BULLETINOF MARINESCIENCE.55(2-3): 418-433. 1994 INFLUENCE OF ARTIFICIAL REEFS ON COASTAL NEKTON ASSEMBLAGES OF THE GULF OF CASTELLAMMARE (NORTHWEST SICILY) C. D'Anna, F. Badalamenti, M. Cristina and C. Pipitone ABSTRACT An artificial reef, of 32 pyramids of concrete block, was built in the coastal watcrs of the Gulf of Castellammare (NW Sicily) during 1986-]989. Our research objective was to com- pare the catches and nekton assemblage structure of this reef with those of surrounding natural habitats. Surveys were carried out in 1990-1991 at three different sites: the artificial reef area (ARA), a natural rocky area (NRA) and a control area on a sandy bottom (CA). A trammel net was used to obtain samples at each site, and a qualitative visual assessment of the fish assemblage of ARA was also carried out. Quantitative analysis showed that ARA had significantly more species and a greater number of specimens than CA but no significant difference in catches was recorded, while a general similarity in the nekton assemblage was found comparing ARA and NRA. Overall, ARA displays features intermediate between the two natural sites tested, and the variability of its assemblage could be attributed to the avail- ability of shelter and to the benthic settlement on the concrete boulders. This study is an attempt to understand the role played by artificial reefs in influencing the composition of nekton assemblages in oligotrophic waters, such as those in the southern Tyrrhenian Sea. The deployment of artificial reefs as a tool for increasing biological resources has grown widely in the last two decades (as reviewed by Bohnsack and Suth- erland, 1985; Doumenge, 1990; Bombace, 1989). In Sicily artificial reefs are being constructed to assist in the development and enhancement of small-scale fisheries and to obstruct illegal inshore trawling. Many scientific and economical features still need to be investigated (Riggio, 1990; Riggio et aI., 1990); for example the role played by artificial reefs in waters characterized by low productivity (Spanier et aI., 1985, 1990; Arculeo et aI., 1990) and by the presence of extensive natural rocky bottoms (Chang, 1985). There are conflicting reports concerning the influence of artificial reefs on near- by natural bottom fish communities. Matthews (1985) and Bohnsack (1989) main- tain that the deployment of an artificial reef close to natural reefs could have detrimental effects on the structure of the populations living on nearby bottoms; Stone et al. (1979) however, did not observe any negative influence. A research program based on experimental fishing surveys was started in March 1990, to further investigate the potential utility of artificial reefs. In this paper we report quantitative and qualitative data which compare the structure of the fish community living in the artificial reef area with those of the surrounding natural habitats. MATERIALS AND METHODS Area. Reef Deployment and Sites.-The Gulf of Castellammare (38°03'N, 12°54'E), located in NW Sicily (Fig. I), with a perimeter of circa 70 km is the widest creek of the Sicilian coast. Three different areas were chosen for the survey (Fig. I). ARTIFICIALREEF AREA(ARA). This includes the artificial reef located off the town of Aleamo Marina on a sand-mud bottom at a depth of 16-20 m. Its construction was started in July 1986 and was completed in June 1989. During this period a total of 448 concrete blocks were assembled to form 32 pyramids (each having 14 elements) occupying a quadrangular area of circa 30 ha. Each block weighs 13 tons, measures 8 m3 (2 X 2 X 2 m) and has holes and cavities of different shapes 418 D'ANNAETAL.:ARTIFICIALREEFSINNWSICILY 4]9 13 00' E GULF OF CASTEu..AMMARE N I 3800'N -=-2.5 Km Figure ]. Map showing steady areas (NRA, CA and ARA) and sampling sites (A, B, C, CA, MF, MT and L). and sizes (Bombace, 1989). The total volume of the artificial reef is 3,584 m3 and the average density in the reef area is circa 120 m3. ha I. Three different sampling sites, named L, MF and MT (Fig. I), were located at a mean distance of 70 m from the outer artificial pyramids, falling within the "influence area of the reef" sensu Bohnsack and Sutherland (1985). CONTROLAREA(CA). This zone encompasses an area of 60 ha and includes a single site (Fig. ]) situated about 1.3 km west of ARA and about 3 km east of the nearest natural rocky bottom. CA (depth 18 m) is characterized by environmental conditions similar to those in ARA. NATURALROCKYAREA (NRA). The natural reef was selected from an area of sublittoral hard substrate that slopes down towards large irregular rocky boulders, scattered on Posidonia oceanica beds and sandy bottoms, at a depth of approximately 25 m. NRA has an area of 200 ha. The three sampling sites, A, Band C (Fig. ]) chosen in NRA are about 4 km and 5 km distant from CA and ARA, respectively. Data Collections.-A total of 119 trammel net collections were made between March ]990 and February ]99], with an average number of 17 samples from each site. The net had the following characteristics: length 500 m; height 3 m; inner mesh sizc 54 mm; outer mesh size 320 mm. Each sample lasted from sunset to sunrise with an average fishing time of 12 h. Twenty-four diurnal underwater observations were carried out in ARA; poor underwater visibility (generally not exceeding 3 m) prevented SCUBAdivers from carrying out an accurate quantitative visual census, These observations allowed us to get a better picture of the diurnal composition of the nekton assemblage closely associated with the concrete boulders and thus not accessible to the net. The results were not included in the analysis of catch data. All species caught wcre classified as pelagic (P), nekto-benthic (NB) or benthic (B) species. The taxonomic list of fishes (Appendix I) was made according to Whitehead et al. (1984/1986). The number of species and the individual size and weight for each specimen were recorded in each sample. Statistical analysis and indices computations were made, taking into account either the total fish assemblage or just the benthic and nekto-benthic species. Species richness was expressed by considering the number of species (S) and by using the Margalef index of species richness (d') (Margalef, 1958). The diversity and homogeneity of the species in each site and for each area were calculated by referring to the Shannon-Weaver diversity index (H' = lNi/N·lnNi/N) and to an expression of evenness (1' = H'IH'max) (Pielou, 1966). 420 BULLETIN OF MARINE SCIENCE. VOL. 55. NO. 2-3. 1994 Table I. Community structure indices; S = No. of species; d' = species richness; H' = diversity index; NRA = Natural Rocky Area; ARA = Artificial Reef Area; CA = Control Area; A, B, C = Sites inside NRA; L, MF, MT = Site inside ARA a) Benthic and Nekto-benthic Community b) Global Community Sites Samples (areas.) (No.) S d' II' l' S d' II' J' A 17 46 7,32 2,59 0,68 54 8,55 2,73 0,68 B 18 40 6,86 2,86 0,78 49 8,34 3,03 0,78 C 18 57 9,88 3,26 0,81 63 10,87 3,34 0,81 (NRA) 53 67 9,48 3,00 0,71 78 ]0,98 3,14 0,72 L 16 39 6,36 2,71 0,74 48 7,40 2,72 0,70 MF 16 30 4,79 2,39 0,70 39 5,90 2,52 0,69 MT 17 32 5,32 2,51 0,72 41 5,85 1,80 0,49 (ARA) 49 47 6,52 2,68 0,70 60 7,69 2,42 0,59 (CA) 17 28 4,97 2,57 0,77 36 5,62 2,15 0,60 The Kruskal-Wallis test (Siegel, 1956) was used to detect significant differences in numerical abun- dance, number of species and fishing yields among sites. Similarity among sampling sites and among species was calculated only for the benthic and nekto- benthic categories. A multivariate ordination technique using the Simple Matching Index (Legendre and Legendre, 1984) was employed; the dendrogram was obtained using the weighted average linkage as the agglomerative algorithm (Pielou, 1984). A two-way coincidence table (Kikkawa, 1968; Smith et aI., 1988) was drawn up to display the percentage of the numerical abundance of each species within each site. RESULTS Species Composition and Diversity.-A total of 3,760 specimens belonging to 47 families and 93 species was caught during the survey, as shown in Appendix 1, which also illustrates the ecological categories (benthic, nekto-benthic or pe- lagic) to which species belong. Benthic species were 41.3% or the overall catch, nekto-benthic species were 42.4% and pelagic species were 16.3%. Considering only the benthic and nekto-benthic assemblage, the highest specific richness was displayed by NRA, followed by ARA and then CA; the maximum and minimum values were obtained in C and CA, respectively (Table 1). The Kruskal-Wallis test performed on the number of benthic and nekto-benthic species showed highly significant differences among the sites of the three areas, between ARA and NRA, and between ARA and CA (Table 2). The maximum value of diversity (H') was obtained in NRA followed by ARA and then CA, whereas the highest value of J' was displayed by CA, followed by NRA and then ARA (Table I). Sparids represented the highest percentage of species in all the areas: 18% in NRA, and 21% both in ARA and CA. Other highly represented families were labrids in NRA, scorpaenids in ARA and solcids in CA.