INTRODUCTION
The musky octopus, Eledone moschata (Lamarck, 1798) is a species inhabiting the whole Mediterranean Sea and the Gulf of Cadiz in the eastern Atlantic (Roper et al., Reference Roper, Sweeney and Nauen1984). It is widely distributed on the continental shelf of the northern and central Adriatic, but it is most abundant at depths down to 80 m (Krstulović Šifner et al., Reference Krstulović Šifner, Dadić, Vrgoč and Kereković2004). Although the catches of the musky octopus in the Adriatic Sea are high and this species is economically very important for the Adriatic demersal fishery (FAO, 2003), studies on its biology and ecology are limited. Distribution and abundance of E. moschata in some Mediterranean areas was described (Casali et al., Reference Casali, Manfrin Piccinetti and Soro1998; Lefkaditou et al., Reference Lefkaditou, Siapatis and Papaconstantinou1998, Reference Lefkaditou, Leodarakis, Papaconstantinou and Tsangridis2001; Salman et al., Reference Salman, Katagan and Gucu2000; Belcari et al., Reference Belcari, Tserpes, Gonzalez, Lefkaditou, Marčeta, Piccinetti Manfrin and Souplet2002; Krstulović Šifner et al., Reference Krstulović Šifner, Dadić, Vrgoč and Kereković2004), as well as some aspects of the reproductive biology (Mangold-Wirz, Reference Mangold-Wirz1963; Mangold, Reference Mangold and Boyle1983a; Manfrin-Piccinetti & Rizzoli, Reference Manfrin-Piccinetti and Rizzoli1984; Soro & Piccinetti-Manfrin, Reference Soro and Piccinetti Manfrin1989; Ezzeddine-Najai, Reference Ezzeddine-Najai1997; Silva et al., Reference Silva, Ramos and Sobrino2004). However, the studies on feeding of the musky octopus are very scarce. The feeding of E. moschata in aquarium conditions was studied (Mangold-Wirz & Boucher-Rodoni, Reference Mangold-Wirz and Boucher-Rodoni1973; Boletzky, Reference Boletzky1975; Repetto et al., Reference Repetto, Wurtz, Palumbo, Minetti, Rebora, Costa and Cavassa1998; Şen, Reference Şen2007), but there are no published data on the diet of this species in the natural environment. The aim of this study is to describe, for the first time, the diet of E. moschata in the northern Adriatic Sea, including the qualitative and quantitative composition of food and food preferences. The paper also deals with the effects of the body size, sex, maturity stage and season on the diet and feeding intensity of the musky octopus.
MATERIALS AND METHODS
A total of 1226 individuals of the musky octopus with mantle length between 38 and 136 mm were examined. The monthly samples were taken in the period between October 2001 and June 2003, at depths between 35 and 50 m. All specimens were caught by commercial bottom-trawls along the western coast of the Istrian peninsula in the northern part of the Adriatic Sea (Figure 1).

Fig. 1. Map showing the area where Eledone moschata was sampled along the western coast of the Istrian peninsula.
Mantle length (ML) was measured to the nearest mm and each individual was weighed (±0.01 g) and sexed. Sexual maturity of each individual was defined by the state of the reproductive organs and sexual outputs, based on the maturity scale with three stages: (1) immature; (2) maturing; and (3) mature (Bertrand, Reference Bertrand1995).
Full and partially full stomachs were dissected and weighed to the nearest ±0.01 g. Stomach contents were sorted, examined microscopically and identified to the lowest possible taxon. Crustaceans were identified by diagnostic exoskeleton fragments, appendages and eyes. Bony fish were recognized from scales, vertebrae, spines, otoliths and other body fragments. Remains of cephalopods were recognized by beaks (Clarke, Reference Clarke1986), gladii, lenses, suckers and hooks. The following indices were used to quantify the diet (Hyslop, Reference Hyslop1980; Castro & Guerra, Reference Castro and Guerra1990; Collins et al., Reference Collins, De Grave, Lordan, Burnell and Rodhouse1994; Coelho et al., Reference Coelho, Domingues, Balguerias, Fernandez and Andrade1997):
percentage of occurrence (%F): the number of stomachs where a determined prey item occurred divided by the total number of stomachs examined, expressed as a percentage;
weight percentage (%W): the wet weight of a particular prey item (g) divided by the total wet weight of all prey (g), in the whole sample, expressed as a percentage;
percentage of number (%N): the number of a determined prey item in relation to the total number of prey, expressed as a percentage;
fullness weight index (FWI): the relationship between the weight of the stomach contents (g) and the total body weight (g) multiplied by 10,000;
emptiness index (EMI): the number of empty stomachs (ISF = 0) compared with the total number of stomachs, expressed as a percentage; and
index of stomach fullness (ISF): subjective method of describing the fullness of the stomach visually assessed and recorded on a scale of zero to four; zero being empty and four fully distended.
Numbers may overestimate the importance of small prey items taken in large numbers, while biomass emphasizes the contribution of single, heavy items (Hyslop, Reference Hyslop1980) therefore two additional feeding measures were used, to reach a more balanced assessment of the diet (Hureau, Reference Hureau1970; Zander, Reference Zander1982):
index of relative importance: IRI = (%N+%W) × %F; and
feeding coefficient: Q = %N × %W.
Using this coefficient prey was categorized as main (Q > 200), additional (Q = 20–200) and accidental (Q < 20).
In order to compare the diet of smaller and larger individuals, animals were divided in two categories: the first included specimens of ML ≤ 80 mm, and the second those of ML > 80 mm, while for seasonal comparison of diet categories ‘warm’, including spring–summer and ‘cold’ comprising autumn–winter data were considered. Comparisons of the diet in males and females, different length-classes, seasons, immature and mature individuals were performed by applying the Chi-square test (χ2) (Sokal & Rohlf, Reference Sokal and Rohlf1969) to the percentage of occurrence index of the prey groups (crustaceans, fish, cephalopods, polychaetes, gastropods and bivalves).
RESULTS
Qualitative and quantitative composition of stomach contents
Empty stomachs were found in 34.8% of individuals, whilst 8.5% of stomachs were fully distended. The major prey categories found in stomachs of the musky octopus with the highest percentage of occurrence (%F) were crustaceans (65.0%), fish (37.8%) and cephalopods (21.8%). The most abundant considering number (%N) were crustaceans (57.1%), followed by fish (28.0%), cephalopods (10.8%), gastropods (2.1%), polychaetes (1.3%) and bivalves (0.6%). Crustaceans were also most important considering weight (40.1%), followed by fish (31.5%) and cephalopods (23.7%). Polychaetes and gastropods were found with weight percentage 2.6% and 1.9%, respectively (Figure 2). The rest of the prey appeared in stomachs only sporadically, accounting for less than 0.5% in weight and number, and they were not considered as prey groups (species belonging to Pteropoda, Porifera, Foraminifera, Trematoda and Xantophyceae).

Fig. 2. Percentage of number (%N) and weight percentage (%W) of the different prey groups in the diet of Eledone moschata.
Based on values of the index of relative importance (IRI = 6323.9) and feeding coefficient (Q = 2292.6) crustaceans were the most important food category in the diet of the musky octopus (Table 1). Among these, decapods were the most abundant, accounting for 98.6% of all crustaceans. Fish were the second most important prey group (IRI = 2251.9; Q = 882.4). All identified fish were Osteichthyes. Among cephalopods (IRI = 754.8; Q = 257.1), species belonging to the families Octopodidae, Loliginidae and Sepiolidae were found. Body parts of E. moschata were identified in 1.46% of stomachs, in individuals with mantle lengths over 80 mm. Gastropods (Q = 3.9), polychaetes (Q = 2.6) and bivalves (Q = 0.2) were occasional prey in the examined stomachs. One prey type was present in 46.3% of the stomachs with food remains, 34.7% of stomachs contained two types of prey, 10.5% three, 6.3% four, and 2.1% five. In stomachs with two prey categories crustaceans and fish were most frequently found (in 56.4% of stomachs with remains of two prey categories).
Table 1. Percentage of occurrence (%F), percentage of number (%N), weight percentage (%W), feeding coefficient (Q) and index of relative importance (IRI) of three most important groups of prey in the diet of Eledone moschata.

Feeding in relation to body size, sex, maturity stage and season
Percentage of stomachs with food was not significantly higher in males (51.3%) than in females (48.7%) (χ2 = 0.552; P < 0.05) and no significant difference (χ2 = 0.360; P = 0.757; df = 5) was found between the qualitative composition of food in males and females (Table 2), so data for both sexes were pooled for further analysis. There was no significant difference in the diet of E. moschata, when the percentage of occurrence for different prey categories in cold and warm season was compared (χ2 = 1.24; 0.90 < P < 0.95; df = 5). When the diet of immature specimens was compared with maturing and mature individuals the significant difference was observed (χ2 = 26.22; P < 0.001; df = 5) (Table 2).
Table 2. Comparison between the diet of Eledone moschat a males and females, two different length-classes (ML ≤ 80 and ML > 80 mm), seasons, and maturity stages. The Chi-square test was applied to test the percentage of occurrence of different prey groups. Seasons: warm (spring–summer) and cold (autumn–winter); maturity stages: immature (1) and maturing/mature (2+3).

*, P < 0.05; **, P < 0.01; ***, P < 0.001.
The diet of E. moschata varied with the size of specimens (Figure 3). Comparison of food composition among different length-classes showed that significant differences existed in the diet of individuals with mantle lengths ≤80 mm and those >80 mm (χ2 = 15.69; P < 0.01; df = 5) (Table 2), so these two size-groups were analysed separately.

Fig. 3. Variation of food composition with size in the diet of Eledone moschata.
A total of 756 stomachs of musky octopus with ML ≤ 80 mm were examined, and 37.1% of them were empty. In 9.3% of stomachs with food the content was fully digested. The highest percentage, considering number, was represented by crustaceans (76.4%), followed by fish (12.4%) and cephalopods (7.1%). The highest percentage, considering weight, also was represented by crustaceans (69.0%) (Table 3; Figure 4). Crustaceans were found in 79.4% of the examined stomachs resulting in the main prey category in the feeding spectrum (Q = 5270.1; IRI = 11540.3). One prey category was present in 33.4% of examined stomachs, two in 31.3%, three in 22.6%, four in 9.7% and five in 3.2% of them.

Fig. 4. Percentage of number (%N) and weight percentage (%W) of the different prey groups in the diet of Eledone moschata sorted by size: (A) ML ≤80 mm; (B) ML > 80 mm.
Table 3. Percentage of occurrence (%F), percentage of number (%N), weight percentage (%W), feeding coefficient (Q) and index of relative importance (IRI) of the food categories for two size-groups of Eledone moschata.

Of a total of 470 stomachs of musky octopus with ML > 80 mm, 31.0% were empty. Fully digested contents were found in 9.5% of the stomachs with food remains. In the remaining 90.5% the most important prey category, by number, were crustaceans (45.6%), fish (35.5%) and cephalopods (15.0%). Considering weight, fish were the most important group (37.3%). They were followed by crustaceans (30.7%) and cephalopods (27.4%) (Table 3; Figure 4). Taking into account the index of relative importance (IRI) and feeding coefficient (Q), these three categories may be considered as the main food of the musky octopus in this size-group (Table 3). One prey type was present in 50.0% of the stomachs with food remains, 36.5% of stomachs contained two types of prey, 7.7% three, 3.8% four, and 1.9% five.
The fullness weight index (FWI) and emptiness index (EMI) varied depending on size and season (Table 4). The mean value of FWI was 36.95 for individuals of ML ≤ 80 mm and 39.94 for specimens of ML > 80 mm. The mean EMI values of smaller and larger individuals were 42.11 and 30.43, respectively. When considering the feeding pattern by season, the lowest mean value of FWI was recorded in summer (37.17) and the highest in spring (39.89). The highest individual value of FWI was registered in autumn (300.75; accounting for 3.0% of the total body weight). Share of individuals with empty stomachs (ISF = 0) ranged between 28.4% in spring and 41.6% in summer. The mean seasonal values of the stomach fullness index (ISF) showed the lowest number of full stomachs in summer, and the highest in winter and spring (Figure 5).

Fig. 5. Percentage of stomachs with different index of the stomach fullness (ISF) of Eledone moschata in different seasons.
Table 4. Values of the fullness weight index (FWI) and emptiness index (EMI) for smaller (ML ≤ 80 mm) and larger individuals (ML > 80 mm) and by season.

DISCUSSION
Several problems are associated with the analysis of cephalopod diet (e.g. Collins et al., Reference Collins, De Grave, Lordan, Burnell and Rodhouse1994). The main one encountered in the present case was the highly macerated status of prey items, and the scarcity of parts normally used for prey identification. The most reliable sources of information were preys' hard structures, so it is possible that the presence of the prey lacking hard parts was underestimated (Nixon, Reference Nixon and Boyle1987).
One of the characteristics of octopods (Hatanaka, Reference Hatanaka1979; Mangold, Reference Mangold and Boyle1983a; Cortez et al., Reference Cortez, Castro and Guerra1995) and cephalopods in general, is that they are opportunistic predators switching easily from one prey to another (Mangold, Reference Mangold and Stone1983b; Boucher-Rodoni et al., Reference Boucher-Rodoni, Bouchard-Camou, Mangold and Boyle1987; Collins et al., Reference Collins, De Grave, Lordan, Burnell and Rodhouse1994). Studies on feeding habits of the related species E. cirrhosa and O. vulgaris also are in favour of this hypothesis (Guerra, Reference Guerra1978; Sanchez, Reference Sanchez1981; Nixon & Budelmann, Reference Nixon and Budelmann1984). In the present study several prey items belonging to different food categories were found in the stomachs of E. moschata including crustaceans, fish, cephalopods, gastropods, polychaetes and bivalves. The ratio of food categories depended on the size of the individuals, both predator and prey, but also clearly on food availability, confirming an opportunistic feeding behaviour.
The observed differences in the food composition between smaller and larger individuals (with crustaceans, more abundant in the diet of smaller animals being substituted by fish and cephalopods in larger animals) are probably related to the onset of sexual maturity, 80 mm ML size being the length at first maturity observed for the species in the Adriatic Sea (Krstulović Šifner, Reference Krstulović Šifner2004). This was proved by testing the differences in food composition between immature and mature individuals, also highly significant. Quite reasonably, during the maturation process there are increased nutritional requirements and fish provide greater energetic profitability than crustaceans so their share is higher in larger specimens (e.g. Collins et al., Reference Collins, De Grave, Lordan, Burnell and Rodhouse1994).
Our observations indicated differences not only in food composition but also in the feeding behaviour between smaller and larger individuals. In smaller individuals (ML ≤ 80 mm) mainly hard parts of decapods were found in the stomachs, indicating the crustaceans were eaten entire. On the other hand, it was rare to find exoskeletons of crustaceans in stomachs of larger specimens (ML > 80 mm): these usually contained only highly macerated soft parts of the prey's body, suggesting that larger individuals drilled holes in the exoskeletons of their preys and sucked out the soft parts (or they just removed the flesh from the exoskeleton to eat it). It was previously reported that juvenile and adult E. moschata drilled the carapaces of crustaceans (Boletzky, Reference Boletzky1975). In aquarium bred E. moschata it was observed that adult animals fed by drilling holes in carapaces of crustaceans more often than juveniles, and in the related species E. cirrhosa the share of crustaceans consumed by hole boring in the exoskeletons increased with the size of the animals (Boyle & Knobloch, Reference Boyle and Knobloch1981). On the other hand, Şen (Reference Şen2007) observed no crab's exoskeleton drilling during his experiment in aquarium conditions with E. moschata: in that case exoskeletons were discharged and soft parts ingested. However, the occurrence of different attitudes is not surprising: data on the feeding behaviour of other octopods confirm that it changes depending on the size of animal and life conditions (Guerra, Reference Guerra1978; Boucher-Rodoni et al., 1983; Nixon, Reference Nixon and Boyle1987).
Based on present results, it seems that the cannibalism occurs only occasionally in this species, and it is probably related more to the availability of juveniles on some occasions (Lefkaditou et al., Reference Lefkaditou, Siapatis and Papaconstantinou1998; Krstulović Šifner, Reference Krstulović Šifner2004), than to the lack of alternative prey.
The seasonal changes in the FWI and EMI values observed are most probably related to the size of the animals and the life cycle phases, which change seasonally. As during spring a very high proportion of individuals are in the phase of the intensive maturation and reproduction, when food requirements are increased, this could explain the high values of FWI recorded. Nevertheless, the relatively low proportion of empty stomachs and the high values of fullness weight index recorded in all seasons indicate that the environmental conditions and the availability of food in different seasons do not vary significantly enough to affect the feeding rate of the species. The high percentage of empty stomachs and the low food weight index observed in summer may be a consequence of the less intensive feeding, but, to some extent, they also reflect the much faster food decomposition rate occurring in warm seasons. The rate of food decomposition depends on several factors, including predator's sex and age, and seawater temperature, higher temperature meaning faster decomposition (Boucher-Rodoni et al., Reference Boucher-Rodoni, Bouchard-Camou, Mangold and Boyle1987). Juveniles of octopods have a high feeding rate but they also digest food faster if compared to adults (Mangold, Reference Mangold and Boyle1983a) and during summer months juveniles are dominant in the catches of E. moschata in the Adriatic Sea (Krstulović Šifner, Reference Krstulović Šifner2004).
In future studies an additional effort should be done to investigate the role of the musky octopus in the trophic chain, as this species, being one of the most abundant demersal species in the area (Krstulović Šifner et al., Reference Krstulovic Šifner, Lefkaditou, Ungaro, Ceriola, Osmani, Kavadas and Vrgoč2005), affects significantly both predators and prey.
ACKNOWLEDGEMENTS
Our gratitude is expressed to all fishermen and technical crew who helped in collection of the biological data. Special thanks to Dr Olja Vidjak for help in determination of planktonic crustaceans. We are also very thankful to the anonymous referees for critical review of the manuscript. This study was supported by the Scientific Research Program of the Ministry of Science, Education and Sports, Republic of Croatia and by the DemMon project.