otter board of the trawl net and also represents a deterrent for this kind of fishing gear, since the net might get entangled in it. Moreover, deep bot- tom gillnets and traps, sometimes employed in this area, might get entangled in the coral branch- es and act as “ghost fishing”. This kind of bio- cenosis is also present in other areas such as the Sicilian Channel and the South Adriatic and in other Mediterranean areas.
Deep-water shrimp fishing grounds are located around the biocenosis of white-corals as well as on the margin of submarine canyons in the western Mediterranean, areas which can act as recruiting grounds and reservoirs of mature specimens also for other deep-water species, such as hake and rockfish. These specimens escape trawl fisheries, reach maturi- ty and renew the exploited portion of the stock annu- ally, sustaining the recruitment under conditions of high fishing mortality on the immature age groups (the spawning refugia paradigm of the Mediterranean fisheries) (Caddy, 1993).
Studies of the effects of fishing on marine deep ecosystems in the Mediterranean Sea have mostly addressed the population structure and dynamics of target species, while little is known of the impact of fishing on benthic communities, non-target species and biodiversity. Discard studies indicate that during the deep-water trawling in the Ionian Sea the discarded catch represented an large frac- tion of the total catch (20-50%). This is almost exclusively due to unwanted fish species, while discards of the target species and other commercial ones are negligible. Discard rates seem to increase with the total catch and depth. The unwanted species are mostly represented by Galeus melasto- mus, Hoplostethus mediterraneus and the species of the Macrouridae family (Trachyrhynchus scabrus) (D’Onghia et al., 2003a).
Fish assemblages and ecology
A number of authors have described the compo- sition of the community supporting the fishery of the deep-sea rose shrimp, Aristeus antennatus, in the western Mediterranean Sea separately for crus- taceans and for fish (Abelló et al., 1988; Stefanescu et al., 1992; Cartes and Sardà, 1993; Stefanescu et al., 1992, 1993 and 1994). This community, dwelling over muddy bottoms on the middle slope, is composed principally of the target species, Aris- teus antennatus, along with other species of no commercial interest, e.g. Geryon longipes, Poly-
cheles typhlops, Lepidion lepidion, Alepochephalus rostratus and Trachyrhynchus scabrus. A. antenna- tus is an interesting species as compared to the other species dwelling in the community because of certain specific biological characteristics, namely: (1) its broad depth distribution, making it a highly eurybathic species, and (2) though fishing pressure has been extremely high over the past 40 years, the population seems to be in a healthy state of exploitation. Cartes and Sardà (1993) defined three main zonations for the deep-sea decapod fauna in the western Mediterranean: the upper middle slope above 670 m; the lower middle slope between 850 and 1200 m; and below this last-mentioned depth, a transition zone to the lower slope community (down to 2000 m). The above-mentioned bound- aries represent genuine barriers to distribution for several decapod crustacean and fish species, but not for A. antennatus, which enjoys a continuous distri- bution between 550 and to at least 3000 m (Sardà et al., 1993; Sardà, 2001). However, we must consid- er here that the definition of boundary is a contro- versial question, often depending on the sampling adequacy and the data employed in the analysis. Koslow (1993), Haedrich and Merrett (1990), Ste- fanescu et al. (1993), Cartes and Sardà (1993) and Moranta et al. (1998) provided different results investigating in the same areas, but only fishes are considered in these studies. In this paper we present clusters including crustaceans and fishes, reaching similar results as Morales-Nin et al. (2003) with a first boundary around 800 m depth .
The community boundary at around 900 m described here is mainly the result of the upper limit to the depth distribution range for such species as Alepocephalus rostratus, Lepidion lepidion, Nezu- mia aequalis, Acanthephyra eximia and Geryon longipes, species with high abundance and biomass levels. At the present time no technical constraints prevent fishing operations at deeper depths, yet fish- ermen seem to be aware that there is a community boundary at that level and thus do not operate at deeper depths, in the knowledge that yields of A. antennatus there will be insufficient.
This mobility pattern indicates the ability of deep-water shrimp to adapt their life cycle to the structure of canyons on the Catalan Coast (Sardà et al., 2003a,b; 2004). Tursi et al. (1996) found similar population structures in the Gulf of Taranto, where submarine canyons are also an important feature of the geomorphological structure. The most active part, regarding particle sedimentation fluxes, are the
MEDITERRANEAN DEEP-SEA BIOLOGY 29