sign in sign up
<<
Home , Sardine, Tetrapturus

Vol. 531: 1–14, 2015 MARINE ECOLOGY PROGRESS SERIES Published July 2 doi: 10.3354/meps11353 Mar Ecol Prog Ser

FREEREE FEATURE ARTICLE ACCESSCCESS The role of sardine as prey for pelagic predators in the western assessed using stable isotopes and fatty acids

Luis Cardona1,*, Laura Martínez-Iñigo1, Rafael Mateo2, Jacob González-Solís1

1Biodiversity Research Institute (IRBio) and Department of Biology, Faculty of Biology, Universitat de Barcelona, Av. Diagonal 643, 08028 Barcelona, 2Instituto de Investigación en Recursos Cinegéticos (IREC), CSIC-UCLM-JCCM, Ronda de Toledo s/n, 13071 Ciudad Real, Spain

ABSTRACT: This study combined the analysis of fatty acids and stable isotopes of nitrogen and carbon to test the hypothesis that the pelagic food web of the south Catalan Sea has a wasp-waist structure sup- ported by sardines Sardina pilchardus. If this hypo- thesis were correct, most predators would be expected to have stable isotope ratios and fatty acid profiles consistent with those derived from a sardine-based diet. However, this was true only for Scomber scomber, blue butterfish Stromateus fiatola, all sea- birds and oceanic loggerhead turtles Caretta caretta. The values of the DHA/EPA index of neritic logger- head turtles and striped dolphins Stenella caeru leo alba were also consistent with a sardine/ diet, but In summer, 2 parallel pelagic food webs coexist in the Medi- terranean Sea. Sardine is the main prey of most seabirds, their trophic positions were too high. On the other whereas and horse mackerel support most of the hand, the DHA/EPA index of most predatory in- predatory fishes. dicated that Engraulis encrasicolus and/or horse mackerel Trachurus trachurus were their main Photos: Seabirds by Pep Arcos (SEO/BirdLife), sardines by Manuel Elices, amberjacks by Manuel Gazo, horse mack- prey. Nevertheless, some amounts of low trophic level erel by Alex Llorente invertebrates were likely to be consumed by some predatory fishes, because their trophic positions where lower than expected from a -based diet only. The heterogeneous distribution of phytoplankton groups above and below the thermocline during the warm season is hypothesized to be the primary reason for INTRODUCTION this food web structure, although the strong reliance of some seabirds on sardines is the likely consequence Marine food webs can be classified into 3 major of a massive consumption of discards from groups according to the mechanisms controlling spe- boats. In short, there is little evidence for a wasp-waist cies abundance. In bottom-up systems, where the structure based on sardine, which may explain why abundance of species at contiguous trophic levels is the populations of predators fluctuate less than the positively correlated, density-dependent mortality is population of sardines. caused by food limitation, which ultimately deter- mines species abundance (Power 1992, Ware & KEY WORDS: Anchovy · Engraulis encrasicolus · Fatty acids · Food web · Sardina pilchardus · Sardine · Thomson 2005). Conversely, in top-down systems, Stable isotope · Top predator which are characterized by an inverse correlation in the biomass of species at contiguous trophic levels, Resale or republication not permitted without written consent of the publisher the major determinant of species abundance is the

*Corresponding author: [email protected] © Inter-Research 2015 · www.int-res.com 2 Mar Ecol Prog Ser 531: 1–14, 2015

density-independent mortality induced by predators bac teria all produce taxon-specific fatty acids that (Power 1992, Paine 2010). Finally, wasp-waist eco- are retained by their predators and can be used to systems are characterized by the capacity of small, identify the sources of fatty acids (Dalsgaard et al. short-lived fishes occupying intermediate trophic 2003, Ramos & González-Solís 2012). Fatty acids levels to simultaneously control the abundance of have even used for quantitative diet reconstruction, through predation as well as that of top although the method is limited by the necessity of predators through resource availability (Brodeur & experimentally assessing the calibration coefficients Pearcy 1992, Cury et al. 2000). Deciphering the con- (Iverson et al. 2004). Because fatty acids are more trol mode of each ecosystem is essential to predict the specific to dietary source compared to δ13C, they can impact of , as these have not only reduced alleviate some of the ambiguities that result from the abundance of top predators (Christensen et al. using stable isotopes alone, particularly in cases 2003), but also eroded the abundance of species at where the differences between δ13C of different car- intermediate trophic levels (Palomera et al. 2007). bon sources are small (Nyssen et al. 2005, Alfaro et This is particularly important in wasp-waist ecosys- al. 2006, Perga et al. 2006, El-Sabaawi et al. 2009). tems, where the whole food web is highly sensitive Recently, Pethybridge et al. (2014) reported that dif- to the of small pelagic fishes (Cury et al. ferences in the fatty acid profiles of sardines and 2000, Shannon et al. 2000, Bakun 2006). anchovies from the Gulf of Lions related to a differ- Although wasp-waist ecosystems are usually asso- ential exploitation of dinoflagellates and diatoms, thus ciated to the world’s eastern boundary currents that suggesting that fatty acids may help to elucidate the undergo seasonal (Cury et al. 2000), they relative importance of sardines and anchovies as for- also exist in other regions with large populations of age for predators. small pelagic fishes (Bakun 2006). In the Mediterran- The main aim of this study was to combine the ean Sea, European sardines Sardina pilchardus and analysis of fatty acids and stable isotopes of nitrogen European anchovies Engraulis encrasicolus occupy and carbon to test the hypothesis that the pelagic slightly different ecological niches but both abound ecosystem in the south Catalan Sea (Mediterranean in the most productive regions of the basin, which are Sea) has a wasp-waist structure, where sardines fertilized either by upwelling or freshwater runoff would be the major prey of most predators (Coll et al. (Palomera et al. 2007, Costalago et al. 2012). Ecosys- 2006, Palomera et al. 2007, Coll et al. 2008). If so, the tem modeling has suggested a possible wasp-waist fatty acid profiles of predators would be closer to structure for those ecosystems, with a key role for those of sardines than to those of anchovies and other sardines in the south Catalan Sea (Coll et al. 2006, low trophic level species. Palomera et al. 2007, Coll et al. 2008). Recent research using stable isotope analysis has revealed that some top predators inhabiting wasp- MATERIALS AND METHODS waist ecosystems consume more crustaceans and gelatinous zooplankton than previously thought Sampling (Cardona et al. 2012, Madigan et al. 2012). This sug- gests a more complex food web structure, since pred- Sampling of most species was conducted in 2006, ators would feed on multiple trophic levels in parallel although that of seabirds extended until 2010 due to food webs (Olsen et al. 2001, Darnaude et al. 2004, the opportunistic nature of sampling. Only samples Frisch et al. 2014), with important implications on collected from May to July were considered for ana - ecosystem stability and resilience (Gross et al. 2009). lysis, to reduce as much as possible the effects of sea- However, the small variability of δ13C values at the sonal variations in the fatty acid profiles of small base of pelagic food webs (Cardona et al. 2012, forage fishes (Bandarra et al. 2001, Zlatanos & Laska- Madigan et al. 2012, Stowasser et al. 2012) impairs ridis 2007, Celik 2008, Pethybridge et al. 2014) and the discriminating power of the isotopic mixing mod- crustaceans (Mayzaud et al. 1999). The fatty acid els used to reconstruct the diet of predators, because profiles and the stable isotope ratios of fish muscle of high uncertainty about the feasible con tribution of integrate those of the diet consumed throughout the potential prey to the diet of predators (Cardona et al. last 3 mo (Robin et al. 2003, Buchheister & Latour 2012, Madigan et al. 2012). 2010). Accordingly, stable isotope ratios and fatty Fatty acids offer an alternative set of intrinsic mark- acid profiles samples are expected to integrate those ers that can help us improve the resolution of food in the local forage species as long as predators web structure. Phytoplankton, microzooplankton and remain within the area for at least 3 mo. Cardona et al.: Mediterranean pelagic food webs 3

Small pelagic fishes, and neritic predatory N isotopes, so the other subsample was used to fishes (see Table 1) were sampled from the catches determine the δ15N value. All of the samples were landed at the harbors of Vilanova i la Geltrú and Sant weighed into tin cups, combusted at 1000°C, and Carles de la Ràpita by commercial vessels operating analyzed in a Flash 1112 IRMS Delta C Series in the south Catalan Sea (Fig. 1). Meganyc- EA Thermo Finnigan continuous flow isotope ratio tiphanes norvegica were collected from the stomach mass spectrometer. Stable isotope abundances were contents of bullet . Blue and some large expressed in δ notation according to the following predatory fishes (see Table 1) were captured by com- expression: mercial longliners operating in a nearby off-shore δX = [(R /R ) − 1] × 1000 area (Fig. 1) and the tissue samples of these species sample standard 13 15 were collected by observers on board. Pink jellyfish where X is C or N and Rsample and Rstandard is the Pelagia noctiluca were collected with a dip net dur- corresponding ratio 13C/12C or 15N/14N of the sample ing the fishing operations. Oceanic loggerhead turtles and the standard. The standards for 13C and 15N were Caretta caretta and seabirds were caught inciden- Vienna Pee Dee Belemnite (VPDB) and atmospheric tally by fishermen operating in those areas and sam- nitrogen (air), respectively. International isotope 13 ples were collected by onboard observers. Samples secondary standards for carbon (IAEA CH6 [δ C = 13 13 from neritic loggerhead turtles and striped dolphin −10.4‰], USGS 24 [δ C = −16.1‰], IAEA CH7 [δ C Stenella caeru leoalba were collected from dead = −31.8‰]) were used to a precision of 0.2‰, and for 15 15 stranded indi viduals (Fig. 1). nitrogen (IAEA NO3 [δ N = +4.7‰], IAEA N2 [δ N = 15 White dorsolateral muscle was sampled from all +20.3‰], IAEA N1 [δ N = +0.4‰]) to a precision of fishes, as well as mantle from the cephalopods and 0.3‰. pectoral muscle from turtles, dolphins and seabirds. Jellyfish and krill were fully homogenized. Samples were stored at −20°C prior to analysis. Sam- ples for fatty acid analysis are usually stored at −80°C to avoid degradation of unsaturated fatty acids (Guitart et al. 1999). However, this is impractical when considering large fishes, sea turtles, dolphins and seabirds, as they are often sampled opportunistically. Nevertheless, it has been shown that most of the ecologically relevant infor- mation is retained in the fatty acid profiles of fishes even when stored at temperatures higher than −80°C (Chaijan et al. 2006).

Stable isotope analysis

Once thawed, tissues were dried for 24 h at 60°C and ground to a fine powder; lipids were then extracted with a chloroform/methanol (2:1) so - lution. Crustacean samples were split Fig. 1. Study region. Small pelagic fishes, squids and neritic predators were into 2 subsamples. One subsample sampled from the catches landed at the harbors of Vilanova i la Geltrú and was treated with 0.5 N HCl to re - Sant Carles de la Ràpita. Dead stranded neritic loggerhead turtles Caretta move the inorganic carbonates of caretta and striped dolphin Stenella caeruleoalba were collected along the the skeleton and avoid any bias in stretch of coastline in bold. The rectangle denotes the area where longliners captured large predatory fishes and where pink jellyish Pelagia noctiluca were δ13 the C. However, acidification may collected during fishing operations. Seabirds were captured incidentally over modify the relative concentration of the entire region 4 Mar Ecol Prog Ser 531: 1–14, 2015

Fatty acid analysis sources of predators. Accordingly, 1-way ANOVA, followed by Tukey post-hoc tests, was used to com- The fatty acid composition was determined after pare the average values of δ13C, δ15N and the methylation following the method used by Mateo et DHA/EPA index of krill, pink jellyfish, longfin squid al. (2003), but using mass spectrometry for the identi- Loligo vulgaris, sardines, anchovies and horse mack- fication and quantification of the wide range of fatty erel Trachurus trachurus. Levene’s test was used to acids present in these samples. Briefly, each sample check for homoscedasticity.

(0.3 g) was homogenized in 9 parts of Na2SO4 (2.7 g) Only fatty acids that had an overall mean of before transferring it into 15 ml Pyrex tubes with >0.4% of total fatty acids were considered for latter teflon caps, at which time 10 µl of internal standard analysis (Iverson et al. 2002). Data were standard- was added (13:0, tridecanoic acid, 10 µg ml−1). Then, ized prior to analysis to have zero mean and unit we added 4 ml of H2SO4 1N in methanol to each tube, variance and principal component analysis (PCA) after which they were topped with N2 and closed, was used to explore the sources of variability with heated in an oven (80°C) for 6 h, and shaken every the fatty acid profiles of the assemblage. Most of hour. Subsequently, the methylation reaction con - the variability was associated with saturated or tinued for 12 h at 60°C. Once cooled, fatty acid monosaturated fatty acids, unlikely to be conserved methylesters (FAMEs) were extracted with 1.7 ml of from predator to predator (Ramos & González-Solís hexane after adding 4 ml of ultrapure water (Milli-Q) 2012), or rare saturated fatty acids of unknown eco- to each tube. For this, the tubes were horizontally logical significance (see ‘Results’). Accordingly, we shaken for 5 min and then centrifuged (626 × g) for focused on the 22:6n-3 to 20:5n-3 ratio (or DHA/ another 5 min. The upper phase of hexane containing EPA index) for further analysis, as they were the 2 the FAMEs was transferred into 2 ml chromato- most abundant and variable polyunsaturated fatty graphic vials, topped with N2 and analyzed by gas acids in the samples and were expected to be con- chromatography coupled to mass spectrometry (GC- served from prey to predator (Ramos & González- MS). The analysis was carried out using an Agilent Solís 2012). Technologies 6890N GC Network System with a Potential consumers of sardines were identified, 7683 series injector and coupled to an electronic assuming that the value of the DHA/EPA does not impact-mass spectrometry detector (EI-MS, 5973N change from prey to predator (Ramos & González- MSD). The capillary column was a BPX70 column Solís 2012) and assuming the following trophic (SGE, 30 m ×, 0.25 mm i.d., 0.25 µm film thickness). enrichment factors: 1.4‰ for δ15N for seabird muscle The injector in split mode (20:1) was set at a tem- (Hobson & Clark 1992) and 2.6‰ δ15N for bony fish perature of 270°C and the oven was maintained for muscle (Pinnegar & Polunin 1999). We are not aware 5 min at 100°C after injection and then increased to of any published trophic enrichment factors experi- 240°C at a ramp rate of 2.5°C min−1. The carrier gas mentally derived for dolphin and sea turtle muscle. was helium at a flow rate of 1.2 ml min−1. The source All calculations were performed with IBM SPSS inlet of the mass spectrometer was held at 230°C Statistics 20. and 70 V. The identification and quantification were achieved by comparison to retention times of FAME reference standards (FAMQ-005, AccuStandard) and RESULTS the mass spectra. Blanks were processed with each batch of samples. This method has been used before Stable isotopes for the analysis of fatty acid composition in several marine species (Guitart et al. 1996, 1999, Mateo et al. Only krill differed in average δ13C from other po -

2004) and the previously obtained chromatographic tential prey (ANOVA: F5,29 = 28.090, p < 0.001; n = 5 information about fatty acids profiles was useful for each species, Tukey as post-hoc test; Fig. 2). Variabil- the identification of the fatty acids in this study. ity between forage species was larger for the δ15N values, as krill and pink jellyfish were significantly depleted in 15N when compared with the other spe- Statistical analysis cies. The average δ15N value of horse mackerel was significantly higher than that of sardines but not dif- If potential prey do not differ in the average values ferent from those of anchovies and longfin squids 13 15 of δ C, δ N and the DHA/EPA index, these bio- (ANOVA: F5,29 = 68.035, p < 0.001; n = 5 each species, chemical markers would not be able to trace the food Tukey as post-hoc test; Fig. 2). Cardona et al.: Mediterranean pelagic food webs 5

–16 a b a a a a the most likely low trophic level prey consumed by Balearic and Mediterranean shearwaters, bullet tuna and mackerel. Scatter of predators along the δ15N axis was larger –18 than along the δ13C axis, although most predatory fishes overlapped within a narrow range (Fig. 3). The average δ15N value of most predators was consistent

C (‰) –20 with a sardine/squid-based diet (Fig. 3), according to 13

δ the trophic discrimination factors reported for sea- birds and fishes (see ‘Materials and Methods’). Only blue , and leerfish had a trophic level –22 higher than expected and bullet tuna and dolphinfish a lower one. However, the fatty acid profiles of most a a b b,c b,c c 14 predators indicated that they relied on other forage species and not sardines (see below).

12

Fatty acids 10

The fatty acid profiles of all the potential prey spe-

N (‰) 8

15 cies were dominated by 16:0, 18:0, 20:5n-3 and δ 22:6n-3 (Table 1; Table S1 in the Supplement at 6 www.int-res. com/articles/ suppl/m531 p001_ supp. xls), but major differences existed between species. Pink 4 jellyfish differed from all other species because of 20 their very high levels of 16:0 and very low levels of a a,b a,b b c d 22:6n-3, whereas krill, longfin squid and sardines 16 were more enriched in 20:5n-3 than anchovies and horse mackerel. Accordingly, the highest average DHA/ EPA index was that of horse mackerel, fol- 12 lowed by that of anchovies and finally those of the

other 4 potential prey species (ANOVA F5,29 = 99.686, 8

DHA/EPA p < 0.001, Tukey as post-hoc test; Fig. 2). The fatty acid profiles of predatory fishes, seabirds, sea turtles 4 and striped dolphins were also dominated by 16:0, 18:0, 20:5n-3 and 22:6n-3, but levels of 18:0 were 0 usually much higher than those in potential prey species (Table 1). Furthermore, loggerhead turtle, Krill Squid Horse Jellyfish Sardine striped dolphin and seabirds differed from inverte- Anchovy mackerel brates and fishes in their low levels of 22:6n-3 and Fig. 2. Biochemical markers of potential prey. Species shar- high levels of 18:2n-6 and 20:4n-6 (Table S1). ing superscript letters within the same panel are not signifi- cantly different based on Tukey’s post-hoc test PCA extracted as many as 8 components with an eigenvalue >1 to explain 76.6% of the overall variability in the fatty acid profiles of the 32 groups Most predatory fishes overlapped within a narrow considered (31 species and 2 groups of loggerhead range of δ13C values, but scatter was larger for air- turtles). According to the loadings of the 3 first compo- breathing predators, with seabirds usually more en- nents, most of that variability was associated with riched in 13C than fishes and sea turtles, and dolphins 14:0, 15:0, 16:0, 16:1-n13, 18:00, 18:3n-3, 20:2n-7, overlapping with fishes (Fig. 3). Considering the 20:4n-6, 20:5n-3 and 22:6n-3 (Table 2). The first average δ13C of krill (−20.8 ± 0.7 ‰), that of the component (PC1) explained 28.4% of the overall vari- remaining potential prey (−17.9 ± 0.6 ‰) and the ability in the fatty acid profiles and opposed the abun- trophic discrimination factors reported for seabirds dance of 16:0 and 22:6n-3 to that of all the other fatty and fishes (see ‘Materials and methods’), krill was acids. The high loading for 20:2n-7 is remarkable 6 Mar Ecol Prog Ser 531: 1–14, 2015

16 When plotted into the PC1–PC2

bluefish space, potential prey species were 15 widely scattered over the PC1 axis and predators were scattered along 14 blue shark both the PC1 and PC2 axis (Fig. 4).

13 leerfish This suggests that fatty acids offered bonito better discrimination between food N (‰) 13 15 12 sources than δ C did. However, δ mackerel amberjack pompano most of the variability was associated 11 shortfin squid blue butterfish with saturated fatty acids (14:0, 16:0 and 18:0) or monosaturated fatty spearfish 10 bluefin tuna acids (16:1-n13), unlikely to be con- served from predator to predator. 9 bullet tuna dolphinfish Furthermore, the dietary signifi- 8 cance of 20:2n-7 and 20:4n-6 was unclear. Accordingly, we focused on 16 the relative abundance of the other major polyunsaturated fatty acids 15 neritic turtles (22:6n-3 and 20:5n-3) and used the

14 DHA/EPA index for further analysis, as its dietary significance is well 13 known. Jellyfish, krill, sardines and striped dolphin squid were characterized by low val- N (‰) Audouin’s gull 15 12 δ storm petrel ues of the DHA/EPA index, anchovies had an intermediate value and horse 11 Balearic shearwater shag mackerel was the species with the yellow-legged gull highest value (ANOVA F5,29 = 10 99.686, p < 0.001, Tukey as post-hoc

9 test; Fig. 2). Mediterranean shearwater oceanic turtles Cory’s shearwater 8 –21 –20 –19 –18 –17 –16 –15 Potential sardine consumers δ13C (‰) As sardines and longfin squids did Fig. 3. Scatterplot of sharks, fishes and large squids (top panel) and air-breath- not differ in their average DHA/EPA δ13 δ15 ing predators (bottom panel) into the C– N space. Arithmetic mean and index and δ15N values (Fig. 2), we standard deviation are shown for each species. Dashed lines show the ex- pected higher and lower threshold of δ15N for predators with an exclusive sar- pooled them to set up the expected dine/squid-based diet, according to the trophic discrimination factors ex- range of values of potential sardine pected for each group. Open triangles denote sharks and fishes, open circles consumers. The values of their DHA/ seabirds, filled triangles turtles, filled circles large squid and filled squares EPA indices ranged from 1.5 to 4.2 dolphins and those of δ15N from 7.75 to 10.23 (Table S1 in the Supplement). Ac- (Table 2), as this fatty acid was scarce in most samples cordingly, only predators with values of the DHA/ except in sardines and mackerel (Table S1). The EPA index lower than 4.15 and δ15N values ranging second component (PC2) explained 13.0% of the from 9.85 to 12.83 (fishes) or from 8.65 to 11.65 (sea- overall variability in the fatty acid profiles and was birds) were likely to have diets based exclusively on strongly influenced by the abundance of 18:0 and sardines/squid. Only mackerel, blue butterfish and 20:4n-6 (positive values) as well as 14:0 and 22:6n-3 all the seabirds fit both requisites (Fig. 5). Neritic log- (negative values).The third component (PC3) ex- gerhead turtles and striped dolphins may also con- plained 10.3% of the overall variability in the fatty sume large amounts of sardines and squids, accord- acid profiles and was strongly influenced by the ing to the DHA/EPA index, but they are at a trophic abundance of 16:0 and 16:1-n13 (positive values) as level higher than expected for a diet based solely on well as 18:3n-3 and 20:5n-3 (negative values). sardines and squid (Fig. 5). Cardona et al.: Mediterranean pelagic food webs 7

DISCUSSION turtles was too high to rely only on sardines and/or longfin squids as their major diet. The results reported here do not support the exis- A second group of predators in cluded blue shark tence of a sardine-based wasp-waist ecosystem in and most predatory fishes and was characterized by the south Catalan Sea. Instead, fatty acid profiles high levels of 22:6n-3 and low levels of 20:5n-3. This revealed that only a handful of predators may rely suggests an anchovy/horse mackerel-based diet, primarily on sardines and/or longfin squids (Fig. 6). although the consumption of some low trophic level Discriminating between these 2 prey species is not invertebrates enriched in 22:6n-3, such as jellyfish possible, because they did not differ in the average and krill, cannot be ruled out on the basis of the δ15N values of the biochemical markers used here. Mack- values of most predators. On the other hand, the erel, blue butterfish, all the seabirds, sea turtles and trophic positon of bluefish, leerfish and blue shark striped dolphins where characterized by high levels was too high to be consistent with an exclusive of 20:5n-3 and low levels of 22:6n-3, consistent with a anchovy/horse mackerel diet. Furthermore, the val- sardine/squid-based diet. Nevertheless, the trophic ues of the DHA/EPA index of amberjack and blue position of striped dolphins and neritic loggerhead shark were much higher than those of any potential

Table 1. Body size and relative abundance (mean ± SD) of the 4 major fatty acids in the species considered. Length comprises total length for fishes and dolphins, mantle length for squids and curved carapace length for sea turtles

Species n Length Relative abundance of major fatty acids (%) Scientific name Common name (cm) 16:00 18:00 20:5n-3 22:6n-3

Potential prey Engraulis encrasicolus Anchovy 5 11−14 28 ± 4 5 ± 1 7 ± 1 46 ± 7 Loligo vulgaris Longfin squid 5 9−11 26 ± 1 4 ± 1 12 ± 1 47 ± 1 Meganyctiphanes norvegica Krill 5 – 21 ± 1 2 ± 1 19 ± 1 39 ± 1 Pelagia noctiluca Pink jellyfish 5 – 69 ± 20 15 ± 10 6 ± 6 2 ± 1 Sardina pilchardus Sardine 5 13−17 23 ± 3 5 ± 1 9 ± 3 24 ± 9 Trachurus trachurus Horse mackerel 5 15−22 20 ± 3 8 ± 1 4 ± 1 60 ± 2 Mesopredators Auxis rochei Bullet tuna 5 35−45 29 ± 7 10 ± 2 6 ± 1 37 ± 12 Sarda sarda Bonito 6 43−60 30 ± 8 13 ± 3 5 ± 4 28 ± 15 Scomber scombrus Mackerel 5 24−28 24 ± 2 7 ± 2 8 ± 1 26 ± 4 Seriola dumerili Amberjack 5 25−27 18 ± 4 13 ± 2 2 ± 1 46 ± 8 Stromateus fiatola Blue butterfish 5 41−55 35 ± 3 15 ± 1 3 ± 1 10 ± 2 Todarodes sagittatus Shortfin squid 5 25−36 23 ± 2 4 ± 1 11 ± 1 56 ± 4 Trachinotus ovatus Pompano 6 33−38 29 ± 7 11 ± 2 3 ± 1 29 ± 15 Large predatory fishes Coryphaena hippurus Dolphinfish 4 47−99 33 ± 9 18 ± 5 2 ± 1 29 ± 16 Euthynnus alletteratus Little tunny 4 65−80 31 ± 4 15 ± 3 3 ± 1 33 ± 11 Lichia amia Leerfish 3 59−60 24 ± 1 15 ± 2 3 ± 1 40 ± 4 Pomatomus saltatrix Bluefish 4 77−95 26 ± 1 10 ± 1 3 ± 1 48 ± 4 Tetrapturus belone Spearfish 5 122−142 22 ± 3 14 ± 4 3 ± 1 47 ± 7 Thunnus alalunga Albacore 5 77−92 28 ± 9 11 ± 2 3 ± 1 36 ± 22 Thunnus thynnus Bluefin tuna 6 91−94 26 ± 7 14 ± 4 3 ± 1 36 ± 13 Xiphias gladius Swordfish 5 120−168 26 ± 1 11 ± 5 2 ± 2 20 ± 9 Sharks Prionace glauca Blue shark 5 200−250 25 ± 2 16 ± 3 1 ± 1 34 ± 3 Seabirds Calonectris diomedea Cory’s shearwater 10 – 21 ± 2 15 ± 2 7 ± 1 14 ± 3 Hydrobates pelagicus Storm petrel 3 – 23 ± 1 20 ± 2 5 ± 1 10 ± 4 Larus audouinii Audouin’s gull 10 – 16 ± 4 18 ± 2 6 ± 2 12 ± 2 Larus michahellis Yellow-legged gull 11 – 22 ± 4 19 ± 3 4 ± 3 7 ± 3 Phalacrocorax aristotelis Shag 3 – 16 ± 1 23 ± 2 6 ± 2 6 ± 1 Puffinus mauretanicus Balearic shearwater 10 – 16 ± 4 22 ± 2 8 ± 2 16 ± 3 Puffinus yelkouan Mediterranean shearwater 10 – 16 ± 3 19 ± 2 9 ± 2 18 ± 2 Marine turtles Caretta caretta Loggerhead turtles (neritic) 4 43−55 14 ± 5 24 ± 5 6 ± 2 5 ± 2 Caretta caretta Loggerhead turtles (oceanic) 5 45−57 27 ± 5 21 ± 3 4 ± 1 4 ± 2 Marine mammals Stenella caeruleoalba Striped dolphin 5 180−220 19 ± 4 21 ± 6 9 ± 3 10 ± 3 8 Mar Ecol Prog Ser 531: 1–14, 2015

Table 2. Component matrix for the 3 first principal com - western Mediterranean, which reveals that sardines ponents (PCs), which shows the correlation coefficients are seldom a major prey item of predatory fishes between the 34 fatty acids selected for analysis and the PCs. The 2 highest (positive) and lowest (negative) PCA scores (Massutí et al. 1998, Sinopoli et al. 2004, Falautano et for each axis are highlighted in bold al. 2007, Mostarda et al. 2007, Castriota et al. 2008, Consoli et al. 2008, Romeo et al. 2009). This is in con- Fatty Acid PC1 PC2 PC3 trast with the high biomass of sardines in study region, where it is the most abundant small pelagic 14:0 0.622 −0.468 −0.112 fish and dominates both landings and dis- 15:0 0.683 −0.456 0.145 cards (Coll et al. 2006). 16:0 −0.166 −0.291 0.476 17:0 0.476 −0.124 0.286 It has to be noted, however, that the biomass of 18:0 0.123 0.837 −0.028 anchovies in the south Catalan Sea is only slightly 20:0 0.869 0.127 0.040 lower than that of sardines (Coll et al. 2006) and 23:0 0.687 0.257 0.075 other 22:6n-3 enriched species, such as juvenile M16:0 0.735 0.015 0.136 anteiso horse mackerel, occur in high numbers over the M16:0 iso 0.773 −0.160 0.360 16:1n-7 0.772 −0.237 0.327 shelf (Coll et al. 2006, 2008). Furthermore, during 16:1n-9 0.610 0.199 −0.260 summer, sardines concentrate in pockets of low sea 16:1n-11 0.383 0.190 0.052 surface temperature and high chlorophyll over the 16:1n-13 0.458 −0.138 0.654 upper con tinental shelf (Tugores et al. 2011, Saraux 17:1n-7 0.732 0.224 0.237 18:1n-7 0.290 0.502 −0.045 et al. 2014), which might limit its detection by most 18:1n-9 0.545 0.547 0.101 predators. On the contrary, anchovies are more 18:1n-13 0.058 0.004 0.412 scattered over the continental shelf, from coastal 20:1n-7 0.266 0.059 0.038 lagoons to the deepest parts of the shelf (Palomera 20:1n-9 0.721 −0.274 −0.084 20:1n-11 0.564 0.007 −0.455 et al. 2007, Cardona et al. 2008, Tugores et al. 2011, 22:1n-7 0.524 −0.371 0.234 Giannoulaki et al. 2013, Saraux et al. 2014). Hence, 22:1n-9 0.597 −0.333 −0.420 during summer, an cho vies and other 22:6n-3 en- 24:1n-9 0.221 −0.145 0.480 riched prey are probably more accessible than sar- 16:3n-3 0.494 −0.441 −0.355 dines (in terms of spatial homo geneity) for a large 18:2n-6 0.414 0.519 −0.382 18:3n-3 0.625 0.017 −0.551 diversity of predators through the warm season, 18:4n-3 0.530 −0.529 −0.294 which may explain the results reported here. 20:2n-7 0.806 0.089 −0.230 The high reliance of seabirds on sardines during the 20:4n-6 0.204 0.807 −0.159 warm season, indicated by their high levels of 20:5n-3 20:5n-3 0.142 −0.280 −0.520 22:4n-6 0.233 0.429 0.467 (this study) and stomach contents analysis (Oro et al. 22:5n-3 0.280 0.011 −0.003 1997, Ramos et al. 2009) may be related to their ca- 22:5n-6 0.230 0.109 0.528 pacity to find the areas where sardines concentrate or 22:6n-3 −0.593 −0.551 0.001 boats are discarding trash fish. Many seabird species, such as Balearic shearwaters, Cory’s shearwater, Audouin’s gulls and yellow-legged gulls aggregate at prey considered here. The fatty acid profiles of some trawlers (Oro et al. 1997, Arcos & Oro 2002, Abelló et fishes are know to vary regionally within the western al. 2003, Ramos et al. 2009), whose discard is domi- Medi terranean (Roncarati et al. 2012) and hence the nated by sardines (Coll et al. 2008). Schools of small high DHA/EPA values re ported for amberjack and pelagic fishes in the south Catalan Sea are usually blue shark may reveal the use of foraging grounds found deeper than 25 m (Iglesias et al. 2003), but gulls beyond the limits of the study region. The standard do not dive at all and Cory’s shearwaters dive much deviation of the DHA/ EPA of other predatory fishes shallower than shearwaters in the Puffinus was also large, although variability remained within (Shealer 2001). This suggests that small the range of potential prey, which highlights the are usually out of reach of gulls and Cory’s shearwa- necessity for a large sample size to better character- ters but within the reach of shearwaters in the genus ize most species. Puffinus. Deeper diving might also ex plain why Despite those shortcomings, the re sults reported Balearic and Mediterranean shearwaters are more here clearly show that, at least in summer, sardines depleted in 13C than Cory’s shearwater, as deeper and longfin squids are minor prey for most of the diving may increase accessibility to 13C-depleted krill. predators studied. This inter pretation is supported by This difference in stable isotope ratios is unlikely available dietary information from elsewhere in the to have a geographic basis, as at least Cory’s and Cardona et al.: Mediterranean pelagic food webs 9

2.5 ine vertebrates usually accumulate n-6 anchovy fatty acids (Guitart et al. 1996, this study), the level of 18:0 increases with 1 trophic level (this study), and some fishes may syn thesize some polyunsat- horse mackerel urated fatty acids (Sargent 1997). –0.5 However, the identification of 2 dis- PC2 jellyfish krill tinct groups of forage species and predators was based the DHA/EPA longfin squid sardine –2 index and un likely to be affected by the physiological factors reported above (Ramos & González-Solís 2012). –3.5 Importantly, turtles, seabirds and dol- 2.5 phins have DHA/ EPA index values close to those of phylogenetically dis-

blue shark tant species such as krill, jellyfish and sardine—hence demonstrating that 1 swordfish dolphinfish diet, and not physiology, was the main leerfish blue butterfish bluefish pompano reason for the pattern reported here. amberjack –0.5 bluefin tuna Seasonality is another potential con- spearfish PC2 albacore little tunny bonito mackerel founding factor (Zlatanos & Laskaridis, bullet tuna 2007, Pethybridge et al. 2014); hence, shortfin squid –2 we considered only samples collected during the warm season. Fatty acid profiles and stable isotope ratios might –3.5 also vary on an annual basis, but the standard deviation of the DHA/EPA 2.5 index of seabirds was usually much shag lower than that of fishes, thus indi - neritic sea turtle Audouin’s oceanic sea turtle gull yellow-legged gull cating that pooling samples from dif- 1 striped dolphin ferent yeas was not a major source of storm petrel variability. Mediterranean shearwater In summary, the evidence reported –0.5 Balearic shearwater Cory’s shearwater here does not support the existence of PC2 a wasp-waist ecosystem based on sar- dine, but a more complex scenario –2 with 2 parallel food webs, one sup- ported primarily by diatoms and the other by dinoflagellates and cyanobac- –3.5 –3 –2 –1 0 1 2 3 teria. The existence of parallel food webs with limited exchange has al - PC1 ready been reported in other marine Fig. 4. Principal component analysis of potential prey (top panel), large squid, ecosystems and is usually related to the blue shark and fishes (central panel) and air-breathing predators (bottom panel) in the space defined by the first 2 principal components (PCs). Arith- existence of a microbial loop (Olsen et metic mean and standard deviation are shown for each species. Open tri- al. 2001) or limited benthic− pelagic angles denote sharks and fishes, open circles seabirds, filled triangles turtles, coupling (Darnaude et al. 2004, Frisch filled circles invertebrates and filled squares dolphins et al. 2014). Here, we hypothesize that the existence of 2 parallel food webs Balearic shearwaters use the same foraging grounds might be related to the heterogeneous distributions in the south Catalan Sea (Arcos et al. 2009). of phytoplankton groups above and below the ther- Although diet is the major determinant of fatty acid mocline during the warm season. profiles (Sargent 1997, Shirai et al. 2002), physiology Mediterranean phytoplankton is dominated by dino- may also play a role. For instance, air-breathing mar- flagellates and cyanobacteria during the thermal 10 Mar Ecol Prog Ser 531: 1–14, 2015

16 web. Although 20:5n-3 has a modest contri- bution to explaining the ob served variabil- 15 bluefish ity in the fatty acid profiles of most of the 14 species studied here, the highest levels blue shark leerfish were observed in sardines, a feature also 13 bonito reported elsewhere in the Mediterranean Sea (Zlatanos & Laskaridis 2007, Pethy- mackerel 12 swordfish N (‰) shortfin albacore pompano bridge et al. 2014). This suggests that dia -

15 squid δ 11 amberjack toms are a more relevant source of fatty acids for sardines than for other species of blue butterfish spearfish 10 small pelagic fishes in the region. This is little tunny bluefin tuna confirmed by the high relevance of diatoms dolphinfish 9 bullet tuna as prey for adult sardines during summer in the western Mediterranean revealed by 8 dietary studies (Costalago et al. 2012). 16 Nevertheless, the existence of 2 parallel 15 food webs may not persist year-round. Firstly, diatoms occur through the photic 14 neritic turtles zone in winter (Marty et al. 2002, Charles et al. 2005), which results in an increase in the 13 relative abundance of 20:5n-3 in anchovies striped dolphin and a fatty acid profile closer to that of sar- 12 Audouin’s gull N (‰) shag dines (Zlatanos & Laskaridis 2007, Pethy- 15

δ storm petrel 11 bridge et al. 2014). Secondly, sardines are yellow-legged gull more widespread over the continental shelf Balearic shearwater 10 in winter (Tugores et al. 2011, Giannoulaki Cory’s shearwater et al. 2013), which increases availability for 9 Mediterranean shearwater a broader spectrum of predators. Further- oceanic turtles 8 more, there is no evidence of coupling in 01020304050the population dynamics of small pelagic DHA/EPA fishes and their predators in the re gion. In a wasp-waist scenario, the populations of top Fig. 5. Distribution of large squids, blue shark and fishes (top panel) and air-breathing predators (bottom panel) in the space defined by the predators are expected to fluctuate accord- DHA/EPA index and δ15N. Arithmetic mean and standard deviation are ingly to changes in biomass of their prey shown for each species. The dashed box shows the region of the graph (Cury et al. 2000). Sardines, anchovies and where predators with a sardine/squid-based diet would be expected. horse mackerel experience high fluctua- Open triangles denote sharks and fishes, open circles seabirds, filled tri- tions in biomass in the south Catalan Seas, angles turtles, filled circles invertebrates and filled squares dolphins linked to variability in winter mixing and freshwater runoff (Lloret et al. 2004, Coll et stratification period (Barlow et al. 1997, Charles et al. al. 2008), but there is little evidence for a coupling 2005), with diatoms present only below the thermo - with predator biomass (Coll et al. 2008). This is prob- cline (Marty et al. 2008). Dinoflagellates and cyano- ably because each species of small pelagic fish fluc- bacteria are depleted in 20:5n-3 and enriched in tuates independently (Lloret et al. 2004, Coll et al. 22:6n-3 or 16:0 respectively, whereas the opposite is 2008) and most predators probably shift prey accord- true for diatoms (Viso & Marty 1993, Sargent 1997, Li ing to availability, except those seabirds highly et al. 1998, Patil et al. 2007). High levels of 22:6n-3 dependent on trawl discard (Navarro et al. 2009). have already been reported for anchovies and horse In conclusion, sardines are probably critical prey mackerel elsewhere in the Mediterranean Sea (Ban- during the summer for seabirds, mackerel, blue but- darra et al. 2001, Zlatanos & Laskaridis 2007, Fernan- terfish and oceanic loggerhead turtles, but not for the dez-Jover et al. 2007, Celik 2008, Ozogul et al. 2008, remaining predators in the system, which rely prima- Roncarati et al. 2012, Pethybridge et al. 2014) and rily on anchovies and horse mackerel. The situation likely reflect the prevalence of phytoplankton might be different in the cold season, when sardines enriched in 22:6n-3 at the base of the summer food are more widespread over the continental shelf. Cardona et al.: Mediterranean pelagic food webs 11

22:6n-3 enriched 20:5n-3 enriched

Bluefish and leerfish Striped dolphin

Blue shark

Bonito Shag Gulls

Swordfish and spearfish Mackerel and butterfish

Shearwaters Amberjack, pompano and dolphinfish

Albacore, bluefin tuna, little tunny and bullet tuna Oceanic loggerhead turtles

Squids Horse mackerel Anchovies Sardines Discarded sardines

Jellyfish Krill

Dinoflagellates and cyanobacteria Diatoms

Fig. 6. Idealized representation of the pelagic food webs in the western Mediterranean Sea. Arrows show the flow of matter and energy and not direct consumption. Trophic level increases from bottom to top and intermediate trophic levels may exists, particularly between phytoplankton and small pelagic fishes. Species not drawn to scale. Swordfish and gull images from OpenClipArt.org (https://openclipart.org). All other images from IAN/UMCES symbol and image libraries (http://ian.umces.edu/imagelibrary/) 12 Mar Ecol Prog Ser 531: 1–14, 2015

Acknowledgements. We thank Pablo R. Camarero (IREC) the picoplanktonic contribution. Estuar Coast Shelf Sci for his technical support in FAMEs analysis and data pro- 65: 199−212 cessing of chromatograms. Alex Aguilar and Asunción Bor- Christensen V, Guénette S, Heymans JJ, Walters CJ, Watson rell supplied the dolphin samples. R, Zeller D, Pauly D (2003) Hundred-year decline of North Atlantic predatory fishes. Fish Fish 4:1−24 Coll M, Palomera I, Tudela S, Sardà F (2006) Trophic flows, LITERATURE CITED ecosystem structure and fishing impacts in the South Catalan Sea, Northwestern Mediterranean. J Mar Syst Abelló P, Arcos JM, Gil de Sola L (2003) Geographical pat- 59: 63−96 terns of seabird attendance to a research trawler along Coll M, Palomera I, Tudela S, Dowd M (2008) Food-web the Iberian Mediterranean coast. Sci Mar 67 (Suppl 2): dynamics in the South Catalan Sea ecosystem (NW 69−75 Mediterranean) for 1978−2003. Ecol Modell 217:95−116 Alfaro AC, Thomas F, Sergent L, Duxbury M (2006) Identifi- Consoli P, Romero T, Battaglia P, Castriota V, Esposito V, cation of trophic interactions within an estuarine food Andaloro F (2008) Feeding habits of the albacore tuna web (northern New Zealand) using fatty acid biomarkers Thunnus alalunga (Perciformes, ) from cen- and stable isotopes. Estuar Coast Shelf Sci 70:271−286 tral Mediterranean Sea. Mar Biol 155: 113−120 Arcos JM, Oro D (2002) Significance of fisheries discards Costalago D, Navarro J, Álvarez-Calleja I, Palomera I (2012) for a threatened Mediterranean seabird, the Balearic Ontogenetic and seasonal changes in the feeding habits shearwater Puffinus mauretanicus. Mar Ecol Prog Ser and trophic levels of two small pelagic fish species. Mar 239: 209−220 Ecol Prog Ser 460: 169−181 Arcos JM, Bécares J, Rodríguez B, Ruiz A (2009) Áreas Cury P, Bakun A, Crawford RJM, Jarre A, Quiñones RA, importantes para la conservación de las aves marinas Shannon LJ, Verheye HM (2000) Small pelagics in en España. Sociedad Española de Ornitologia (SEO/ upwelling systems: patterns of interaction and structural BirdLife), Madrid changes in ‘wasp-waist’ ecosystems. ICES J Mar Sci 57: Bakun A (2006) Wasp-waist populations and marine eco- 603−618 system dynamics: navigating the ‘predator pit’ topo - Dalsgaard J, St. John M, Kattner G, Müller-Navarra D, graphies. Prog Oceanogr 68:271−288 Hagen W (2003) Fatty acid trophic markers in the pelagic Bandarra NM, Batista I, Nunes ML, Empis JM (2001) Sea- marine environment. Adv Mar Biol 46: 225−340 sonal variation in the chemical composition of horse- Darnaude AM, Salen-Picard C, Polunin NVC, Harmelin- mackerel (Trachurus trachurus). Eur Food Res Technol Vivien ML (2004) Trophodynamic linkage between river 212: 535−539 runoff and coastal fishery yield elucidated by stable iso- Barlow RG, Mantoura RFC, Cummings DG, Fileman TW tope data in the Gulf of Lions (NW Mediterranean). (1997) Pigment chemotaxonomic distributions of phyto- Oecologia 138: 325−332 during summer in the western Mediterranean. El-Sabaawi R, Dower JF, Kainz M, Mazumder A (2009) Deep-Sea Res II 44:833−850 Characterizing dietary variability and trophic positions Brodeur RD, Pearcy WG (1992) Effects of environmental of coastal calanoid copepods: insight from stable isotopes variability on trophic interactions and food web structure and fatty acids. Mar Biol 156: 225−237 in a pelagic upwelling ecosystem. Mar Ecol Prog Ser 84: Falautano M, Castriota L, Finoia MG, Andaloro F (2007) 101−119 Feeding ecology of little tunny Euthynnus alletteratus in Buchheister A, Latour RJ (2010) Turnover and fractionation the central Mediterranean Sea. J Mar Biol Assoc UK 87: of carbon and nitrogen stable isotopes in tissues of a 999−1005 migratory coastal predator, summer (Para - Fernandez-Jover D, Lopez Jimenez JA, Sanchez-Jerez P, lichthys dentatus). Can J Fish Aquat Sci 67: 445−461 Bayle-Sempere J, Gimenez Casalduero F, Martinez Lopez Cardona L, Hereu B, Torras X (2008) Juvenile bottlenecks FJ, Dempster T (2007) Changes in body condition and and salinity shape grey assemblages in Mediter- fatty acid composition of wild Mediterranean horse ranean estuaries. Estuar Coast Shelf Sci 77: 623−632 mackerel (Trachurus mediterraneus, Steindachner, 1868) Cardona L, Álvarez De Quevedo I, Borrell A, Aguilar A associated to sea cage fish farms. Mar Environ Res 63: (2012) Massive consumption of gelatinous plankton by 1−18 Mediterranean apex predators. PLoS ONE 7: e31329 Frisch AJ, Ireland M, Baker R (2014) Trophic ecology of Castriota L, Finoia MG, Campagnuolo S, Romeo T, Potoschi large predatory reef fishes: energy pathways, trophic A, Andaloro F (2008) Diet of Tetrapturus belone (Istio- level, and implications for fisheries in a changing cli- phoridae) in the central Mediterranean Sea. J Mar Biol mate. Mar Biol 161:61−73 Assoc UK 88: 183−187 Giannoulaki M, Iglesias M, Tugores MP, Bonanno A and oth- Celik M (2008) Seasonal changes in the proximate chemical ers (2013) Characterizing the potential habitat of Euro- compositions and fatty acids of chub mackerel (Scomber pean anchovy Engraulis encrasicolus in the Mediterran- japonicus) and horse mackerel (Trachurus trachurus) ean Sea, at different life stages. Fish Oceanogr 22:69−89 from the north eastern Mediterranean Sea. Int J Food Sci Gross T, Rudolf L, Levin SA, Dieckmann U (2009) Gen - Technol 43: 933−938 eralized models reveal stabilizing factors in food webs. Chaijan M, Benjakul S, Visessanguan W, Faustman C (2006) Science 325:747−750 Changes of lipids in sardine ( gibbosa) muscle Guitart R, Guerrero X, Silvestre AM, Gutiérrez JM, Mateo R during iced storage. Food Chem 99: 83−91 (1996) Organochlorine residues in tissues of striped Charles F, Lantoine F, Brugel S, Chrétiennot-Dinet MJ, dolphins affected by the 1990 Mediterranean epizootic: Quiroga I, Rivière B (2005) Seasonal survey of the phyto- relationships with the fatty acid composition. Arch Envi- plankton biomass, composition and production in a lit- ron Contam Toxicol 30:79−83 toral NW Mediterranean site, with special emphasis on Guitart R, Silvestre AM, Guerrero X, Mateo R (1999) Com- Cardona et al.: Mediterranean pelagic food webs 13

parative study on the fatty acid composition of two mar- Nyssen F, Brey T, Dauby P, Graeve M (2005) Trophic ine vertebrates: striped dolphins and loggerhead turtles. position of Antarctic amphipods—enhanced analysis by Comp Biochem Physiol B Biochem Mol Bio 124: 439−443 a 2-dimensional biomarker assay. Mar Ecol Prog Ser 300: Hobson KA, Clark RG (1992) Assessing avian diet using sta- 135−145 ble isotopes II: factors influencing diet-tissue fractiona- Olsen Y, Reinertsen H, Vadstein O, Andersen T and others tion. Condor 94:189−197 (2001) Comparative analysis of food webs based on flow Iglesias M, Carrera P, Muiño R (2003) Spatio-temporal pat- networks: effects of nutrient supply on structure and terns and morphological characterisation of multispecies function of coastal plankton communities. Cont Shelf Res pelagic fish schools in the north-western Mediterranean 21: 2043−2053 Sea. Aquat Living Res 16: 541−548 Oro D, Ruiz X, Jover L, Pedrocchi V, González-Solís J (1997) Iverson SJ, Frost KJ, Lang SLC (2002) content and fatty Diet and adult time budgets of Audouin’s gull Larus acid composition of and invertebrates in audouinii in response to changes in commercial fisheries. Prince William Sound, Alaska: factors contributing to Ibis 139: 631−637 among and within species variability. Mar Ecol Prog Ser Ozogul Y, Duysak O, Ozogul F, Özkütük AS, Türeli C (2008) 241: 161−181 Seasonal effects in the nutritional quality of the body Iverson SJ, Field C, Bowen WD, Blanchard W (2004) Quan- structural tissue of cephalopods. Food Chem 108: 847−852 titative fatty acid signature analysis: a new method of Paine RT (2010) Food chain dynamics and trophic cascades estimating predator diets. Ecol Monogr 74: 211−235 in intertidal habitats. In: Terborgh J, Estes JA (eds) Li R, Yokota A, Sugiyama J, Watanabe M, Hiroki M, Watan- Trophic cascades. Island Press, , DC, p 21−53 abe MM (1998) Chemotaxonomy of planktonic cyano- Palomera I, Olivar MP, Salat J, Sabatés A, Coll M, García A, bacteria based on non-polar and 3-hydroxy fatty acid Morales-Nin B (2007) Small pelagic fish in the NW Medi- composition. Phycol Res 46: 21−28 terranean Sea: an ecological review. Prog Oceanogr 74: Lloret J, Palomera I, Salat J, Solé I (2004) Impact of fresh- 377−396 water input and wind on landings of anchovy (Engraulis Patil V, Kallqvist T, Olsen E, Vogt G, Gislerød HR (2007) encrasicolus) and sardine (Sardina pilchardus) in shelf Fatty acid composition of 12 microalgae for possible use waters surrounding the Ebre (Ebro) River delta (north- in feed. Aquacult Int 15: 1−9 western Mediterranean). Fish Oceanogr 13: 102−110 Perga ME, Kainz M, Matthews B, Mazumder A (2006) Car- Madigan DJ, Carlisle AB, Dewar H, Snodgrass OE, Litvin bon pathways to zooplankton: insights from the com- SY, Micheli F, Block BA (2012) Stable isotope analysis bined use of stable isotope and fatty acid biomarkers. challenges wasp-waist food web assumptions in an up - Freshw Biol 51: 2041−2051 welling pelagic ecosystem. Sci Rep 2:654 Pethybridge H, Bodin N, Arsenault-Pernet EJ, Bourdeix JH Marty JC, Chiaverini J, Pizay M, Avril B (2002) Seasonal and and others (2014) Temporal and inter-specific variations interannual dynamics of nutrients and phytoplankton in forage fish feeding conditions in the NW Mediterran- pigments in the western Mediterranean Sea at the ean: lipid content and fatty acid compositional changes. DYFAMED time-series station (1991−1999). Deep-Sea Mar Ecol Prog Ser 512: 39−54 Res II 49:1965−1985 Pinnegar JK, Polunin NVC (1999) Differential fractionation Marty JC, Garcia N, Raimbault P (2008) Phytoplankton of δ13C and δ15N among fish tissues: implications for the dynamics and primary production under late summer study of trophic interactions. Funct Ecol 13: 225−231 conditions in the NW Mediterranean Sea. Deep-Sea Res Power ME (1992) Top-down and bottom-up forces in food I 55: 1131−1149 webs: Do plants have primacy? Ecology 73: 733−746 Massutí E, Deudero S, Sánchez P, Morales-Nin B (1998) Diet Ramos R, González-Solís J (2012) Trace me if you can: the and feeding of dolphin (Coryphaena hippurus) in the use of intrinsic biogeochemical markers in marine top western Mediterranean waters. Bull Mar Sci 63: 329−341 predators. Front Ecol Environ 10:258−266 Mateo R, Beyer WN, Spann JW, Hoffman DJ (2003) Relation Ramos R, Ramírez R, Sanpera C, Jover L, Ruiz X (2009) Diet of fatty acid composition in lead-exposed mallards to of yellow-legged gull (Larus michahellis) chicks along fat mobilization, lipid peroxidation and alkaline phos- the Spanish Western Mediterranean coast: the relevance phatase activity. Comp Biochem Physiol C Toxicol Phar- of refuse dumps. J Ornithol 150: 265−272 macol 135: 451−458 Robin JH, Regost C, Arzel J, Kaushik SJ (2003) Fatty acid Mateo R, Gil C, Badia-Vila M, Gitart R, Hernández-Matías profile of fish following a change in dietary fatty acid A, Sanpera C, Ruiz X (2004) Use of fatty acids to explain source: model of fatty acid composition with a dilution variability of organochlorine concentrations in eggs and hypothesis. Aquaculture 225: 283−293 plasma of common terns (Sterna hirundo). Ecotoxicology Romeo T, Consoli P, Castriota L, Andaloro F (2009) An eval- 13: 545−554 uation of resource partitioning between two , Mayzaud P, Virtue P, Albessard E (1999) Seasonal variations Tetrapturus belone and Xiphias gladius, in the central in the lipid and fatty acid composition of the euphausiid Mediterranean Sea. J Mar Biol Assoc UK 89:849−857 Meganyctiphanes norvegica from the Ligurian Sea. Mar Roncarati A, Brambilla G, Meluzzi A, Iamiceli AL and others Ecol Prog Ser 186: 199−210 (2012) Fatty acid profile and proximate composition of Mostarda E, Campo D, Castriota L, Esposito V, Scarabello fillets from Engraulis encrasicholus, Mullus barbatus, MP, Andaloro F (2007) Feeding habits of the bullet tuna and Sarda sarda caught in Tyr - Auxis rochei in the southern Tyrrhenian Sea. J Mar Biol rhenian, Adriatic and Ionian seas. J Appl 28: Assoc UK 87:1007−1012 545−552 Navarro J, Louzao M, Igual JM, Oro D and others (2009) Saraux C, Fromentin JM, Bigot JL, Bourdeix JH and others Seasonal changes in the diet of a critically endangered (2014) Spatial structure and distribution of small pelagic seabird and the importance of trawling discards. Mar fish in the Northwestern Mediterranean Sea. PLoS ONE Biol 156: 2571−2578 9: e111211 14 Mar Ecol Prog Ser 531: 1–14, 2015

➤ Sargent JR (1997) Fish oils and human diet. Br J Nutr ➤ Stowasser G, Atkinson A, McGill RAR, Phillips RA, Collins 78(Suppl 1):S5−S13 MA, Pond DW (2012) Food web dynamics in the Scotia ➤ Shannon LJ, Cury PM, Jarre A (2000) Modelling effects of Sea in summer: a stable isotope study. Deep-Sea Res II fishing in the Southern Benguela ecosystem. ICES J Mar 59-60: 208−221 Sci 57:720−722 ➤ Tugores MP, Giannoulaki M, Iglesias M, Bonanno A and Shealer D (2001) Foraging behavior and food of seabirds. In: others (2011) Habitat suitability modelling for sardine Schreiber EA, Burger J (eds) Biology of marine birds. Sardina pilchardus in a highly diverse ecosystem: the CRC Press, Boca Raton, FL, p 137−178 Mediterranean Sea. Mar Ecol Prog Ser 443: 181−205 ➤ Shirai N, Terayama M, Takeda H (2002) Effect of season on ➤ Viso AC, Marty JC (1993) Fatty acids from 28 marine micro- the fatty acid composition and free amino acid content algae. Phytochemistry 34:1521–1533 of the sardine melanostictus. Comp Biochem ➤ Ware DM, Thomson RE (2005) Bottom-up ecosystem trophic Physiol B 131:387−393 dynamics determine fish production in the Northeast Sinopoli M, Pipitone C, Campagnoulo S, Campo D, Castriota Pacific. Science 308: 1280−1284 L, Mostarda E, Andaloro F (2004) Diet of young-of-the- ➤ Zlatanos S, Laskaridis K (2007) Seasonal variation in the fatty year bluefin tuna, Thunnus thynnus (Linnaeus, 1758), in acid composition of three Mediterranean fish—sardine the southern Tyrrhenian (Mediterranean). Sea J Appl (Sardina pilchardus), anchovy (Engraulis encrasicholus) Ichthyol 20:310−313 and picarel (Spicara smaris). Food Chem 103:725−728

Editorial responsibility: Katherine Richardson, Submitted: December 1, 2014; Accepted: May 13, 2015 Copenhagen, Denmark Proofs received from author(s): June 19, 2015