*2.2. Field Collections*

Samples of basal resources (primary producers and detritus), invertebrates and fish were collected in 4 sites per lake between April and May 2012, when primary productivity and invertebrate abundances were high. The sampling sites within each lake were selected from areas with heterogeneous physical and biotic characteristics and a range of anthropogenic impacts deriving from the surrounding areas [2,3]. The sampling sites were located at the northern and southern ends of each lake, and both on the landward and seaward sides (see also Santoro et al. [2]). Macrophytes, algae, and detritus samples were collected by hand and invertebrates by Van Veen grab (volume: 3.5 L) in three replicates per sampling site. The dominant macrophytes were *Ruppia* sp. and *Cymodocea nodosa* (Ucria) Ascherson, while the macroalgae were represented by taxa of the genera *Chetomorpha*, *Chondria*, *Gracilaria*, *Rytiphloea* and *Ulva*. The detritus was mostly composed of fragments of dead leaves delicately scraped to remove any epibionts and rinsed in distilled water. Phytoplankton samples were collected using a plankton net (20-μm mesh size) and concentrated by centrifugation (2000 rpm for 20 min).

Samples of fish were collected once a day for 3 days in each site. In order to collect pelagic, benthic, resident and migratory fish species, fish samples were collected using fixed weirs and fishing traps placed on the bottom. The fishing traps, made of very fine mesh (0.5 cm), were 1.5 m in diameter at the mouth and were composed of four consecutive chambers of decreasing diameter with a total length of 3.6 m. In addition to *A. anguilla* and *D. annularis*, the sampled fish community included the sand smelt *Atherina boyeri* (Risso, 1810), black goby *Gobius niger* (Linnaeus, 1758) and the mullets *Chelon ramada* (Risso, 1827), *C. aurata* (Risso, 1810), *C. saliens* (Risso, 1810) and *C. labrosus* (Risso, 1827), which are known to be prey species of *A. anguilla* and *D. annularis* [19,24,25,42,43]. Further fish samples included species belonging to the Sparidae, Scorpaenidae, Clupeidae, Cyprinodontidae, Blenniidae and Belonidae, Gobiidae, Labridae, Moronidae, Mugilidae, Soleidae and Syngnathidae families. Standard length measured in centimetres was recorded for each fish specimen. For each fish species, individuals of different sizes were collected in order to reduce the effects of size variability on isotopic signals. From the sampled fish specimens, including *A. anguilla* and *D. annularis*, samples of dorsal white muscle were taken. This tissue provides a long-term (several months) integrated indicator of food sources due to its slow turnover with respect to other tissues (e.g., liver and blood) [32].

After collection, all samples were transported to the laboratory, where specimens were sorted, counted, and identified to the lowest possible taxonomic level and processed for the stable isotope analysis.

## *2.3. Stable Isotope Analysis (SIA)*

Samples were individually stored at −80 ◦C and freeze-dried for 24 h. Fish specimens were considered individually for isotopic analysis. Muscle samples were also taken from large invertebrates such as crustaceans, for which the tissue was taken from the claws, and bivalves and sea snails, whose tissue was taken from the feet [7]. When present, shells, valves and other exoskeletal parts of animals were removed under dissection microscopes in order to avoid tissue acidification before the stable isotope analysis. For small invertebrates (such as amphipods and polychaetes), the whole body was used. Samples were individually analysed. Plankton biomass was analysed as a whole due to the di fficulty of obtaining su fficient biomass for isotopic analysis.

Before the stable isotope analysis, each sample was homogenised to a fine powder using a ball mill (Mini-Mill Fritsch Pulverisette 23: Fritsch Instruments, Idar-Oberstein, Germany). When necessary, samples were pre-acidified using 1M HCl according to the drop-by-drop method [48] in order to eliminate inorganic carbon and re-dried (60 ◦C) for 72 h to remove the remaining moisture. δ<sup>15</sup> N signatures were measured in un-acidified powders to prevent acidification from interfering with the nitrogen analysis [30,49].

Aliquots of 0.25 ± 0.10 mg for the animals and 2.00 ± 0.10 mg for basal resources were placed into tin capsules for C and N stable isotopic analysis (SIA). Each sample was analysed in two replicates. The analyses were carried out using a continuous flow mass spectrometer (IsoPrime100, Isoprime Ltd., Cheadle Hulme, United Kingdom) coupled with an elemental analyser (Elementar Vario Micro-Cube, Elementar Analysensysteme GmbH, Germany).

The isotopic signatures of each sample were expressed in δ units (δ15N; δ13C) as parts per thousand (%) deviations from international standards (atmospheric N2 for N; PD-belemnite (PDB) carbonate for C), in accordance with the formula:

δX (%) = [(Rsample − Rstandard)/Rstandard] × 10<sup>3</sup> [50], where X is 13C or 15N and R is the corresponding ratio of heavy to light isotope for the element (13 C/12C or 15 N/14N). Outputs were standardised with the internal laboratory standard Ca ffeine IAEA-600 (C8H10 N4O2). Measurement errors were found to be typically smaller than 0.05%.
