Ligas et at: Modeling the growth of recruits of Merluccius mer/uccius in the northwestern Mediterranean Sea 
75 
14 A 
12 
_ 10 
E 
u, 
_] 
H 8 
50 
o (no o© 
o o © oo © o 
©0u©00 OB© OO 
0©©©OCX3EO© (3D OO 
o ooannnx©© oo 
o tmnrrm ®) ons©©oO O 
o o GnmD® © o ©o o 
O 0©0©©0 © (© O 
oanxDO © oo o 
o o o o o ®® ©o o o 
o oooo © 
oocoo O O © 00 
o ©oo o 
o o © o 
100 
150 
Age (days) 
200 
14 
12 
10 
8 
6 4 
4 I 
B 
0 
0 1008 1378 
2003 3197 3571 6835 
Density (individuals krrr 2 ) 
8319 
Figure 4 
For the subsample of 271 European hake (Merluccius merluc- 
cius) recruits caught in June 2011 during the Mediterranean 
International [bottom] Trawl Survey (MEDITS) in the Ligurian 
Sea and northern Tyrrhenian Sea, (A) length-at-age data distri- 
bution and (B) a boxplot of length data conditional on density. 
part of the age range. However, in the older part of the 
age range, the smoothing function that describes the 
relationship between age and length flattened, indicat- 
ing a decrease in growth rate. The effect of density on 
the growth rate was positive up to 3000 individuals 
km -2 , and had a weak negative effect at higher densi- 
ties (Fig. 6). At the stations that had a recruit density 
higher than 3000 individuals km -2 , the length at age 
was on average 0.5 cm TL greater than the length at 
age observed at stations with lower densities. 
The value of k selected for the final analysis of the 
effect of density, by comparison of AIC scores, was 4. 
The AIC was higher when values of k >4 were used, 
and a “dome-shaped” effect caused by the lack of obser- 
vations of densities between 3700 and 6000 individuals 
km -2 was evident. With k set at a value of 4, the effect 
of density showed an increasing trend up to 
3700 individuals km -2 and then a rather con- 
stant pattern at higher densities, limiting the 
effect caused by fitting the model in the area 
where values of density were missing in the 
database. Further, the use of the variable den- 
sity as a factor (3-level factor) was tested. The 
results were similar to those obtained with 
the model that had a continuous variable of 
density: increasing growth rate with density, 
although with a slight decrease at very high 
densities. However, this model had a higher 
AIC and, therefore, was less suitable in its fit 
to the data than the model that used density 
as a continuous variable and a k value of 4. An 
inspection of the graphs for the model (Fig. 7) 
shows that there was no evidence of a pattern 
within the model residuals. 
Discussion 
Despite the ongoing debate about general is- 
sues of age estimation in European hake, read- 
ing daily growth rings on otoliths has proved 
to be a useful tool in the understanding of the 
first year of growth (Arneri and Morales-Nin, 
2000). The use of annual rings on otoliths led 
to the assumption that the European hake 
grows slowly (Drouineau et ah, 2010), but fur- 
ther studies on adult growth that have used 
tagging campaigns (de Pontual et ah, 2003; Pi- 
neiro et ah, 2007) and analysis of otolith daily 
increments in juveniles (Arneri and Morales- 
Nin, 2000; Morales-Nin and Moranta, 2004; 
Kacher and Amara, 2005, Belcari et ah, 2006; 
Pineiro et ah, 2008; Otxotorena et ah, 2010) 
have revealed that the growth rate is probably 
much faster than previously thought. 
Results from the present study are consis- 
tent with the main findings of recent studies 
on juvenile European hake growth in both the 
Atlantic and the Mediterranean Sea, which in- 
dicate a fast growth rate of about 0.6 mm day -1 during 
the first year of life (Morales-Nin and Moranta, 2004; 
Kacher and Amara, 2005; Belcari et ah, 2006; Otxo- 
torena et ah, 2010). Also, the results of back-calcula- 
tions of birth dates are in agreement with the available 
knowledge on spawning and recruitment of European 
hake in the western Mediterranean Sea (Maynou et ah, 
2003; Abella et ah, 2005; Belcari et ah, 2006; Recasens 
et ah, 2008). 
The present study is one of the first attempts to fit 
a recruit growth model for European hake with factors 
that could potentially affect growth dynamics. Through 
the application of a GAM on the length-age relation- 
ship, recruit density (number of individuals per square 
kilometer) was found to have a significant effect on the 
growth dynamics of European hake recruits; however. 
