Introduction

From an ecological perspective, sharks and rays can be keystone species (i.e., species that despite having relatively low biomass have a disproportional effect on the food web; Power et al. 1996). Marine habitats are often characterized by strong top-down interactions (Power et al. 1996; Valls et al. 2015), and chondrichthyans play central roles with effects that can propagate through the food web (Baum and Worm 2009; Bornatowski et al. 2014). As a consequence, any change in their abundance and distribution can have widespread consequences along food webs, potentially impacting their prey, coexisting predators, and competitors (Dunne 2009). Moreover, the ecological effects of reducing population numbers or eliminating top predators such as large sharks can trigger cascading effects that travel both up and down marine food webs (Bornatowski et al. 2014; McCauley et al. 2015). However, those effects can be attenuated in highly diverse systems (Grubbs et al. 2016; Rasher et al. 2017), where prey-predator interactions are weaker and bottom-up drivers can have larger impacts (Desbiens et al. 2021).

After the depletion of key species, the likelihood of trophic cascades will depend on the degree of omnivory and complexity of the system (Bascompte et al. 2005; Grubbs et al. 2016). The role of these indirect effects is crucial to understanding food web energy flow and the structure and functioning of ecosystems (Bornatowski et al. 2014). Although food webs have shown to be resilient to the removal of random nodes (Montoya et al. 2006), the loss of species as a consequence of anthropogenic stressors tends to be directed towards species that play key roles in ecosystems and are unable to adapt to the respective pressure (Bascompte et al. 2005), which is the case for most chondrichthyans. As they are important predators (i.e. they are connected to many species), predicting the effects of their removal is complex (Field et al. 2009). In fact, the strength of predator–prey interactions influences the stability of marine communities (Bascompte et al. 2005) and, thus, effective management of shark populations should consider how different drivers affect particular species declines (Field et al. 2009). However, determining how individual traits and interactions among species contribute to community functioning and ecosystem resilience is still a challenge (Marbà and Coll 2021).

As predators, chondrichthyans developed a range of feeding strategies, which vary from scavenging to active predation and from opportunism to high specialization, and thus, they consume a wide prey spectrum, extending from plankton to marine mammals. As an important ecological trait, diet affects many biological aspects of a species, including its vulnerability to extinction (Machado et al. 2022). In general, chondrichthyans have a long-lived/K-selected life history strategy (Stevens et al. 2000; Ferretti et al. 2010; Hutchings et al. 2012). This is linked to low population recovery rates following human disturbance, and therefore a high extinction risk, which, combined with the global increase in fishing effort (considered the main threat to chondrichthyans), has triggered the depletion of populations of these marine predators around the world (Aldebert 1997; Worm et al. 2013; Walker et al. 2021). Particularly in the Mediterranean Sea, where fleet is multi-species and uses a large variety of fishing gears, chondrichthyan species are highly vulnerable (Cavanagh and Gibson 2007), with reported impacts from small demersal to large pelagic species (Ferretti et al. 2008; Cartes et al. 2013; Nuez et al. 2021; Ruiz-García et al. 2023). Indeed, the Mediterranean Sea has been identified as one of the main hotspots where the biodiversity of sharks and rays is particularly threatened (Field et al. 2009; Dulvy et al. 2014). Furthermore, the impacts of fishing are, in most cases, combined with the synergistic effects of climate change and habitat loss (Dulvy et al. 2014, 2021). Cumulative impacts range from spatial distribution shifts due to ocean warming and habitat degradation (Fowler et al. 2005), to direct impacts on the physiology, behavior, and survival due to other anthropogenic stressors (such as ocean acidification and pollution; Walker et al. 2021). These can lead to population declines both in terms of abundance and viability (Wheeler et al. 2020; Jorgensen et al. 2022). According to the most recent International Union for Conservation of Nature (IUCN) assessment for the Mediterranean, half of the chondrichthyan species there are threatened, and a quarter of the species have not been assessed or have been classified as “Data Deficient (DD)”, highlighting a risky combination of high threat, low safety and high uncertainty in the threat status of sharks and rays (Dulvy et al. 2021).

The decline of chondrichthyan populations has left a handful of ecological systems worldwide where their ecological role is still functional, particularly in the case of large species that usually act as top predators. This limitation, coupled with the challenges associated with targeting, handling, or conducting experiments on these species, highlights the need of theoretical approaches to understand their ecological roles. The main objective of this study was to investigate the ecological role of chondrichthyans inhabiting the Mediterranean Sea, focusing on their trophic (prey-predator) interactions and the consequences of their extinction at the food web level. This overarching goal was divided into three specific objectives: (1) to review the current knowledge and gaps in the trophic ecology of chondrichthyans in the Mediterranean Sea, (2) to represent the chondrichthyan species within the Mediterranean Sea food web, and (3) to quantify the vulnerability of the Mediterranean food web to selective removals of different sharks and rays’ groups by applying a qualitative network approach. This study contributes knowledge to the urgent need for an ecosystem-based management approach to halt and prevent Mediterranean chondrichthyan populations from local extinctions, with large direct and indirect ecosystem implications.

Material and methods

We first reviewed all available published data on the feeding ecology and conservation status of Mediterranean chondrichthyans, followed by an examination of the multivariate structure of their diets (Fig. 1). We then used a meta-web approach to investigate the topology of the food web in the basin, with an emphasis on chondrichthyan groups (i.e., different functional groups according to body size, habitat type, and conservation status). Finally, we tested different extinction scenarios to examine the ecosystem consequences of removing threatened shark and ray groups.

Fig. 1
figure 1

Overview of the methodology followed in the present study to investigate the ecological role of chondrichthyans inhabiting the Mediterranean Sea

Full size image

Bibliographic review of conservation status and trophic information

An updated list of chondrichthyan species inhabiting the Mediterranean was compiled including frequent, rare, and vagrant species reported in previous occurrence lists from Dulvy et al. (2016) and Serena et al. (2020). Each species was then classified according to its conservation status and population trend, both at the Mediterranean and global level based on IUCN assessments (IUCN 2021), adding the predicted status by Walls and Dulvy (2020) for Data Deficient species. Centrophorus granulosus diets were reclassified as Centrophorus uyato according to White et al. (2022).

To model the Mediterranean food web, all available information on trophic links among marine species was compiled through an extensive literature review of the diet of the species inhabiting the Mediterranean Sea, including all the available data up to the cutoff date, December 2021. This review updated previous efforts by Stergiou and Karpouzi (2002) and Karachle and Stergiou (2017), and the compilation previously built to develop a Mediterranean meta-web (Coll et al. 2019b), which was substantially complemented with additional references. For each selected publication, we recorded:

  1. 1.

    Species taxonomy;

  2. 2.

    Information related to the life stage: juveniles or adults;

  3. 3.

    Spatial information: divisions of the Mediterranean Sea were defined following Notarbartolo di Sciara and Agardy (2010): Alboran Sea, Algero-Provençal basin, Tyrrhenian Sea, Adriatic Sea, Strait of Sicily/Tunisian Plateau/Gulf of Sirte, Ionian Sea, Aegean Sea and Levantine Sea;

  4. 4.

    Temporal information: years, season, and months of sampling; and

  5. 5.

    Information related to the diet: prey species, diet metrics (contribution by weight, %W; contribution by number, %N; contribution by volume, %V; frequency of occurrence, %F; Index of Relative Importance, %IRI; mean contribution to the Stable Isotope Analysis, SIA hereinafter) and type of analysis (e.g., stomach content analysis or SIA).

Species identified in the literature and coded as prey or predators in the database were also classified in terms of life history and distribution using information from FishBase (Froese and Pauly 2021) and SeaLifeBase (Palomares and Pauly 2021). Once the database was completed, the distribution of chondrichthyan trophic studies was examined by predator species and families, conservation status, area and years of the study, population life stage, and methodology used.

Trophic ecology of Mediterranean chondrichthyans

An overview of principal prey groups of the chondrichthyan species was developed using a quantitative approach. To standardize all the reports, we gathered all the trophic indexes: when the %W of prey was not available (the main metric to characterize diet), the %V was used, followed by the mean contribution based on stable isotopes mixing models, %IRI, %F, and %N. To compare diet composition among species, all prey taxa were categorized as ‘Marine mammals’, ‘Seabirds’, ‘Sea turtles’, ‘Chondrichthyans’, ‘Teleosts’, ‘Cephalopods’, ‘Mollusks’, ‘Crustaceans’, ‘Polychaeta’, ‘Other invertebrates’ and ‘Seagrasses and algae’. Chondrichthyan species were classified according to the main habitats in which they occur (pelagic, benthopelagic or demersal) and according to body size (small, medium or large). A hierarchical clustering analysis using the maximum length (ML) reported in FishBase (Froese and Pauly 2021) was used to define the size classes. The clustering analysis resulted in 6 groups: we considered as ‘small’ those species with ML < 150 cm (cluster 1, n = 34), ‘medium’ between 150 and 250 cm (cluster 2, n = 20), and the remaining clusters were grouped into ‘large’ ≥ 250 cm (n = 27) in order to have similar sample sizes. The overview analyses of the diet were conducted in R (R Core Team 2020) and figures were produced using the ggplot2 package (Wickham 2016).

Using the composition of chondrichthyan diets, we created a prey-predator diet matrix where the columns represented predator species and rows represented prey groups. Different statistical approaches were used to compare the trophic ecology of the chondrichthyan species. The diet matrix was fourth-root transformed and converted to a resemblance matrix using the Bray–Curtis similarity. The multivariate structure of chondrichthyans trophic ecology was then examined using non-metric Multidimensional Scaling analysis (nMDS). Permutational multivariate analysis of variance (PERMANOVA) tests were applied to test for differences among taxonomic groups (Selachimorpha hereinafter referred to as sharks, Batoidea hereinafter referred to as rays, and Holocephali hereinafter referred to as chimaeras), sizes (small, medium, or large), and habitats (pelagic, benthopelagic, or demersal). The homogeneity of multivariate dispersions was tested for all factors using the permutational analysis of multivariate dispersions (PERMDISP; Anderson 2006). When the evidence of difference was strong (i.e., p value < 0.05), pairwise tests were performed, and a similarity percentage procedure (SIMPER) was used to identify discriminant prey groups, indicating the average contribution of prey to the dissimilarity between groups. For PERMANOVA tests, significance was determined using unrestricted permutation of the raw data with 9999 permutations. PERMANOVA, PERMDISP, SIMPER, and nMDS tests were conducted in PRIMER version 7 software (Clarke and Gorley 2015) and the PERMANOVA + add-in (Anderson et al. 2015).

Mediterranean meta-web analysis

A meta-web is defined as a compilation of species and their potential feeding interactions within a specific geographical area and time period, which does not represent observed realizations of trophic interactions at a given time step (Kortsch et al. 2021). However, understanding all trophic interactions in a food web is only possible in simplified communities (Sala 2004). Following a previously established methodology (Planque et al. 2014), we used the information obtained from the literature review of chondrichthyan species to modify the structure of the available Mediterranean meta-web (Coll et al. 2019b). All species identified as predators or prey in our trophic database were classified into functional groups representing the different trophic levels of the marine ecosystem: from primary producers to top predators, based on previous Mediterranean food-web models (Corrales et al. 2015; Piroddi et al. 2015; Coll et al. 2019b). In addition to the taxonomic group, the body size, and the habitat type, chondrichthyan species were classified considering their conservation status, dividing them between low extinction risk (Least Concern, LC; and Near Threatened NT) and high extinction risk (Vulnerable, VU; Endangered, EN; and Critically Endangered, CR). These groups were then used to perform the extinction simulations. A detailed description of the functional groups is given in Online Resource 1.

The meta-web was analyzed using a binary network model that considered only the presence/absence of prey species in the diet of predators to reduce bias caused by the usage of different diet indexes (e.g., %N or %IRI). Qualitative models provide a useful approach to assessing the potential importance of species in ecosystems, which requires fewer assumptions and has lower data requirements for parametrization than quantitative models (Dunne et al. 2008). Although they are conservative approaches, results in terms of food web degradation can be informative on ecosystem structure and functioning when used in relative terms (Coll et al. 2008, 2019a; Lotze et al. 2011). The meta-web was represented using the cheddar package (v0.1-636; Hudson et al. 2013, 2020) in R (R Core Team 2020), which provides a flexible and extendable representation of an ecological community and a range of functions for analysis and visualization.

Extinction scenarios

To compare the influence of the chondrichthyan species on the community by habitat, taxonomic group, and size, we tested eight scenarios of the functional extinction of chondrichthyan groups (Table 1) by using qualitative modelling (Hudson et al. 2013).

Table 1 Description of the different scenarios tested for the functional extinction of chondrichthyan groups
Full size table

To quantify changes in community structural complexity, we calculated linkage density (L/N, where L is the number of links and N is the number of nodes), and connectance (measured as the proportion of all possible trophic links that are actually realized, L/N2). The depletion of highly connected species has major implications on network stability, and connectance is therefore used as an indicator of robustness and structural complexity (Gilbert 2009; Sánchez-Carmona et al. 2013; Bornatowski et al. 2014). Omnivory, which is related to the magnitude and likelihood of trophic cascades (Bascompte et al. 2005), was measured as the proportion of species feeding at multiple trophic levels. In addition, trophic similarity, which measures trophic overlap among functional groups, was averaged across each species’ highest trophic similarity to another species to determine the mean maximum trophic similarity (Dunne et al. 2004).

The calculation of food web properties was based on a binary network model that considered the presence or absence of prey species in the diet of predators. Additionally, the strength of predation based on diet data (quantitative) was used to identify the least certain links (i.e., trophic links with the weakest strength of interaction). In order to assess if the changes in the properties were consistent when the least certain links were eliminated, we ran the simulations (Table 1) with the full dataset, without the trophic links with a strength of interaction lower than 0.5%, and without the trophic links with a strength of interaction lower than 1%.

Results

Bibliographic review of conservation status and trophic information

A total of 81 species from 27 different families were considered: 34 species of rays, 46 species of sharks, and 1 chimaera (Chimaera monstrosa; Table S1). We identified 119 studies describing the trophic ecology of 53 different species, including 61% of the sharks, 71% of the rays, and 100% of the chimaeras’ species (see Online Resource 2). The most studied Families were the rays of the family Rajidae (30% of the reports) and the sharks of the family Scyliorhinidae (24%). The most studied species were the small-spotted catshark (Scyliorhinus canicula) (18%), the blackmouth catshark Galeus melastomus (15%), and the thornback ray (Raja clavata) (13%).

According to the Mediterranean assessment by the IUCN, half of the chondrichthyan species are threatened, with 25% of them classified as CR, which increases to 30% with Walls and Dulvy (2020) predicted status for DD species. When compared to their global status, the percentage of threatened species was similar (53% versus 57% in the Mediterranean), but the extinction threat was lower globally, with more species classified as VU but fewer species considered CR (14% and 30% at the global and Mediterranean assessment, respectively). Although the number of species classified as threatened in the Mediterranean was high, these only accounted for 37.6% of the reported diets (Fig. 2).

Fig. 2
figure 2

Percentage of Mediterranean species (left) and trophic studies (right), by taxonomic group and IUCN Red List status according to the Mediterranean assessment (IUCN 2021) and Walls and Dulvy (2020). CR, Critically Endangered; EN, Endangered; VU, Vulnerable; NT, Near Threatened; LC, Least Concern; DD, Data Deficient; NA, Not Assessed

Full size image

Chondrichthyan trophic information was heterogeneously distributed between Mediterranean areas, with most studies focusing on the Algero-Provençal basin, the Aegean Sea, and the Strait of Sicily/Tunisian Plateau/Gulf of Sirte (Fig. 3). Information was scarce regarding the Alboran Sea, with only one published study.

Fig. 3
figure 3

Number of studies reporting chondrichthyans diet information per area between 1973 and 2021. Divisions of the Mediterranean Sea were defined following Notarbartolo di Sciara and Agardy (2010): Alboran Sea (ALBO), Algero-Provençal Basin (ALPR), Tyrrhenian Sea (TYRR), Adriatic Sea (ADRI), Strait of Sicily/Tunisian Plateau/Gulf of Sirte (STPS), Ionian Sea (IONI), Aegean Sea (AEGA) and Levantine Sea (LEVA)

Full size image

Regarding the temporal distribution of trophic information, most of the sampling of Mediterranean sharks, rays, and chimaeras started between 2000 and 2010 (48%), with the first study beginning in 1973 (Fig. 4). Only a few articles included trophic information about chondrichthyans before the 1990s (12.5%). Different methodologies have been used to describe chondrichthyan diets, but the most used method was stomach content analysis (SCA), being used in 90.3% of the studies, followed by SIA (5.6%) and other methods (e.g., visual observation; 4.1%). The studies included different life stages: 17.5% of them described juveniles’ diets and 22.1% described adults’ diets, although most of the studies gathered information from adults and juveniles together (57.1%).

Fig. 4
figure 4

Number of studies reporting chondrichthyans diet information per year and methodology. SIA, stable isotopes analysis; SCA, stomach content analysis

Full size image

Trophic ecology of Mediterranean chondrichthyans

All prey group categories considered were reported in the diet of at least one species. On a presence-absence basis, fish were the most frequent prey, being present in 94% of the species’ diets, followed by cephalopods (91%) and crustaceans (80%). In contrast, marine mammals, seabirds, and sea turtles were present in less than 10% of the diets.

Trophic differences were observed between demersal sharks and demersal rays, with crustaceans being more consumed by rays, while cephalopods being more consumed by demersal sharks (Fig. 5). Cephalopods also made up an important part of pelagic and benthopelagic sharks’ diets, except for the white shark (Carcharodon carcharias), whose diet contained mostly fish and other vertebrates (e.g., marine mammals). In the case of benthopelagic rays, mollusks and crustaceans played particularly important roles in the diet of myliobatids [the bull ray (Aetomylaeus bovinus) and the common eagle ray (Myliobatis aquila)], while the electric ray (Tetronarce nobiliana) and the pelagic rays, [the devil fish (Mobula mobular) and the pelagic stingray (Pteroplatytrygon violacea)], preyed mainly on fish (Fig. 5). The diet of rabbitfish (Chimaera monstrosa), the only chimera present in the Mediterranean, was dominated by crustaceans.

Fig. 5
figure 5

Prey contribution to the diet of Mediterranean chondrichthyan species summarized from the diet literature review. References included are listed in Online Resource 2

Full size image

The results of the multidimensional scaling analysis did not show well-defined clusters (Fig. 6). However, shark species were more related to a diet composed of cephalopods and less of crustaceans than rays, except for the starry smooth-hound (Mustelus asterias), which mainly feeds on crustaceans, and the angular roughshark (Oxynotus centrina), in whose diet polychaetes and mollusks were more abundant (Fig. 5). This was in accordance with the SIMPER results (Table S2), which highlighted crustaceans and cephalopods as the main groups contributing to the dissimilarity between sharks and rays. Chimaeras were grouped next to demersal rays, which mainly preyed on mollusks and crustaceans [e.g., the common stingray (Dasyatis pastinaca)]. Mollusks were the main prey causing dissimilarity between chimaeras from one side and sharks and rays from the other (Table S2).

Fig. 6
figure 6

NMDS plot of 53 chondrichthyan species diets grouped according to their taxonomic group and habitat. The plots were generated using Bray–Curtis similarity (2D-stress = 0.17). Prey taxa are identified as: CEPH, Cephalopods; CHON, Chondrichthyan; CRUS, Crustaceans; MOLL, Mollusks; INVE, Other invertebrates; POLY, Polychaeta; SALG, Seagrasses and algae; TELE, Teleosts; VERT, Other vertebrates

Full size image

Accordingly, our results revealed strong evidence for differences in prey composition among taxonomic groups, but there was little or no evidence for a relationship between main prey groups and chondrichthyan habitat (pelagic, benthopelagic and demersal) or size (small, medium and large; Table 2). Pairwise comparisons showed strong evidence for prey composition differences between sharks and rays (p < 0.01). The interaction between factors was not significant. PERMDISP results showed that the differences obtained with PERMANOVA were not due to multivariate dispersion (p > 0.05 for all factors).

Table 2 Summary of PERMANOVA tests examining differences between taxonomic group, habitat, and size (and their interactions) in prey composition of 53 chondrichthyan species diet
Full size table

Mediterranean meta-web and extinction scenarios

A total of 612 studies were included in the Mediterranean meta-web. The meta-web included information on 401 different predator species and a total of 1726 taxa were identified as prey, with the resulting topology of 1508 trophic links between 79 nodes (Table 3, Fig. 7). A detailed description of the meta-web including functional groups description, trophic links, and references is presented in Online Resource 1. Chondrichthyan species were classified in 19 nodes with 411 links, representing 27% of the total trophic links and occupying high trophic levels (Fig. 7).

Table 3 Functional groups of the meta-web of the Mediterranean Sea. Chondrichthyans are represented in dark blue, teleosts in light blue, other vertebrates in purple, cephalopods in orange, other invertebrates in red, producers in green, and others in black. (For interpretation of the references to color, the reader is referred to the web version of this article.)
Full size table
Fig. 7
figure 7

Representation of the Mediterranean meta-web with emphasis on chondrichthyan species. Functional group numbers are presented in Table 3, but see Online Resource 1 for a detailed description of the meta-web

Full size image

Results from the extinction simulations showed values of connectance, omnivory, and trophic similarity within the range described for marine food webs (Fig. 8 and Tables S3, S4, and S5). Major changes were detected when all threatened chondrichthyans were removed (simulation 8), resulting in a food web with higher connectance and trophic similarity, and lower density and omnivory than the non-extinction scenario. Omnivory levels descended consistently with the number of nodes deleted, and omnivory did not change when the weakest links were removed (Fig. 8).

Fig. 8
figure 8

Food web properties of the different extinction scenarios when all the dataset was used and when the trophic links with lower strength than 0.5% and 1% were removed. Extinction scenarios by habitat are colored in red, by taxonomic group in blue, and by size in orange. Note the different scales in the y axis. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article)

Full size image

When comparing the extinction of demersal, benthopelagic, and pelagic species (Fig. 8; simulations 1, 2, and 3, respectively), the scenario where demersal species were removed was the one losing more nodes and therefore changes in density, connectance, omnivory, and trophic similarity were greater compared to the extinction simulations of benthopelagic and pelagic species. More nodes were lost with the extinction of sharks (simulation 5) than with the extinction of rays (simulation 4), and the reduction of trophic links and, therefore, of density, was greater when sharks were removed. Omnivory decreased more with the loss of threatened shark species versus the loss of threatened ray species (simulation 5 vs. simulation 4), and changes in connectance and trophic similarity were similar between the simulations. The removal of large chondrichthyans (simulation 6) had a higher effect on trophic similarity and connectance than the removal of medium and small-sized ones (simulation 7), even though more nodes were lost with the extinction of the medium and small-sized species.

Discussion

In this study, we reviewed the trophic information available for the chondrichthyan species inhabiting the Mediterranean Sea. We then used this information to describe the trophic role of these species using two complementary methodologies: first, we compared their diets by using multivariate analysis techniques; and then we examined the role of these species within a food web context. We complemented a previously available Mediterranean meta-web (Coll et al. 2019b) with detailed information on chondrichthyan species, and then tested different extinction scenarios and explored the consequences of removing different groups of threatened chondrichthyan species. Our results are a key step towards information required to make progress on the conservation and management of the chondrichthyan species of the Mediterranean Sea (Gračan et al. 2017).

Conservation status and trophic information of Mediterranean chondrichthyans

Our study shows that limited trophic data are available for most Mediterranean chondrichthyan species and points to the future research priorities for this group. Moreover, available information is usually obtained by opportunistic fisheries-related captures. Since most of the chondrichthyan species are not primarily targeted by fishing fleets in the Mediterranean Sea, they are mainly caught as bycatch and, therefore, underreported (Coll et al. 2013; Bradai et al. 2018). As Stevens et al. (2000) and Bradai et al. (2018) pointed out, the commercial value of species conditions research priorities, and thus scientific knowledge is less extensive for species that are legally protected, such as the blue skate (Dipturus batis), whose fishing is prohibited in the Mediterranean by the Specially Protected Areas / Biological Diversity (SPA/BD) protocol and for which diet reports were not found. Besides, threatened species tend to be scarce and, therefore, less likely to be fished.

This lack of trophic ecology information hinders the representation of endangered species in ecosystem models, with the consequent underrepresentation of their role (Barría et al. 2015). Scientific surveys targeting these species could help understand their trophic ecology. However, the impacts on their low abundances could be detrimental. More effort is needed in areas with a high abundance of chondrichthyans such as the central Mediterranean, which reported more than twice the elasmobranchs incidental catch than the rest areas of the Mediterranean (FAO 2020). Yet, the Algero-Provençal basin was the area that contributed with more reports, and the most studied species corresponded to abundant species in the western Mediterranean such as the blackmouth catshark and velvet belly lantern shark (Etmopterus spinax; Giménez et al. 2020), two species of small demersal sharks commonly captured by bottom trawl survey campaigns like the Mediterranean International Trawling Survey (MEDITS; Bertrand et al. 2002; Follesa et al. 2019).

Trophic ecology of Mediterranean chondrichthyans

Trophic ecology studies usually rely on stomach content analysis, requiring large sample sizes to obtain an accurate representation of prey composition, especially in chondrichthyan species, which often present empty stomachs (Navarro et al. 2014 and references therein). These samples are difficult to obtain, particularly for threatened species such as the little sleeper shark (Somniosus rostratus), which was only reported in one study that analyzed 3 individuals (Barría et al. 2015). In addition, the lack of samples forces many studies to analyze both juveniles and adults together, even though ontogenetic shifts in the diet have been already reported in sharks (Barría et al. 2018) and rays (e.g., skates showing diet shifts from crustaceans in smaller individuals to teleost fishes in larger ones; Ebert and Bizzarro 2007). For the same reasons, few studies take into account seasonal variability in chondrichthyans trophic ecology (e.g., Romanelli et al. 2006; Filiz 2009). Overall, adequate representation of the diet of chondrichthyans gets complicated by differences in diet that occur within species, among individuals of different sizes, geographical locations, and during different seasons (Carrier et al. 2006). Besides, soft prey like gelatinous plankton are usually underestimated because of their rapid degradation (Cardona et al. 2012; Fernández-Corredor et al. 2023), and prey with hard structures (e.g., cephalopod beaks) can accumulate and be overrepresented in number (Stergiou and Karpouzi 2002). These limitations could be partially overcome with other methods such as stable isotope analyses or DNA metabarcoding from cloacal swabs, which allows an estimation of the relative contribution of each prey in their predator diet without the need for a lethal approach (Barría et al. 2018; van Zinnicq Bergmann et al. 2021).

Although stomach content allows higher taxonomic resolution than other methods like SIA, we found high percentages of unidentified prey or low taxonomic resolution (e.g., teleosts) in some studies using stomach contents, such as in the case of the nursehound (Scyliorhinus stellaris) diet reported by Yemisken et al. (2019). Overall, the difference in prey taxonomic resolution, together with the different number of reports per species and the use of different metrics when presenting prey proportions, hinders standardization of species trophic ecology.

Since the present study aimed to obtain an overview of the ecological role of sharks, rays, and chimaeras in the Mediterranean food web, we used a low taxonomic resolution of the diet composition, which is likely conditioning our results. We found evidence for differences in prey composition among taxonomic groups: sharks fed mainly on teleost fishes and cephalopods, as stated by Cortés (1999); and rays fed more on crustaceans and teleost fishes, as found by Ebert and Bizzarro (2007). However, trophic segregation among sizes and habitats was not statistically significant, which could be due to not considering the habitat and size of prey in the analyses. For many marine predators, prey is conditioned by mouth size (Morlon et al. 2014), thus, although the consumption of particular prey groups showed no clear differences among sizes, the specific composition of the diet did. For example, small demersal sharks showed different trophic links than large demersal sharks (Online Resource 1). Besides, size and habitat do relate to conservation status, with larger species being more likely to be threatened than smaller ones (Dulvy et al. 2021) and pelagic species suffering a higher extinction risk than demersal ones (Walls and Dulvy 2021).

Mediterranean meta-web and extinction scenarios

Chondrichthyans’ predatory role can modulate prey diversity and size distributions, or even the foraging behavior of prey, by regulating ecosystem functions such as nutrient recycling and structural habitat complexity (Field et al. 2009). Results from our extinction simulations showed that the disappearance of chondrichthyan species from the Mediterranean marine food web can have impacts at the ecosystem level, which are evident at the topology level, as has been shown previously for other Mediterranean organisms (Sala 2004). Our results also pointed out the important role that chondrichthyan species can play in Mediterranean marine food webs according to the number of trophic links they can be related with and according to the potential high trophic levels achieved in these ecosystems.

Overall, levels of omnivory found in this study were around 80%, which was in accordance with Dunne et al. (2004), who found that marine food webs have higher levels of omnivory than terrestrial ones (which range between 21 and 76%). The proportion of species feeding at multiple trophic levels was consistent and omnivory values did not change when the most uncertain links were removed. However, according to our scenarios, the extinction of threatened chondrichthyans may lead to a reduction in omnivory, increasing the likelihood and magnitude of trophic cascades, that are buffered in the presence of strong omnivory (Bascompte et al. 2005). Higher values of connectance were found when removing threatened chondrichthyans. Changes in connectance were greater when more nodes were removed except for the removal of large species (simulation 6) when compared with the removal of small and medium ones. These changes can be interpreted differently depending on their origin: low connectance can reflect ecosystem degradation if the changes are due to a loss of trophic links between the nodes (Dunne et al. 2002). However, when changes in connectance are due to the loss of nodes in the food web, as in our extinction scenarios, higher values of connectance are related to lower structural complexity (Bornatowski et al. 2014). These alterations produced by species loss constitute a decrease in robustness, which relates to the maintenance of network integrity and also has consequences for stability (Sánchez-Carmona et al. 2013). Therefore, food web robustness could be more affected by the extinction of large-sized chondrichthyans compared to small and medium ones.

On the other hand, large species also showed greater contribution to trophic dissimilarity when compared to small and medium-sized species. Therefore, our results suggest that a special focus on the conservation of large species is needed: higher similarity between functional groups could enhance competition and lead to a less diverse system (Morlon et al. 2014). In fact, all Mediterranean large pelagic sharks are considered endangered or critically endangered (Table S1; IUCN 2021), and species composition is changing: larger sharks such as the thresher shark (Alopias vulpinus) and hammerhead sharks (Sphyrna spp.) which were frequently reported by fishers in the past, have been replaced by smaller species such the blue shark (Prionace glauca) (Coll et al. 2014). This is in line with the historical change in populations of large sharks in the Mediterranean Sea (Ferretti et al. 2008).

Our scenarios of the extinction of sharks and rays were tested using qualitative modelling, and, since neither biomasses nor abundances were considered, trophic cascade effects could not be investigated. Our analysis of extinction consequences was limited to the qualitative topological changes, such as changes in connectance and omnivory of the system. Future studies using quantitative modelling could extend our study to evaluate the quantitative effects of our qualitative simulations.

Conclusions

Whether it is due to an increase in scientific knowledge, a genuinely worsened status, or a combination of both, the number of threatened chondrichthyans keeps growing (Dulvy et al. 2021). Actual exploitation rates are unsustainable over the long term, suggesting that the majority of chondrichthyan populations will continue to decline under current fishing pressure (Worm et al. 2013; Dulvy et al. 2021; Juan-Jordá et al. 2022). Most populations are ecologically extinct and the effects of their decline are difficult to assess or study in situ, that is why theoretical/qualitative approaches are relevant.

Overall, our analyses showed an increase in trophic similarity and a reduction in omnivory with the extinction of threatened chondrichthyans, highlighting the contribution of large species to a more resilient and diverse system. Thus, conserving these large species is not only a matter of ecological conservation of individual species but also a key strategy for maintaining the overall health and functionality of the ecosystem. Some common measures such as the implementation of MPAs are not enough to protect large mobile species therefore other dynamic measures should be explored (Heupel et al. 2014). Our results underline the need for decisive and effective measures to improve fisheries management that focus on chondrichthyan protection to prevent regional species extinctions, which are already happening in the Mediterranean Sea (Dulvy et al. 2021; Nuez et al. 2021; Walls and Dulvy 2021); and to preserve ecosystem diversity, complexity, and resilience (Field et al. 2009). The present results provide a comprehensive picture of the trophic ecology of Mediterranean chondrichthyans and highlight that further research is needed to consider all the species and incorporate explicitly in the analysis the ontogenetic variations of their diets, which could lead to a better understanding and representation of their key role in marine food webs. We present a first step towards the analysis of food-web complex interactions where Mediterranean chondrichthyan species intervene. Future iterations of this work can be used to test additional typologies of sub-systems within the Mediterranean meta-web and additional extinction scenarios, where the fate of their prey should also be considered. In addition, this study can be the baseline from where quantitative modelling applications can be developed to assess the quantitative effects of our extinction scenarios.