70 
Fishery Bulletin 113(1) 
et al., 2007; Bartolino et al., 2011). This pattern results 
in high exploitation rates across the entire population, 
from small individuals (< 15 cm TL) to large females 
(>60 cm TL) in spawning aggregations (Aldebert et ah, 
1993) and highlights the importance of management 
measures to reduce the catch of both juveniles and 
spawning adults (Drouineau et al., 2010). 
The very high concentration of juvenile European 
hake along the coasts of Italy in the Ligurian and Tyr- 
rhenian Seas indicates that these areas are the main 
European hake nurseries in the northwestern Mediter- 
ranean basin (Orsi Relini et ah, 2002; Colloca et al., 
2009). Orsi Relini et al. (2002) estimated mean den- 
sities of European hake recruits to be up to 8 times 
higher than densities reported for other nurseries in 
the Mediterranean Sea. In the Tyrrhenian Sea, at the 
boundary between the continental shelf and the upper 
slope, densities >25,000 individuals km -2 have been ob- 
served (Colloca et ah, 2009). 
Despite many studies of the biology of European 
hake in the Mediterranean Sea, the growth pattern of 
juveniles is still controversial (Drouineau et ah, 2010). 
It is not clear whether the reported variability in the 
growth rate of European hake juveniles is due to meth- 
odological differences in the approaches used for age 
determination or due to natural variability in growth 
processes. 
Atmospheric processes and oceanographic features 
are known to have a key role in recruit condition and 
recruitment strength of other fishes. A link between en- 
vironmental variables (e.g., climate variability, temper- 
ature, and phytoplankton production) and recruitment 
of Atlantic cod ( Gadus morhua) in the North Sea and 
northeastern Atlantic has been described previously 
(Koster et al., 2005; Steingrund and Gaard, 2005; Stige 
et ah, 2006; Beggs et ah, 2014). Furthermore, increases 
in temperature in the North Sea have been reported 
to affect growth dynamics of haddock ( Melanogram - 
mus aeglefinus) (Baudron et al., 2011). Thermal con- 
ditions have been reported to affect recruitment and 
distribution of Pacific cod (Gadus macrocephalus ) in 
the eastern Barents Sea (Hurst et ah, 2012), of Pacific 
hake ( Mei'luccius productus ) along the western coast of 
North America (Agostini et al., 2008), and of Chilean 
hake (Merluccius gayi gayi) in the south Pacific (San 
Martin et ah, 2013). Within-year variability in growth 
of recruits of Argentine hake (Merluccius hubbsi) in re- 
sponse to environmental variables was found by Norbis 
et al. (1999) along Uruguayan coasts, and interannual 
variability in growth was found in Pacific hake (Wood- 
bury et al., 1995). 
Density-related growth relationships also have been 
described for a range of species: Bromley (1989) found 
a negative relationship between growth and fish densi- 
ty in both 1-year-old (I group) and 2-year-old (II group) 
Atlantic cod, whiting (Merlangius merlangus), and 
haddock in the North Sea: fish in areas of low density 
were larger than fish in areas of high density, possibly 
because of feeding competition in high density areas. 
Although environmental and oceanographic features 
are known to affect recruit condition and recruitment 
strength of European hake (Alvarez et al., 2001; May- 
nou et al., 2003; Olivar et ah, 2003; Abella et ah, 2008; 
Bartolino et ah, 2008a; Hidalgo et ah, 2008), under- 
standing of the effect of those factors on the growth 
dynamics of juvenile European hake is still limited. 
The aim of the present study was to model the 
growth of European hake juveniles to determine the 
effects of environmental variables (i.e., sea-surface 
temperature, bottom temperature, depth, scalar wind 
speed, chlorophyll-a) and population factors (i.e., fish 
density) using a generalized additive model (GAM) 
(Hastie and Tibshirani, 1990). 
Materials and methods 
Study area 
For this study, the growth pattern of juvenile Europe- 
an hake was analyzed in northwestern Italian waters, 
which include the Ligurian Sea and the northern Tyr- 
rhenian Sea (Fig. 1). The Tyrrhenian Sea is generally 
considered a distinct entity within the western Medi- 
terranean basin because it is semi-enclosed between 
the islands of Corsica, Sardinia, and Elba and the 
mainland (Italy) and is separated from the rest of the 
western basin by a channel of moderate depth, about 
1500 m (Orfila et al., 2005). Along the central western 
Italian coasts, the Tyrrhenian Current, also called the 
Eastern Corsica Current, flows northward through the 
Corsica Channel into the Ligurian Sea. The Corsica 
Channel is the passage between the islands of Corsica 
and Elba that connects the northern Tyrrhenian Sea to 
the southern Ligurian Sea. It plays a key role for water 
circulation in the northwestern Mediterranean Sea be- 
cause the water exchange that runs through it involves 
the whole water column (Gasparini et al., 1999). 
The general seasonal pattern of phytoplankton dy- 
namics is typical of subtropical areas, with a bloom 
period of maximum productivity from February to 
April and a period of minimum productivity in sum- 
mer months. The intensity of this winter-spring bloom 
varies significantly between years. In the Ligurian Sea, 
a substantial positive correlation links the intensity of 
the phytoplankton winter-spring bloom with a strong 
autumn-winter water turbulence (which is mainly 
driven by winds), and reduced wind mixing in March 
(Nezlin et ah, 2004). 
Trawl sampling and environmental data 
Specimens of juvenile European hake were collected 
from 13 of the 120 trawl stations sampled in June 
2011 during an experimental bottom trawl survey, the 
Mediterranean International Trawl Survey (MEDITS; 
see Bertrand et al., 2002, for technical specifications) in 
the Ligurian and northern Tyrrhenian Seas (Fig. 1). To 
