and early juvenile Micropogonias undulatus 
25 
40 
Age (dah) 
Figure 3 
Relationships of estimates of standard length (SL) versus 
age in days after hatching (dah) determined with a Laird- 
Gompertz growth model, for larval Atlantic croaker ( Micropo- 
gonias undulatus) collected from Bayou Tartellan, Louisiana, 
from October 2006 through March 2007 and from September 
2007 through March 2008. This model was forced through 
the intercept at the estimated 1.5-mm notochord length at 
hatching to accurately reflect length at hatching. Boxplots 
show the median, 25% and 75% quantiles, and 95% confi- 
dence intervals, and outliers are provided for each axis. The 
Laird-Gompertz model was parameterized with the following 
equation: SL = 1 . 5 .^. 6132720 ^-°™^ 
Length, age, and hatching dates determined from counts 
of otolith increments 
Sagittal otoliths were removed from 203 larvae and 
early juveniles of Atlantic croaker. Of those removed, 
13 otoliths were not readable and were excluded from 
analysis. The length frequency plot of all larval Atlan- 
tic croaker that were aged (A^= 190 ) followed a normal 
distribution (Shapiro-Wilk: P=0.43), with a mean of 8.3 
mm SL (SD 5.4) and range from 3.5 to 15.3 mm SL). 
However, the distribution of lengths was flatter in year 
1 (SD=2.5 mm SL) than in year 2 (SD=2.0 mm SL). 
All ages of larval Atlantic croaker, from both direct 
measurement and estimation, also followed a normal 
distribution (Shapiro-Wilk: P=0.11) with a mean age 
of 41 dph (SD 10.1), a median of 39 dah, and a range 
of 20-70 dah. 
In general, SL increased as the spawning and re- 
cruitment season progressed. Smaller larvae (<7 mm 
SL) were most prevalent in October for sampling year 1 
and in September and October for sampling year 2. For 
both sampling periods, the median SL of larvae sampled 
began to stabilize in late November and early Decem- 
ber and remained relatively constant through the 
spring. In both sampling years, the highest num- 
bers of hatching dates occurred between 16 Sep- 
tember and 31 October (Table 1). In year 1, cumu- 
latively, more than 95% of all calculated hatching 
dates had occurred before 15 January 2007, but, 
during year 2, the 95% cumulative distribution 
was not reached until early February 2008, after 
a small, secondary peak in late January. 
Modeled growth rates and dates of estuarine 
ingress 
There was a significant linear relationship be- 
tween measured lengths and estimated ages for 
larval Atlantic croaker (P<0.001; coefficient of de- 
termination [r 2 ]=0.76). The linear model indicated 
that the average growth rate was 0.20 mm/d, and 
it underestimated the hatching length (at 0.97 
mm NL). The Laird-Gompertz growth model for 
the entirety of the data set, forced through the 
1.5-mm-NL hatching length, provided a model fit 
that accounted for changes in growth rate due to 
sensory and organ development better than the 
fit of the linear model, which had nonstatistically 
independent and nonhomoscedastic errors that 
indicated a nonlinear relationship of growth rates 
that were variable through time. However, it only 
partially accounted for a slower initial growth 
rate at ages less than 20 dah (Fig. 3). 
The sum of squared residuals was significantly 
reduced when models were fitted by season and 
sampling year than when a model was fitted with 
data pooled from both sampling years (P=0.007). 
Length, age, and growth were significantly differ- 
ent for seasons within years (P<0.001). Larvae 
collected during the peak spawning and recruit- 
ment season from September through December of 
2006 and 2007 were shorter and younger than those 
collected from January through March of 2007 and 
2008 (Fig. 4A; Chambers et ah, 1983). Initial growth 
rates in the fall (September-December) of 2006 and 
2007 were lower than those in the spring (January- 
March) of 2007 and 2008, but they steadily increased 
and did not level off at older ages. In contrast, larvae 
collected in the spring of 2007 and 2008 had higher 
initial growth than those collected in the previous fall 
seasons, but rates in both spring seasons leveled off 
quickly and resulted in shorter fish at ages greater 
than 50 dah and 60 dah, respectively (Fig. 4B). Com- 
pared with patterns observed in year 1, the difference 
in patterns of larval growth rates in fall compared with 
those in spring was more pronounced in year 2, when 
there were generally warmer water temperatures and 
higher salinities. 
The ages and magnitudes of maximum growth rates 
for larval Atlantic croaker differed significantly within 
sampling year by season in the Laird-Gompertz models 
(P<0.001). Overall, the Laird-Gompertz models fitted 
by year and season showed that the maximum growth 
