INTRODUCTION
Chelidonichthys lucerna (Linnaeus, 1758) occurs in the Mediterranean Sea, Black Sea and eastern Atlantic from Norway to Senegal. It is a nectobenthic fish mainly distributed on the soft bottoms of the continental shelf and is more abundant in coastal areas (Serena et al., Reference Serena, Voliani and Auteri1998). Commercial Triglidae catches have greatly diminished worldwide in recent years, dropping from about 38,000 tonnes in 2000 to 11,000 in 2004 (FAO, 2007), a negative trend that underscores the need for adopting suitable fishery management policies. Despite its importance for the Mediterranean fishery and the increasingly intense fishing pressure (FAO, 2007), information on its feeding habits are few and sparse (Atlantic Ocean: Nouvel, Reference Nouvel1950; Costa, Reference Costa1988; Mediterranean: Reys, Reference Reys1960; Faltas, Reference Faltas1996; Morte et al., Reference Morte, Redon and Sanz-Brau1997), particularly in Italian Seas (central Adriatic Sea: Froglia, Reference Froglia1976; Tyrrhenian Sea: Colloca et al., Reference Colloca, Ardizzone and Gravina1994).
With growth, the species changes its diet both in terms of prey size and type of prey, a fact that has also been attributed to its bathymetric migratory behaviour (Colloca et al., Reference Colloca, Ardizzone and Gravina1994; Morte et al., Reference Morte, Redon and Sanz-Brau1997). Smaller individuals feed upon benthic Crustacea, mainly Mysidacea (North Sea: Hostens & Mees, Reference Hostens and Mees1999), Amphipoda and Decapoda Natantia (Philocheras monacanthus). As they grow larger, C. lucerna increasingly prey on decapods (mainly Portunidae and Crangonidae) and teleosts (mainly Callionymidae) (Froglia, Reference Froglia1976; Colloca et al., Reference Colloca, Ardizzone and Gravina1994, Stagioni et al., Reference Stagioni, Mazzoni, Montanini and Vallisneri2007).
Determining fish diet and knowing trophic behaviour is necessary information for fishery resource management and for assessing fishing activity impact on the ecosystem. Acknowledging ecological interactions, such as predation, is essential for an ecosystem approach to fisheries (Bascompte et al., Reference Bascompte, Melian and Sala2005; Cury et al., Reference Cury, Shannon, Roux, Daskalov, Jarre, Moloney and Pauly2005). The inadequacy of traditional single-species models is amplified by the lack of knowledge as to long-term inter-species relationships. A ‘multi-species fisheries assessment model’ requires biological data on fish for a better understanding of how species relate to one another (Adriamed Project, 2001; Eldredge, Reference Eldredge and Eldredge2002; Stergiou & Karpouzi, Reference Stergiou and Karpouzi2002; DCR Medits Working Group, 2007).
The aim of this paper was to assess the diet of tub gurnard in the north-east Mediterranean Sea, introduced from 2006 in the list of ‘reference species from the Coordination of the International bottom trawl survey in the Mediterranean Sea (Medits)’ (Relini et al., Reference Relini, Carpentieri and Murenu2008), in order to increase the knowledge about the trophic biology of this scarcely studied species (comparing the diet among sex, size groups, depth and season).
MATERIALS AND METHODS
Sampling oceanographic surveys and samples processing
A total of 1096 specimens of tub gurnard were collected seasonally (winter and summer) during several international bottom trawl surveys (MEDITS project) using as sampling gear a bottom trawl made of four panels (Relini et al., Reference Relini, Carpentieri and Murenu2008). The surveys were carried out between May 2005 and March 2007 at depths ranging from 10 to 260 m in the north-east Mediterranean (Adriatic Sea from the Gulf of Trieste (13°37′E and 45°40′N) to the Tremiti Islands (15°16′E and 42°08′N)) (Figure 1).
Fig. 1. Map of sampling sites in northern-central Adriatic Sea.
In the laboratory, total length (TL) was measured to the nearest mm and specimens were weighed (W) to the nearest 0.1 g. Macroscopic analysis of gonads was performed to determine sex, with specimens classified as females, males, juveniles and not determined (sex not distinguished). The stomachs were immediately removed and preserved in 70% ethanol solution for stomach content analysis. Preys were identified to the lowest possible taxonomic level, counted, and weighed to the nearest 0.1 mg after removal of surface water using blotting paper.
Data analysis
Stomach contents, according to size-class, sex, season (of international bottom trawl survey in the Mediterranean Sea: winter and summer) and water depth, were investigated by qualitative and quantitative analyses. Size-classes were defined based on the results of cluster analysis of diet composition, using the Bray–Curtis similarity index (Clarke & Warwick, Reference Clarke and Warwick1994) that assessed differences between predator length and number (%N) of ingested food categories.
The diet was described as follows:
%N = the number of individuals of each prey type expressed as a percentage of the total number of prey items from all stomachs;
%W = the weight of individuals of each prey type expressed as a percentage of the total weight of prey items from all stomachs;
%F = the number of stomachs in which a food type occurred expressed as a percentage of the total number of stomachs that contained prey;
%IRI = the index of relative importance expressed as a percentage where IRI (Pinkas et al., Reference Pinkas, Oliphant and Iverson1971), is expressed as: IRI = (%N + %W) %F where %N is the percentage of numerical composition, %W is the percentage in weight, %F is the percentage of frequency of occurrence.
Prey-specific abundance (PSA) is defined as the percentage of a prey taxon comprising all prey items in only those predators in which the actual prey occurs, or in mathematical terms (Amundsen et al., Reference Amundsen, Gabler and Staldvik1996):
PSA = Pi = ∑Si/∑Sti where Pi is the prey-specific abundance of prey i, Si the stomach content (number) comprising prey i, Sti the total stomach content of only those predator specimens with prey i in their stomachs.
The Shannon index (H′) was employed to measure trophic diversity as: H′= ΣNi·ln(Ni).
Multivariate analyses were performed on prey classes in order to evaluate environmental and ontogenetic patterns: non-metric multidimensional scaling (nMDS) plot, analysis of similarity (ANOSIM) and multiple response permutation procedure (MRPP) were calculated on numerical abundance Bray–Curtis dissimilarity matrix.
Feeding trends were performed with R software version 2.10 base and Vegan package (R Development Core Team, 2010).
RESULTS
The TL of the 1096 tub gurnard examined ranged from 63 to 415 mm (mean size = 207 mm; median size = 218 mm). According to sex determination, 466 females, 373 males, 235 juveniles and 22 not determined were recorded.
Overall diet description
Qualitative analysis permitted to identify 55 food items belonging to 6 main taxa (Table 1). Tub gurnard diet chiefly consisted of Crustacea (especially Decapoda) and Teleostei. In quantity terms, Crustacea (%N = 89.7, %W = 58) were the most abundant prey taxon followed by Teleostei (%N = 6.7, %W = 39.7). Other preys such as Mollusca (%N = 1.6, %W = 0.6) and Anellida: Polychaeta (%N = 0.03, %W = 0.01) were occasionally recorded. In terms of number and weight abundance (%N and %W) and of relative importance (%IRI), the most important preys were: Goneplax rhomboides (%N = 14.2, %W = 29.5, %IRI = 50.3), Liocarcinus depurator (%N = 6, %W = 20.3, %IRI = 10.5) and Liocarcinus spp. (%N = 13, %W = 24.8, %IRI = 5) for Decapoda: Reptantia; Philocheras spp. (%N = 59.5, %W = 2.5, %IRI = 26.5) and Philocheras bispinosus (%N = 15.1, %W = 1, %IRI = 3) for Decapoda: Natantia; Engraulis encrasicolus (%N = 0.6, %W = 11.9, %IRI = 1.1) and Gobius niger (%N = 0.6, %W = 8.8, %IRI = 0.8) for fish.
Table 1. Diet composition of Chelidonichthys lucerna. %N, percentage in number; %W, percentage in weight; %F, frequency of occurrence; %IRI, percentage of index of relative importance of prey items; PSA, % prey taxon over all preys in all predators.
In terms of PSA, which is a food preference index, the most important preys were found to be Corystes cassivelaunus and Munida spp. for crustaceans, and Merlangius merlangius, Callionymus risso, Merluccius merluccius, Pomatoschistus minutus, Trisopterus minutus, Cepola macrophthalma and Deltentosteus quadrimaculatus for fish.
Diet variation with sex
No significant difference was found between feeding habits and sex (Crustacea F–M: χ2 = 0.0011, df = 1, P = 0.974; Teleostei: F–M: χ2 = 3.6085, df = 1, P = 0.05749). No trophic diversity between sex was found (H′= 0.219 for both). Multivariate analyses showed no difference: no pattern in nMDS plot (Figure 6), ANOSIM R = –0.0018 P = 0.635, MRPP A = 0.0013 P = 0.069.
Diet variation with fish size-class
Diet variation as a function of length was appreciable. Cluster analysis based on similarity in the diet showed a first dichotomy that discriminated two main groups with a total length superior or lower than 180 mm (Figures 2 & 3).
Fig. 2. Cluster analysis results based on the feeding habits of Chelidonichthys lucerna per size-class of predators (mm).
Fig. 3. Class size-wise numeric and gravimetric prey abundance percentage of the most abundant prey item.
Juveniles (≤180 mm) were more related to Crustacea, while adults were more related to (>180 mm) to fish. Trophic diversity was lower in juveniles (H′ = 0.098) than adults (H′ = 0.229).
Multivariate analyses showed difference between juveniles and adults (Figure 7), ANOSIM R = 0.1039 P = 0.001, MRPP A = 0.0499 P = 0.001.
Diet variation with season
Seasonal diet showed an increase of fish in winter (E. encrasicolus and G. niger) and Crustacea in summer (Philocheras sp.). However, some Crustacea such as L. depurator and G. rhomboides were found in high abundance in both seasons (Figure 4).
Fig. 4. Season-wise numeric and gravimetric prey abundance percentage of the most abundant prey item.
Trophic diversity was lower in summer (H′= 0.099) than winter (H′ = 0.229).
Multivariate analyses showed difference between winter and summer (Figure 8), ANOSIM R = 0.1064 P = 0.001, MRPP A = 0.0509 P = 0.001.
Diet variation with depth
Three depth strata (I: 10–25 m; II: 25–50 m; III: 50–260 m) were considered. No noticeable differences were found between large taxonomic groups (H′I = 0.187; H′II = 0.205; H′III = 0.170), but there were differences in a specific level. For instance, G. niger prevailed in first and second depth strata, while E. encrasicolus prevailed in second and third strata (Figure 5).
Fig. 5. Depth-wise numeric and gravimetric prey abundance percentage of the most abundant prey item.
Fig. 6. Non-metric multidimensional scaling plot between sexes (females and males) with superimposed 95% confidence ellipses.
Fig. 7. Non-metric multidimensional scaling plot between two size-classes (Class I ≤ 180; Class II > 180) with superimposed 95% confidence ellipses.
Fig. 8. Non-metric multidimensional scaling plot between seasons (winter and summer) with superimposed 95% confidence ellipses.
Multivariate analyses showed difference between strata (Figure 9), ANOSIM R = 0.0767 P = 0.001, MRPP A = 0.0465 P = 0.001.
Fig. 9. Non-metric multidimensional scaling plot among three depth strata (I: 10–25 m; II: 25–50 m; III: 50–260 m) with superimposed 95% confidence ellipses.
DISCUSSION
The diet of C. lucerna in the study area mainly consisted of epibenthic crustacea such as Goneplax rhomboides, Liocarcinus spp. and Philocheras spp. There were also nectobenthic preys, mainly fish while pelagic preys such as Engraulis encrasicolus could probably result from being active predation or discarded food, although Olaso et al. (Reference Olaso, Sanchez, Rodriguez-Cabello and Velasco2002) indicates that C. lucerna does not show scavenging behaviour.
Crustacea and Teleostei were the two principal prey categories, while Mollusca and Polychaeta were found in low numbers. The results of this study are in accordance with previous studies (Atlantic Ocean: Nouvel, Reference Nouvel1950; Costa, Reference Costa1988; Hostens & Mees, Reference Hostens and Mees1999; Mediterranean Sea: Reys, Reference Reys1960 (Gulf of Lions); Froglia, Reference Froglia1976 (juveniles, central Adriatic Sea); Colloca et al., Reference Colloca, Ardizzone and Gravina1994 (Tyrrhenian Sea); Faltas, Reference Faltas1996 (Egyptian waters); Morte et al., Reference Morte, Redon and Sanz-Brau1997 (Spain)).
The feeding of the species changes substantially as a function of size. It is in fact greater, as measured by amount of prey in the stomach, in juveniles, during summer and at shallow depths. Crustacea were found to be the main prey regardless of predator size. They were however found to prevail in the diet of juveniles, while in the stomachs of adult specimens, Crustacea were almost equal, weight-wise, to fish, a fact that has been previously reported for this species in the study area (Froglia, Reference Froglia1976; Colloca et al., Reference Colloca, Ardizzone and Gravina1994). Particularly, the predator length of 180 mm was found to correspond to predator growth ‘critical size’, that is to the onset of the start of sexual maturity in the Adriatic Sea (Montanini et al., Reference Montanini, Stagioni and Vallisneri2008; Vallisneri et al., Reference Vallisneri, Montanini and Stagioniin press). Variation observed between juveniles and adults could be attributed mainly to differences in gonad maturity and in the biological cycle, that requires greater energy. So, this critical size was related to the start of sexual maturity, the tendency to migrate to greater depths (generally muddy substrate), a change of diet from crustaceans to fish and an increase of variety of food items eaten (Montanini et al., Reference Montanini, Stagioni and Vallisneri2008).
Variation observed between seasons and depths reflected the spawning period of this species (Tsikliras et al., Reference Tsikliras, Antonopoulou and Stergiou2010): juveniles were more abundant during summer along the Italian coast, adults in winter at greater depth.
Based on the results presented here, it may be said that C. lucerna is an ‘opportunistic predator’, since its feeding habits are not species-specific. Its trophic spectrum is in fact very wide, being characterized by a high degree of biodiversity and correlated with changes in feeding habits during growth. Stergiou & Karpouzi (Reference Stergiou and Karpouzi2002), who identified Triglidae as ‘omnivores with preference for animal material’, are not in agreement with the present study, where only only nectobenthic fish and epibenthic invertebrates were found in its diet.
Diet composition is considered to reflect the biocenosis typical of the area (e.g. Colloca et al., Reference Colloca, Ardizzone and Gravina1994; Serena et al., Reference Serena, Voliani and Auteri1998; Costa & Cabral, Reference Costa and Cabral1999; Link, Reference Link2004; Morte et al., Reference Morte, Redon and Sanz-Brau1997). For instance, E. encrasicolus plays an essential role in adult diet and is found in large quantities on the same relatively shallow beds (<100 m) of the western Adriatic Sea (Piccinetti & Piccinetti Manfrin, Reference Piccinetti and Piccinetti Manfrin1971), which is also the favourite habitat of C. lucerna.
Moreover, it is necessary to consider the possible impact among C. lucerna and other commercial fish and Crustacea (such as mantis prawn and shrimps). It in fact preys intensively on these species with marked effects on food webs, and this alone justifies the importance of a ‘multispecific approach’ in terms of analyses and management.
Despite the considerable increase in fishing activity, the inadequacy of traditional single-species models is aggravated by the lack of long-term routine fishery data and scarce scientific information (Stergiou & Karpouzi, Reference Stergiou and Karpouzi2002; Cury et al., Reference Cury, Shannon, Roux, Daskalov, Jarre, Moloney and Pauly2005; Vallisneri et al., Reference Vallisneri, Montanini and Stagioni2010). Food webs depict the feeding relationships (who eat whom) in communities, but for the large number of species and interactions, our knowledge of real food webs is limited (Bascompte et al., Reference Bascompte, Melian and Sala2005).
In conclusion, the goal of feeding behaviour studies for fishery management applications is to improve knowledge as to fish population biology and ecology in view of an ecosystem-based management of commercially important stocks.
ACKNOWLEDGEMENTS
Sampling was performed under the MIPAAF project ‘Trophic demersal population structure’. We thank Professor Corrado Piccinetti for his assistance in bottom trawl surveys and for his valuable suggestions in the drafting of the paper and Dr Emanuela Mazzoni for his precious help regarding stomach contents sampling.
