Method step description:

1. A systematic literature search was conducted according to the PRISMA methodology (Moher et al., 2009). We collected studies using the web database Google Scholar (https://scholar.google.com/), dividing the search into two phases. In the first phase, we queried the database for research and review articles published between 2015 and 2025, using the filter embedded in the Google Scholar page option panels. The query was structured as follows: Mediterranean trophic sea OR food OR web OR diet OR prey OR stomach OR gut -freshwater -lake -river. From the resulting list, we selected studies reporting stomach content data of Mediterranean marine organisms. We focused on studies in English, French, Italian, and Spanish that used qualitative or quantitative data and had a clear taxonomic scope. On the contrary, we excluded reviews without primary data references, studies focused on aquaculture or freshwater species, and papers with unclear or non-standard dietary methods (e.g., emetics) (Kamler & Pope, 2001). In the second phase, given the recognized importance of fish in structuring the marine food webs (Pauly et al., 1998) and the abundance of literature in the Mediterranean area (Karachle & Stergiou, 2017), we queried again the web database by extending the period of study condition and incorporating species from a list of 759 fish species recorded in the Mediterranean Sea (Kovačić et al., 2021). The inclusion and exclusion criteria remained the same as in the first phase. We prioritized studies conducted in the Mediterranean Sea, but we also included relevant studies conducted in other regions. For each study, the species name, sampling location, and sample size were recorded. When available, the species’ length data was also recorded. Then, dietary data were extracted from the studies, including the prey’s scientific name, and their relative importance (e.g., frequency of occurrence, weight percentage, and number percentage). The locations, longitudes, and latitudes of the sampling sites listed in the original studies have been reported for each data entry. When the original studies reported multiple sampling points, the midpoint coordinates of the indicated region were entered using either the region name or the sampling-point map. The sampling year was reported for each data entry (Format YYYY). When the original studies reported several years, the truncated midyear within the covered range was reported.