Ligas et al.: Modeling the growth of recruits of Merluccius merluccius in the northwestern Mediterranean Sea 
77 
Table 2 
Summary of the backward selection of the best model in the generalized additive model. The best model, with 
smoothers that are all significant and that has the lowest AIC value, is highlighted in bold. Explanatory vari- 
ables included mean bottom temperature (temp.); mean depth; recruit density (number of individuals per square 
kilometer); mean concentration of chlorophyll-a (chl-a); and age (days). Depth:density is the interaction; “y” and 
“n” indicate the explanatory variables included and excluded in GAMs. Measures of model fitness were deviance 
explained (Dev. expl.), coefficient of determination (r 2 ), generalized cross-validation score (GCV), and Akaike’s 
information criterion (AIC); ns=not significant: P>0.05; **=P<0.01; ***=P<0.001. 
Mean 
bottom Mean Recruit Depth: Dev. expl. 
temp. 
depth 
Chl-a 
density 
Age 
density 
(%) 
r 2 
GCV 
AIC 
y ns 
y ns 
y ns 
y** 
y*** 
y ns 
80.7 
0.795 
0.731 
692.4 
y ns 
yns 
yns 
y** 
y*** 
n 
80.7 
0.794 
0.734 
691.1 
n 
yns 
y ns 
y** 
y*** 
n 
80.6 
0.794 
0.737 
692.3 
n 
n 
yns 
y*** 
y*** 
n 
80.6 
0.795 
0.728 
690.1 
n 
n 
n 
y*** 
y*** 
n 
80.7 
0.795 
0.727 
689.3 
impaired. Bromley (1989) 
i found a negative 
ent study; also, see 
Belcari et al., 2006), Europeai 
relationship between density and growth in some ga- 
doid (Atlantic cod, whiting, and haddock) juveniles in 
the North Sea. 
The ecological factors that affect differences in 
growth between areas deserve further multiyear inves- 
tigations, for example, through sampling to examine 
the abundance of prey, such as macrozooplankton, in 
the main nursery areas (Ferraton et ah, 2007; Cartes 
et ah, 2009). Indirect evidence of the environmental 
quality of European hake nurseries in the Mediterra- 
nean Sea has been inferred from their high spatiotem- 
poral stability (Fiorentino et ah, 2003; Colloca et ah, 
2009). In the Ligurian Sea and Tyrrhenian Sea, after 2 
months spent in the pelagic environment (in the pres- 
larvae were transported by eddies and frontal systems 
to areas of relatively high planktonic production that 
resulted from upwelling (Abella et ah, 2008). These ar- 
eas are located along the shelf break (Colloca et ah, 
2004), where production is enhanced by upwelling and 
water turbulence, both of which increase the transport 
and nutrient input of organic matter into the water 
column (Pinazo et ah, 1996). The entire trophic chain, 
therefore, is increased, including a positive effect on the 
main prey of European hake recruits, in particular eu- 
phausiids and mysids, which reach their highest diur- 
nal abundance on the shelf break (Colloca et ah, 2004). 
The high spatial stability over time of the main 
hake nurseries in the Ligurian and Tyrrhenian Seas 
